diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,120016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.8823843082998257,
+  "global_step": 200000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999956764235376e-05,
+      "loss": 1.0119,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999908724496906e-05,
+      "loss": 0.5547,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999860684758435e-05,
+      "loss": 0.4887,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999812645019964e-05,
+      "loss": 0.4381,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999976460528149e-05,
+      "loss": 0.374,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999971656554302e-05,
+      "loss": 0.3704,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999966852580455e-05,
+      "loss": 0.3445,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999620486066074e-05,
+      "loss": 0.3367,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999572446327604e-05,
+      "loss": 0.328,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999524406589133e-05,
+      "loss": 0.3112,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999476366850662e-05,
+      "loss": 0.294,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999428327112188e-05,
+      "loss": 0.279,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999380287373717e-05,
+      "loss": 0.2838,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999332247635247e-05,
+      "loss": 0.279,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999284207896773e-05,
+      "loss": 0.2581,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999236168158302e-05,
+      "loss": 0.2651,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999918812841983e-05,
+      "loss": 0.2498,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999914008868136e-05,
+      "loss": 0.2378,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999092048942886e-05,
+      "loss": 0.237,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999044009204415e-05,
+      "loss": 0.2462,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998995969465945e-05,
+      "loss": 0.2412,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998947929727474e-05,
+      "loss": 0.2202,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998899889989e-05,
+      "loss": 0.2176,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999885185025053e-05,
+      "loss": 0.2145,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999880381051206e-05,
+      "loss": 0.1972,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998755770773584e-05,
+      "loss": 0.2118,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998707731035114e-05,
+      "loss": 0.2123,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998659691296643e-05,
+      "loss": 0.1982,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998611651558172e-05,
+      "loss": 0.2082,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998563611819698e-05,
+      "loss": 0.1963,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998515572081227e-05,
+      "loss": 0.177,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998467532342757e-05,
+      "loss": 0.1931,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998419492604282e-05,
+      "loss": 0.1811,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999837145286581e-05,
+      "loss": 0.182,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999832341312734e-05,
+      "loss": 0.1795,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999827537338887e-05,
+      "loss": 0.1669,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998227333650396e-05,
+      "loss": 0.1752,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998179293911925e-05,
+      "loss": 0.182,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998131254173455e-05,
+      "loss": 0.1644,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998083214434984e-05,
+      "loss": 0.1659,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999803517469651e-05,
+      "loss": 0.1618,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999798713495804e-05,
+      "loss": 0.1606,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999793909521957e-05,
+      "loss": 0.1658,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997891055481094e-05,
+      "loss": 0.1701,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997843015742623e-05,
+      "loss": 0.1586,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997794976004153e-05,
+      "loss": 0.1506,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997746936265682e-05,
+      "loss": 0.1644,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997698896527208e-05,
+      "loss": 0.1466,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997650856788737e-05,
+      "loss": 0.1512,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997602817050266e-05,
+      "loss": 0.1576,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997554777311792e-05,
+      "loss": 0.1502,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999750673757332e-05,
+      "loss": 0.1619,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999745869783485e-05,
+      "loss": 0.1455,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999741065809638e-05,
+      "loss": 0.1399,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997362618357906e-05,
+      "loss": 0.1423,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997314578619435e-05,
+      "loss": 0.1413,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997266538880965e-05,
+      "loss": 0.149,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997218499142494e-05,
+      "loss": 0.1356,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999717045940402e-05,
+      "loss": 0.141,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999712241966555e-05,
+      "loss": 0.137,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997074379927078e-05,
+      "loss": 0.142,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997026340188604e-05,
+      "loss": 0.1388,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996978300450133e-05,
+      "loss": 0.1413,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996930260711663e-05,
+      "loss": 0.1419,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996882220973192e-05,
+      "loss": 0.1461,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996834181234718e-05,
+      "loss": 0.1357,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996786141496247e-05,
+      "loss": 0.1354,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996738101757776e-05,
+      "loss": 0.1363,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996690062019302e-05,
+      "loss": 0.1374,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999664202228083e-05,
+      "loss": 0.1385,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999659398254236e-05,
+      "loss": 0.1491,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999654594280389e-05,
+      "loss": 0.126,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996497903065416e-05,
+      "loss": 0.1295,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996449863326945e-05,
+      "loss": 0.1405,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996401823588474e-05,
+      "loss": 0.1267,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996353783850004e-05,
+      "loss": 0.1265,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999630574411153e-05,
+      "loss": 0.121,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999625770437306e-05,
+      "loss": 0.124,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996209664634588e-05,
+      "loss": 0.1277,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996161624896114e-05,
+      "loss": 0.1166,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996113585157643e-05,
+      "loss": 0.1203,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996065545419173e-05,
+      "loss": 0.1244,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996017505680702e-05,
+      "loss": 0.1293,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995969465942228e-05,
+      "loss": 0.1309,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995921426203757e-05,
+      "loss": 0.1197,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995873386465286e-05,
+      "loss": 0.131,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995825346726812e-05,
+      "loss": 0.133,
+      "step": 870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999577730698834e-05,
+      "loss": 0.114,
+      "step": 880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999572926724987e-05,
+      "loss": 0.1241,
+      "step": 890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99956812275114e-05,
+      "loss": 0.1213,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995633187772926e-05,
+      "loss": 0.1275,
+      "step": 910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995585148034455e-05,
+      "loss": 0.1218,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995537108295984e-05,
+      "loss": 0.1239,
+      "step": 930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995489068557514e-05,
+      "loss": 0.1247,
+      "step": 940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999544102881904e-05,
+      "loss": 0.1091,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999539298908057e-05,
+      "loss": 0.1174,
+      "step": 960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995344949342098e-05,
+      "loss": 0.1211,
+      "step": 970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995296909603624e-05,
+      "loss": 0.1151,
+      "step": 980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995248869865153e-05,
+      "loss": 0.1269,
+      "step": 990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995200830126683e-05,
+      "loss": 0.118,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995152790388212e-05,
+      "loss": 0.116,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995104750649738e-05,
+      "loss": 0.1079,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995056710911267e-05,
+      "loss": 0.1105,
+      "step": 1030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995008671172796e-05,
+      "loss": 0.1084,
+      "step": 1040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994960631434322e-05,
+      "loss": 0.1122,
+      "step": 1050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999491259169585e-05,
+      "loss": 0.1102,
+      "step": 1060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999486455195738e-05,
+      "loss": 0.117,
+      "step": 1070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999481651221891e-05,
+      "loss": 0.1096,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994768472480436e-05,
+      "loss": 0.1154,
+      "step": 1090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994720432741965e-05,
+      "loss": 0.11,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994672393003494e-05,
+      "loss": 0.1084,
+      "step": 1110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994624353265024e-05,
+      "loss": 0.103,
+      "step": 1120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999457631352655e-05,
+      "loss": 0.1033,
+      "step": 1130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999452827378808e-05,
+      "loss": 0.1185,
+      "step": 1140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994480234049608e-05,
+      "loss": 0.1157,
+      "step": 1150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994432194311134e-05,
+      "loss": 0.1047,
+      "step": 1160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994384154572663e-05,
+      "loss": 0.1023,
+      "step": 1170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994336114834192e-05,
+      "loss": 0.115,
+      "step": 1180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994288075095722e-05,
+      "loss": 0.1172,
+      "step": 1190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994240035357248e-05,
+      "loss": 0.1033,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994191995618777e-05,
+      "loss": 0.1044,
+      "step": 1210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994143955880306e-05,
+      "loss": 0.1042,
+      "step": 1220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994095916141832e-05,
+      "loss": 0.1156,
+      "step": 1230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999404787640336e-05,
+      "loss": 0.1081,
+      "step": 1240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999399983666489e-05,
+      "loss": 0.0975,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999395179692642e-05,
+      "loss": 0.0989,
+      "step": 1260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993903757187946e-05,
+      "loss": 0.105,
+      "step": 1270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993855717449475e-05,
+      "loss": 0.1129,
+      "step": 1280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993807677711004e-05,
+      "loss": 0.0925,
+      "step": 1290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993759637972533e-05,
+      "loss": 0.1051,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999371159823406e-05,
+      "loss": 0.117,
+      "step": 1310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999366355849559e-05,
+      "loss": 0.1076,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993615518757118e-05,
+      "loss": 0.1056,
+      "step": 1330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993567479018644e-05,
+      "loss": 0.1068,
+      "step": 1340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993519439280173e-05,
+      "loss": 0.1155,
+      "step": 1350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993471399541702e-05,
+      "loss": 0.0998,
+      "step": 1360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999342335980323e-05,
+      "loss": 0.1004,
+      "step": 1370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993375320064757e-05,
+      "loss": 0.1006,
+      "step": 1380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993327280326287e-05,
+      "loss": 0.0929,
+      "step": 1390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993279240587816e-05,
+      "loss": 0.1018,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993231200849342e-05,
+      "loss": 0.0979,
+      "step": 1410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999318316111087e-05,
+      "loss": 0.1063,
+      "step": 1420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99931351213724e-05,
+      "loss": 0.1032,
+      "step": 1430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999308708163393e-05,
+      "loss": 0.094,
+      "step": 1440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993039041895456e-05,
+      "loss": 0.1073,
+      "step": 1450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992991002156985e-05,
+      "loss": 0.0988,
+      "step": 1460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992942962418514e-05,
+      "loss": 0.1095,
+      "step": 1470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992894922680043e-05,
+      "loss": 0.1055,
+      "step": 1480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999284688294157e-05,
+      "loss": 0.0908,
+      "step": 1490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99927988432031e-05,
+      "loss": 0.0924,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992750803464628e-05,
+      "loss": 0.0949,
+      "step": 1510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992702763726154e-05,
+      "loss": 0.097,
+      "step": 1520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992654723987683e-05,
+      "loss": 0.0939,
+      "step": 1530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992606684249212e-05,
+      "loss": 0.0922,
+      "step": 1540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999255864451074e-05,
+      "loss": 0.092,
+      "step": 1550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992510604772267e-05,
+      "loss": 0.0963,
+      "step": 1560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992462565033797e-05,
+      "loss": 0.0912,
+      "step": 1570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992414525295326e-05,
+      "loss": 0.0963,
+      "step": 1580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992366485556852e-05,
+      "loss": 0.0883,
+      "step": 1590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999231844581838e-05,
+      "loss": 0.1008,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999227040607991e-05,
+      "loss": 0.0844,
+      "step": 1610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999222236634144e-05,
+      "loss": 0.09,
+      "step": 1620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992174326602966e-05,
+      "loss": 0.0904,
+      "step": 1630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992126286864495e-05,
+      "loss": 0.1015,
+      "step": 1640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992078247126024e-05,
+      "loss": 0.1012,
+      "step": 1650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999203020738755e-05,
+      "loss": 0.0849,
+      "step": 1660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999198216764908e-05,
+      "loss": 0.0955,
+      "step": 1670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999193412791061e-05,
+      "loss": 0.0961,
+      "step": 1680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991886088172138e-05,
+      "loss": 0.0949,
+      "step": 1690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991838048433664e-05,
+      "loss": 0.0878,
+      "step": 1700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991790008695193e-05,
+      "loss": 0.1,
+      "step": 1710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991741968956722e-05,
+      "loss": 0.0926,
+      "step": 1720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999169392921825e-05,
+      "loss": 0.0911,
+      "step": 1730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991645889479777e-05,
+      "loss": 0.099,
+      "step": 1740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991597849741307e-05,
+      "loss": 0.0947,
+      "step": 1750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991549810002836e-05,
+      "loss": 0.0949,
+      "step": 1760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991501770264362e-05,
+      "loss": 0.0981,
+      "step": 1770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999145373052589e-05,
+      "loss": 0.0919,
+      "step": 1780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999140569078742e-05,
+      "loss": 0.0926,
+      "step": 1790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999135765104895e-05,
+      "loss": 0.0852,
+      "step": 1800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999130961131048e-05,
+      "loss": 0.0913,
+      "step": 1810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991261571572008e-05,
+      "loss": 0.0868,
+      "step": 1820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991213531833534e-05,
+      "loss": 0.0932,
+      "step": 1830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991165492095063e-05,
+      "loss": 0.0954,
+      "step": 1840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991117452356593e-05,
+      "loss": 0.092,
+      "step": 1850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991069412618122e-05,
+      "loss": 0.0874,
+      "step": 1860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991021372879648e-05,
+      "loss": 0.0942,
+      "step": 1870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990973333141177e-05,
+      "loss": 0.0858,
+      "step": 1880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990925293402706e-05,
+      "loss": 0.0915,
+      "step": 1890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990877253664232e-05,
+      "loss": 0.0853,
+      "step": 1900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999082921392576e-05,
+      "loss": 0.0848,
+      "step": 1910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999078117418729e-05,
+      "loss": 0.0869,
+      "step": 1920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999073313444882e-05,
+      "loss": 0.0881,
+      "step": 1930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990685094710346e-05,
+      "loss": 0.0914,
+      "step": 1940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990637054971875e-05,
+      "loss": 0.0862,
+      "step": 1950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990589015233404e-05,
+      "loss": 0.0851,
+      "step": 1960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990540975494934e-05,
+      "loss": 0.0813,
+      "step": 1970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999049293575646e-05,
+      "loss": 0.0733,
+      "step": 1980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999044489601799e-05,
+      "loss": 0.0935,
+      "step": 1990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990396856279518e-05,
+      "loss": 0.0975,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990348816541044e-05,
+      "loss": 0.0877,
+      "step": 2010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990300776802573e-05,
+      "loss": 0.0968,
+      "step": 2020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990252737064102e-05,
+      "loss": 0.0858,
+      "step": 2030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990204697325632e-05,
+      "loss": 0.0851,
+      "step": 2040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990156657587158e-05,
+      "loss": 0.0803,
+      "step": 2050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990108617848687e-05,
+      "loss": 0.0874,
+      "step": 2060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990060578110216e-05,
+      "loss": 0.0882,
+      "step": 2070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990012538371742e-05,
+      "loss": 0.0789,
+      "step": 2080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998996449863327e-05,
+      "loss": 0.0842,
+      "step": 2090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99899164588948e-05,
+      "loss": 0.0806,
+      "step": 2100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998986841915633e-05,
+      "loss": 0.08,
+      "step": 2110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989820379417856e-05,
+      "loss": 0.0834,
+      "step": 2120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989772339679385e-05,
+      "loss": 0.0851,
+      "step": 2130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989724299940914e-05,
+      "loss": 0.0844,
+      "step": 2140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989676260202443e-05,
+      "loss": 0.0804,
+      "step": 2150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998962822046397e-05,
+      "loss": 0.0792,
+      "step": 2160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99895801807255e-05,
+      "loss": 0.0747,
+      "step": 2170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989532140987028e-05,
+      "loss": 0.0795,
+      "step": 2180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989484101248554e-05,
+      "loss": 0.0721,
+      "step": 2190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989436061510083e-05,
+      "loss": 0.0859,
+      "step": 2200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989388021771612e-05,
+      "loss": 0.0844,
+      "step": 2210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998933998203314e-05,
+      "loss": 0.0813,
+      "step": 2220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989291942294667e-05,
+      "loss": 0.0885,
+      "step": 2230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989243902556197e-05,
+      "loss": 0.0873,
+      "step": 2240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989195862817726e-05,
+      "loss": 0.0803,
+      "step": 2250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989147823079252e-05,
+      "loss": 0.0779,
+      "step": 2260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998909978334078e-05,
+      "loss": 0.0899,
+      "step": 2270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998905174360231e-05,
+      "loss": 0.0789,
+      "step": 2280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998900370386384e-05,
+      "loss": 0.0929,
+      "step": 2290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988955664125366e-05,
+      "loss": 0.0861,
+      "step": 2300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988907624386895e-05,
+      "loss": 0.0717,
+      "step": 2310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988859584648424e-05,
+      "loss": 0.0856,
+      "step": 2320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988811544909953e-05,
+      "loss": 0.075,
+      "step": 2330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998876350517148e-05,
+      "loss": 0.0906,
+      "step": 2340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998871546543301e-05,
+      "loss": 0.0801,
+      "step": 2350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988667425694538e-05,
+      "loss": 0.0752,
+      "step": 2360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988619385956064e-05,
+      "loss": 0.0732,
+      "step": 2370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988571346217593e-05,
+      "loss": 0.073,
+      "step": 2380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988523306479122e-05,
+      "loss": 0.0783,
+      "step": 2390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998847526674065e-05,
+      "loss": 0.0744,
+      "step": 2400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988427227002177e-05,
+      "loss": 0.0792,
+      "step": 2410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988379187263707e-05,
+      "loss": 0.0863,
+      "step": 2420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988331147525236e-05,
+      "loss": 0.0776,
+      "step": 2430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988283107786762e-05,
+      "loss": 0.0765,
+      "step": 2440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998823506804829e-05,
+      "loss": 0.0743,
+      "step": 2450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998818702830982e-05,
+      "loss": 0.0788,
+      "step": 2460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998813898857135e-05,
+      "loss": 0.0747,
+      "step": 2470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988090948832876e-05,
+      "loss": 0.0801,
+      "step": 2480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988042909094405e-05,
+      "loss": 0.0736,
+      "step": 2490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987994869355934e-05,
+      "loss": 0.0783,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987946829617463e-05,
+      "loss": 0.0763,
+      "step": 2510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998789878987899e-05,
+      "loss": 0.0783,
+      "step": 2520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998785075014052e-05,
+      "loss": 0.0867,
+      "step": 2530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987802710402048e-05,
+      "loss": 0.0722,
+      "step": 2540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987754670663574e-05,
+      "loss": 0.0727,
+      "step": 2550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987706630925103e-05,
+      "loss": 0.0746,
+      "step": 2560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987658591186632e-05,
+      "loss": 0.0809,
+      "step": 2570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998761055144816e-05,
+      "loss": 0.0802,
+      "step": 2580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987562511709687e-05,
+      "loss": 0.0796,
+      "step": 2590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987514471971217e-05,
+      "loss": 0.0827,
+      "step": 2600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987466432232746e-05,
+      "loss": 0.085,
+      "step": 2610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987418392494272e-05,
+      "loss": 0.0726,
+      "step": 2620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99873703527558e-05,
+      "loss": 0.0782,
+      "step": 2630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998732231301733e-05,
+      "loss": 0.0826,
+      "step": 2640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998727427327886e-05,
+      "loss": 0.074,
+      "step": 2650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987226233540385e-05,
+      "loss": 0.0737,
+      "step": 2660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987178193801915e-05,
+      "loss": 0.0787,
+      "step": 2670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987130154063444e-05,
+      "loss": 0.0734,
+      "step": 2680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987082114324973e-05,
+      "loss": 0.0648,
+      "step": 2690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99870340745865e-05,
+      "loss": 0.0752,
+      "step": 2700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998698603484803e-05,
+      "loss": 0.0712,
+      "step": 2710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986937995109558e-05,
+      "loss": 0.0642,
+      "step": 2720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986889955371084e-05,
+      "loss": 0.0724,
+      "step": 2730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986841915632613e-05,
+      "loss": 0.0683,
+      "step": 2740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986793875894142e-05,
+      "loss": 0.0712,
+      "step": 2750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998674583615567e-05,
+      "loss": 0.069,
+      "step": 2760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986697796417197e-05,
+      "loss": 0.0701,
+      "step": 2770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986649756678727e-05,
+      "loss": 0.0737,
+      "step": 2780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986601716940256e-05,
+      "loss": 0.0756,
+      "step": 2790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998655367720178e-05,
+      "loss": 0.088,
+      "step": 2800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998650563746331e-05,
+      "loss": 0.0657,
+      "step": 2810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998645759772484e-05,
+      "loss": 0.0743,
+      "step": 2820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998640955798637e-05,
+      "loss": 0.0728,
+      "step": 2830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986361518247895e-05,
+      "loss": 0.0715,
+      "step": 2840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986313478509425e-05,
+      "loss": 0.075,
+      "step": 2850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986265438770954e-05,
+      "loss": 0.0736,
+      "step": 2860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986217399032483e-05,
+      "loss": 0.0653,
+      "step": 2870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998616935929401e-05,
+      "loss": 0.0654,
+      "step": 2880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998612131955554e-05,
+      "loss": 0.0771,
+      "step": 2890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986073279817068e-05,
+      "loss": 0.058,
+      "step": 2900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986025240078593e-05,
+      "loss": 0.0688,
+      "step": 2910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985977200340123e-05,
+      "loss": 0.0753,
+      "step": 2920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985929160601652e-05,
+      "loss": 0.0743,
+      "step": 2930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998588112086318e-05,
+      "loss": 0.0647,
+      "step": 2940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985833081124707e-05,
+      "loss": 0.0616,
+      "step": 2950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985785041386236e-05,
+      "loss": 0.0753,
+      "step": 2960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985737001647766e-05,
+      "loss": 0.0712,
+      "step": 2970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998568896190929e-05,
+      "loss": 0.0682,
+      "step": 2980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998564092217082e-05,
+      "loss": 0.0679,
+      "step": 2990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998559288243235e-05,
+      "loss": 0.0764,
+      "step": 3000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998554484269388e-05,
+      "loss": 0.0644,
+      "step": 3010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985496802955405e-05,
+      "loss": 0.0593,
+      "step": 3020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985448763216935e-05,
+      "loss": 0.0652,
+      "step": 3030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985400723478464e-05,
+      "loss": 0.0677,
+      "step": 3040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998535268373999e-05,
+      "loss": 0.0658,
+      "step": 3050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998530464400152e-05,
+      "loss": 0.0669,
+      "step": 3060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985256604263048e-05,
+      "loss": 0.0674,
+      "step": 3070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985208564524577e-05,
+      "loss": 0.0722,
+      "step": 3080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985160524786103e-05,
+      "loss": 0.0668,
+      "step": 3090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985112485047633e-05,
+      "loss": 0.0759,
+      "step": 3100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985064445309162e-05,
+      "loss": 0.0736,
+      "step": 3110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998501640557069e-05,
+      "loss": 0.0766,
+      "step": 3120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984968365832217e-05,
+      "loss": 0.0764,
+      "step": 3130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984920326093746e-05,
+      "loss": 0.0701,
+      "step": 3140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984872286355276e-05,
+      "loss": 0.0764,
+      "step": 3150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99848242466168e-05,
+      "loss": 0.0708,
+      "step": 3160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998477620687833e-05,
+      "loss": 0.0749,
+      "step": 3170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998472816713986e-05,
+      "loss": 0.0758,
+      "step": 3180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998468012740139e-05,
+      "loss": 0.0694,
+      "step": 3190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984632087662915e-05,
+      "loss": 0.0718,
+      "step": 3200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984584047924444e-05,
+      "loss": 0.0747,
+      "step": 3210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984536008185974e-05,
+      "loss": 0.0613,
+      "step": 3220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99844879684475e-05,
+      "loss": 0.0575,
+      "step": 3230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998443992870903e-05,
+      "loss": 0.0625,
+      "step": 3240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984391888970558e-05,
+      "loss": 0.0616,
+      "step": 3250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984343849232087e-05,
+      "loss": 0.071,
+      "step": 3260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984295809493613e-05,
+      "loss": 0.0646,
+      "step": 3270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984247769755143e-05,
+      "loss": 0.064,
+      "step": 3280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984199730016672e-05,
+      "loss": 0.0683,
+      "step": 3290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99841516902782e-05,
+      "loss": 0.0686,
+      "step": 3300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984103650539727e-05,
+      "loss": 0.0715,
+      "step": 3310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984055610801256e-05,
+      "loss": 0.0574,
+      "step": 3320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984007571062786e-05,
+      "loss": 0.068,
+      "step": 3330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998395953132431e-05,
+      "loss": 0.0644,
+      "step": 3340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998391149158584e-05,
+      "loss": 0.0634,
+      "step": 3350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998386345184737e-05,
+      "loss": 0.0646,
+      "step": 3360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99838154121089e-05,
+      "loss": 0.0602,
+      "step": 3370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983767372370425e-05,
+      "loss": 0.0685,
+      "step": 3380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983719332631954e-05,
+      "loss": 0.0589,
+      "step": 3390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983671292893484e-05,
+      "loss": 0.0703,
+      "step": 3400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998362325315501e-05,
+      "loss": 0.0633,
+      "step": 3410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998357521341654e-05,
+      "loss": 0.0676,
+      "step": 3420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983527173678068e-05,
+      "loss": 0.065,
+      "step": 3430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983479133939597e-05,
+      "loss": 0.0658,
+      "step": 3440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983431094201123e-05,
+      "loss": 0.0694,
+      "step": 3450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983383054462652e-05,
+      "loss": 0.0647,
+      "step": 3460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983335014724182e-05,
+      "loss": 0.0721,
+      "step": 3470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998328697498571e-05,
+      "loss": 0.0626,
+      "step": 3480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983238935247237e-05,
+      "loss": 0.058,
+      "step": 3490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983190895508766e-05,
+      "loss": 0.0653,
+      "step": 3500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983142855770295e-05,
+      "loss": 0.0698,
+      "step": 3510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998309481603182e-05,
+      "loss": 0.0687,
+      "step": 3520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998304677629335e-05,
+      "loss": 0.0709,
+      "step": 3530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998299873655488e-05,
+      "loss": 0.0729,
+      "step": 3540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998295069681641e-05,
+      "loss": 0.0632,
+      "step": 3550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982902657077935e-05,
+      "loss": 0.0803,
+      "step": 3560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982854617339464e-05,
+      "loss": 0.0678,
+      "step": 3570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982806577600994e-05,
+      "loss": 0.0627,
+      "step": 3580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998275853786252e-05,
+      "loss": 0.064,
+      "step": 3590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998271049812405e-05,
+      "loss": 0.0764,
+      "step": 3600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982662458385578e-05,
+      "loss": 0.065,
+      "step": 3610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982614418647107e-05,
+      "loss": 0.0621,
+      "step": 3620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982566378908633e-05,
+      "loss": 0.0664,
+      "step": 3630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982518339170162e-05,
+      "loss": 0.064,
+      "step": 3640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998247029943169e-05,
+      "loss": 0.0676,
+      "step": 3650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998242225969322e-05,
+      "loss": 0.0579,
+      "step": 3660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982374219954747e-05,
+      "loss": 0.0542,
+      "step": 3670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982326180216276e-05,
+      "loss": 0.0571,
+      "step": 3680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982278140477805e-05,
+      "loss": 0.0676,
+      "step": 3690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998223010073933e-05,
+      "loss": 0.0692,
+      "step": 3700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998218206100086e-05,
+      "loss": 0.0666,
+      "step": 3710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998213402126239e-05,
+      "loss": 0.0583,
+      "step": 3720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998208598152392e-05,
+      "loss": 0.0654,
+      "step": 3730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982037941785445e-05,
+      "loss": 0.0656,
+      "step": 3740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981989902046974e-05,
+      "loss": 0.0686,
+      "step": 3750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981941862308503e-05,
+      "loss": 0.0697,
+      "step": 3760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998189382257003e-05,
+      "loss": 0.064,
+      "step": 3770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998184578283156e-05,
+      "loss": 0.0662,
+      "step": 3780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981797743093088e-05,
+      "loss": 0.0612,
+      "step": 3790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981749703354617e-05,
+      "loss": 0.0714,
+      "step": 3800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981701663616143e-05,
+      "loss": 0.0663,
+      "step": 3810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9981653623877672e-05,
+      "loss": 0.0666,
+      "step": 3820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99816055841392e-05,
+      "loss": 0.0666,
+      "step": 3830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998155754440073e-05,
+      "loss": 0.0649,
+      "step": 3840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9981509504662257e-05,
+      "loss": 0.0685,
+      "step": 3850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9981461464923786e-05,
+      "loss": 0.0612,
+      "step": 3860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9981413425185315e-05,
+      "loss": 0.0579,
+      "step": 3870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998136538544684e-05,
+      "loss": 0.0651,
+      "step": 3880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998131734570837e-05,
+      "loss": 0.0526,
+      "step": 3890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99812693059699e-05,
+      "loss": 0.0697,
+      "step": 3900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998122126623143e-05,
+      "loss": 0.0682,
+      "step": 3910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9981173226492955e-05,
+      "loss": 0.0614,
+      "step": 3920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9981125186754484e-05,
+      "loss": 0.0614,
+      "step": 3930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9981077147016013e-05,
+      "loss": 0.0504,
+      "step": 3940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998102910727754e-05,
+      "loss": 0.0563,
+      "step": 3950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998098106753907e-05,
+      "loss": 0.0621,
+      "step": 3960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980933027800598e-05,
+      "loss": 0.055,
+      "step": 3970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980884988062127e-05,
+      "loss": 0.0556,
+      "step": 3980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980836948323653e-05,
+      "loss": 0.0649,
+      "step": 3990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980788908585182e-05,
+      "loss": 0.063,
+      "step": 4000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998074086884671e-05,
+      "loss": 0.0658,
+      "step": 4010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998069282910824e-05,
+      "loss": 0.0626,
+      "step": 4020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980644789369767e-05,
+      "loss": 0.0612,
+      "step": 4030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980596749631296e-05,
+      "loss": 0.056,
+      "step": 4040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980548709892825e-05,
+      "loss": 0.0626,
+      "step": 4050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998050067015435e-05,
+      "loss": 0.0567,
+      "step": 4060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998045263041588e-05,
+      "loss": 0.0656,
+      "step": 4070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998040459067741e-05,
+      "loss": 0.054,
+      "step": 4080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998035655093894e-05,
+      "loss": 0.0657,
+      "step": 4090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980308511200465e-05,
+      "loss": 0.0664,
+      "step": 4100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980260471461994e-05,
+      "loss": 0.0675,
+      "step": 4110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980212431723523e-05,
+      "loss": 0.062,
+      "step": 4120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998016439198505e-05,
+      "loss": 0.053,
+      "step": 4130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998011635224658e-05,
+      "loss": 0.0583,
+      "step": 4140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980068312508108e-05,
+      "loss": 0.0536,
+      "step": 4150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980020272769637e-05,
+      "loss": 0.064,
+      "step": 4160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979972233031163e-05,
+      "loss": 0.057,
+      "step": 4170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979924193292692e-05,
+      "loss": 0.062,
+      "step": 4180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997987615355422e-05,
+      "loss": 0.0606,
+      "step": 4190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997982811381575e-05,
+      "loss": 0.0527,
+      "step": 4200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979780074077277e-05,
+      "loss": 0.0577,
+      "step": 4210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979732034338806e-05,
+      "loss": 0.0548,
+      "step": 4220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979683994600335e-05,
+      "loss": 0.0536,
+      "step": 4230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997963595486186e-05,
+      "loss": 0.0575,
+      "step": 4240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997958791512339e-05,
+      "loss": 0.0577,
+      "step": 4250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997953987538492e-05,
+      "loss": 0.0554,
+      "step": 4260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997949183564645e-05,
+      "loss": 0.0569,
+      "step": 4270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979443795907975e-05,
+      "loss": 0.0634,
+      "step": 4280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979395756169504e-05,
+      "loss": 0.0582,
+      "step": 4290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979347716431033e-05,
+      "loss": 0.0589,
+      "step": 4300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997929967669256e-05,
+      "loss": 0.0601,
+      "step": 4310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997925163695409e-05,
+      "loss": 0.054,
+      "step": 4320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979203597215618e-05,
+      "loss": 0.0581,
+      "step": 4330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979155557477147e-05,
+      "loss": 0.0753,
+      "step": 4340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979107517738673e-05,
+      "loss": 0.0696,
+      "step": 4350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979059478000205e-05,
+      "loss": 0.0651,
+      "step": 4360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997901143826173e-05,
+      "loss": 0.0585,
+      "step": 4370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997896339852326e-05,
+      "loss": 0.0659,
+      "step": 4380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997891535878479e-05,
+      "loss": 0.064,
+      "step": 4390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997886731904632e-05,
+      "loss": 0.0696,
+      "step": 4400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9978819279307845e-05,
+      "loss": 0.0572,
+      "step": 4410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9978771239569374e-05,
+      "loss": 0.0525,
+      "step": 4420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9978723199830904e-05,
+      "loss": 0.0565,
+      "step": 4430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9978675160092433e-05,
+      "loss": 0.0559,
+      "step": 4440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997862712035396e-05,
+      "loss": 0.0615,
+      "step": 4450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9978579080615488e-05,
+      "loss": 0.0638,
+      "step": 4460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9978531040877017e-05,
+      "loss": 0.0624,
+      "step": 4470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9978483001138543e-05,
+      "loss": 0.053,
+      "step": 4480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9978434961400072e-05,
+      "loss": 0.0482,
+      "step": 4490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99783869216616e-05,
+      "loss": 0.0562,
+      "step": 4500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997833888192313e-05,
+      "loss": 0.0558,
+      "step": 4510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9978290842184657e-05,
+      "loss": 0.0481,
+      "step": 4520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9978242802446186e-05,
+      "loss": 0.0557,
+      "step": 4530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9978194762707715e-05,
+      "loss": 0.059,
+      "step": 4540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997814672296924e-05,
+      "loss": 0.0578,
+      "step": 4550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997809868323077e-05,
+      "loss": 0.0562,
+      "step": 4560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99780506434923e-05,
+      "loss": 0.056,
+      "step": 4570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997800260375383e-05,
+      "loss": 0.0553,
+      "step": 4580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977954564015355e-05,
+      "loss": 0.0585,
+      "step": 4590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977906524276884e-05,
+      "loss": 0.0534,
+      "step": 4600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977858484538413e-05,
+      "loss": 0.0532,
+      "step": 4610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997781044479994e-05,
+      "loss": 0.0582,
+      "step": 4620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997776240506147e-05,
+      "loss": 0.063,
+      "step": 4630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977714365322998e-05,
+      "loss": 0.058,
+      "step": 4640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977666325584527e-05,
+      "loss": 0.062,
+      "step": 4650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977618285846053e-05,
+      "loss": 0.0618,
+      "step": 4660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977570246107582e-05,
+      "loss": 0.0612,
+      "step": 4670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997752220636911e-05,
+      "loss": 0.0475,
+      "step": 4680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997747416663064e-05,
+      "loss": 0.0631,
+      "step": 4690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977426126892167e-05,
+      "loss": 0.0588,
+      "step": 4700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977378087153696e-05,
+      "loss": 0.0639,
+      "step": 4710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977330047415225e-05,
+      "loss": 0.0517,
+      "step": 4720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997728200767675e-05,
+      "loss": 0.0451,
+      "step": 4730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997723396793828e-05,
+      "loss": 0.0506,
+      "step": 4740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997718592819981e-05,
+      "loss": 0.0548,
+      "step": 4750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997713788846134e-05,
+      "loss": 0.0586,
+      "step": 4760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977089848722865e-05,
+      "loss": 0.0528,
+      "step": 4770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977041808984394e-05,
+      "loss": 0.0564,
+      "step": 4780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976993769245923e-05,
+      "loss": 0.0496,
+      "step": 4790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997694572950745e-05,
+      "loss": 0.0547,
+      "step": 4800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997689768976898e-05,
+      "loss": 0.0568,
+      "step": 4810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976849650030508e-05,
+      "loss": 0.0575,
+      "step": 4820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976801610292037e-05,
+      "loss": 0.0668,
+      "step": 4830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976753570553563e-05,
+      "loss": 0.0633,
+      "step": 4840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976705530815092e-05,
+      "loss": 0.0522,
+      "step": 4850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997665749107662e-05,
+      "loss": 0.0492,
+      "step": 4860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997660945133815e-05,
+      "loss": 0.0626,
+      "step": 4870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976561411599677e-05,
+      "loss": 0.0522,
+      "step": 4880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976513371861206e-05,
+      "loss": 0.0573,
+      "step": 4890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976465332122735e-05,
+      "loss": 0.0586,
+      "step": 4900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997641729238426e-05,
+      "loss": 0.0611,
+      "step": 4910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997636925264579e-05,
+      "loss": 0.0599,
+      "step": 4920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997632121290732e-05,
+      "loss": 0.0555,
+      "step": 4930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997627317316885e-05,
+      "loss": 0.0523,
+      "step": 4940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976225133430375e-05,
+      "loss": 0.0577,
+      "step": 4950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976177093691904e-05,
+      "loss": 0.0465,
+      "step": 4960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976129053953433e-05,
+      "loss": 0.0589,
+      "step": 4970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997608101421496e-05,
+      "loss": 0.0622,
+      "step": 4980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997603297447649e-05,
+      "loss": 0.0654,
+      "step": 4990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975984934738018e-05,
+      "loss": 0.05,
+      "step": 5000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975936894999547e-05,
+      "loss": 0.0524,
+      "step": 5010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975888855261073e-05,
+      "loss": 0.0544,
+      "step": 5020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975840815522602e-05,
+      "loss": 0.0508,
+      "step": 5030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997579277578413e-05,
+      "loss": 0.0563,
+      "step": 5040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997574473604566e-05,
+      "loss": 0.0604,
+      "step": 5050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975696696307187e-05,
+      "loss": 0.0568,
+      "step": 5060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975648656568716e-05,
+      "loss": 0.0589,
+      "step": 5070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975600616830245e-05,
+      "loss": 0.0493,
+      "step": 5080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997555257709177e-05,
+      "loss": 0.0477,
+      "step": 5090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99755045373533e-05,
+      "loss": 0.0531,
+      "step": 5100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997545649761483e-05,
+      "loss": 0.0584,
+      "step": 5110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997540845787636e-05,
+      "loss": 0.0457,
+      "step": 5120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975360418137885e-05,
+      "loss": 0.0541,
+      "step": 5130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975312378399414e-05,
+      "loss": 0.0469,
+      "step": 5140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975264338660943e-05,
+      "loss": 0.0593,
+      "step": 5150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997521629892247e-05,
+      "loss": 0.0527,
+      "step": 5160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975168259184e-05,
+      "loss": 0.0573,
+      "step": 5170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975120219445528e-05,
+      "loss": 0.0562,
+      "step": 5180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975072179707057e-05,
+      "loss": 0.0562,
+      "step": 5190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975024139968583e-05,
+      "loss": 0.0537,
+      "step": 5200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9974976100230112e-05,
+      "loss": 0.054,
+      "step": 5210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997492806049164e-05,
+      "loss": 0.0531,
+      "step": 5220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997488002075317e-05,
+      "loss": 0.0584,
+      "step": 5230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9974831981014696e-05,
+      "loss": 0.0542,
+      "step": 5240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9974783941276226e-05,
+      "loss": 0.0541,
+      "step": 5250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9974735901537755e-05,
+      "loss": 0.0481,
+      "step": 5260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997468786179928e-05,
+      "loss": 0.0572,
+      "step": 5270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997463982206081e-05,
+      "loss": 0.0489,
+      "step": 5280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997459178232234e-05,
+      "loss": 0.0593,
+      "step": 5290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997454374258387e-05,
+      "loss": 0.0523,
+      "step": 5300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9974495702845395e-05,
+      "loss": 0.0537,
+      "step": 5310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9974447663106924e-05,
+      "loss": 0.0496,
+      "step": 5320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9974399623368453e-05,
+      "loss": 0.0532,
+      "step": 5330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997435158362998e-05,
+      "loss": 0.06,
+      "step": 5340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9974303543891508e-05,
+      "loss": 0.0558,
+      "step": 5350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9974255504153038e-05,
+      "loss": 0.0551,
+      "step": 5360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9974207464414567e-05,
+      "loss": 0.0479,
+      "step": 5370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9974159424676093e-05,
+      "loss": 0.0519,
+      "step": 5380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9974111384937622e-05,
+      "loss": 0.0583,
+      "step": 5390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997406334519915e-05,
+      "loss": 0.0559,
+      "step": 5400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997401530546068e-05,
+      "loss": 0.0487,
+      "step": 5410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973967265722206e-05,
+      "loss": 0.0474,
+      "step": 5420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973919225983736e-05,
+      "loss": 0.0575,
+      "step": 5430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973871186245265e-05,
+      "loss": 0.0562,
+      "step": 5440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997382314650679e-05,
+      "loss": 0.0477,
+      "step": 5450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997377510676832e-05,
+      "loss": 0.0542,
+      "step": 5460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997372706702985e-05,
+      "loss": 0.0696,
+      "step": 5470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997367902729138e-05,
+      "loss": 0.0496,
+      "step": 5480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973630987552904e-05,
+      "loss": 0.0574,
+      "step": 5490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973582947814434e-05,
+      "loss": 0.0507,
+      "step": 5500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973534908075963e-05,
+      "loss": 0.0557,
+      "step": 5510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997348686833749e-05,
+      "loss": 0.0507,
+      "step": 5520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973438828599018e-05,
+      "loss": 0.0516,
+      "step": 5530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973390788860547e-05,
+      "loss": 0.0497,
+      "step": 5540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973342749122077e-05,
+      "loss": 0.056,
+      "step": 5550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973294709383603e-05,
+      "loss": 0.0487,
+      "step": 5560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973246669645132e-05,
+      "loss": 0.0487,
+      "step": 5570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997319862990666e-05,
+      "loss": 0.0473,
+      "step": 5580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997315059016819e-05,
+      "loss": 0.0504,
+      "step": 5590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973102550429716e-05,
+      "loss": 0.051,
+      "step": 5600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973054510691246e-05,
+      "loss": 0.052,
+      "step": 5610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973006470952775e-05,
+      "loss": 0.0567,
+      "step": 5620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99729584312143e-05,
+      "loss": 0.0458,
+      "step": 5630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997291039147583e-05,
+      "loss": 0.0565,
+      "step": 5640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997286235173736e-05,
+      "loss": 0.0568,
+      "step": 5650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997281431199889e-05,
+      "loss": 0.0459,
+      "step": 5660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972766272260414e-05,
+      "loss": 0.0536,
+      "step": 5670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972718232521944e-05,
+      "loss": 0.0568,
+      "step": 5680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972670192783473e-05,
+      "loss": 0.0608,
+      "step": 5690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972622153045e-05,
+      "loss": 0.0505,
+      "step": 5700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972574113306528e-05,
+      "loss": 0.042,
+      "step": 5710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972526073568057e-05,
+      "loss": 0.0417,
+      "step": 5720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972478033829587e-05,
+      "loss": 0.05,
+      "step": 5730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972429994091113e-05,
+      "loss": 0.0432,
+      "step": 5740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972381954352642e-05,
+      "loss": 0.041,
+      "step": 5750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997233391461417e-05,
+      "loss": 0.0657,
+      "step": 5760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99722858748757e-05,
+      "loss": 0.0502,
+      "step": 5770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972237835137226e-05,
+      "loss": 0.0515,
+      "step": 5780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972189795398755e-05,
+      "loss": 0.0487,
+      "step": 5790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972141755660285e-05,
+      "loss": 0.0561,
+      "step": 5800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997209371592181e-05,
+      "loss": 0.046,
+      "step": 5810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997204567618334e-05,
+      "loss": 0.0466,
+      "step": 5820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997199763644487e-05,
+      "loss": 0.0511,
+      "step": 5830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99719495967064e-05,
+      "loss": 0.0585,
+      "step": 5840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9971901556967924e-05,
+      "loss": 0.0512,
+      "step": 5850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9971853517229454e-05,
+      "loss": 0.0563,
+      "step": 5860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9971805477490983e-05,
+      "loss": 0.0491,
+      "step": 5870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997175743775251e-05,
+      "loss": 0.0507,
+      "step": 5880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9971709398014038e-05,
+      "loss": 0.0406,
+      "step": 5890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9971661358275567e-05,
+      "loss": 0.0632,
+      "step": 5900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9971613318537097e-05,
+      "loss": 0.0485,
+      "step": 5910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9971565278798622e-05,
+      "loss": 0.0404,
+      "step": 5920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9971517239060152e-05,
+      "loss": 0.0485,
+      "step": 5930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997146919932168e-05,
+      "loss": 0.0594,
+      "step": 5940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997142115958321e-05,
+      "loss": 0.0534,
+      "step": 5950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9971373119844736e-05,
+      "loss": 0.0448,
+      "step": 5960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9971325080106265e-05,
+      "loss": 0.0442,
+      "step": 5970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9971277040367795e-05,
+      "loss": 0.0471,
+      "step": 5980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997122900062932e-05,
+      "loss": 0.0494,
+      "step": 5990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997118096089085e-05,
+      "loss": 0.0588,
+      "step": 6000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997113292115238e-05,
+      "loss": 0.0425,
+      "step": 6010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997108488141391e-05,
+      "loss": 0.0469,
+      "step": 6020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9971036841675434e-05,
+      "loss": 0.0537,
+      "step": 6030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970988801936964e-05,
+      "loss": 0.0436,
+      "step": 6040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970940762198493e-05,
+      "loss": 0.0605,
+      "step": 6050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997089272246002e-05,
+      "loss": 0.0514,
+      "step": 6060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970844682721548e-05,
+      "loss": 0.047,
+      "step": 6070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970796642983077e-05,
+      "loss": 0.0664,
+      "step": 6080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970748603244606e-05,
+      "loss": 0.0542,
+      "step": 6090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970700563506132e-05,
+      "loss": 0.0481,
+      "step": 6100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997065252376766e-05,
+      "loss": 0.0444,
+      "step": 6110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997060448402919e-05,
+      "loss": 0.0446,
+      "step": 6120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997055644429072e-05,
+      "loss": 0.0471,
+      "step": 6130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970508404552246e-05,
+      "loss": 0.0514,
+      "step": 6140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970460364813775e-05,
+      "loss": 0.0555,
+      "step": 6150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970412325075305e-05,
+      "loss": 0.0489,
+      "step": 6160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997036428533683e-05,
+      "loss": 0.0492,
+      "step": 6170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997031624559836e-05,
+      "loss": 0.0506,
+      "step": 6180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997026820585989e-05,
+      "loss": 0.0496,
+      "step": 6190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970220166121418e-05,
+      "loss": 0.0425,
+      "step": 6200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970172126382944e-05,
+      "loss": 0.053,
+      "step": 6210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970124086644473e-05,
+      "loss": 0.051,
+      "step": 6220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9970076046906003e-05,
+      "loss": 0.05,
+      "step": 6230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997002800716753e-05,
+      "loss": 0.0429,
+      "step": 6240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969979967429058e-05,
+      "loss": 0.0474,
+      "step": 6250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969931927690587e-05,
+      "loss": 0.044,
+      "step": 6260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969883887952116e-05,
+      "loss": 0.0501,
+      "step": 6270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969835848213642e-05,
+      "loss": 0.0462,
+      "step": 6280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996979261244902e-05,
+      "loss": 0.0448,
+      "step": 6290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969744572710546e-05,
+      "loss": 0.0505,
+      "step": 6300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969696532972075e-05,
+      "loss": 0.0444,
+      "step": 6310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969648493233605e-05,
+      "loss": 0.0462,
+      "step": 6320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969600453495134e-05,
+      "loss": 0.0447,
+      "step": 6330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996955241375666e-05,
+      "loss": 0.044,
+      "step": 6340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996950437401819e-05,
+      "loss": 0.0509,
+      "step": 6350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996945633427972e-05,
+      "loss": 0.0543,
+      "step": 6360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969408294541244e-05,
+      "loss": 0.051,
+      "step": 6370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969360254802774e-05,
+      "loss": 0.0424,
+      "step": 6380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969312215064303e-05,
+      "loss": 0.041,
+      "step": 6390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969264175325832e-05,
+      "loss": 0.0449,
+      "step": 6400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969216135587358e-05,
+      "loss": 0.0462,
+      "step": 6410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969168095848887e-05,
+      "loss": 0.0427,
+      "step": 6420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969120056110417e-05,
+      "loss": 0.0518,
+      "step": 6430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9969072016371946e-05,
+      "loss": 0.0437,
+      "step": 6440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996902397663347e-05,
+      "loss": 0.0428,
+      "step": 6450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9968975936895e-05,
+      "loss": 0.0564,
+      "step": 6460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996892789715653e-05,
+      "loss": 0.0461,
+      "step": 6470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9968879857418056e-05,
+      "loss": 0.0488,
+      "step": 6480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9968831817679585e-05,
+      "loss": 0.0444,
+      "step": 6490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9968783777941115e-05,
+      "loss": 0.0509,
+      "step": 6500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9968735738202644e-05,
+      "loss": 0.0517,
+      "step": 6510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996868769846417e-05,
+      "loss": 0.0446,
+      "step": 6520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99686396587257e-05,
+      "loss": 0.0414,
+      "step": 6530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996859161898723e-05,
+      "loss": 0.0485,
+      "step": 6540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9968543579248754e-05,
+      "loss": 0.0524,
+      "step": 6550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9968495539510283e-05,
+      "loss": 0.0501,
+      "step": 6560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9968447499771813e-05,
+      "loss": 0.0487,
+      "step": 6570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9968399460033342e-05,
+      "loss": 0.0457,
+      "step": 6580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9968351420294868e-05,
+      "loss": 0.0546,
+      "step": 6590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9968303380556397e-05,
+      "loss": 0.0521,
+      "step": 6600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9968255340817926e-05,
+      "loss": 0.0442,
+      "step": 6610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9968207301079456e-05,
+      "loss": 0.0516,
+      "step": 6620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996815926134098e-05,
+      "loss": 0.0448,
+      "step": 6630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996811122160251e-05,
+      "loss": 0.0498,
+      "step": 6640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996806318186404e-05,
+      "loss": 0.0415,
+      "step": 6650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9968015142125566e-05,
+      "loss": 0.0555,
+      "step": 6660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967967102387095e-05,
+      "loss": 0.0499,
+      "step": 6670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967919062648625e-05,
+      "loss": 0.0481,
+      "step": 6680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967871022910154e-05,
+      "loss": 0.0443,
+      "step": 6690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996782298317168e-05,
+      "loss": 0.0433,
+      "step": 6700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996777494343321e-05,
+      "loss": 0.0612,
+      "step": 6710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967726903694738e-05,
+      "loss": 0.051,
+      "step": 6720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967678863956264e-05,
+      "loss": 0.0422,
+      "step": 6730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967630824217793e-05,
+      "loss": 0.0419,
+      "step": 6740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967582784479323e-05,
+      "loss": 0.0437,
+      "step": 6750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967534744740852e-05,
+      "loss": 0.0589,
+      "step": 6760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967486705002378e-05,
+      "loss": 0.0416,
+      "step": 6770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967438665263907e-05,
+      "loss": 0.0452,
+      "step": 6780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967390625525436e-05,
+      "loss": 0.0441,
+      "step": 6790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967342585786966e-05,
+      "loss": 0.0534,
+      "step": 6800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996729454604849e-05,
+      "loss": 0.0483,
+      "step": 6810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996724650631002e-05,
+      "loss": 0.0471,
+      "step": 6820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996719846657155e-05,
+      "loss": 0.045,
+      "step": 6830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967150426833076e-05,
+      "loss": 0.0399,
+      "step": 6840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967102387094605e-05,
+      "loss": 0.044,
+      "step": 6850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967054347356134e-05,
+      "loss": 0.0435,
+      "step": 6860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9967006307617664e-05,
+      "loss": 0.0397,
+      "step": 6870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996695826787919e-05,
+      "loss": 0.045,
+      "step": 6880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996691022814072e-05,
+      "loss": 0.0432,
+      "step": 6890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966862188402248e-05,
+      "loss": 0.0482,
+      "step": 6900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966814148663774e-05,
+      "loss": 0.0412,
+      "step": 6910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966766108925303e-05,
+      "loss": 0.0557,
+      "step": 6920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966718069186833e-05,
+      "loss": 0.0487,
+      "step": 6930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966670029448362e-05,
+      "loss": 0.0367,
+      "step": 6940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966621989709888e-05,
+      "loss": 0.0468,
+      "step": 6950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966573949971417e-05,
+      "loss": 0.0492,
+      "step": 6960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966525910232946e-05,
+      "loss": 0.0416,
+      "step": 6970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966477870494476e-05,
+      "loss": 0.049,
+      "step": 6980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966429830756e-05,
+      "loss": 0.0469,
+      "step": 6990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996638179101753e-05,
+      "loss": 0.0433,
+      "step": 7000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996633375127906e-05,
+      "loss": 0.0456,
+      "step": 7010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966285711540586e-05,
+      "loss": 0.045,
+      "step": 7020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966237671802115e-05,
+      "loss": 0.0491,
+      "step": 7030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966189632063644e-05,
+      "loss": 0.0439,
+      "step": 7040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966141592325174e-05,
+      "loss": 0.0424,
+      "step": 7050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99660935525867e-05,
+      "loss": 0.0514,
+      "step": 7060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996604551284823e-05,
+      "loss": 0.0394,
+      "step": 7070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965997473109758e-05,
+      "loss": 0.0393,
+      "step": 7080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965949433371284e-05,
+      "loss": 0.0475,
+      "step": 7090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965901393632813e-05,
+      "loss": 0.0492,
+      "step": 7100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965853353894343e-05,
+      "loss": 0.0449,
+      "step": 7110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965805314155872e-05,
+      "loss": 0.0393,
+      "step": 7120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965757274417398e-05,
+      "loss": 0.0469,
+      "step": 7130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965709234678927e-05,
+      "loss": 0.0465,
+      "step": 7140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965661194940456e-05,
+      "loss": 0.0542,
+      "step": 7150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965613155201985e-05,
+      "loss": 0.0456,
+      "step": 7160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996556511546351e-05,
+      "loss": 0.0498,
+      "step": 7170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996551707572504e-05,
+      "loss": 0.0513,
+      "step": 7180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996546903598657e-05,
+      "loss": 0.043,
+      "step": 7190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965420996248096e-05,
+      "loss": 0.0432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965372956509625e-05,
+      "loss": 0.0541,
+      "step": 7210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965324916771154e-05,
+      "loss": 0.0506,
+      "step": 7220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965276877032684e-05,
+      "loss": 0.0481,
+      "step": 7230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996522883729421e-05,
+      "loss": 0.0442,
+      "step": 7240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996518079755574e-05,
+      "loss": 0.0476,
+      "step": 7250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965132757817268e-05,
+      "loss": 0.0409,
+      "step": 7260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965084718078794e-05,
+      "loss": 0.044,
+      "step": 7270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965036678340323e-05,
+      "loss": 0.0565,
+      "step": 7280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964988638601852e-05,
+      "loss": 0.0496,
+      "step": 7290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996494059886338e-05,
+      "loss": 0.051,
+      "step": 7300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964892559124908e-05,
+      "loss": 0.0415,
+      "step": 7310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964844519386437e-05,
+      "loss": 0.0451,
+      "step": 7320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964796479647966e-05,
+      "loss": 0.0418,
+      "step": 7330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964748439909495e-05,
+      "loss": 0.0414,
+      "step": 7340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996470040017102e-05,
+      "loss": 0.0397,
+      "step": 7350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996465236043255e-05,
+      "loss": 0.0422,
+      "step": 7360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996460432069408e-05,
+      "loss": 0.0431,
+      "step": 7370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964556280955606e-05,
+      "loss": 0.0442,
+      "step": 7380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964508241217135e-05,
+      "loss": 0.0445,
+      "step": 7390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964460201478664e-05,
+      "loss": 0.0502,
+      "step": 7400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964412161740193e-05,
+      "loss": 0.0482,
+      "step": 7410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996436412200172e-05,
+      "loss": 0.0429,
+      "step": 7420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996431608226325e-05,
+      "loss": 0.0532,
+      "step": 7430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964268042524778e-05,
+      "loss": 0.0401,
+      "step": 7440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964220002786307e-05,
+      "loss": 0.0354,
+      "step": 7450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964171963047836e-05,
+      "loss": 0.038,
+      "step": 7460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964123923309366e-05,
+      "loss": 0.0425,
+      "step": 7470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996407588357089e-05,
+      "loss": 0.054,
+      "step": 7480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996402784383242e-05,
+      "loss": 0.0407,
+      "step": 7490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996397980409395e-05,
+      "loss": 0.0421,
+      "step": 7500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963931764355476e-05,
+      "loss": 0.0517,
+      "step": 7510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963883724617005e-05,
+      "loss": 0.0491,
+      "step": 7520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963835684878535e-05,
+      "loss": 0.0425,
+      "step": 7530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963787645140064e-05,
+      "loss": 0.0374,
+      "step": 7540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996373960540159e-05,
+      "loss": 0.0467,
+      "step": 7550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996369156566312e-05,
+      "loss": 0.0467,
+      "step": 7560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963643525924648e-05,
+      "loss": 0.0496,
+      "step": 7570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963595486186174e-05,
+      "loss": 0.0522,
+      "step": 7580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963547446447703e-05,
+      "loss": 0.0439,
+      "step": 7590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963499406709233e-05,
+      "loss": 0.0417,
+      "step": 7600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963451366970762e-05,
+      "loss": 0.0456,
+      "step": 7610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963403327232288e-05,
+      "loss": 0.0461,
+      "step": 7620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963355287493817e-05,
+      "loss": 0.0454,
+      "step": 7630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963307247755346e-05,
+      "loss": 0.0557,
+      "step": 7640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963259208016876e-05,
+      "loss": 0.0447,
+      "step": 7650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99632111682784e-05,
+      "loss": 0.0451,
+      "step": 7660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996316312853993e-05,
+      "loss": 0.0338,
+      "step": 7670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996311508880146e-05,
+      "loss": 0.0426,
+      "step": 7680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963067049062986e-05,
+      "loss": 0.0508,
+      "step": 7690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963019009324515e-05,
+      "loss": 0.0413,
+      "step": 7700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962970969586044e-05,
+      "loss": 0.0433,
+      "step": 7710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962922929847574e-05,
+      "loss": 0.048,
+      "step": 7720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99628748901091e-05,
+      "loss": 0.0504,
+      "step": 7730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996282685037063e-05,
+      "loss": 0.0401,
+      "step": 7740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962778810632158e-05,
+      "loss": 0.0398,
+      "step": 7750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962730770893684e-05,
+      "loss": 0.0437,
+      "step": 7760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962682731155213e-05,
+      "loss": 0.0541,
+      "step": 7770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962634691416743e-05,
+      "loss": 0.0475,
+      "step": 7780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962586651678272e-05,
+      "loss": 0.046,
+      "step": 7790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962538611939798e-05,
+      "loss": 0.0426,
+      "step": 7800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962490572201327e-05,
+      "loss": 0.0422,
+      "step": 7810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962442532462856e-05,
+      "loss": 0.0392,
+      "step": 7820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962394492724386e-05,
+      "loss": 0.0428,
+      "step": 7830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996234645298591e-05,
+      "loss": 0.0395,
+      "step": 7840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996229841324744e-05,
+      "loss": 0.055,
+      "step": 7850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996225037350897e-05,
+      "loss": 0.0427,
+      "step": 7860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962202333770496e-05,
+      "loss": 0.044,
+      "step": 7870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962154294032025e-05,
+      "loss": 0.0369,
+      "step": 7880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962106254293554e-05,
+      "loss": 0.0486,
+      "step": 7890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962058214555084e-05,
+      "loss": 0.0445,
+      "step": 7900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996201017481661e-05,
+      "loss": 0.0426,
+      "step": 7910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996196213507814e-05,
+      "loss": 0.0466,
+      "step": 7920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961914095339668e-05,
+      "loss": 0.0526,
+      "step": 7930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961866055601194e-05,
+      "loss": 0.0444,
+      "step": 7940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961818015862723e-05,
+      "loss": 0.0438,
+      "step": 7950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961769976124253e-05,
+      "loss": 0.0434,
+      "step": 7960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961721936385782e-05,
+      "loss": 0.0478,
+      "step": 7970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961673896647308e-05,
+      "loss": 0.0366,
+      "step": 7980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961625856908837e-05,
+      "loss": 0.043,
+      "step": 7990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961577817170366e-05,
+      "loss": 0.0416,
+      "step": 8000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961529777431895e-05,
+      "loss": 0.0388,
+      "step": 8010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996148173769342e-05,
+      "loss": 0.0423,
+      "step": 8020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996143369795495e-05,
+      "loss": 0.0425,
+      "step": 8030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996138565821648e-05,
+      "loss": 0.0483,
+      "step": 8040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961337618478006e-05,
+      "loss": 0.0497,
+      "step": 8050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961289578739535e-05,
+      "loss": 0.0394,
+      "step": 8060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961241539001064e-05,
+      "loss": 0.0461,
+      "step": 8070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961193499262594e-05,
+      "loss": 0.0401,
+      "step": 8080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996114545952412e-05,
+      "loss": 0.0447,
+      "step": 8090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996109741978565e-05,
+      "loss": 0.0402,
+      "step": 8100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961049380047178e-05,
+      "loss": 0.0508,
+      "step": 8110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961001340308704e-05,
+      "loss": 0.0549,
+      "step": 8120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960953300570233e-05,
+      "loss": 0.0408,
+      "step": 8130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960905260831762e-05,
+      "loss": 0.0425,
+      "step": 8140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996085722109329e-05,
+      "loss": 0.0511,
+      "step": 8150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960809181354818e-05,
+      "loss": 0.0448,
+      "step": 8160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960761141616347e-05,
+      "loss": 0.0363,
+      "step": 8170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960713101877876e-05,
+      "loss": 0.0381,
+      "step": 8180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960665062139405e-05,
+      "loss": 0.0415,
+      "step": 8190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996061702240093e-05,
+      "loss": 0.0524,
+      "step": 8200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996056898266246e-05,
+      "loss": 0.0385,
+      "step": 8210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996052094292399e-05,
+      "loss": 0.0407,
+      "step": 8220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960472903185516e-05,
+      "loss": 0.0479,
+      "step": 8230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960424863447045e-05,
+      "loss": 0.0437,
+      "step": 8240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960376823708574e-05,
+      "loss": 0.0442,
+      "step": 8250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960328783970103e-05,
+      "loss": 0.0432,
+      "step": 8260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996028074423163e-05,
+      "loss": 0.0479,
+      "step": 8270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996023270449316e-05,
+      "loss": 0.0445,
+      "step": 8280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960184664754688e-05,
+      "loss": 0.0353,
+      "step": 8290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960136625016214e-05,
+      "loss": 0.0427,
+      "step": 8300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960088585277743e-05,
+      "loss": 0.045,
+      "step": 8310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960040545539272e-05,
+      "loss": 0.0514,
+      "step": 8320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99599925058008e-05,
+      "loss": 0.0409,
+      "step": 8330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959944466062327e-05,
+      "loss": 0.0408,
+      "step": 8340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959896426323857e-05,
+      "loss": 0.0421,
+      "step": 8350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959848386585386e-05,
+      "loss": 0.0452,
+      "step": 8360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959800346846915e-05,
+      "loss": 0.0477,
+      "step": 8370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995975230710844e-05,
+      "loss": 0.0444,
+      "step": 8380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995970426736997e-05,
+      "loss": 0.0442,
+      "step": 8390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99596562276315e-05,
+      "loss": 0.0408,
+      "step": 8400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959608187893026e-05,
+      "loss": 0.0459,
+      "step": 8410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959560148154555e-05,
+      "loss": 0.0411,
+      "step": 8420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959512108416084e-05,
+      "loss": 0.0498,
+      "step": 8430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959464068677613e-05,
+      "loss": 0.039,
+      "step": 8440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995941602893914e-05,
+      "loss": 0.0347,
+      "step": 8450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995936798920067e-05,
+      "loss": 0.0432,
+      "step": 8460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959319949462198e-05,
+      "loss": 0.046,
+      "step": 8470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959271909723724e-05,
+      "loss": 0.0435,
+      "step": 8480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959223869985253e-05,
+      "loss": 0.0349,
+      "step": 8490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959175830246782e-05,
+      "loss": 0.0446,
+      "step": 8500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995912779050831e-05,
+      "loss": 0.0397,
+      "step": 8510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959079750769837e-05,
+      "loss": 0.0368,
+      "step": 8520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959031711031367e-05,
+      "loss": 0.0359,
+      "step": 8530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958983671292896e-05,
+      "loss": 0.0403,
+      "step": 8540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958935631554425e-05,
+      "loss": 0.0395,
+      "step": 8550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995888759181595e-05,
+      "loss": 0.0476,
+      "step": 8560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995883955207748e-05,
+      "loss": 0.043,
+      "step": 8570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995879151233901e-05,
+      "loss": 0.0383,
+      "step": 8580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958743472600536e-05,
+      "loss": 0.0438,
+      "step": 8590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958695432862065e-05,
+      "loss": 0.0433,
+      "step": 8600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958647393123594e-05,
+      "loss": 0.0385,
+      "step": 8610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958599353385123e-05,
+      "loss": 0.0391,
+      "step": 8620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995855131364665e-05,
+      "loss": 0.0447,
+      "step": 8630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995850327390818e-05,
+      "loss": 0.0401,
+      "step": 8640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958455234169708e-05,
+      "loss": 0.0377,
+      "step": 8650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958407194431234e-05,
+      "loss": 0.0411,
+      "step": 8660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958359154692763e-05,
+      "loss": 0.0423,
+      "step": 8670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9958311114954292e-05,
+      "loss": 0.0352,
+      "step": 8680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995826307521582e-05,
+      "loss": 0.0338,
+      "step": 8690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9958215035477347e-05,
+      "loss": 0.0465,
+      "step": 8700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9958166995738877e-05,
+      "loss": 0.0409,
+      "step": 8710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9958118956000406e-05,
+      "loss": 0.0457,
+      "step": 8720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9958070916261935e-05,
+      "loss": 0.0362,
+      "step": 8730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995802287652346e-05,
+      "loss": 0.0379,
+      "step": 8740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995797483678499e-05,
+      "loss": 0.0332,
+      "step": 8750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995792679704652e-05,
+      "loss": 0.0459,
+      "step": 8760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957878757308045e-05,
+      "loss": 0.0331,
+      "step": 8770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957830717569575e-05,
+      "loss": 0.0431,
+      "step": 8780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957782677831104e-05,
+      "loss": 0.038,
+      "step": 8790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957734638092633e-05,
+      "loss": 0.0493,
+      "step": 8800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995768659835416e-05,
+      "loss": 0.0391,
+      "step": 8810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995763855861569e-05,
+      "loss": 0.038,
+      "step": 8820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957590518877218e-05,
+      "loss": 0.0407,
+      "step": 8830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957542479138744e-05,
+      "loss": 0.0413,
+      "step": 8840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957494439400273e-05,
+      "loss": 0.0458,
+      "step": 8850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957446399661802e-05,
+      "loss": 0.0401,
+      "step": 8860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995739835992333e-05,
+      "loss": 0.045,
+      "step": 8870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957350320184857e-05,
+      "loss": 0.0501,
+      "step": 8880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957302280446386e-05,
+      "loss": 0.0392,
+      "step": 8890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957254240707916e-05,
+      "loss": 0.0405,
+      "step": 8900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957206200969445e-05,
+      "loss": 0.0369,
+      "step": 8910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995715816123097e-05,
+      "loss": 0.0368,
+      "step": 8920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99571101214925e-05,
+      "loss": 0.0409,
+      "step": 8930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995706208175403e-05,
+      "loss": 0.0451,
+      "step": 8940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9957014042015555e-05,
+      "loss": 0.0412,
+      "step": 8950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956966002277085e-05,
+      "loss": 0.036,
+      "step": 8960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956917962538614e-05,
+      "loss": 0.0414,
+      "step": 8970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956869922800143e-05,
+      "loss": 0.042,
+      "step": 8980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995682188306167e-05,
+      "loss": 0.0406,
+      "step": 8990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956773843323198e-05,
+      "loss": 0.042,
+      "step": 9000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956725803584728e-05,
+      "loss": 0.0486,
+      "step": 9010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956677763846253e-05,
+      "loss": 0.0418,
+      "step": 9020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956629724107783e-05,
+      "loss": 0.0431,
+      "step": 9030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956581684369312e-05,
+      "loss": 0.0377,
+      "step": 9040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995653364463084e-05,
+      "loss": 0.0403,
+      "step": 9050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956485604892367e-05,
+      "loss": 0.0447,
+      "step": 9060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956437565153896e-05,
+      "loss": 0.0449,
+      "step": 9070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956389525415426e-05,
+      "loss": 0.0344,
+      "step": 9080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956341485676955e-05,
+      "loss": 0.0404,
+      "step": 9090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995629824991233e-05,
+      "loss": 0.0409,
+      "step": 9100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995625021017386e-05,
+      "loss": 0.0363,
+      "step": 9110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956202170435385e-05,
+      "loss": 0.0358,
+      "step": 9120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956154130696914e-05,
+      "loss": 0.0489,
+      "step": 9130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956106090958443e-05,
+      "loss": 0.0407,
+      "step": 9140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995605805121997e-05,
+      "loss": 0.0374,
+      "step": 9150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99560100114815e-05,
+      "loss": 0.0385,
+      "step": 9160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955961971743028e-05,
+      "loss": 0.037,
+      "step": 9170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955913932004557e-05,
+      "loss": 0.0413,
+      "step": 9180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955865892266083e-05,
+      "loss": 0.0393,
+      "step": 9190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955817852527612e-05,
+      "loss": 0.0405,
+      "step": 9200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995576981278914e-05,
+      "loss": 0.0468,
+      "step": 9210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995572177305067e-05,
+      "loss": 0.0381,
+      "step": 9220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955673733312197e-05,
+      "loss": 0.0403,
+      "step": 9230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955625693573726e-05,
+      "loss": 0.0412,
+      "step": 9240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955577653835255e-05,
+      "loss": 0.0403,
+      "step": 9250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995552961409678e-05,
+      "loss": 0.0368,
+      "step": 9260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995548157435831e-05,
+      "loss": 0.0504,
+      "step": 9270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995543353461984e-05,
+      "loss": 0.0404,
+      "step": 9280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995538549488137e-05,
+      "loss": 0.0357,
+      "step": 9290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955337455142895e-05,
+      "loss": 0.0374,
+      "step": 9300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955289415404424e-05,
+      "loss": 0.0325,
+      "step": 9310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955241375665953e-05,
+      "loss": 0.0378,
+      "step": 9320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995519333592748e-05,
+      "loss": 0.0473,
+      "step": 9330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995514529618901e-05,
+      "loss": 0.0405,
+      "step": 9340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955097256450538e-05,
+      "loss": 0.041,
+      "step": 9350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955049216712067e-05,
+      "loss": 0.0404,
+      "step": 9360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9955001176973593e-05,
+      "loss": 0.0355,
+      "step": 9370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954953137235122e-05,
+      "loss": 0.0373,
+      "step": 9380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995490509749665e-05,
+      "loss": 0.0339,
+      "step": 9390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995485705775818e-05,
+      "loss": 0.0362,
+      "step": 9400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954809018019706e-05,
+      "loss": 0.0511,
+      "step": 9410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954760978281236e-05,
+      "loss": 0.0361,
+      "step": 9420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954712938542765e-05,
+      "loss": 0.0442,
+      "step": 9430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995466489880429e-05,
+      "loss": 0.0384,
+      "step": 9440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995461685906582e-05,
+      "loss": 0.0372,
+      "step": 9450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995456881932735e-05,
+      "loss": 0.0455,
+      "step": 9460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995452077958888e-05,
+      "loss": 0.0366,
+      "step": 9470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954472739850405e-05,
+      "loss": 0.0411,
+      "step": 9480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954424700111934e-05,
+      "loss": 0.0447,
+      "step": 9490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954376660373463e-05,
+      "loss": 0.0458,
+      "step": 9500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995432862063499e-05,
+      "loss": 0.0373,
+      "step": 9510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954280580896518e-05,
+      "loss": 0.0407,
+      "step": 9520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954232541158048e-05,
+      "loss": 0.0417,
+      "step": 9530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954184501419577e-05,
+      "loss": 0.0409,
+      "step": 9540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954136461681103e-05,
+      "loss": 0.0417,
+      "step": 9550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9954088421942632e-05,
+      "loss": 0.0463,
+      "step": 9560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995404038220416e-05,
+      "loss": 0.039,
+      "step": 9570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995399234246569e-05,
+      "loss": 0.0364,
+      "step": 9580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953944302727216e-05,
+      "loss": 0.0365,
+      "step": 9590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953896262988746e-05,
+      "loss": 0.0381,
+      "step": 9600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953848223250275e-05,
+      "loss": 0.0366,
+      "step": 9610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99538001835118e-05,
+      "loss": 0.0432,
+      "step": 9620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995375214377333e-05,
+      "loss": 0.0373,
+      "step": 9630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995370410403486e-05,
+      "loss": 0.0332,
+      "step": 9640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995365606429639e-05,
+      "loss": 0.0389,
+      "step": 9650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953608024557914e-05,
+      "loss": 0.0426,
+      "step": 9660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953559984819444e-05,
+      "loss": 0.0382,
+      "step": 9670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953511945080973e-05,
+      "loss": 0.0342,
+      "step": 9680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99534639053425e-05,
+      "loss": 0.0372,
+      "step": 9690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953415865604028e-05,
+      "loss": 0.0373,
+      "step": 9700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953367825865557e-05,
+      "loss": 0.0424,
+      "step": 9710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953319786127087e-05,
+      "loss": 0.0437,
+      "step": 9720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953271746388613e-05,
+      "loss": 0.0403,
+      "step": 9730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953223706650142e-05,
+      "loss": 0.0422,
+      "step": 9740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995317566691167e-05,
+      "loss": 0.0395,
+      "step": 9750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99531276271732e-05,
+      "loss": 0.0366,
+      "step": 9760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953079587434726e-05,
+      "loss": 0.0396,
+      "step": 9770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9953031547696256e-05,
+      "loss": 0.0406,
+      "step": 9780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9952983507957785e-05,
+      "loss": 0.0416,
+      "step": 9790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995293546821931e-05,
+      "loss": 0.0351,
+      "step": 9800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995288742848084e-05,
+      "loss": 0.0507,
+      "step": 9810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995283938874237e-05,
+      "loss": 0.0338,
+      "step": 9820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99527913490039e-05,
+      "loss": 0.0462,
+      "step": 9830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9952743309265424e-05,
+      "loss": 0.0434,
+      "step": 9840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9952695269526954e-05,
+      "loss": 0.0372,
+      "step": 9850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9952647229788483e-05,
+      "loss": 0.0418,
+      "step": 9860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995259919005001e-05,
+      "loss": 0.0332,
+      "step": 9870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9952551150311538e-05,
+      "loss": 0.0385,
+      "step": 9880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9952503110573067e-05,
+      "loss": 0.0375,
+      "step": 9890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9952455070834597e-05,
+      "loss": 0.0393,
+      "step": 9900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9952407031096123e-05,
+      "loss": 0.041,
+      "step": 9910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9952358991357652e-05,
+      "loss": 0.0381,
+      "step": 9920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995231095161918e-05,
+      "loss": 0.033,
+      "step": 9930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995226291188071e-05,
+      "loss": 0.046,
+      "step": 9940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9952214872142236e-05,
+      "loss": 0.0372,
+      "step": 9950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9952166832403765e-05,
+      "loss": 0.0368,
+      "step": 9960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9952118792665295e-05,
+      "loss": 0.0387,
+      "step": 9970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995207075292682e-05,
+      "loss": 0.0472,
+      "step": 9980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995202271318835e-05,
+      "loss": 0.0386,
+      "step": 9990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995197467344988e-05,
+      "loss": 0.0333,
+      "step": 10000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995192663371141e-05,
+      "loss": 0.0411,
+      "step": 10010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9951878593972934e-05,
+      "loss": 0.0378,
+      "step": 10020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9951830554234464e-05,
+      "loss": 0.0461,
+      "step": 10030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9951782514495993e-05,
+      "loss": 0.0416,
+      "step": 10040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.995173447475752e-05,
+      "loss": 0.0481,
+      "step": 10050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9951686435019048e-05,
+      "loss": 0.0439,
+      "step": 10060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9951638395280577e-05,
+      "loss": 0.0345,
+      "step": 10070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9951590355542107e-05,
+      "loss": 0.0425,
+      "step": 10080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9951542315803632e-05,
+      "loss": 0.0345,
+      "step": 10090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9951494276065162e-05,
+      "loss": 0.0333,
+      "step": 10100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995144623632669e-05,
+      "loss": 0.034,
+      "step": 10110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995139819658822e-05,
+      "loss": 0.0418,
+      "step": 10120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9951350156849746e-05,
+      "loss": 0.0348,
+      "step": 10130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9951302117111275e-05,
+      "loss": 0.0391,
+      "step": 10140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9951254077372805e-05,
+      "loss": 0.039,
+      "step": 10150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995120603763433e-05,
+      "loss": 0.0368,
+      "step": 10160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995115799789586e-05,
+      "loss": 0.0335,
+      "step": 10170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995110995815739e-05,
+      "loss": 0.0325,
+      "step": 10180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995106191841892e-05,
+      "loss": 0.0346,
+      "step": 10190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9951013878680444e-05,
+      "loss": 0.0363,
+      "step": 10200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950965838941974e-05,
+      "loss": 0.04,
+      "step": 10210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950917799203503e-05,
+      "loss": 0.0407,
+      "step": 10220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995086975946503e-05,
+      "loss": 0.0375,
+      "step": 10230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950821719726558e-05,
+      "loss": 0.0392,
+      "step": 10240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950773679988087e-05,
+      "loss": 0.0288,
+      "step": 10250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950725640249616e-05,
+      "loss": 0.0356,
+      "step": 10260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950677600511142e-05,
+      "loss": 0.0522,
+      "step": 10270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995062956077267e-05,
+      "loss": 0.039,
+      "step": 10280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99505815210342e-05,
+      "loss": 0.0373,
+      "step": 10290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995053348129573e-05,
+      "loss": 0.0329,
+      "step": 10300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950485441557256e-05,
+      "loss": 0.0397,
+      "step": 10310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950437401818785e-05,
+      "loss": 0.0361,
+      "step": 10320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950389362080315e-05,
+      "loss": 0.0427,
+      "step": 10330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995034132234184e-05,
+      "loss": 0.0359,
+      "step": 10340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995029328260337e-05,
+      "loss": 0.0366,
+      "step": 10350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99502452428649e-05,
+      "loss": 0.0371,
+      "step": 10360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950197203126428e-05,
+      "loss": 0.037,
+      "step": 10370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950149163387954e-05,
+      "loss": 0.0372,
+      "step": 10380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950101123649483e-05,
+      "loss": 0.0374,
+      "step": 10390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950053083911013e-05,
+      "loss": 0.0432,
+      "step": 10400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995000504417254e-05,
+      "loss": 0.0376,
+      "step": 10410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949957004434068e-05,
+      "loss": 0.0347,
+      "step": 10420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949908964695597e-05,
+      "loss": 0.0429,
+      "step": 10430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949860924957126e-05,
+      "loss": 0.04,
+      "step": 10440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949812885218652e-05,
+      "loss": 0.0395,
+      "step": 10450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.994976484548018e-05,
+      "loss": 0.0384,
+      "step": 10460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.994971680574171e-05,
+      "loss": 0.0374,
+      "step": 10470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.994966876600324e-05,
+      "loss": 0.0426,
+      "step": 10480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949620726264766e-05,
+      "loss": 0.0335,
+      "step": 10490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949572686526295e-05,
+      "loss": 0.0344,
+      "step": 10500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949524646787824e-05,
+      "loss": 0.0397,
+      "step": 10510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949476607049354e-05,
+      "loss": 0.0481,
+      "step": 10520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949428567310883e-05,
+      "loss": 0.0426,
+      "step": 10530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949380527572412e-05,
+      "loss": 0.0368,
+      "step": 10540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949332487833938e-05,
+      "loss": 0.0357,
+      "step": 10550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949284448095467e-05,
+      "loss": 0.0379,
+      "step": 10560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949236408356997e-05,
+      "loss": 0.0353,
+      "step": 10570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949188368618523e-05,
+      "loss": 0.0379,
+      "step": 10580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9949140328880052e-05,
+      "loss": 0.0453,
+      "step": 10590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.994909228914158e-05,
+      "loss": 0.0351,
+      "step": 10600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.994904424940311e-05,
+      "loss": 0.0378,
+      "step": 10610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9948996209664636e-05,
+      "loss": 0.037,
+      "step": 10620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9948948169926166e-05,
+      "loss": 0.0377,
+      "step": 10630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9948900130187695e-05,
+      "loss": 0.0414,
+      "step": 10640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.994885209044922e-05,
+      "loss": 0.0476,
+      "step": 10650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.994880405071075e-05,
+      "loss": 0.0301,
+      "step": 10660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.994875601097228e-05,
+      "loss": 0.0385,
+      "step": 10670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.994870797123381e-05,
+      "loss": 0.0371,
+      "step": 10680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9948659931495334e-05,
+      "loss": 0.0408,
+      "step": 10690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9948611891756864e-05,
+      "loss": 0.0384,
+      "step": 10700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9948563852018393e-05,
+      "loss": 0.0321,
+      "step": 10710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.994851581227992e-05,
+      "loss": 0.035,
+      "step": 10720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9948467772541448e-05,
+      "loss": 0.0408,
+      "step": 10730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9948419732802977e-05,
+      "loss": 0.0352,
+      "step": 10740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9948371693064507e-05,
+      "loss": 0.041,
+      "step": 10750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9948323653326033e-05,
+      "loss": 0.0406,
+      "step": 10760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9948275613587562e-05,
+      "loss": 0.0326,
+      "step": 10770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994822757384909e-05,
+      "loss": 0.037,
+      "step": 10780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994817953411062e-05,
+      "loss": 0.041,
+      "step": 10790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9948131494372146e-05,
+      "loss": 0.0379,
+      "step": 10800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9948083454633675e-05,
+      "loss": 0.052,
+      "step": 10810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9948035414895205e-05,
+      "loss": 0.0443,
+      "step": 10820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994798737515673e-05,
+      "loss": 0.0359,
+      "step": 10830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994793933541826e-05,
+      "loss": 0.0461,
+      "step": 10840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994789129567979e-05,
+      "loss": 0.0324,
+      "step": 10850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994784325594132e-05,
+      "loss": 0.0394,
+      "step": 10860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9947795216202844e-05,
+      "loss": 0.0474,
+      "step": 10870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9947747176464374e-05,
+      "loss": 0.0341,
+      "step": 10880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9947699136725903e-05,
+      "loss": 0.0293,
+      "step": 10890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994765109698743e-05,
+      "loss": 0.0353,
+      "step": 10900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9947603057248958e-05,
+      "loss": 0.0389,
+      "step": 10910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9947555017510487e-05,
+      "loss": 0.0385,
+      "step": 10920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9947506977772017e-05,
+      "loss": 0.0386,
+      "step": 10930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9947458938033542e-05,
+      "loss": 0.0347,
+      "step": 10940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9947410898295072e-05,
+      "loss": 0.0394,
+      "step": 10950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99473628585566e-05,
+      "loss": 0.0398,
+      "step": 10960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994731481881813e-05,
+      "loss": 0.033,
+      "step": 10970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9947266779079656e-05,
+      "loss": 0.0292,
+      "step": 10980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9947218739341185e-05,
+      "loss": 0.0347,
+      "step": 10990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9947170699602715e-05,
+      "loss": 0.0402,
+      "step": 11000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994712265986424e-05,
+      "loss": 0.0319,
+      "step": 11010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994707462012577e-05,
+      "loss": 0.0309,
+      "step": 11020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99470265803873e-05,
+      "loss": 0.0364,
+      "step": 11030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994697854064883e-05,
+      "loss": 0.0413,
+      "step": 11040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946930500910354e-05,
+      "loss": 0.0392,
+      "step": 11050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946882461171884e-05,
+      "loss": 0.0379,
+      "step": 11060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946834421433413e-05,
+      "loss": 0.0356,
+      "step": 11070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994678638169494e-05,
+      "loss": 0.0373,
+      "step": 11080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946738341956468e-05,
+      "loss": 0.0287,
+      "step": 11090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946690302217997e-05,
+      "loss": 0.0387,
+      "step": 11100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946642262479526e-05,
+      "loss": 0.0418,
+      "step": 11110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946594222741052e-05,
+      "loss": 0.0359,
+      "step": 11120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994654618300258e-05,
+      "loss": 0.0369,
+      "step": 11130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994649814326411e-05,
+      "loss": 0.0378,
+      "step": 11140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994645010352564e-05,
+      "loss": 0.0371,
+      "step": 11150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946402063787166e-05,
+      "loss": 0.0288,
+      "step": 11160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946354024048695e-05,
+      "loss": 0.0386,
+      "step": 11170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946305984310225e-05,
+      "loss": 0.0333,
+      "step": 11180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994625794457175e-05,
+      "loss": 0.044,
+      "step": 11190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994620990483328e-05,
+      "loss": 0.0448,
+      "step": 11200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994616186509481e-05,
+      "loss": 0.0385,
+      "step": 11210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946113825356338e-05,
+      "loss": 0.0346,
+      "step": 11220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946065785617864e-05,
+      "loss": 0.0468,
+      "step": 11230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9946017745879393e-05,
+      "loss": 0.0406,
+      "step": 11240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945969706140923e-05,
+      "loss": 0.0362,
+      "step": 11250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994592166640245e-05,
+      "loss": 0.031,
+      "step": 11260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945873626663978e-05,
+      "loss": 0.0385,
+      "step": 11270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945825586925507e-05,
+      "loss": 0.036,
+      "step": 11280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945777547187036e-05,
+      "loss": 0.037,
+      "step": 11290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945729507448562e-05,
+      "loss": 0.0311,
+      "step": 11300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994568146771009e-05,
+      "loss": 0.0387,
+      "step": 11310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994563342797162e-05,
+      "loss": 0.0373,
+      "step": 11320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994558538823315e-05,
+      "loss": 0.0416,
+      "step": 11330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945537348494676e-05,
+      "loss": 0.0295,
+      "step": 11340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945489308756205e-05,
+      "loss": 0.0388,
+      "step": 11350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945441269017734e-05,
+      "loss": 0.0408,
+      "step": 11360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994539322927926e-05,
+      "loss": 0.0435,
+      "step": 11370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994534518954079e-05,
+      "loss": 0.0402,
+      "step": 11380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994529714980232e-05,
+      "loss": 0.0294,
+      "step": 11390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945249110063848e-05,
+      "loss": 0.0466,
+      "step": 11400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945201070325374e-05,
+      "loss": 0.0317,
+      "step": 11410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945153030586903e-05,
+      "loss": 0.0395,
+      "step": 11420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945104990848433e-05,
+      "loss": 0.0416,
+      "step": 11430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994505695110996e-05,
+      "loss": 0.033,
+      "step": 11440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9945008911371488e-05,
+      "loss": 0.0355,
+      "step": 11450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9944960871633017e-05,
+      "loss": 0.0574,
+      "step": 11460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9944912831894546e-05,
+      "loss": 0.0338,
+      "step": 11470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9944864792156072e-05,
+      "loss": 0.0383,
+      "step": 11480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99448167524176e-05,
+      "loss": 0.0423,
+      "step": 11490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994476871267913e-05,
+      "loss": 0.0452,
+      "step": 11500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994472067294066e-05,
+      "loss": 0.0365,
+      "step": 11510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9944672633202186e-05,
+      "loss": 0.0337,
+      "step": 11520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9944624593463715e-05,
+      "loss": 0.0388,
+      "step": 11530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9944576553725244e-05,
+      "loss": 0.0304,
+      "step": 11540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994452851398677e-05,
+      "loss": 0.041,
+      "step": 11550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99444804742483e-05,
+      "loss": 0.0346,
+      "step": 11560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994443243450983e-05,
+      "loss": 0.0407,
+      "step": 11570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9944384394771358e-05,
+      "loss": 0.0355,
+      "step": 11580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9944336355032884e-05,
+      "loss": 0.0432,
+      "step": 11590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9944288315294413e-05,
+      "loss": 0.0407,
+      "step": 11600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9944240275555943e-05,
+      "loss": 0.0387,
+      "step": 11610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994419223581747e-05,
+      "loss": 0.036,
+      "step": 11620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9944149000052846e-05,
+      "loss": 0.0318,
+      "step": 11630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9944100960314376e-05,
+      "loss": 0.0349,
+      "step": 11640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99440529205759e-05,
+      "loss": 0.0417,
+      "step": 11650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994400488083743e-05,
+      "loss": 0.0344,
+      "step": 11660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994395684109896e-05,
+      "loss": 0.0352,
+      "step": 11670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943908801360486e-05,
+      "loss": 0.0338,
+      "step": 11680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943860761622015e-05,
+      "loss": 0.0336,
+      "step": 11690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943812721883545e-05,
+      "loss": 0.0316,
+      "step": 11700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943764682145074e-05,
+      "loss": 0.0452,
+      "step": 11710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99437166424066e-05,
+      "loss": 0.0367,
+      "step": 11720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994366860266813e-05,
+      "loss": 0.0391,
+      "step": 11730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943620562929658e-05,
+      "loss": 0.0379,
+      "step": 11740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943572523191188e-05,
+      "loss": 0.032,
+      "step": 11750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943524483452713e-05,
+      "loss": 0.039,
+      "step": 11760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943476443714243e-05,
+      "loss": 0.0346,
+      "step": 11770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943428403975772e-05,
+      "loss": 0.031,
+      "step": 11780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943380364237298e-05,
+      "loss": 0.0324,
+      "step": 11790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943332324498827e-05,
+      "loss": 0.0357,
+      "step": 11800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943284284760356e-05,
+      "loss": 0.0351,
+      "step": 11810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943236245021886e-05,
+      "loss": 0.0351,
+      "step": 11820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994318820528341e-05,
+      "loss": 0.041,
+      "step": 11830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994314016554494e-05,
+      "loss": 0.0386,
+      "step": 11840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994309212580647e-05,
+      "loss": 0.0348,
+      "step": 11850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9943044086067996e-05,
+      "loss": 0.0314,
+      "step": 11860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942996046329525e-05,
+      "loss": 0.0375,
+      "step": 11870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942948006591054e-05,
+      "loss": 0.0328,
+      "step": 11880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942899966852584e-05,
+      "loss": 0.0401,
+      "step": 11890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994285192711411e-05,
+      "loss": 0.0387,
+      "step": 11900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994280388737564e-05,
+      "loss": 0.0325,
+      "step": 11910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942755847637168e-05,
+      "loss": 0.034,
+      "step": 11920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942707807898697e-05,
+      "loss": 0.0271,
+      "step": 11930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942659768160223e-05,
+      "loss": 0.0348,
+      "step": 11940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942611728421753e-05,
+      "loss": 0.0367,
+      "step": 11950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942563688683282e-05,
+      "loss": 0.036,
+      "step": 11960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942515648944808e-05,
+      "loss": 0.0322,
+      "step": 11970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942467609206337e-05,
+      "loss": 0.0318,
+      "step": 11980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942419569467866e-05,
+      "loss": 0.0334,
+      "step": 11990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942371529729396e-05,
+      "loss": 0.0426,
+      "step": 12000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994232348999092e-05,
+      "loss": 0.038,
+      "step": 12010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994227545025245e-05,
+      "loss": 0.038,
+      "step": 12020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994222741051398e-05,
+      "loss": 0.0379,
+      "step": 12030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942179370775506e-05,
+      "loss": 0.0396,
+      "step": 12040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942131331037035e-05,
+      "loss": 0.0324,
+      "step": 12050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942083291298564e-05,
+      "loss": 0.0323,
+      "step": 12060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9942035251560094e-05,
+      "loss": 0.0439,
+      "step": 12070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994198721182162e-05,
+      "loss": 0.0373,
+      "step": 12080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.994193917208315e-05,
+      "loss": 0.0356,
+      "step": 12090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9941891132344678e-05,
+      "loss": 0.0304,
+      "step": 12100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9941843092606207e-05,
+      "loss": 0.0384,
+      "step": 12110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9941795052867733e-05,
+      "loss": 0.0407,
+      "step": 12120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9941747013129262e-05,
+      "loss": 0.035,
+      "step": 12130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9941698973390792e-05,
+      "loss": 0.035,
+      "step": 12140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9941650933652318e-05,
+      "loss": 0.031,
+      "step": 12150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9941602893913847e-05,
+      "loss": 0.0382,
+      "step": 12160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9941554854175376e-05,
+      "loss": 0.0312,
+      "step": 12170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9941506814436905e-05,
+      "loss": 0.0407,
+      "step": 12180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994145877469843e-05,
+      "loss": 0.04,
+      "step": 12190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994141073495996e-05,
+      "loss": 0.0396,
+      "step": 12200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994136269522149e-05,
+      "loss": 0.0364,
+      "step": 12210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9941314655483016e-05,
+      "loss": 0.0314,
+      "step": 12220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9941266615744545e-05,
+      "loss": 0.0347,
+      "step": 12230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9941218576006074e-05,
+      "loss": 0.0358,
+      "step": 12240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9941170536267604e-05,
+      "loss": 0.036,
+      "step": 12250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994112249652913e-05,
+      "loss": 0.0383,
+      "step": 12260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994107445679066e-05,
+      "loss": 0.0305,
+      "step": 12270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9941026417052188e-05,
+      "loss": 0.0269,
+      "step": 12280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940978377313714e-05,
+      "loss": 0.0315,
+      "step": 12290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940930337575243e-05,
+      "loss": 0.0429,
+      "step": 12300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940882297836772e-05,
+      "loss": 0.0358,
+      "step": 12310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99408342580983e-05,
+      "loss": 0.0396,
+      "step": 12320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940786218359828e-05,
+      "loss": 0.0327,
+      "step": 12330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940738178621357e-05,
+      "loss": 0.0355,
+      "step": 12340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940690138882886e-05,
+      "loss": 0.0341,
+      "step": 12350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940642099144415e-05,
+      "loss": 0.0331,
+      "step": 12360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994059405940594e-05,
+      "loss": 0.0393,
+      "step": 12370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994054601966747e-05,
+      "loss": 0.0347,
+      "step": 12380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940497979929e-05,
+      "loss": 0.0423,
+      "step": 12390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940449940190526e-05,
+      "loss": 0.0383,
+      "step": 12400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940401900452055e-05,
+      "loss": 0.0343,
+      "step": 12410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940353860713584e-05,
+      "loss": 0.0442,
+      "step": 12420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940305820975113e-05,
+      "loss": 0.0459,
+      "step": 12430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994025778123664e-05,
+      "loss": 0.0339,
+      "step": 12440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994020974149817e-05,
+      "loss": 0.042,
+      "step": 12450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940161701759698e-05,
+      "loss": 0.0409,
+      "step": 12460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940113662021224e-05,
+      "loss": 0.0346,
+      "step": 12470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940065622282753e-05,
+      "loss": 0.0361,
+      "step": 12480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9940017582544282e-05,
+      "loss": 0.0351,
+      "step": 12490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993996954280581e-05,
+      "loss": 0.0334,
+      "step": 12500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939921503067337e-05,
+      "loss": 0.0356,
+      "step": 12510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939873463328867e-05,
+      "loss": 0.0345,
+      "step": 12520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939825423590396e-05,
+      "loss": 0.0335,
+      "step": 12530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939777383851925e-05,
+      "loss": 0.034,
+      "step": 12540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993972934411345e-05,
+      "loss": 0.0353,
+      "step": 12550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993968130437498e-05,
+      "loss": 0.0315,
+      "step": 12560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993963326463651e-05,
+      "loss": 0.0358,
+      "step": 12570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939585224898036e-05,
+      "loss": 0.0343,
+      "step": 12580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939537185159565e-05,
+      "loss": 0.0364,
+      "step": 12590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939489145421094e-05,
+      "loss": 0.039,
+      "step": 12600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939441105682623e-05,
+      "loss": 0.0471,
+      "step": 12610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993939306594415e-05,
+      "loss": 0.0384,
+      "step": 12620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993934502620568e-05,
+      "loss": 0.034,
+      "step": 12630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939296986467208e-05,
+      "loss": 0.0381,
+      "step": 12640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939248946728734e-05,
+      "loss": 0.0415,
+      "step": 12650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939200906990263e-05,
+      "loss": 0.0341,
+      "step": 12660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939152867251792e-05,
+      "loss": 0.0349,
+      "step": 12670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993910482751332e-05,
+      "loss": 0.0404,
+      "step": 12680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939056787774847e-05,
+      "loss": 0.0423,
+      "step": 12690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9939008748036377e-05,
+      "loss": 0.0445,
+      "step": 12700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9938960708297906e-05,
+      "loss": 0.0426,
+      "step": 12710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9938912668559435e-05,
+      "loss": 0.0461,
+      "step": 12720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993886462882096e-05,
+      "loss": 0.0323,
+      "step": 12730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993881658908249e-05,
+      "loss": 0.0333,
+      "step": 12740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993876854934402e-05,
+      "loss": 0.0337,
+      "step": 12750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9938720509605546e-05,
+      "loss": 0.0357,
+      "step": 12760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9938672469867075e-05,
+      "loss": 0.0402,
+      "step": 12770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9938624430128604e-05,
+      "loss": 0.0388,
+      "step": 12780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9938576390390133e-05,
+      "loss": 0.0316,
+      "step": 12790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993852835065166e-05,
+      "loss": 0.0336,
+      "step": 12800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.993848031091319e-05,
+      "loss": 0.0385,
+      "step": 12810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9938432271174718e-05,
+      "loss": 0.0356,
+      "step": 12820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9938384231436244e-05,
+      "loss": 0.0342,
+      "step": 12830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938336191697773e-05,
+      "loss": 0.0376,
+      "step": 12840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938288151959302e-05,
+      "loss": 0.0311,
+      "step": 12850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993824011222083e-05,
+      "loss": 0.0286,
+      "step": 12860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938192072482357e-05,
+      "loss": 0.0336,
+      "step": 12870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938144032743887e-05,
+      "loss": 0.0334,
+      "step": 12880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938095993005416e-05,
+      "loss": 0.0339,
+      "step": 12890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938047953266945e-05,
+      "loss": 0.0341,
+      "step": 12900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993799991352847e-05,
+      "loss": 0.0345,
+      "step": 12910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993795187379e-05,
+      "loss": 0.0367,
+      "step": 12920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993790383405153e-05,
+      "loss": 0.0278,
+      "step": 12930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937855794313055e-05,
+      "loss": 0.0344,
+      "step": 12940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937807754574585e-05,
+      "loss": 0.0307,
+      "step": 12950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937759714836114e-05,
+      "loss": 0.035,
+      "step": 12960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937711675097643e-05,
+      "loss": 0.0333,
+      "step": 12970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993766363535917e-05,
+      "loss": 0.0318,
+      "step": 12980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99376155956207e-05,
+      "loss": 0.0314,
+      "step": 12990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937567555882228e-05,
+      "loss": 0.0343,
+      "step": 13000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937519516143754e-05,
+      "loss": 0.0396,
+      "step": 13010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937471476405283e-05,
+      "loss": 0.0428,
+      "step": 13020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937423436666812e-05,
+      "loss": 0.0353,
+      "step": 13030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993737539692834e-05,
+      "loss": 0.0355,
+      "step": 13040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937327357189867e-05,
+      "loss": 0.0361,
+      "step": 13050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937279317451396e-05,
+      "loss": 0.0323,
+      "step": 13060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937231277712926e-05,
+      "loss": 0.0311,
+      "step": 13070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937183237974455e-05,
+      "loss": 0.0308,
+      "step": 13080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993713519823598e-05,
+      "loss": 0.037,
+      "step": 13090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993708715849751e-05,
+      "loss": 0.0316,
+      "step": 13100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993703911875904e-05,
+      "loss": 0.0357,
+      "step": 13110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936991079020565e-05,
+      "loss": 0.0345,
+      "step": 13120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936943039282095e-05,
+      "loss": 0.0391,
+      "step": 13130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936894999543624e-05,
+      "loss": 0.032,
+      "step": 13140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936846959805153e-05,
+      "loss": 0.0345,
+      "step": 13150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993679892006668e-05,
+      "loss": 0.0286,
+      "step": 13160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936750880328208e-05,
+      "loss": 0.0291,
+      "step": 13170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936702840589738e-05,
+      "loss": 0.0391,
+      "step": 13180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936654800851263e-05,
+      "loss": 0.0357,
+      "step": 13190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936606761112793e-05,
+      "loss": 0.0316,
+      "step": 13200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936558721374322e-05,
+      "loss": 0.0367,
+      "step": 13210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993651068163585e-05,
+      "loss": 0.0315,
+      "step": 13220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936462641897377e-05,
+      "loss": 0.0353,
+      "step": 13230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936414602158906e-05,
+      "loss": 0.0324,
+      "step": 13240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936366562420436e-05,
+      "loss": 0.0418,
+      "step": 13250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936318522681965e-05,
+      "loss": 0.0397,
+      "step": 13260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993627048294349e-05,
+      "loss": 0.0363,
+      "step": 13270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993622244320502e-05,
+      "loss": 0.0353,
+      "step": 13280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993617440346655e-05,
+      "loss": 0.0363,
+      "step": 13290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936126363728075e-05,
+      "loss": 0.0331,
+      "step": 13300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936078323989605e-05,
+      "loss": 0.0271,
+      "step": 13310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936030284251134e-05,
+      "loss": 0.0362,
+      "step": 13320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935982244512663e-05,
+      "loss": 0.0409,
+      "step": 13330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993593420477419e-05,
+      "loss": 0.0343,
+      "step": 13340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935886165035718e-05,
+      "loss": 0.0436,
+      "step": 13350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935838125297247e-05,
+      "loss": 0.0305,
+      "step": 13360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935790085558773e-05,
+      "loss": 0.0389,
+      "step": 13370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935742045820303e-05,
+      "loss": 0.0308,
+      "step": 13380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993569881005568e-05,
+      "loss": 0.0349,
+      "step": 13390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935650770317207e-05,
+      "loss": 0.0342,
+      "step": 13400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935602730578736e-05,
+      "loss": 0.0348,
+      "step": 13410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935554690840265e-05,
+      "loss": 0.0333,
+      "step": 13420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993550665110179e-05,
+      "loss": 0.0356,
+      "step": 13430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993545861136332e-05,
+      "loss": 0.0336,
+      "step": 13440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993541057162485e-05,
+      "loss": 0.0307,
+      "step": 13450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993536253188638e-05,
+      "loss": 0.0388,
+      "step": 13460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935314492147905e-05,
+      "loss": 0.031,
+      "step": 13470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935266452409434e-05,
+      "loss": 0.0343,
+      "step": 13480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935218412670963e-05,
+      "loss": 0.0275,
+      "step": 13490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935170372932492e-05,
+      "loss": 0.0328,
+      "step": 13500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993512233319402e-05,
+      "loss": 0.0356,
+      "step": 13510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935074293455548e-05,
+      "loss": 0.0377,
+      "step": 13520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935026253717077e-05,
+      "loss": 0.0309,
+      "step": 13530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934978213978603e-05,
+      "loss": 0.0271,
+      "step": 13540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934930174240132e-05,
+      "loss": 0.0362,
+      "step": 13550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993488213450166e-05,
+      "loss": 0.0386,
+      "step": 13560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993483409476319e-05,
+      "loss": 0.0301,
+      "step": 13570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934786055024716e-05,
+      "loss": 0.0328,
+      "step": 13580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934738015286246e-05,
+      "loss": 0.0291,
+      "step": 13590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934689975547775e-05,
+      "loss": 0.0338,
+      "step": 13600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99346419358093e-05,
+      "loss": 0.0378,
+      "step": 13610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993459389607083e-05,
+      "loss": 0.0385,
+      "step": 13620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993454585633236e-05,
+      "loss": 0.0425,
+      "step": 13630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993449781659389e-05,
+      "loss": 0.0315,
+      "step": 13640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934449776855415e-05,
+      "loss": 0.0335,
+      "step": 13650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934401737116944e-05,
+      "loss": 0.0359,
+      "step": 13660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934353697378473e-05,
+      "loss": 0.0384,
+      "step": 13670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934305657640002e-05,
+      "loss": 0.0382,
+      "step": 13680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934257617901528e-05,
+      "loss": 0.0254,
+      "step": 13690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934209578163058e-05,
+      "loss": 0.0351,
+      "step": 13700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934161538424587e-05,
+      "loss": 0.0293,
+      "step": 13710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934113498686113e-05,
+      "loss": 0.0325,
+      "step": 13720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934065458947642e-05,
+      "loss": 0.0371,
+      "step": 13730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993401741920917e-05,
+      "loss": 0.0387,
+      "step": 13740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99339693794707e-05,
+      "loss": 0.034,
+      "step": 13750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933921339732226e-05,
+      "loss": 0.0365,
+      "step": 13760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933873299993756e-05,
+      "loss": 0.0343,
+      "step": 13770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933825260255285e-05,
+      "loss": 0.0319,
+      "step": 13780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993377722051681e-05,
+      "loss": 0.0349,
+      "step": 13790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993372918077834e-05,
+      "loss": 0.0388,
+      "step": 13800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993368114103987e-05,
+      "loss": 0.0328,
+      "step": 13810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99336331013014e-05,
+      "loss": 0.0345,
+      "step": 13820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933585061562924e-05,
+      "loss": 0.04,
+      "step": 13830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933537021824454e-05,
+      "loss": 0.0338,
+      "step": 13840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933488982085983e-05,
+      "loss": 0.0374,
+      "step": 13850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993344094234751e-05,
+      "loss": 0.0358,
+      "step": 13860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933392902609038e-05,
+      "loss": 0.0329,
+      "step": 13870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933344862870567e-05,
+      "loss": 0.0282,
+      "step": 13880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933296823132097e-05,
+      "loss": 0.0333,
+      "step": 13890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933248783393623e-05,
+      "loss": 0.0296,
+      "step": 13900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933200743655152e-05,
+      "loss": 0.038,
+      "step": 13910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993315270391668e-05,
+      "loss": 0.0318,
+      "step": 13920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993310466417821e-05,
+      "loss": 0.0346,
+      "step": 13930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933056624439736e-05,
+      "loss": 0.0346,
+      "step": 13940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933008584701266e-05,
+      "loss": 0.028,
+      "step": 13950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932960544962795e-05,
+      "loss": 0.0325,
+      "step": 13960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993291250522432e-05,
+      "loss": 0.0297,
+      "step": 13970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993286446548585e-05,
+      "loss": 0.0361,
+      "step": 13980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993281642574738e-05,
+      "loss": 0.0318,
+      "step": 13990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993276838600891e-05,
+      "loss": 0.0335,
+      "step": 14000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932720346270434e-05,
+      "loss": 0.0355,
+      "step": 14010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932672306531964e-05,
+      "loss": 0.0337,
+      "step": 14020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932624266793493e-05,
+      "loss": 0.0278,
+      "step": 14030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993257622705502e-05,
+      "loss": 0.0386,
+      "step": 14040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932528187316548e-05,
+      "loss": 0.0393,
+      "step": 14050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932480147578077e-05,
+      "loss": 0.0321,
+      "step": 14060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932432107839607e-05,
+      "loss": 0.0269,
+      "step": 14070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932384068101133e-05,
+      "loss": 0.0273,
+      "step": 14080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932336028362662e-05,
+      "loss": 0.0335,
+      "step": 14090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993228798862419e-05,
+      "loss": 0.0387,
+      "step": 14100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993223994888572e-05,
+      "loss": 0.0391,
+      "step": 14110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993219190914725e-05,
+      "loss": 0.0342,
+      "step": 14120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993214386940878e-05,
+      "loss": 0.0326,
+      "step": 14130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932095829670305e-05,
+      "loss": 0.0237,
+      "step": 14140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932047789931834e-05,
+      "loss": 0.0319,
+      "step": 14150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931999750193363e-05,
+      "loss": 0.0314,
+      "step": 14160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931951710454893e-05,
+      "loss": 0.0299,
+      "step": 14170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931908474690264e-05,
+      "loss": 0.0358,
+      "step": 14180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931860434951793e-05,
+      "loss": 0.0324,
+      "step": 14190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931812395213322e-05,
+      "loss": 0.0269,
+      "step": 14200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931764355474848e-05,
+      "loss": 0.0304,
+      "step": 14210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931716315736378e-05,
+      "loss": 0.0336,
+      "step": 14220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9931668275997907e-05,
+      "loss": 0.027,
+      "step": 14230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9931620236259436e-05,
+      "loss": 0.0328,
+      "step": 14240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9931572196520962e-05,
+      "loss": 0.0388,
+      "step": 14250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993152415678249e-05,
+      "loss": 0.0282,
+      "step": 14260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993147611704402e-05,
+      "loss": 0.0332,
+      "step": 14270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9931428077305546e-05,
+      "loss": 0.0266,
+      "step": 14280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9931380037567076e-05,
+      "loss": 0.0287,
+      "step": 14290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9931331997828605e-05,
+      "loss": 0.0313,
+      "step": 14300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9931283958090134e-05,
+      "loss": 0.0357,
+      "step": 14310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993123591835166e-05,
+      "loss": 0.0299,
+      "step": 14320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993118787861319e-05,
+      "loss": 0.0308,
+      "step": 14330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993113983887472e-05,
+      "loss": 0.0344,
+      "step": 14340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9931091799136248e-05,
+      "loss": 0.0298,
+      "step": 14350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9931043759397774e-05,
+      "loss": 0.039,
+      "step": 14360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930995719659303e-05,
+      "loss": 0.0315,
+      "step": 14370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930947679920832e-05,
+      "loss": 0.0345,
+      "step": 14380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930899640182358e-05,
+      "loss": 0.0411,
+      "step": 14390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930851600443887e-05,
+      "loss": 0.0367,
+      "step": 14400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930803560705417e-05,
+      "loss": 0.0352,
+      "step": 14410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930755520966946e-05,
+      "loss": 0.0301,
+      "step": 14420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930707481228472e-05,
+      "loss": 0.0354,
+      "step": 14430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993065944149e-05,
+      "loss": 0.0337,
+      "step": 14440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993061140175153e-05,
+      "loss": 0.0335,
+      "step": 14450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930563362013056e-05,
+      "loss": 0.0268,
+      "step": 14460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930515322274586e-05,
+      "loss": 0.0318,
+      "step": 14470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930467282536115e-05,
+      "loss": 0.0334,
+      "step": 14480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930419242797644e-05,
+      "loss": 0.028,
+      "step": 14490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993037120305917e-05,
+      "loss": 0.036,
+      "step": 14500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99303231633207e-05,
+      "loss": 0.0408,
+      "step": 14510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.993027512358223e-05,
+      "loss": 0.033,
+      "step": 14520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930227083843758e-05,
+      "loss": 0.0311,
+      "step": 14530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930179044105284e-05,
+      "loss": 0.0376,
+      "step": 14540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930131004366813e-05,
+      "loss": 0.0324,
+      "step": 14550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930082964628342e-05,
+      "loss": 0.0409,
+      "step": 14560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9930034924889868e-05,
+      "loss": 0.031,
+      "step": 14570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929986885151397e-05,
+      "loss": 0.0375,
+      "step": 14580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929938845412927e-05,
+      "loss": 0.0304,
+      "step": 14590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929890805674456e-05,
+      "loss": 0.0364,
+      "step": 14600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929842765935982e-05,
+      "loss": 0.0361,
+      "step": 14610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929794726197514e-05,
+      "loss": 0.0318,
+      "step": 14620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992974668645904e-05,
+      "loss": 0.0339,
+      "step": 14630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992969864672057e-05,
+      "loss": 0.0322,
+      "step": 14640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99296506069821e-05,
+      "loss": 0.0349,
+      "step": 14650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929602567243628e-05,
+      "loss": 0.0296,
+      "step": 14660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929554527505154e-05,
+      "loss": 0.0341,
+      "step": 14670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929506487766683e-05,
+      "loss": 0.0317,
+      "step": 14680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929458448028213e-05,
+      "loss": 0.0318,
+      "step": 14690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992941040828974e-05,
+      "loss": 0.0381,
+      "step": 14700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929362368551268e-05,
+      "loss": 0.0317,
+      "step": 14710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929314328812797e-05,
+      "loss": 0.0236,
+      "step": 14720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929266289074326e-05,
+      "loss": 0.0282,
+      "step": 14730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929218249335852e-05,
+      "loss": 0.026,
+      "step": 14740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992917020959738e-05,
+      "loss": 0.0308,
+      "step": 14750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992912216985891e-05,
+      "loss": 0.0286,
+      "step": 14760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992907413012044e-05,
+      "loss": 0.0329,
+      "step": 14770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9929026090381966e-05,
+      "loss": 0.0291,
+      "step": 14780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9928978050643495e-05,
+      "loss": 0.0342,
+      "step": 14790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9928930010905024e-05,
+      "loss": 0.0334,
+      "step": 14800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992888197116655e-05,
+      "loss": 0.0287,
+      "step": 14810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992883393142808e-05,
+      "loss": 0.0287,
+      "step": 14820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992878589168961e-05,
+      "loss": 0.032,
+      "step": 14830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9928737851951138e-05,
+      "loss": 0.0296,
+      "step": 14840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9928689812212664e-05,
+      "loss": 0.0333,
+      "step": 14850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9928641772474193e-05,
+      "loss": 0.0356,
+      "step": 14860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9928593732735722e-05,
+      "loss": 0.0375,
+      "step": 14870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992854569299725e-05,
+      "loss": 0.0398,
+      "step": 14880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9928497653258778e-05,
+      "loss": 0.0251,
+      "step": 14890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9928449613520307e-05,
+      "loss": 0.025,
+      "step": 14900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9928401573781836e-05,
+      "loss": 0.0369,
+      "step": 14910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9928353534043362e-05,
+      "loss": 0.0268,
+      "step": 14920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992830549430489e-05,
+      "loss": 0.0306,
+      "step": 14930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992825745456642e-05,
+      "loss": 0.0358,
+      "step": 14940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9928209414827946e-05,
+      "loss": 0.0396,
+      "step": 14950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9928161375089476e-05,
+      "loss": 0.0326,
+      "step": 14960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9928113335351005e-05,
+      "loss": 0.0307,
+      "step": 14970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9928065295612534e-05,
+      "loss": 0.0299,
+      "step": 14980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992801725587406e-05,
+      "loss": 0.0364,
+      "step": 14990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992796921613559e-05,
+      "loss": 0.0341,
+      "step": 15000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992792117639712e-05,
+      "loss": 0.0248,
+      "step": 15010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927873136658648e-05,
+      "loss": 0.032,
+      "step": 15020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927825096920174e-05,
+      "loss": 0.0328,
+      "step": 15030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927777057181703e-05,
+      "loss": 0.033,
+      "step": 15040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927729017443232e-05,
+      "loss": 0.0314,
+      "step": 15050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927680977704758e-05,
+      "loss": 0.0293,
+      "step": 15060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927632937966288e-05,
+      "loss": 0.0284,
+      "step": 15070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927584898227817e-05,
+      "loss": 0.0245,
+      "step": 15080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927536858489346e-05,
+      "loss": 0.0317,
+      "step": 15090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927488818750872e-05,
+      "loss": 0.0331,
+      "step": 15100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99274407790124e-05,
+      "loss": 0.0312,
+      "step": 15110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992739273927393e-05,
+      "loss": 0.0345,
+      "step": 15120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927344699535456e-05,
+      "loss": 0.0285,
+      "step": 15130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927296659796986e-05,
+      "loss": 0.0355,
+      "step": 15140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927248620058515e-05,
+      "loss": 0.0341,
+      "step": 15150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927200580320044e-05,
+      "loss": 0.0315,
+      "step": 15160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992715254058157e-05,
+      "loss": 0.0318,
+      "step": 15170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99271045008431e-05,
+      "loss": 0.0347,
+      "step": 15180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992705646110463e-05,
+      "loss": 0.0329,
+      "step": 15190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9927008421366158e-05,
+      "loss": 0.0322,
+      "step": 15200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926960381627684e-05,
+      "loss": 0.0315,
+      "step": 15210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926912341889213e-05,
+      "loss": 0.0392,
+      "step": 15220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926864302150742e-05,
+      "loss": 0.0314,
+      "step": 15230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926816262412268e-05,
+      "loss": 0.0336,
+      "step": 15240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926768222673797e-05,
+      "loss": 0.029,
+      "step": 15250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926720182935327e-05,
+      "loss": 0.0327,
+      "step": 15260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926672143196856e-05,
+      "loss": 0.0305,
+      "step": 15270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926624103458382e-05,
+      "loss": 0.0348,
+      "step": 15280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992657606371991e-05,
+      "loss": 0.0361,
+      "step": 15290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992652802398144e-05,
+      "loss": 0.0327,
+      "step": 15300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926479984242966e-05,
+      "loss": 0.0311,
+      "step": 15310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926431944504496e-05,
+      "loss": 0.0319,
+      "step": 15320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926383904766025e-05,
+      "loss": 0.0298,
+      "step": 15330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926335865027554e-05,
+      "loss": 0.0321,
+      "step": 15340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992628782528908e-05,
+      "loss": 0.0326,
+      "step": 15350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992623978555061e-05,
+      "loss": 0.0311,
+      "step": 15360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992619174581214e-05,
+      "loss": 0.0241,
+      "step": 15370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926143706073668e-05,
+      "loss": 0.0334,
+      "step": 15380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926095666335194e-05,
+      "loss": 0.0313,
+      "step": 15390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9926047626596723e-05,
+      "loss": 0.0334,
+      "step": 15400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925999586858252e-05,
+      "loss": 0.0324,
+      "step": 15410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925951547119778e-05,
+      "loss": 0.0286,
+      "step": 15420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925903507381307e-05,
+      "loss": 0.0253,
+      "step": 15430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925855467642837e-05,
+      "loss": 0.0363,
+      "step": 15440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925807427904366e-05,
+      "loss": 0.0316,
+      "step": 15450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925759388165892e-05,
+      "loss": 0.036,
+      "step": 15460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992571134842742e-05,
+      "loss": 0.0314,
+      "step": 15470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992566330868895e-05,
+      "loss": 0.0343,
+      "step": 15480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925615268950476e-05,
+      "loss": 0.028,
+      "step": 15490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925567229212005e-05,
+      "loss": 0.0242,
+      "step": 15500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925519189473535e-05,
+      "loss": 0.0269,
+      "step": 15510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925471149735064e-05,
+      "loss": 0.0306,
+      "step": 15520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992542310999659e-05,
+      "loss": 0.0412,
+      "step": 15530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992537507025812e-05,
+      "loss": 0.0319,
+      "step": 15540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.992532703051965e-05,
+      "loss": 0.0333,
+      "step": 15550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925278990781178e-05,
+      "loss": 0.0256,
+      "step": 15560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925230951042704e-05,
+      "loss": 0.024,
+      "step": 15570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925182911304233e-05,
+      "loss": 0.0313,
+      "step": 15580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925134871565762e-05,
+      "loss": 0.0319,
+      "step": 15590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925086831827288e-05,
+      "loss": 0.0278,
+      "step": 15600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9925038792088817e-05,
+      "loss": 0.0322,
+      "step": 15610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924990752350347e-05,
+      "loss": 0.0261,
+      "step": 15620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924942712611876e-05,
+      "loss": 0.0298,
+      "step": 15630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99248946728734e-05,
+      "loss": 0.0351,
+      "step": 15640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992484663313493e-05,
+      "loss": 0.0317,
+      "step": 15650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992479859339646e-05,
+      "loss": 0.0334,
+      "step": 15660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924750553657986e-05,
+      "loss": 0.0345,
+      "step": 15670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924702513919515e-05,
+      "loss": 0.0302,
+      "step": 15680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924654474181045e-05,
+      "loss": 0.0379,
+      "step": 15690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924606434442574e-05,
+      "loss": 0.0268,
+      "step": 15700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99245583947041e-05,
+      "loss": 0.0314,
+      "step": 15710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992451035496563e-05,
+      "loss": 0.0271,
+      "step": 15720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992446231522716e-05,
+      "loss": 0.0325,
+      "step": 15730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924414275488688e-05,
+      "loss": 0.0387,
+      "step": 15740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924366235750213e-05,
+      "loss": 0.0308,
+      "step": 15750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924318196011743e-05,
+      "loss": 0.0337,
+      "step": 15760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924270156273272e-05,
+      "loss": 0.0289,
+      "step": 15770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924222116534798e-05,
+      "loss": 0.032,
+      "step": 15780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924174076796327e-05,
+      "loss": 0.0344,
+      "step": 15790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924126037057856e-05,
+      "loss": 0.0264,
+      "step": 15800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924077997319386e-05,
+      "loss": 0.038,
+      "step": 15810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992402995758091e-05,
+      "loss": 0.0322,
+      "step": 15820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992398191784244e-05,
+      "loss": 0.0225,
+      "step": 15830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992393387810397e-05,
+      "loss": 0.0276,
+      "step": 15840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923885838365496e-05,
+      "loss": 0.0261,
+      "step": 15850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923837798627025e-05,
+      "loss": 0.0333,
+      "step": 15860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923789758888555e-05,
+      "loss": 0.031,
+      "step": 15870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923741719150084e-05,
+      "loss": 0.0308,
+      "step": 15880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992369367941161e-05,
+      "loss": 0.0286,
+      "step": 15890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992364563967314e-05,
+      "loss": 0.0304,
+      "step": 15900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923597599934668e-05,
+      "loss": 0.0294,
+      "step": 15910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923549560196198e-05,
+      "loss": 0.0283,
+      "step": 15920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923501520457723e-05,
+      "loss": 0.0248,
+      "step": 15930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923453480719253e-05,
+      "loss": 0.0395,
+      "step": 15940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923405440980782e-05,
+      "loss": 0.0324,
+      "step": 15950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923357401242308e-05,
+      "loss": 0.0374,
+      "step": 15960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923309361503837e-05,
+      "loss": 0.0311,
+      "step": 15970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923261321765366e-05,
+      "loss": 0.037,
+      "step": 15980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923213282026896e-05,
+      "loss": 0.0244,
+      "step": 15990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992316524228842e-05,
+      "loss": 0.0337,
+      "step": 16000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992311720254995e-05,
+      "loss": 0.0309,
+      "step": 16010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992306916281148e-05,
+      "loss": 0.0293,
+      "step": 16020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923021123073006e-05,
+      "loss": 0.0271,
+      "step": 16030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922973083334535e-05,
+      "loss": 0.0303,
+      "step": 16040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922925043596064e-05,
+      "loss": 0.0325,
+      "step": 16050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922877003857594e-05,
+      "loss": 0.0309,
+      "step": 16060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992282896411912e-05,
+      "loss": 0.0302,
+      "step": 16070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992278092438065e-05,
+      "loss": 0.0312,
+      "step": 16080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922732884642178e-05,
+      "loss": 0.0295,
+      "step": 16090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922684844903707e-05,
+      "loss": 0.0274,
+      "step": 16100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922636805165233e-05,
+      "loss": 0.0263,
+      "step": 16110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922588765426763e-05,
+      "loss": 0.0237,
+      "step": 16120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922540725688292e-05,
+      "loss": 0.0357,
+      "step": 16130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922492685949818e-05,
+      "loss": 0.0332,
+      "step": 16140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922444646211347e-05,
+      "loss": 0.0302,
+      "step": 16150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922396606472876e-05,
+      "loss": 0.0376,
+      "step": 16160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922348566734406e-05,
+      "loss": 0.0334,
+      "step": 16170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992230052699593e-05,
+      "loss": 0.0327,
+      "step": 16180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992225248725746e-05,
+      "loss": 0.0351,
+      "step": 16190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992220444751899e-05,
+      "loss": 0.0282,
+      "step": 16200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922156407780516e-05,
+      "loss": 0.0357,
+      "step": 16210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922108368042045e-05,
+      "loss": 0.0353,
+      "step": 16220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922060328303574e-05,
+      "loss": 0.025,
+      "step": 16230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922012288565104e-05,
+      "loss": 0.0307,
+      "step": 16240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992196424882663e-05,
+      "loss": 0.0348,
+      "step": 16250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992191620908816e-05,
+      "loss": 0.0324,
+      "step": 16260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9921868169349688e-05,
+      "loss": 0.031,
+      "step": 16270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9921820129611217e-05,
+      "loss": 0.0279,
+      "step": 16280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9921772089872743e-05,
+      "loss": 0.0316,
+      "step": 16290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9921724050134272e-05,
+      "loss": 0.0273,
+      "step": 16300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9921676010395802e-05,
+      "loss": 0.0279,
+      "step": 16310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9921627970657328e-05,
+      "loss": 0.0292,
+      "step": 16320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9921579930918857e-05,
+      "loss": 0.033,
+      "step": 16330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9921531891180386e-05,
+      "loss": 0.0326,
+      "step": 16340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9921483851441915e-05,
+      "loss": 0.0294,
+      "step": 16350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.992143581170344e-05,
+      "loss": 0.0406,
+      "step": 16360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.992138777196497e-05,
+      "loss": 0.0352,
+      "step": 16370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99213397322265e-05,
+      "loss": 0.0297,
+      "step": 16380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9921291692488026e-05,
+      "loss": 0.0262,
+      "step": 16390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9921243652749555e-05,
+      "loss": 0.0256,
+      "step": 16400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9921195613011084e-05,
+      "loss": 0.0305,
+      "step": 16410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9921147573272614e-05,
+      "loss": 0.0333,
+      "step": 16420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.992109953353414e-05,
+      "loss": 0.0362,
+      "step": 16430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.992105149379567e-05,
+      "loss": 0.0266,
+      "step": 16440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9921003454057198e-05,
+      "loss": 0.0262,
+      "step": 16450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920955414318727e-05,
+      "loss": 0.0285,
+      "step": 16460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920907374580253e-05,
+      "loss": 0.0387,
+      "step": 16470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920859334841782e-05,
+      "loss": 0.026,
+      "step": 16480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.992081129510331e-05,
+      "loss": 0.0274,
+      "step": 16490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920763255364838e-05,
+      "loss": 0.0341,
+      "step": 16500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920715215626367e-05,
+      "loss": 0.0311,
+      "step": 16510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920667175887896e-05,
+      "loss": 0.0203,
+      "step": 16520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920619136149425e-05,
+      "loss": 0.0282,
+      "step": 16530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.992057109641095e-05,
+      "loss": 0.0387,
+      "step": 16540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.992052305667248e-05,
+      "loss": 0.0341,
+      "step": 16550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.992047501693401e-05,
+      "loss": 0.0393,
+      "step": 16560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920426977195536e-05,
+      "loss": 0.0312,
+      "step": 16570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920378937457065e-05,
+      "loss": 0.0326,
+      "step": 16580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920330897718594e-05,
+      "loss": 0.034,
+      "step": 16590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920282857980123e-05,
+      "loss": 0.0301,
+      "step": 16600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.992023481824165e-05,
+      "loss": 0.0267,
+      "step": 16610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.992018677850318e-05,
+      "loss": 0.0352,
+      "step": 16620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920138738764708e-05,
+      "loss": 0.0292,
+      "step": 16630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920090699026237e-05,
+      "loss": 0.0323,
+      "step": 16640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920042659287763e-05,
+      "loss": 0.0302,
+      "step": 16650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919994619549292e-05,
+      "loss": 0.0283,
+      "step": 16660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991994657981082e-05,
+      "loss": 0.0326,
+      "step": 16670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919898540072347e-05,
+      "loss": 0.029,
+      "step": 16680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919850500333877e-05,
+      "loss": 0.033,
+      "step": 16690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919802460595406e-05,
+      "loss": 0.0325,
+      "step": 16700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919754420856935e-05,
+      "loss": 0.0265,
+      "step": 16710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991970638111846e-05,
+      "loss": 0.03,
+      "step": 16720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991965834137999e-05,
+      "loss": 0.0306,
+      "step": 16730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991961030164152e-05,
+      "loss": 0.0319,
+      "step": 16740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919562261903046e-05,
+      "loss": 0.0272,
+      "step": 16750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919514222164575e-05,
+      "loss": 0.0341,
+      "step": 16760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919466182426104e-05,
+      "loss": 0.0299,
+      "step": 16770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919418142687633e-05,
+      "loss": 0.0348,
+      "step": 16780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991937010294916e-05,
+      "loss": 0.0254,
+      "step": 16790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991932206321069e-05,
+      "loss": 0.0266,
+      "step": 16800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919274023472218e-05,
+      "loss": 0.0325,
+      "step": 16810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919225983733744e-05,
+      "loss": 0.0252,
+      "step": 16820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919177943995273e-05,
+      "loss": 0.0297,
+      "step": 16830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919129904256802e-05,
+      "loss": 0.0291,
+      "step": 16840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991908186451833e-05,
+      "loss": 0.0252,
+      "step": 16850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919033824779857e-05,
+      "loss": 0.0322,
+      "step": 16860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9918985785041387e-05,
+      "loss": 0.0353,
+      "step": 16870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9918937745302916e-05,
+      "loss": 0.0327,
+      "step": 16880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9918889705564445e-05,
+      "loss": 0.0227,
+      "step": 16890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991884166582597e-05,
+      "loss": 0.0257,
+      "step": 16900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99187936260875e-05,
+      "loss": 0.0331,
+      "step": 16910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991874558634903e-05,
+      "loss": 0.032,
+      "step": 16920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9918697546610556e-05,
+      "loss": 0.031,
+      "step": 16930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9918649506872085e-05,
+      "loss": 0.0328,
+      "step": 16940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9918601467133614e-05,
+      "loss": 0.0287,
+      "step": 16950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9918553427395143e-05,
+      "loss": 0.0299,
+      "step": 16960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991850538765667e-05,
+      "loss": 0.0279,
+      "step": 16970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99184573479182e-05,
+      "loss": 0.0315,
+      "step": 16980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9918409308179728e-05,
+      "loss": 0.0315,
+      "step": 16990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9918361268441254e-05,
+      "loss": 0.0242,
+      "step": 17000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9918313228702783e-05,
+      "loss": 0.027,
+      "step": 17010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9918265188964312e-05,
+      "loss": 0.0273,
+      "step": 17020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991821714922584e-05,
+      "loss": 0.0286,
+      "step": 17030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9918169109487367e-05,
+      "loss": 0.0307,
+      "step": 17040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9918121069748897e-05,
+      "loss": 0.0316,
+      "step": 17050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9918073030010426e-05,
+      "loss": 0.0245,
+      "step": 17060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9918024990271955e-05,
+      "loss": 0.0326,
+      "step": 17070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991797695053348e-05,
+      "loss": 0.0299,
+      "step": 17080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991792891079501e-05,
+      "loss": 0.0251,
+      "step": 17090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991788087105654e-05,
+      "loss": 0.0278,
+      "step": 17100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9917832831318065e-05,
+      "loss": 0.0208,
+      "step": 17110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9917784791579595e-05,
+      "loss": 0.029,
+      "step": 17120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9917736751841124e-05,
+      "loss": 0.0321,
+      "step": 17130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9917688712102653e-05,
+      "loss": 0.0284,
+      "step": 17140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991764067236418e-05,
+      "loss": 0.0258,
+      "step": 17150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991759263262571e-05,
+      "loss": 0.028,
+      "step": 17160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9917544592887238e-05,
+      "loss": 0.0287,
+      "step": 17170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9917496553148764e-05,
+      "loss": 0.0276,
+      "step": 17180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9917448513410296e-05,
+      "loss": 0.0332,
+      "step": 17190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9917400473671825e-05,
+      "loss": 0.0313,
+      "step": 17200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991735243393335e-05,
+      "loss": 0.0309,
+      "step": 17210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991730439419488e-05,
+      "loss": 0.0254,
+      "step": 17220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991725635445641e-05,
+      "loss": 0.0276,
+      "step": 17230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9917208314717936e-05,
+      "loss": 0.0315,
+      "step": 17240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9917160274979465e-05,
+      "loss": 0.0229,
+      "step": 17250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9917112235240994e-05,
+      "loss": 0.0288,
+      "step": 17260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9917064195502524e-05,
+      "loss": 0.0291,
+      "step": 17270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991701615576405e-05,
+      "loss": 0.0296,
+      "step": 17280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991696811602558e-05,
+      "loss": 0.0287,
+      "step": 17290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916920076287108e-05,
+      "loss": 0.0283,
+      "step": 17300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916872036548637e-05,
+      "loss": 0.0326,
+      "step": 17310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916823996810163e-05,
+      "loss": 0.0283,
+      "step": 17320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916775957071692e-05,
+      "loss": 0.0373,
+      "step": 17330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991672791733322e-05,
+      "loss": 0.0256,
+      "step": 17340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916679877594748e-05,
+      "loss": 0.0312,
+      "step": 17350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916631837856277e-05,
+      "loss": 0.0338,
+      "step": 17360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916583798117806e-05,
+      "loss": 0.0265,
+      "step": 17370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916535758379335e-05,
+      "loss": 0.0259,
+      "step": 17380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991648771864086e-05,
+      "loss": 0.0282,
+      "step": 17390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991643967890239e-05,
+      "loss": 0.0276,
+      "step": 17400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991639163916392e-05,
+      "loss": 0.0279,
+      "step": 17410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916343599425446e-05,
+      "loss": 0.0333,
+      "step": 17420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916295559686975e-05,
+      "loss": 0.0386,
+      "step": 17430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916247519948504e-05,
+      "loss": 0.0276,
+      "step": 17440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916199480210033e-05,
+      "loss": 0.0279,
+      "step": 17450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991615144047156e-05,
+      "loss": 0.0269,
+      "step": 17460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991610340073309e-05,
+      "loss": 0.0331,
+      "step": 17470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916055360994618e-05,
+      "loss": 0.0291,
+      "step": 17480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9916007321256147e-05,
+      "loss": 0.0339,
+      "step": 17490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915959281517673e-05,
+      "loss": 0.0358,
+      "step": 17500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915911241779202e-05,
+      "loss": 0.0287,
+      "step": 17510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991586320204073e-05,
+      "loss": 0.0296,
+      "step": 17520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915815162302257e-05,
+      "loss": 0.0299,
+      "step": 17530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915767122563787e-05,
+      "loss": 0.0287,
+      "step": 17540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915719082825316e-05,
+      "loss": 0.0357,
+      "step": 17550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915671043086845e-05,
+      "loss": 0.0307,
+      "step": 17560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991562300334837e-05,
+      "loss": 0.0239,
+      "step": 17570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99155749636099e-05,
+      "loss": 0.0304,
+      "step": 17580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991552692387143e-05,
+      "loss": 0.0313,
+      "step": 17590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915478884132956e-05,
+      "loss": 0.0216,
+      "step": 17600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915430844394485e-05,
+      "loss": 0.032,
+      "step": 17610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915382804656014e-05,
+      "loss": 0.0393,
+      "step": 17620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915334764917543e-05,
+      "loss": 0.03,
+      "step": 17630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.991528672517907e-05,
+      "loss": 0.0188,
+      "step": 17640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99152386854406e-05,
+      "loss": 0.0334,
+      "step": 17650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915190645702128e-05,
+      "loss": 0.031,
+      "step": 17660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915142605963657e-05,
+      "loss": 0.0269,
+      "step": 17670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915094566225183e-05,
+      "loss": 0.0273,
+      "step": 17680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9915046526486712e-05,
+      "loss": 0.0305,
+      "step": 17690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991499848674824e-05,
+      "loss": 0.0329,
+      "step": 17700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914950447009767e-05,
+      "loss": 0.0255,
+      "step": 17710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914902407271297e-05,
+      "loss": 0.0329,
+      "step": 17720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914854367532826e-05,
+      "loss": 0.0273,
+      "step": 17730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914806327794355e-05,
+      "loss": 0.0326,
+      "step": 17740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991475828805588e-05,
+      "loss": 0.0419,
+      "step": 17750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991471024831741e-05,
+      "loss": 0.0326,
+      "step": 17760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991466220857894e-05,
+      "loss": 0.0281,
+      "step": 17770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914614168840466e-05,
+      "loss": 0.0288,
+      "step": 17780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914566129101995e-05,
+      "loss": 0.0299,
+      "step": 17790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914518089363524e-05,
+      "loss": 0.0236,
+      "step": 17800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914470049625053e-05,
+      "loss": 0.0286,
+      "step": 17810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991442200988658e-05,
+      "loss": 0.0296,
+      "step": 17820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991437397014811e-05,
+      "loss": 0.0238,
+      "step": 17830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914325930409638e-05,
+      "loss": 0.0328,
+      "step": 17840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914277890671167e-05,
+      "loss": 0.0303,
+      "step": 17850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914229850932693e-05,
+      "loss": 0.0299,
+      "step": 17860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914181811194222e-05,
+      "loss": 0.0372,
+      "step": 17870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991413377145575e-05,
+      "loss": 0.0352,
+      "step": 17880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914085731717277e-05,
+      "loss": 0.0257,
+      "step": 17890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9914037691978807e-05,
+      "loss": 0.0322,
+      "step": 17900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913989652240336e-05,
+      "loss": 0.0375,
+      "step": 17910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913941612501865e-05,
+      "loss": 0.0231,
+      "step": 17920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991389357276339e-05,
+      "loss": 0.0307,
+      "step": 17930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991384553302492e-05,
+      "loss": 0.0319,
+      "step": 17940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991379749328645e-05,
+      "loss": 0.0271,
+      "step": 17950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913749453547975e-05,
+      "loss": 0.0266,
+      "step": 17960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913701413809505e-05,
+      "loss": 0.0274,
+      "step": 17970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913653374071034e-05,
+      "loss": 0.0295,
+      "step": 17980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913605334332563e-05,
+      "loss": 0.0256,
+      "step": 17990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991355729459409e-05,
+      "loss": 0.0242,
+      "step": 18000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991350925485562e-05,
+      "loss": 0.0246,
+      "step": 18010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913461215117148e-05,
+      "loss": 0.0294,
+      "step": 18020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913413175378677e-05,
+      "loss": 0.0244,
+      "step": 18030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913365135640203e-05,
+      "loss": 0.0291,
+      "step": 18040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913317095901732e-05,
+      "loss": 0.0259,
+      "step": 18050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991326905616326e-05,
+      "loss": 0.0321,
+      "step": 18060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913221016424787e-05,
+      "loss": 0.0305,
+      "step": 18070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913172976686316e-05,
+      "loss": 0.0283,
+      "step": 18080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913124936947846e-05,
+      "loss": 0.0247,
+      "step": 18090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9913076897209375e-05,
+      "loss": 0.0285,
+      "step": 18100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99130288574709e-05,
+      "loss": 0.0343,
+      "step": 18110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991298081773243e-05,
+      "loss": 0.0237,
+      "step": 18120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991293277799396e-05,
+      "loss": 0.0283,
+      "step": 18130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912884738255485e-05,
+      "loss": 0.0262,
+      "step": 18140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912836698517015e-05,
+      "loss": 0.032,
+      "step": 18150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912788658778544e-05,
+      "loss": 0.0257,
+      "step": 18160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912740619040073e-05,
+      "loss": 0.0281,
+      "step": 18170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99126925793016e-05,
+      "loss": 0.0312,
+      "step": 18180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912644539563128e-05,
+      "loss": 0.0322,
+      "step": 18190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912596499824658e-05,
+      "loss": 0.028,
+      "step": 18200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912548460086183e-05,
+      "loss": 0.0238,
+      "step": 18210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912500420347713e-05,
+      "loss": 0.0253,
+      "step": 18220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912452380609242e-05,
+      "loss": 0.0272,
+      "step": 18230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991240434087077e-05,
+      "loss": 0.0266,
+      "step": 18240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912356301132297e-05,
+      "loss": 0.0322,
+      "step": 18250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912308261393826e-05,
+      "loss": 0.0274,
+      "step": 18260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912260221655356e-05,
+      "loss": 0.0292,
+      "step": 18270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912212181916885e-05,
+      "loss": 0.029,
+      "step": 18280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991216414217841e-05,
+      "loss": 0.0327,
+      "step": 18290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991211610243994e-05,
+      "loss": 0.0256,
+      "step": 18300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991206806270147e-05,
+      "loss": 0.0283,
+      "step": 18310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9912020022962995e-05,
+      "loss": 0.0333,
+      "step": 18320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9911971983224525e-05,
+      "loss": 0.0269,
+      "step": 18330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9911923943486054e-05,
+      "loss": 0.0298,
+      "step": 18340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9911875903747583e-05,
+      "loss": 0.0335,
+      "step": 18350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.991182786400911e-05,
+      "loss": 0.0243,
+      "step": 18360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9911779824270638e-05,
+      "loss": 0.0258,
+      "step": 18370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9911731784532167e-05,
+      "loss": 0.0313,
+      "step": 18380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9911683744793693e-05,
+      "loss": 0.0267,
+      "step": 18390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9911635705055223e-05,
+      "loss": 0.0356,
+      "step": 18400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9911587665316752e-05,
+      "loss": 0.0307,
+      "step": 18410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991153962557828e-05,
+      "loss": 0.026,
+      "step": 18420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9911491585839807e-05,
+      "loss": 0.0298,
+      "step": 18430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9911443546101336e-05,
+      "loss": 0.0236,
+      "step": 18440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9911395506362866e-05,
+      "loss": 0.0328,
+      "step": 18450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9911347466624395e-05,
+      "loss": 0.0295,
+      "step": 18460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991129942688592e-05,
+      "loss": 0.0346,
+      "step": 18470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991125138714745e-05,
+      "loss": 0.0377,
+      "step": 18480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991120334740898e-05,
+      "loss": 0.0322,
+      "step": 18490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9911155307670505e-05,
+      "loss": 0.0299,
+      "step": 18500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9911107267932034e-05,
+      "loss": 0.0333,
+      "step": 18510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9911059228193564e-05,
+      "loss": 0.0292,
+      "step": 18520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9911011188455093e-05,
+      "loss": 0.0224,
+      "step": 18530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991096314871662e-05,
+      "loss": 0.0281,
+      "step": 18540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910915108978148e-05,
+      "loss": 0.0337,
+      "step": 18550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910867069239677e-05,
+      "loss": 0.0273,
+      "step": 18560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910819029501203e-05,
+      "loss": 0.0322,
+      "step": 18570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910770989762733e-05,
+      "loss": 0.0332,
+      "step": 18580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910722950024262e-05,
+      "loss": 0.0262,
+      "step": 18590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991067491028579e-05,
+      "loss": 0.0305,
+      "step": 18600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910626870547317e-05,
+      "loss": 0.0303,
+      "step": 18610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910578830808846e-05,
+      "loss": 0.0322,
+      "step": 18620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910530791070376e-05,
+      "loss": 0.0261,
+      "step": 18630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910482751331905e-05,
+      "loss": 0.0266,
+      "step": 18640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991043471159343e-05,
+      "loss": 0.0379,
+      "step": 18650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991038667185496e-05,
+      "loss": 0.0318,
+      "step": 18660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991033863211649e-05,
+      "loss": 0.0282,
+      "step": 18670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910290592378015e-05,
+      "loss": 0.0354,
+      "step": 18680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910242552639544e-05,
+      "loss": 0.0373,
+      "step": 18690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910194512901074e-05,
+      "loss": 0.0338,
+      "step": 18700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910146473162603e-05,
+      "loss": 0.0359,
+      "step": 18710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991009843342413e-05,
+      "loss": 0.0255,
+      "step": 18720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910050393685658e-05,
+      "loss": 0.0282,
+      "step": 18730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910002353947187e-05,
+      "loss": 0.0292,
+      "step": 18740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909954314208713e-05,
+      "loss": 0.0261,
+      "step": 18750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909906274470242e-05,
+      "loss": 0.0317,
+      "step": 18760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909858234731772e-05,
+      "loss": 0.0246,
+      "step": 18770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99098101949933e-05,
+      "loss": 0.0232,
+      "step": 18780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909762155254827e-05,
+      "loss": 0.029,
+      "step": 18790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909714115516356e-05,
+      "loss": 0.032,
+      "step": 18800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909666075777885e-05,
+      "loss": 0.0312,
+      "step": 18810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909618036039415e-05,
+      "loss": 0.0308,
+      "step": 18820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.990956999630094e-05,
+      "loss": 0.0273,
+      "step": 18830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.990952195656247e-05,
+      "loss": 0.0287,
+      "step": 18840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909473916824e-05,
+      "loss": 0.0214,
+      "step": 18850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909425877085525e-05,
+      "loss": 0.0237,
+      "step": 18860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909377837347054e-05,
+      "loss": 0.0314,
+      "step": 18870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909329797608584e-05,
+      "loss": 0.0331,
+      "step": 18880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909281757870113e-05,
+      "loss": 0.029,
+      "step": 18890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.990923371813164e-05,
+      "loss": 0.0358,
+      "step": 18900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909185678393168e-05,
+      "loss": 0.0256,
+      "step": 18910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909137638654697e-05,
+      "loss": 0.0295,
+      "step": 18920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909089598916223e-05,
+      "loss": 0.0312,
+      "step": 18930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909041559177752e-05,
+      "loss": 0.0277,
+      "step": 18940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.990899351943928e-05,
+      "loss": 0.0313,
+      "step": 18950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.990894547970081e-05,
+      "loss": 0.0326,
+      "step": 18960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9908897439962337e-05,
+      "loss": 0.0246,
+      "step": 18970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9908849400223866e-05,
+      "loss": 0.0326,
+      "step": 18980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9908801360485395e-05,
+      "loss": 0.0278,
+      "step": 18990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9908753320746925e-05,
+      "loss": 0.0225,
+      "step": 19000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.990870528100845e-05,
+      "loss": 0.0258,
+      "step": 19010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.990865724126998e-05,
+      "loss": 0.0286,
+      "step": 19020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.990860920153151e-05,
+      "loss": 0.0253,
+      "step": 19030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9908561161793035e-05,
+      "loss": 0.0413,
+      "step": 19040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9908513122054564e-05,
+      "loss": 0.0256,
+      "step": 19050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9908465082316093e-05,
+      "loss": 0.0301,
+      "step": 19060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9908417042577623e-05,
+      "loss": 0.0253,
+      "step": 19070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.990836900283915e-05,
+      "loss": 0.0315,
+      "step": 19080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9908320963100678e-05,
+      "loss": 0.0289,
+      "step": 19090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9908272923362207e-05,
+      "loss": 0.0264,
+      "step": 19100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9908224883623733e-05,
+      "loss": 0.0373,
+      "step": 19110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9908176843885262e-05,
+      "loss": 0.0318,
+      "step": 19120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990812880414679e-05,
+      "loss": 0.0246,
+      "step": 19130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990808076440832e-05,
+      "loss": 0.0256,
+      "step": 19140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9908032724669847e-05,
+      "loss": 0.034,
+      "step": 19150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907984684931376e-05,
+      "loss": 0.027,
+      "step": 19160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907936645192905e-05,
+      "loss": 0.0269,
+      "step": 19170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907888605454435e-05,
+      "loss": 0.0262,
+      "step": 19180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990784056571596e-05,
+      "loss": 0.0289,
+      "step": 19190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990779252597749e-05,
+      "loss": 0.0281,
+      "step": 19200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990774448623902e-05,
+      "loss": 0.0243,
+      "step": 19210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907696446500545e-05,
+      "loss": 0.0208,
+      "step": 19220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907648406762074e-05,
+      "loss": 0.024,
+      "step": 19230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907600367023603e-05,
+      "loss": 0.0331,
+      "step": 19240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907552327285133e-05,
+      "loss": 0.0277,
+      "step": 19250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990750428754666e-05,
+      "loss": 0.0229,
+      "step": 19260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907456247808188e-05,
+      "loss": 0.0264,
+      "step": 19270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907408208069717e-05,
+      "loss": 0.0329,
+      "step": 19280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907360168331243e-05,
+      "loss": 0.0306,
+      "step": 19290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907312128592772e-05,
+      "loss": 0.0284,
+      "step": 19300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.99072640888543e-05,
+      "loss": 0.0283,
+      "step": 19310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990721604911583e-05,
+      "loss": 0.0291,
+      "step": 19320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907168009377357e-05,
+      "loss": 0.0261,
+      "step": 19330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907119969638886e-05,
+      "loss": 0.0268,
+      "step": 19340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907071929900415e-05,
+      "loss": 0.0267,
+      "step": 19350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9907023890161944e-05,
+      "loss": 0.0283,
+      "step": 19360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990697585042347e-05,
+      "loss": 0.0287,
+      "step": 19370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906927810685e-05,
+      "loss": 0.0297,
+      "step": 19380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990687977094653e-05,
+      "loss": 0.0342,
+      "step": 19390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906831731208055e-05,
+      "loss": 0.0316,
+      "step": 19400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906783691469584e-05,
+      "loss": 0.0264,
+      "step": 19410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906735651731113e-05,
+      "loss": 0.0289,
+      "step": 19420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906687611992643e-05,
+      "loss": 0.0271,
+      "step": 19430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990663957225417e-05,
+      "loss": 0.0275,
+      "step": 19440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906591532515698e-05,
+      "loss": 0.0279,
+      "step": 19450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906543492777227e-05,
+      "loss": 0.0325,
+      "step": 19460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906495453038753e-05,
+      "loss": 0.0276,
+      "step": 19470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906447413300282e-05,
+      "loss": 0.0308,
+      "step": 19480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990639937356181e-05,
+      "loss": 0.0317,
+      "step": 19490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990635133382334e-05,
+      "loss": 0.0285,
+      "step": 19500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906303294084867e-05,
+      "loss": 0.0225,
+      "step": 19510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906255254346396e-05,
+      "loss": 0.0205,
+      "step": 19520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906207214607925e-05,
+      "loss": 0.0301,
+      "step": 19530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9906159174869454e-05,
+      "loss": 0.0296,
+      "step": 19540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990611113513098e-05,
+      "loss": 0.0238,
+      "step": 19550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990606309539251e-05,
+      "loss": 0.0267,
+      "step": 19560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990601505565404e-05,
+      "loss": 0.0224,
+      "step": 19570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905967015915565e-05,
+      "loss": 0.0292,
+      "step": 19580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905918976177094e-05,
+      "loss": 0.0287,
+      "step": 19590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905870936438623e-05,
+      "loss": 0.0199,
+      "step": 19600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905822896700152e-05,
+      "loss": 0.0289,
+      "step": 19610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990577485696168e-05,
+      "loss": 0.027,
+      "step": 19620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905726817223208e-05,
+      "loss": 0.0283,
+      "step": 19630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905678777484737e-05,
+      "loss": 0.0245,
+      "step": 19640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905630737746263e-05,
+      "loss": 0.027,
+      "step": 19650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905582698007792e-05,
+      "loss": 0.0263,
+      "step": 19660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990553465826932e-05,
+      "loss": 0.0257,
+      "step": 19670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990548661853085e-05,
+      "loss": 0.0275,
+      "step": 19680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905438578792376e-05,
+      "loss": 0.0364,
+      "step": 19690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905390539053906e-05,
+      "loss": 0.0312,
+      "step": 19700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905342499315435e-05,
+      "loss": 0.0271,
+      "step": 19710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905294459576964e-05,
+      "loss": 0.0269,
+      "step": 19720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990524641983849e-05,
+      "loss": 0.0235,
+      "step": 19730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990519838010002e-05,
+      "loss": 0.0254,
+      "step": 19740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990515034036155e-05,
+      "loss": 0.0242,
+      "step": 19750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905102300623078e-05,
+      "loss": 0.0239,
+      "step": 19760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9905054260884607e-05,
+      "loss": 0.025,
+      "step": 19770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9905006221146133e-05,
+      "loss": 0.0302,
+      "step": 19780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904958181407662e-05,
+      "loss": 0.0199,
+      "step": 19790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990491014166919e-05,
+      "loss": 0.0313,
+      "step": 19800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990486210193072e-05,
+      "loss": 0.0235,
+      "step": 19810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904814062192247e-05,
+      "loss": 0.0273,
+      "step": 19820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904766022453776e-05,
+      "loss": 0.0214,
+      "step": 19830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904717982715305e-05,
+      "loss": 0.0325,
+      "step": 19840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904669942976835e-05,
+      "loss": 0.0316,
+      "step": 19850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990462190323836e-05,
+      "loss": 0.0325,
+      "step": 19860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990457386349989e-05,
+      "loss": 0.0322,
+      "step": 19870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990452582376142e-05,
+      "loss": 0.0278,
+      "step": 19880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904477784022945e-05,
+      "loss": 0.0299,
+      "step": 19890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904429744284474e-05,
+      "loss": 0.0247,
+      "step": 19900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904381704546003e-05,
+      "loss": 0.0314,
+      "step": 19910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904333664807533e-05,
+      "loss": 0.0275,
+      "step": 19920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904290429042904e-05,
+      "loss": 0.0335,
+      "step": 19930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904242389304433e-05,
+      "loss": 0.0237,
+      "step": 19940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904194349565963e-05,
+      "loss": 0.0276,
+      "step": 19950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990414630982749e-05,
+      "loss": 0.0293,
+      "step": 19960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904098270089018e-05,
+      "loss": 0.0253,
+      "step": 19970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904050230350547e-05,
+      "loss": 0.0278,
+      "step": 19980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904002190612076e-05,
+      "loss": 0.0244,
+      "step": 19990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903954150873602e-05,
+      "loss": 0.0245,
+      "step": 20000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990390611113513e-05,
+      "loss": 0.0264,
+      "step": 20010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990385807139666e-05,
+      "loss": 0.0263,
+      "step": 20020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990381003165819e-05,
+      "loss": 0.0272,
+      "step": 20030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903761991919716e-05,
+      "loss": 0.0209,
+      "step": 20040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903713952181245e-05,
+      "loss": 0.0329,
+      "step": 20050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903665912442774e-05,
+      "loss": 0.0294,
+      "step": 20060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.99036178727043e-05,
+      "loss": 0.0224,
+      "step": 20070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990356983296583e-05,
+      "loss": 0.0268,
+      "step": 20080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990352179322736e-05,
+      "loss": 0.0297,
+      "step": 20090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903473753488888e-05,
+      "loss": 0.0279,
+      "step": 20100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903425713750414e-05,
+      "loss": 0.0249,
+      "step": 20110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903377674011943e-05,
+      "loss": 0.0361,
+      "step": 20120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903329634273472e-05,
+      "loss": 0.0226,
+      "step": 20130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903281594535e-05,
+      "loss": 0.0301,
+      "step": 20140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903233554796528e-05,
+      "loss": 0.0271,
+      "step": 20150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903185515058057e-05,
+      "loss": 0.028,
+      "step": 20160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903137475319586e-05,
+      "loss": 0.0247,
+      "step": 20170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903089435581112e-05,
+      "loss": 0.0293,
+      "step": 20180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990304139584264e-05,
+      "loss": 0.0231,
+      "step": 20190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990299335610417e-05,
+      "loss": 0.0294,
+      "step": 20200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.99029453163657e-05,
+      "loss": 0.0302,
+      "step": 20210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9902897276627226e-05,
+      "loss": 0.0296,
+      "step": 20220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9902849236888755e-05,
+      "loss": 0.0292,
+      "step": 20230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9902801197150284e-05,
+      "loss": 0.0334,
+      "step": 20240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990275315741181e-05,
+      "loss": 0.0265,
+      "step": 20250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9902705117673343e-05,
+      "loss": 0.0308,
+      "step": 20260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9902657077934872e-05,
+      "loss": 0.0235,
+      "step": 20270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9902609038196398e-05,
+      "loss": 0.026,
+      "step": 20280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9902560998457927e-05,
+      "loss": 0.0287,
+      "step": 20290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9902512958719456e-05,
+      "loss": 0.0232,
+      "step": 20300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9902464918980982e-05,
+      "loss": 0.0288,
+      "step": 20310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990241687924251e-05,
+      "loss": 0.0232,
+      "step": 20320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990236883950404e-05,
+      "loss": 0.0257,
+      "step": 20330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990232079976557e-05,
+      "loss": 0.0252,
+      "step": 20340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9902272760027096e-05,
+      "loss": 0.0195,
+      "step": 20350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9902224720288625e-05,
+      "loss": 0.0314,
+      "step": 20360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9902176680550155e-05,
+      "loss": 0.0332,
+      "step": 20370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990212864081168e-05,
+      "loss": 0.0322,
+      "step": 20380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990208060107321e-05,
+      "loss": 0.0262,
+      "step": 20390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990203256133474e-05,
+      "loss": 0.0301,
+      "step": 20400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9901984521596268e-05,
+      "loss": 0.0234,
+      "step": 20410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9901936481857794e-05,
+      "loss": 0.0258,
+      "step": 20420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9901888442119323e-05,
+      "loss": 0.0297,
+      "step": 20430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9901840402380853e-05,
+      "loss": 0.0316,
+      "step": 20440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9901792362642382e-05,
+      "loss": 0.0258,
+      "step": 20450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9901744322903908e-05,
+      "loss": 0.0283,
+      "step": 20460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9901696283165437e-05,
+      "loss": 0.0283,
+      "step": 20470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9901648243426966e-05,
+      "loss": 0.0299,
+      "step": 20480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9901600203688492e-05,
+      "loss": 0.0317,
+      "step": 20490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.990155216395002e-05,
+      "loss": 0.0301,
+      "step": 20500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.990150412421155e-05,
+      "loss": 0.0353,
+      "step": 20510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.990145608447308e-05,
+      "loss": 0.0236,
+      "step": 20520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9901408044734606e-05,
+      "loss": 0.0259,
+      "step": 20530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9901360004996135e-05,
+      "loss": 0.0253,
+      "step": 20540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9901311965257664e-05,
+      "loss": 0.036,
+      "step": 20550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.990126392551919e-05,
+      "loss": 0.024,
+      "step": 20560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.990121588578072e-05,
+      "loss": 0.0274,
+      "step": 20570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.990116784604225e-05,
+      "loss": 0.0305,
+      "step": 20580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9901119806303778e-05,
+      "loss": 0.0318,
+      "step": 20590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9901071766565304e-05,
+      "loss": 0.0252,
+      "step": 20600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9901023726826833e-05,
+      "loss": 0.0234,
+      "step": 20610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900975687088363e-05,
+      "loss": 0.0265,
+      "step": 20620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900927647349892e-05,
+      "loss": 0.0257,
+      "step": 20630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900879607611418e-05,
+      "loss": 0.0247,
+      "step": 20640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900831567872947e-05,
+      "loss": 0.0339,
+      "step": 20650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900783528134476e-05,
+      "loss": 0.0212,
+      "step": 20660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900735488396002e-05,
+      "loss": 0.0218,
+      "step": 20670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.990068744865753e-05,
+      "loss": 0.0352,
+      "step": 20680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.990063940891906e-05,
+      "loss": 0.0314,
+      "step": 20690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.990059136918059e-05,
+      "loss": 0.0343,
+      "step": 20700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900543329442116e-05,
+      "loss": 0.0289,
+      "step": 20710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900495289703645e-05,
+      "loss": 0.0232,
+      "step": 20720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900447249965174e-05,
+      "loss": 0.0276,
+      "step": 20730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99003992102267e-05,
+      "loss": 0.036,
+      "step": 20740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.990035117048823e-05,
+      "loss": 0.0327,
+      "step": 20750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.990030313074976e-05,
+      "loss": 0.026,
+      "step": 20760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900255091011288e-05,
+      "loss": 0.0272,
+      "step": 20770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900207051272814e-05,
+      "loss": 0.0255,
+      "step": 20780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900159011534343e-05,
+      "loss": 0.027,
+      "step": 20790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900110971795873e-05,
+      "loss": 0.0254,
+      "step": 20800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900062932057402e-05,
+      "loss": 0.0281,
+      "step": 20810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9900014892318928e-05,
+      "loss": 0.0243,
+      "step": 20820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899966852580457e-05,
+      "loss": 0.0244,
+      "step": 20830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899918812841986e-05,
+      "loss": 0.0305,
+      "step": 20840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899870773103512e-05,
+      "loss": 0.0319,
+      "step": 20850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989982273336504e-05,
+      "loss": 0.0276,
+      "step": 20860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989977469362657e-05,
+      "loss": 0.0386,
+      "step": 20870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.98997266538881e-05,
+      "loss": 0.0253,
+      "step": 20880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899678614149626e-05,
+      "loss": 0.0244,
+      "step": 20890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899630574411155e-05,
+      "loss": 0.0257,
+      "step": 20900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899582534672684e-05,
+      "loss": 0.0265,
+      "step": 20910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989953449493421e-05,
+      "loss": 0.0277,
+      "step": 20920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989948645519574e-05,
+      "loss": 0.0248,
+      "step": 20930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989943841545727e-05,
+      "loss": 0.0314,
+      "step": 20940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899390375718798e-05,
+      "loss": 0.0257,
+      "step": 20950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899342335980324e-05,
+      "loss": 0.0318,
+      "step": 20960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899294296241853e-05,
+      "loss": 0.0243,
+      "step": 20970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899246256503382e-05,
+      "loss": 0.0302,
+      "step": 20980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899198216764912e-05,
+      "loss": 0.0271,
+      "step": 20990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899150177026438e-05,
+      "loss": 0.0343,
+      "step": 21000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899102137287967e-05,
+      "loss": 0.028,
+      "step": 21010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899054097549496e-05,
+      "loss": 0.0242,
+      "step": 21020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899006057811022e-05,
+      "loss": 0.0253,
+      "step": 21030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989895801807255e-05,
+      "loss": 0.0274,
+      "step": 21040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989890997833408e-05,
+      "loss": 0.0281,
+      "step": 21050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989886193859561e-05,
+      "loss": 0.0279,
+      "step": 21060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9898813898857136e-05,
+      "loss": 0.0296,
+      "step": 21070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9898765859118665e-05,
+      "loss": 0.0285,
+      "step": 21080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9898717819380194e-05,
+      "loss": 0.0267,
+      "step": 21090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989866977964172e-05,
+      "loss": 0.0298,
+      "step": 21100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989862173990325e-05,
+      "loss": 0.0256,
+      "step": 21110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989857370016478e-05,
+      "loss": 0.02,
+      "step": 21120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9898525660426308e-05,
+      "loss": 0.0201,
+      "step": 21130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9898477620687834e-05,
+      "loss": 0.0317,
+      "step": 21140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9898429580949363e-05,
+      "loss": 0.0392,
+      "step": 21150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9898381541210892e-05,
+      "loss": 0.0258,
+      "step": 21160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989833350147242e-05,
+      "loss": 0.0267,
+      "step": 21170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9898285461733948e-05,
+      "loss": 0.0247,
+      "step": 21180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9898237421995477e-05,
+      "loss": 0.0269,
+      "step": 21190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9898189382257006e-05,
+      "loss": 0.0292,
+      "step": 21200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9898141342518532e-05,
+      "loss": 0.0264,
+      "step": 21210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989809330278006e-05,
+      "loss": 0.0237,
+      "step": 21220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989804526304159e-05,
+      "loss": 0.0239,
+      "step": 21230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989799722330312e-05,
+      "loss": 0.0221,
+      "step": 21240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897949183564646e-05,
+      "loss": 0.0249,
+      "step": 21250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897901143826175e-05,
+      "loss": 0.0264,
+      "step": 21260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897853104087704e-05,
+      "loss": 0.0202,
+      "step": 21270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989780506434923e-05,
+      "loss": 0.0291,
+      "step": 21280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989775702461076e-05,
+      "loss": 0.029,
+      "step": 21290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989770898487229e-05,
+      "loss": 0.026,
+      "step": 21300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897660945133818e-05,
+      "loss": 0.0199,
+      "step": 21310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897612905395344e-05,
+      "loss": 0.0252,
+      "step": 21320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897564865656873e-05,
+      "loss": 0.0268,
+      "step": 21330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897516825918402e-05,
+      "loss": 0.0237,
+      "step": 21340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897468786179928e-05,
+      "loss": 0.0341,
+      "step": 21350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897420746441457e-05,
+      "loss": 0.0239,
+      "step": 21360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897372706702987e-05,
+      "loss": 0.033,
+      "step": 21370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897324666964516e-05,
+      "loss": 0.0312,
+      "step": 21380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897276627226042e-05,
+      "loss": 0.0211,
+      "step": 21390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989722858748757e-05,
+      "loss": 0.0244,
+      "step": 21400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.98971805477491e-05,
+      "loss": 0.0214,
+      "step": 21410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989713250801063e-05,
+      "loss": 0.0278,
+      "step": 21420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897084468272156e-05,
+      "loss": 0.0245,
+      "step": 21430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9897036428533685e-05,
+      "loss": 0.0326,
+      "step": 21440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896988388795214e-05,
+      "loss": 0.0258,
+      "step": 21450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989694034905674e-05,
+      "loss": 0.0246,
+      "step": 21460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989689230931827e-05,
+      "loss": 0.0287,
+      "step": 21470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.98968442695798e-05,
+      "loss": 0.0249,
+      "step": 21480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896796229841328e-05,
+      "loss": 0.0265,
+      "step": 21490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896748190102854e-05,
+      "loss": 0.0207,
+      "step": 21500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896700150364383e-05,
+      "loss": 0.0224,
+      "step": 21510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896652110625912e-05,
+      "loss": 0.0301,
+      "step": 21520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896604070887438e-05,
+      "loss": 0.0248,
+      "step": 21530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896556031148967e-05,
+      "loss": 0.0305,
+      "step": 21540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896507991410497e-05,
+      "loss": 0.0191,
+      "step": 21550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896459951672026e-05,
+      "loss": 0.0251,
+      "step": 21560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896411911933552e-05,
+      "loss": 0.0237,
+      "step": 21570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989636387219508e-05,
+      "loss": 0.0255,
+      "step": 21580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989631583245661e-05,
+      "loss": 0.032,
+      "step": 21590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989626779271814e-05,
+      "loss": 0.0263,
+      "step": 21600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896219752979665e-05,
+      "loss": 0.0311,
+      "step": 21610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896171713241195e-05,
+      "loss": 0.0222,
+      "step": 21620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9896123673502724e-05,
+      "loss": 0.0381,
+      "step": 21630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989607563376425e-05,
+      "loss": 0.024,
+      "step": 21640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989602759402578e-05,
+      "loss": 0.0202,
+      "step": 21650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989597955428731e-05,
+      "loss": 0.0276,
+      "step": 21660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9895931514548838e-05,
+      "loss": 0.0273,
+      "step": 21670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9895883474810364e-05,
+      "loss": 0.0289,
+      "step": 21680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9895835435071893e-05,
+      "loss": 0.0248,
+      "step": 21690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9895787395333422e-05,
+      "loss": 0.0199,
+      "step": 21700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9895739355594948e-05,
+      "loss": 0.0279,
+      "step": 21710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9895691315856477e-05,
+      "loss": 0.0253,
+      "step": 21720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9895643276118007e-05,
+      "loss": 0.0285,
+      "step": 21730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9895595236379536e-05,
+      "loss": 0.0219,
+      "step": 21740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989554719664106e-05,
+      "loss": 0.0227,
+      "step": 21750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989549915690259e-05,
+      "loss": 0.0244,
+      "step": 21760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989545111716412e-05,
+      "loss": 0.0237,
+      "step": 21770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989540307742565e-05,
+      "loss": 0.0264,
+      "step": 21780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9895355037687175e-05,
+      "loss": 0.0311,
+      "step": 21790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9895306997948705e-05,
+      "loss": 0.0209,
+      "step": 21800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9895258958210234e-05,
+      "loss": 0.0304,
+      "step": 21810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989521091847176e-05,
+      "loss": 0.0216,
+      "step": 21820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989516287873329e-05,
+      "loss": 0.0287,
+      "step": 21830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.989511483899482e-05,
+      "loss": 0.0336,
+      "step": 21840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9895066799256348e-05,
+      "loss": 0.0302,
+      "step": 21850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9895018759517873e-05,
+      "loss": 0.0421,
+      "step": 21860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894970719779403e-05,
+      "loss": 0.0276,
+      "step": 21870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894922680040932e-05,
+      "loss": 0.0317,
+      "step": 21880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894874640302458e-05,
+      "loss": 0.029,
+      "step": 21890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894826600563987e-05,
+      "loss": 0.0365,
+      "step": 21900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894778560825516e-05,
+      "loss": 0.0315,
+      "step": 21910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894730521087046e-05,
+      "loss": 0.031,
+      "step": 21920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989468248134857e-05,
+      "loss": 0.026,
+      "step": 21930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.98946344416101e-05,
+      "loss": 0.0261,
+      "step": 21940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989458640187163e-05,
+      "loss": 0.0308,
+      "step": 21950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989453836213316e-05,
+      "loss": 0.0316,
+      "step": 21960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894490322394685e-05,
+      "loss": 0.0276,
+      "step": 21970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894442282656215e-05,
+      "loss": 0.0213,
+      "step": 21980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894394242917744e-05,
+      "loss": 0.0252,
+      "step": 21990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989434620317927e-05,
+      "loss": 0.0265,
+      "step": 22000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.98942981634408e-05,
+      "loss": 0.0286,
+      "step": 22010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894250123702328e-05,
+      "loss": 0.0276,
+      "step": 22020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894202083963858e-05,
+      "loss": 0.0311,
+      "step": 22030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894154044225383e-05,
+      "loss": 0.026,
+      "step": 22040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894106004486913e-05,
+      "loss": 0.036,
+      "step": 22050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894057964748442e-05,
+      "loss": 0.0241,
+      "step": 22060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9894009925009968e-05,
+      "loss": 0.0267,
+      "step": 22070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893961885271497e-05,
+      "loss": 0.0237,
+      "step": 22080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893913845533026e-05,
+      "loss": 0.0207,
+      "step": 22090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893865805794556e-05,
+      "loss": 0.0295,
+      "step": 22100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989382257002993e-05,
+      "loss": 0.0192,
+      "step": 22110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989377453029146e-05,
+      "loss": 0.0276,
+      "step": 22120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893726490552985e-05,
+      "loss": 0.0191,
+      "step": 22130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893678450814515e-05,
+      "loss": 0.0254,
+      "step": 22140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893630411076044e-05,
+      "loss": 0.0241,
+      "step": 22150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893582371337573e-05,
+      "loss": 0.0249,
+      "step": 22160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.98935343315991e-05,
+      "loss": 0.0303,
+      "step": 22170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989348629186063e-05,
+      "loss": 0.0266,
+      "step": 22180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893438252122158e-05,
+      "loss": 0.0252,
+      "step": 22190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893390212383687e-05,
+      "loss": 0.0279,
+      "step": 22200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893342172645213e-05,
+      "loss": 0.0226,
+      "step": 22210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893294132906742e-05,
+      "loss": 0.0265,
+      "step": 22220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989324609316827e-05,
+      "loss": 0.0274,
+      "step": 22230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893198053429797e-05,
+      "loss": 0.0234,
+      "step": 22240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893150013691326e-05,
+      "loss": 0.0305,
+      "step": 22250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893101973952856e-05,
+      "loss": 0.0276,
+      "step": 22260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9893053934214385e-05,
+      "loss": 0.0191,
+      "step": 22270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989300589447591e-05,
+      "loss": 0.0278,
+      "step": 22280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989295785473744e-05,
+      "loss": 0.0272,
+      "step": 22290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989290981499897e-05,
+      "loss": 0.0292,
+      "step": 22300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892861775260495e-05,
+      "loss": 0.0241,
+      "step": 22310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892813735522025e-05,
+      "loss": 0.0331,
+      "step": 22320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892765695783554e-05,
+      "loss": 0.0254,
+      "step": 22330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892717656045083e-05,
+      "loss": 0.0236,
+      "step": 22340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989266961630661e-05,
+      "loss": 0.0245,
+      "step": 22350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892621576568138e-05,
+      "loss": 0.0204,
+      "step": 22360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892573536829668e-05,
+      "loss": 0.0295,
+      "step": 22370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892525497091197e-05,
+      "loss": 0.0259,
+      "step": 22380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892477457352723e-05,
+      "loss": 0.0284,
+      "step": 22390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892429417614252e-05,
+      "loss": 0.0261,
+      "step": 22400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989238137787578e-05,
+      "loss": 0.0268,
+      "step": 22410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892333338137307e-05,
+      "loss": 0.0241,
+      "step": 22420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892285298398836e-05,
+      "loss": 0.0264,
+      "step": 22430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892237258660366e-05,
+      "loss": 0.0256,
+      "step": 22440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9892189218921895e-05,
+      "loss": 0.0266,
+      "step": 22450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989214117918342e-05,
+      "loss": 0.0303,
+      "step": 22460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989209313944495e-05,
+      "loss": 0.027,
+      "step": 22470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989204509970648e-05,
+      "loss": 0.0214,
+      "step": 22480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9891997059968005e-05,
+      "loss": 0.0268,
+      "step": 22490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9891949020229535e-05,
+      "loss": 0.0269,
+      "step": 22500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9891900980491064e-05,
+      "loss": 0.0243,
+      "step": 22510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9891852940752593e-05,
+      "loss": 0.0235,
+      "step": 22520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.989180490101412e-05,
+      "loss": 0.031,
+      "step": 22530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9891756861275648e-05,
+      "loss": 0.0221,
+      "step": 22540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9891708821537177e-05,
+      "loss": 0.0283,
+      "step": 22550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9891660781798707e-05,
+      "loss": 0.0244,
+      "step": 22560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9891612742060233e-05,
+      "loss": 0.0297,
+      "step": 22570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9891564702321762e-05,
+      "loss": 0.0256,
+      "step": 22580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.989151666258329e-05,
+      "loss": 0.0265,
+      "step": 22590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9891468622844817e-05,
+      "loss": 0.0253,
+      "step": 22600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9891420583106346e-05,
+      "loss": 0.037,
+      "step": 22610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9891372543367876e-05,
+      "loss": 0.033,
+      "step": 22620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9891324503629405e-05,
+      "loss": 0.0272,
+      "step": 22630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.989127646389093e-05,
+      "loss": 0.0244,
+      "step": 22640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.989122842415246e-05,
+      "loss": 0.0274,
+      "step": 22650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.989118038441399e-05,
+      "loss": 0.0281,
+      "step": 22660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9891132344675515e-05,
+      "loss": 0.025,
+      "step": 22670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9891084304937044e-05,
+      "loss": 0.0294,
+      "step": 22680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9891036265198574e-05,
+      "loss": 0.023,
+      "step": 22690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890988225460103e-05,
+      "loss": 0.0192,
+      "step": 22700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.989094018572163e-05,
+      "loss": 0.0264,
+      "step": 22710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890892145983158e-05,
+      "loss": 0.03,
+      "step": 22720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890844106244687e-05,
+      "loss": 0.0314,
+      "step": 22730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890796066506217e-05,
+      "loss": 0.0214,
+      "step": 22740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890748026767743e-05,
+      "loss": 0.0222,
+      "step": 22750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890699987029272e-05,
+      "loss": 0.0228,
+      "step": 22760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.98906519472908e-05,
+      "loss": 0.0222,
+      "step": 22770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890603907552327e-05,
+      "loss": 0.0285,
+      "step": 22780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890555867813856e-05,
+      "loss": 0.0278,
+      "step": 22790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890507828075386e-05,
+      "loss": 0.0283,
+      "step": 22800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890459788336915e-05,
+      "loss": 0.0312,
+      "step": 22810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.989041174859844e-05,
+      "loss": 0.0237,
+      "step": 22820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.989036370885997e-05,
+      "loss": 0.0253,
+      "step": 22830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.98903156691215e-05,
+      "loss": 0.023,
+      "step": 22840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890267629383025e-05,
+      "loss": 0.0291,
+      "step": 22850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890219589644554e-05,
+      "loss": 0.0336,
+      "step": 22860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890171549906084e-05,
+      "loss": 0.0342,
+      "step": 22870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890123510167613e-05,
+      "loss": 0.0307,
+      "step": 22880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.989007547042914e-05,
+      "loss": 0.0263,
+      "step": 22890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9890027430690668e-05,
+      "loss": 0.0296,
+      "step": 22900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889979390952197e-05,
+      "loss": 0.0274,
+      "step": 22910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889931351213723e-05,
+      "loss": 0.0284,
+      "step": 22920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889883311475252e-05,
+      "loss": 0.0259,
+      "step": 22930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889835271736782e-05,
+      "loss": 0.0266,
+      "step": 22940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.988978723199831e-05,
+      "loss": 0.0255,
+      "step": 22950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889739192259837e-05,
+      "loss": 0.0219,
+      "step": 22960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889691152521366e-05,
+      "loss": 0.0293,
+      "step": 22970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889643112782895e-05,
+      "loss": 0.0237,
+      "step": 22980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889595073044425e-05,
+      "loss": 0.0266,
+      "step": 22990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.988954703330595e-05,
+      "loss": 0.0282,
+      "step": 23000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.988949899356748e-05,
+      "loss": 0.0308,
+      "step": 23010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.988945095382901e-05,
+      "loss": 0.029,
+      "step": 23020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889402914090535e-05,
+      "loss": 0.0251,
+      "step": 23030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889354874352064e-05,
+      "loss": 0.0289,
+      "step": 23040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889306834613594e-05,
+      "loss": 0.0217,
+      "step": 23050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889258794875123e-05,
+      "loss": 0.0234,
+      "step": 23060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.988921075513665e-05,
+      "loss": 0.0316,
+      "step": 23070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889162715398178e-05,
+      "loss": 0.0233,
+      "step": 23080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889114675659707e-05,
+      "loss": 0.0204,
+      "step": 23090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889066635921233e-05,
+      "loss": 0.0291,
+      "step": 23100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9889018596182762e-05,
+      "loss": 0.0341,
+      "step": 23110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.988897055644429e-05,
+      "loss": 0.0246,
+      "step": 23120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.988892251670582e-05,
+      "loss": 0.0235,
+      "step": 23130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9888874476967347e-05,
+      "loss": 0.0242,
+      "step": 23140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9888826437228876e-05,
+      "loss": 0.0233,
+      "step": 23150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9888778397490405e-05,
+      "loss": 0.0217,
+      "step": 23160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9888730357751935e-05,
+      "loss": 0.0316,
+      "step": 23170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.988868231801346e-05,
+      "loss": 0.0276,
+      "step": 23180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.988863427827499e-05,
+      "loss": 0.0224,
+      "step": 23190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.988858623853652e-05,
+      "loss": 0.0237,
+      "step": 23200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9888538198798045e-05,
+      "loss": 0.0303,
+      "step": 23210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9888490159059574e-05,
+      "loss": 0.0271,
+      "step": 23220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9888442119321103e-05,
+      "loss": 0.0219,
+      "step": 23230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9888394079582633e-05,
+      "loss": 0.0234,
+      "step": 23240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988834603984416e-05,
+      "loss": 0.0279,
+      "step": 23250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9888298000105688e-05,
+      "loss": 0.027,
+      "step": 23260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9888249960367217e-05,
+      "loss": 0.0264,
+      "step": 23270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9888201920628743e-05,
+      "loss": 0.0233,
+      "step": 23280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9888153880890272e-05,
+      "loss": 0.0252,
+      "step": 23290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.98881058411518e-05,
+      "loss": 0.0254,
+      "step": 23300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988805780141333e-05,
+      "loss": 0.0266,
+      "step": 23310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9888009761674857e-05,
+      "loss": 0.0323,
+      "step": 23320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887961721936386e-05,
+      "loss": 0.0323,
+      "step": 23330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887913682197915e-05,
+      "loss": 0.0307,
+      "step": 23340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887865642459445e-05,
+      "loss": 0.0262,
+      "step": 23350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887817602720974e-05,
+      "loss": 0.0168,
+      "step": 23360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887769562982503e-05,
+      "loss": 0.0194,
+      "step": 23370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988772152324403e-05,
+      "loss": 0.0273,
+      "step": 23380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887673483505558e-05,
+      "loss": 0.0286,
+      "step": 23390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887625443767087e-05,
+      "loss": 0.0275,
+      "step": 23400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887577404028617e-05,
+      "loss": 0.0229,
+      "step": 23410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887529364290143e-05,
+      "loss": 0.0261,
+      "step": 23420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887481324551672e-05,
+      "loss": 0.0243,
+      "step": 23430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.98874332848132e-05,
+      "loss": 0.0264,
+      "step": 23440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887385245074727e-05,
+      "loss": 0.0203,
+      "step": 23450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887337205336256e-05,
+      "loss": 0.0284,
+      "step": 23460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887289165597786e-05,
+      "loss": 0.0251,
+      "step": 23470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887241125859315e-05,
+      "loss": 0.024,
+      "step": 23480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988719308612084e-05,
+      "loss": 0.0254,
+      "step": 23490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988714504638237e-05,
+      "loss": 0.0228,
+      "step": 23500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.98870970066439e-05,
+      "loss": 0.0235,
+      "step": 23510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887048966905425e-05,
+      "loss": 0.0205,
+      "step": 23520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9887000927166954e-05,
+      "loss": 0.0216,
+      "step": 23530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886952887428484e-05,
+      "loss": 0.026,
+      "step": 23540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886904847690013e-05,
+      "loss": 0.029,
+      "step": 23550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988685680795154e-05,
+      "loss": 0.0222,
+      "step": 23560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886808768213068e-05,
+      "loss": 0.0231,
+      "step": 23570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886760728474597e-05,
+      "loss": 0.0321,
+      "step": 23580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886712688736127e-05,
+      "loss": 0.0268,
+      "step": 23590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886664648997653e-05,
+      "loss": 0.0216,
+      "step": 23600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886616609259182e-05,
+      "loss": 0.0297,
+      "step": 23610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988656856952071e-05,
+      "loss": 0.0275,
+      "step": 23620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886520529782237e-05,
+      "loss": 0.0274,
+      "step": 23630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886472490043766e-05,
+      "loss": 0.0259,
+      "step": 23640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886424450305296e-05,
+      "loss": 0.0288,
+      "step": 23650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886376410566825e-05,
+      "loss": 0.0216,
+      "step": 23660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988632837082835e-05,
+      "loss": 0.0228,
+      "step": 23670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988628033108988e-05,
+      "loss": 0.0217,
+      "step": 23680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988623229135141e-05,
+      "loss": 0.0281,
+      "step": 23690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886184251612935e-05,
+      "loss": 0.0293,
+      "step": 23700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886136211874464e-05,
+      "loss": 0.0268,
+      "step": 23710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886088172135994e-05,
+      "loss": 0.0263,
+      "step": 23720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9886040132397523e-05,
+      "loss": 0.0289,
+      "step": 23730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988599209265905e-05,
+      "loss": 0.0252,
+      "step": 23740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885944052920578e-05,
+      "loss": 0.0294,
+      "step": 23750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885896013182107e-05,
+      "loss": 0.0291,
+      "step": 23760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885847973443637e-05,
+      "loss": 0.0242,
+      "step": 23770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885799933705162e-05,
+      "loss": 0.0216,
+      "step": 23780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885751893966692e-05,
+      "loss": 0.0233,
+      "step": 23790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988570385422822e-05,
+      "loss": 0.0176,
+      "step": 23800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885655814489747e-05,
+      "loss": 0.0341,
+      "step": 23810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885607774751276e-05,
+      "loss": 0.0211,
+      "step": 23820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885559735012805e-05,
+      "loss": 0.02,
+      "step": 23830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885511695274335e-05,
+      "loss": 0.0265,
+      "step": 23840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988546365553586e-05,
+      "loss": 0.0287,
+      "step": 23850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988541561579739e-05,
+      "loss": 0.025,
+      "step": 23860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988536757605892e-05,
+      "loss": 0.0242,
+      "step": 23870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885319536320445e-05,
+      "loss": 0.0177,
+      "step": 23880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885271496581974e-05,
+      "loss": 0.028,
+      "step": 23890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885223456843504e-05,
+      "loss": 0.0232,
+      "step": 23900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885175417105033e-05,
+      "loss": 0.0285,
+      "step": 23910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.988512737736656e-05,
+      "loss": 0.0316,
+      "step": 23920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885079337628088e-05,
+      "loss": 0.0219,
+      "step": 23930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9885031297889617e-05,
+      "loss": 0.0298,
+      "step": 23940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884983258151146e-05,
+      "loss": 0.0234,
+      "step": 23950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884935218412672e-05,
+      "loss": 0.0298,
+      "step": 23960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.98848871786742e-05,
+      "loss": 0.0229,
+      "step": 23970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988483913893573e-05,
+      "loss": 0.0204,
+      "step": 23980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884791099197257e-05,
+      "loss": 0.0234,
+      "step": 23990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884743059458786e-05,
+      "loss": 0.0242,
+      "step": 24000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884695019720315e-05,
+      "loss": 0.0266,
+      "step": 24010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884646979981845e-05,
+      "loss": 0.0189,
+      "step": 24020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988459894024337e-05,
+      "loss": 0.0275,
+      "step": 24030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.98845509005049e-05,
+      "loss": 0.0291,
+      "step": 24040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988450286076643e-05,
+      "loss": 0.0226,
+      "step": 24050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884454821027955e-05,
+      "loss": 0.0257,
+      "step": 24060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884406781289484e-05,
+      "loss": 0.0225,
+      "step": 24070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884358741551013e-05,
+      "loss": 0.0244,
+      "step": 24080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884310701812543e-05,
+      "loss": 0.0204,
+      "step": 24090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988426266207407e-05,
+      "loss": 0.0254,
+      "step": 24100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884214622335598e-05,
+      "loss": 0.0194,
+      "step": 24110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884166582597127e-05,
+      "loss": 0.025,
+      "step": 24120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884118542858656e-05,
+      "loss": 0.0239,
+      "step": 24130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9884070503120182e-05,
+      "loss": 0.0267,
+      "step": 24140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988402246338171e-05,
+      "loss": 0.0243,
+      "step": 24150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988397442364324e-05,
+      "loss": 0.0354,
+      "step": 24160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883926383904767e-05,
+      "loss": 0.0223,
+      "step": 24170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883878344166296e-05,
+      "loss": 0.0259,
+      "step": 24180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883830304427825e-05,
+      "loss": 0.0227,
+      "step": 24190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883782264689355e-05,
+      "loss": 0.0211,
+      "step": 24200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988373422495088e-05,
+      "loss": 0.0269,
+      "step": 24210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988368618521241e-05,
+      "loss": 0.0217,
+      "step": 24220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988363814547394e-05,
+      "loss": 0.0213,
+      "step": 24230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883590105735465e-05,
+      "loss": 0.0258,
+      "step": 24240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883542065996994e-05,
+      "loss": 0.0256,
+      "step": 24250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883494026258523e-05,
+      "loss": 0.0201,
+      "step": 24260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883445986520053e-05,
+      "loss": 0.029,
+      "step": 24270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988339794678158e-05,
+      "loss": 0.0273,
+      "step": 24280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883349907043108e-05,
+      "loss": 0.0277,
+      "step": 24290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883301867304637e-05,
+      "loss": 0.0278,
+      "step": 24300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883253827566163e-05,
+      "loss": 0.0268,
+      "step": 24310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883205787827692e-05,
+      "loss": 0.0261,
+      "step": 24320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988315774808922e-05,
+      "loss": 0.0237,
+      "step": 24330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988310970835075e-05,
+      "loss": 0.0209,
+      "step": 24340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883061668612277e-05,
+      "loss": 0.0249,
+      "step": 24350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9883013628873806e-05,
+      "loss": 0.0261,
+      "step": 24360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882965589135335e-05,
+      "loss": 0.0263,
+      "step": 24370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882917549396864e-05,
+      "loss": 0.0326,
+      "step": 24380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988286950965839e-05,
+      "loss": 0.0307,
+      "step": 24390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988282146991992e-05,
+      "loss": 0.0201,
+      "step": 24400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988277343018145e-05,
+      "loss": 0.0188,
+      "step": 24410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882725390442975e-05,
+      "loss": 0.0226,
+      "step": 24420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882677350704504e-05,
+      "loss": 0.0219,
+      "step": 24430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882629310966033e-05,
+      "loss": 0.0229,
+      "step": 24440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882581271227563e-05,
+      "loss": 0.0203,
+      "step": 24450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988253323148909e-05,
+      "loss": 0.0313,
+      "step": 24460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882485191750618e-05,
+      "loss": 0.0259,
+      "step": 24470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882437152012147e-05,
+      "loss": 0.0332,
+      "step": 24480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882389112273673e-05,
+      "loss": 0.0322,
+      "step": 24490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882341072535202e-05,
+      "loss": 0.0237,
+      "step": 24500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988229303279673e-05,
+      "loss": 0.0264,
+      "step": 24510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988224499305826e-05,
+      "loss": 0.025,
+      "step": 24520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882196953319787e-05,
+      "loss": 0.0207,
+      "step": 24530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882148913581316e-05,
+      "loss": 0.0245,
+      "step": 24540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882100873842845e-05,
+      "loss": 0.0219,
+      "step": 24550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9882052834104374e-05,
+      "loss": 0.023,
+      "step": 24560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.98820047943659e-05,
+      "loss": 0.033,
+      "step": 24570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988195675462743e-05,
+      "loss": 0.0275,
+      "step": 24580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.988190871488896e-05,
+      "loss": 0.0285,
+      "step": 24590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9881860675150485e-05,
+      "loss": 0.0236,
+      "step": 24600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9881812635412014e-05,
+      "loss": 0.0247,
+      "step": 24610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9881764595673543e-05,
+      "loss": 0.0248,
+      "step": 24620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9881716555935072e-05,
+      "loss": 0.0266,
+      "step": 24630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.98816685161966e-05,
+      "loss": 0.029,
+      "step": 24640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9881620476458128e-05,
+      "loss": 0.0271,
+      "step": 24650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9881572436719657e-05,
+      "loss": 0.0273,
+      "step": 24660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9881524396981183e-05,
+      "loss": 0.0223,
+      "step": 24670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9881476357242712e-05,
+      "loss": 0.0285,
+      "step": 24680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.988142831750424e-05,
+      "loss": 0.0232,
+      "step": 24690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.988138027776577e-05,
+      "loss": 0.0258,
+      "step": 24700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9881332238027296e-05,
+      "loss": 0.0325,
+      "step": 24710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9881284198288826e-05,
+      "loss": 0.0248,
+      "step": 24720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9881236158550355e-05,
+      "loss": 0.0261,
+      "step": 24730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9881188118811884e-05,
+      "loss": 0.0237,
+      "step": 24740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.988114007907341e-05,
+      "loss": 0.025,
+      "step": 24750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.988109203933494e-05,
+      "loss": 0.0267,
+      "step": 24760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.988104399959647e-05,
+      "loss": 0.0233,
+      "step": 24770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880995959857995e-05,
+      "loss": 0.0304,
+      "step": 24780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880947920119524e-05,
+      "loss": 0.0256,
+      "step": 24790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880899880381053e-05,
+      "loss": 0.0243,
+      "step": 24800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880851840642582e-05,
+      "loss": 0.0245,
+      "step": 24810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880803800904108e-05,
+      "loss": 0.0237,
+      "step": 24820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880755761165638e-05,
+      "loss": 0.0261,
+      "step": 24830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880707721427167e-05,
+      "loss": 0.0208,
+      "step": 24840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880659681688693e-05,
+      "loss": 0.0214,
+      "step": 24850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880611641950222e-05,
+      "loss": 0.0191,
+      "step": 24860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.988056360221175e-05,
+      "loss": 0.0234,
+      "step": 24870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.988051556247328e-05,
+      "loss": 0.0291,
+      "step": 24880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880467522734806e-05,
+      "loss": 0.0276,
+      "step": 24890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880419482996336e-05,
+      "loss": 0.0191,
+      "step": 24900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880371443257865e-05,
+      "loss": 0.0242,
+      "step": 24910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880323403519394e-05,
+      "loss": 0.0243,
+      "step": 24920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.988027536378092e-05,
+      "loss": 0.0268,
+      "step": 24930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.988022732404245e-05,
+      "loss": 0.0338,
+      "step": 24940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.988017928430398e-05,
+      "loss": 0.0275,
+      "step": 24950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880131244565504e-05,
+      "loss": 0.024,
+      "step": 24960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880083204827034e-05,
+      "loss": 0.0343,
+      "step": 24970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9880035165088563e-05,
+      "loss": 0.0222,
+      "step": 24980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879987125350092e-05,
+      "loss": 0.028,
+      "step": 24990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879939085611618e-05,
+      "loss": 0.0247,
+      "step": 25000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879891045873147e-05,
+      "loss": 0.0289,
+      "step": 25010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879843006134677e-05,
+      "loss": 0.0254,
+      "step": 25020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879794966396203e-05,
+      "loss": 0.0212,
+      "step": 25030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879746926657732e-05,
+      "loss": 0.029,
+      "step": 25040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.987969888691926e-05,
+      "loss": 0.026,
+      "step": 25050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.987965084718079e-05,
+      "loss": 0.0273,
+      "step": 25060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879602807442316e-05,
+      "loss": 0.0227,
+      "step": 25070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879554767703846e-05,
+      "loss": 0.0198,
+      "step": 25080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879506727965375e-05,
+      "loss": 0.0227,
+      "step": 25090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879458688226904e-05,
+      "loss": 0.0287,
+      "step": 25100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.987941064848843e-05,
+      "loss": 0.0225,
+      "step": 25110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.987936260874996e-05,
+      "loss": 0.0267,
+      "step": 25120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.987931456901149e-05,
+      "loss": 0.0256,
+      "step": 25130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879266529273014e-05,
+      "loss": 0.0219,
+      "step": 25140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879218489534544e-05,
+      "loss": 0.026,
+      "step": 25150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879170449796073e-05,
+      "loss": 0.0285,
+      "step": 25160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879122410057602e-05,
+      "loss": 0.021,
+      "step": 25170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879074370319128e-05,
+      "loss": 0.0256,
+      "step": 25180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9879026330580657e-05,
+      "loss": 0.0254,
+      "step": 25190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9878978290842187e-05,
+      "loss": 0.0231,
+      "step": 25200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9878930251103713e-05,
+      "loss": 0.0268,
+      "step": 25210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9878882211365242e-05,
+      "loss": 0.0244,
+      "step": 25220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.987883417162677e-05,
+      "loss": 0.027,
+      "step": 25230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.98787861318883e-05,
+      "loss": 0.0273,
+      "step": 25240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9878738092149826e-05,
+      "loss": 0.0242,
+      "step": 25250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9878690052411355e-05,
+      "loss": 0.023,
+      "step": 25260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9878642012672885e-05,
+      "loss": 0.0242,
+      "step": 25270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9878593972934414e-05,
+      "loss": 0.0382,
+      "step": 25280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.987854593319594e-05,
+      "loss": 0.0255,
+      "step": 25290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.987849789345747e-05,
+      "loss": 0.0242,
+      "step": 25300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9878449853719e-05,
+      "loss": 0.0257,
+      "step": 25310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9878401813980524e-05,
+      "loss": 0.0201,
+      "step": 25320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9878353774242054e-05,
+      "loss": 0.024,
+      "step": 25330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9878305734503583e-05,
+      "loss": 0.0313,
+      "step": 25340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9878257694765112e-05,
+      "loss": 0.0286,
+      "step": 25350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9878209655026638e-05,
+      "loss": 0.0243,
+      "step": 25360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9878161615288167e-05,
+      "loss": 0.0228,
+      "step": 25370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9878113575549697e-05,
+      "loss": 0.0329,
+      "step": 25380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9878065535811222e-05,
+      "loss": 0.0296,
+      "step": 25390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987801749607275e-05,
+      "loss": 0.0269,
+      "step": 25400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987796945633428e-05,
+      "loss": 0.0269,
+      "step": 25410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987792141659581e-05,
+      "loss": 0.0286,
+      "step": 25420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877873376857336e-05,
+      "loss": 0.0242,
+      "step": 25430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877825337118865e-05,
+      "loss": 0.0273,
+      "step": 25440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877777297380395e-05,
+      "loss": 0.0244,
+      "step": 25450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877729257641924e-05,
+      "loss": 0.025,
+      "step": 25460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987768121790345e-05,
+      "loss": 0.0237,
+      "step": 25470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987763317816498e-05,
+      "loss": 0.0298,
+      "step": 25480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987758513842651e-05,
+      "loss": 0.0261,
+      "step": 25490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877537098688034e-05,
+      "loss": 0.0223,
+      "step": 25500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877489058949563e-05,
+      "loss": 0.0288,
+      "step": 25510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877441019211093e-05,
+      "loss": 0.0247,
+      "step": 25520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877392979472622e-05,
+      "loss": 0.025,
+      "step": 25530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877344939734148e-05,
+      "loss": 0.0279,
+      "step": 25540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877296899995677e-05,
+      "loss": 0.0194,
+      "step": 25550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877248860257206e-05,
+      "loss": 0.0249,
+      "step": 25560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877200820518732e-05,
+      "loss": 0.026,
+      "step": 25570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987715278078026e-05,
+      "loss": 0.0284,
+      "step": 25580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987710474104179e-05,
+      "loss": 0.0254,
+      "step": 25590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987705670130332e-05,
+      "loss": 0.0218,
+      "step": 25600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9877008661564846e-05,
+      "loss": 0.0234,
+      "step": 25610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876960621826375e-05,
+      "loss": 0.0264,
+      "step": 25620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876912582087905e-05,
+      "loss": 0.0328,
+      "step": 25630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876864542349434e-05,
+      "loss": 0.0243,
+      "step": 25640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987681650261096e-05,
+      "loss": 0.0239,
+      "step": 25650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987676846287249e-05,
+      "loss": 0.0254,
+      "step": 25660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876720423134018e-05,
+      "loss": 0.0235,
+      "step": 25670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876672383395544e-05,
+      "loss": 0.0227,
+      "step": 25680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876624343657073e-05,
+      "loss": 0.0222,
+      "step": 25690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876576303918603e-05,
+      "loss": 0.0243,
+      "step": 25700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876528264180132e-05,
+      "loss": 0.0235,
+      "step": 25710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876480224441658e-05,
+      "loss": 0.0325,
+      "step": 25720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876432184703187e-05,
+      "loss": 0.0258,
+      "step": 25730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876384144964716e-05,
+      "loss": 0.0265,
+      "step": 25740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876336105226242e-05,
+      "loss": 0.0203,
+      "step": 25750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987628806548777e-05,
+      "loss": 0.0287,
+      "step": 25760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.98762400257493e-05,
+      "loss": 0.0281,
+      "step": 25770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987619198601083e-05,
+      "loss": 0.0271,
+      "step": 25780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876143946272356e-05,
+      "loss": 0.025,
+      "step": 25790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876095906533885e-05,
+      "loss": 0.025,
+      "step": 25800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9876047866795414e-05,
+      "loss": 0.0245,
+      "step": 25810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875999827056944e-05,
+      "loss": 0.0239,
+      "step": 25820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987595178731847e-05,
+      "loss": 0.0278,
+      "step": 25830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987590374758e-05,
+      "loss": 0.0232,
+      "step": 25840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875855707841528e-05,
+      "loss": 0.0322,
+      "step": 25850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875807668103054e-05,
+      "loss": 0.0228,
+      "step": 25860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875759628364583e-05,
+      "loss": 0.0239,
+      "step": 25870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875711588626113e-05,
+      "loss": 0.0223,
+      "step": 25880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875663548887642e-05,
+      "loss": 0.0272,
+      "step": 25890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987561550914917e-05,
+      "loss": 0.0263,
+      "step": 25900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.98755674694107e-05,
+      "loss": 0.0215,
+      "step": 25910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875519429672226e-05,
+      "loss": 0.026,
+      "step": 25920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875471389933756e-05,
+      "loss": 0.0265,
+      "step": 25930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875423350195285e-05,
+      "loss": 0.0214,
+      "step": 25940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875375310456814e-05,
+      "loss": 0.0272,
+      "step": 25950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987532727071834e-05,
+      "loss": 0.0224,
+      "step": 25960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.987527923097987e-05,
+      "loss": 0.0277,
+      "step": 25970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.98752311912414e-05,
+      "loss": 0.0187,
+      "step": 25980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875183151502924e-05,
+      "loss": 0.0248,
+      "step": 25990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875135111764454e-05,
+      "loss": 0.0227,
+      "step": 26000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875087072025983e-05,
+      "loss": 0.0318,
+      "step": 26010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9875039032287512e-05,
+      "loss": 0.0275,
+      "step": 26020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874990992549038e-05,
+      "loss": 0.0172,
+      "step": 26030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874942952810567e-05,
+      "loss": 0.0243,
+      "step": 26040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874894913072097e-05,
+      "loss": 0.0285,
+      "step": 26050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874846873333623e-05,
+      "loss": 0.0239,
+      "step": 26060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874798833595152e-05,
+      "loss": 0.0221,
+      "step": 26070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987475079385668e-05,
+      "loss": 0.0272,
+      "step": 26080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987470275411821e-05,
+      "loss": 0.0208,
+      "step": 26090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874654714379736e-05,
+      "loss": 0.0301,
+      "step": 26100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874606674641265e-05,
+      "loss": 0.0255,
+      "step": 26110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874558634902795e-05,
+      "loss": 0.0231,
+      "step": 26120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874510595164324e-05,
+      "loss": 0.0228,
+      "step": 26130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987446255542585e-05,
+      "loss": 0.0212,
+      "step": 26140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987441451568738e-05,
+      "loss": 0.0249,
+      "step": 26150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874371279922754e-05,
+      "loss": 0.0278,
+      "step": 26160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987432324018428e-05,
+      "loss": 0.0234,
+      "step": 26170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987427520044581e-05,
+      "loss": 0.0187,
+      "step": 26180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874227160707338e-05,
+      "loss": 0.0252,
+      "step": 26190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874179120968868e-05,
+      "loss": 0.021,
+      "step": 26200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874131081230393e-05,
+      "loss": 0.0238,
+      "step": 26210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874083041491923e-05,
+      "loss": 0.0269,
+      "step": 26220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9874035001753452e-05,
+      "loss": 0.0308,
+      "step": 26230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873986962014978e-05,
+      "loss": 0.0295,
+      "step": 26240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873938922276507e-05,
+      "loss": 0.0203,
+      "step": 26250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873895686511885e-05,
+      "loss": 0.0255,
+      "step": 26260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987384764677341e-05,
+      "loss": 0.0252,
+      "step": 26270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987379960703494e-05,
+      "loss": 0.02,
+      "step": 26280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987375156729647e-05,
+      "loss": 0.0176,
+      "step": 26290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873703527557995e-05,
+      "loss": 0.0292,
+      "step": 26300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873655487819525e-05,
+      "loss": 0.0219,
+      "step": 26310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873607448081054e-05,
+      "loss": 0.0238,
+      "step": 26320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873559408342583e-05,
+      "loss": 0.0209,
+      "step": 26330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987351136860411e-05,
+      "loss": 0.0243,
+      "step": 26340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987346332886564e-05,
+      "loss": 0.0231,
+      "step": 26350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873415289127168e-05,
+      "loss": 0.0189,
+      "step": 26360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873367249388697e-05,
+      "loss": 0.0201,
+      "step": 26370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873319209650223e-05,
+      "loss": 0.0239,
+      "step": 26380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873271169911752e-05,
+      "loss": 0.0314,
+      "step": 26390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987322313017328e-05,
+      "loss": 0.0261,
+      "step": 26400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873175090434807e-05,
+      "loss": 0.0308,
+      "step": 26410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873127050696336e-05,
+      "loss": 0.0259,
+      "step": 26420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873079010957866e-05,
+      "loss": 0.02,
+      "step": 26430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9873030971219395e-05,
+      "loss": 0.0295,
+      "step": 26440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987298293148092e-05,
+      "loss": 0.0255,
+      "step": 26450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987293489174245e-05,
+      "loss": 0.0265,
+      "step": 26460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987288685200398e-05,
+      "loss": 0.0215,
+      "step": 26470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872838812265505e-05,
+      "loss": 0.0174,
+      "step": 26480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872790772527035e-05,
+      "loss": 0.0261,
+      "step": 26490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872742732788564e-05,
+      "loss": 0.0258,
+      "step": 26500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872694693050093e-05,
+      "loss": 0.0281,
+      "step": 26510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987264665331162e-05,
+      "loss": 0.0194,
+      "step": 26520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872598613573148e-05,
+      "loss": 0.0244,
+      "step": 26530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872550573834678e-05,
+      "loss": 0.0229,
+      "step": 26540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872502534096207e-05,
+      "loss": 0.0239,
+      "step": 26550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872454494357733e-05,
+      "loss": 0.0297,
+      "step": 26560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872406454619262e-05,
+      "loss": 0.0265,
+      "step": 26570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987235841488079e-05,
+      "loss": 0.0238,
+      "step": 26580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872310375142317e-05,
+      "loss": 0.0188,
+      "step": 26590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872262335403846e-05,
+      "loss": 0.0282,
+      "step": 26600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872214295665376e-05,
+      "loss": 0.0304,
+      "step": 26610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9872166255926905e-05,
+      "loss": 0.0289,
+      "step": 26620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987211821618843e-05,
+      "loss": 0.0223,
+      "step": 26630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987207017644996e-05,
+      "loss": 0.0196,
+      "step": 26640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987202213671149e-05,
+      "loss": 0.024,
+      "step": 26650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9871974096973015e-05,
+      "loss": 0.0221,
+      "step": 26660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9871926057234545e-05,
+      "loss": 0.022,
+      "step": 26670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9871878017496074e-05,
+      "loss": 0.0222,
+      "step": 26680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9871829977757603e-05,
+      "loss": 0.0253,
+      "step": 26690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.987178193801913e-05,
+      "loss": 0.0186,
+      "step": 26700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9871733898280658e-05,
+      "loss": 0.0208,
+      "step": 26710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9871685858542187e-05,
+      "loss": 0.0206,
+      "step": 26720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9871637818803717e-05,
+      "loss": 0.0296,
+      "step": 26730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9871589779065243e-05,
+      "loss": 0.02,
+      "step": 26740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9871541739326772e-05,
+      "loss": 0.027,
+      "step": 26750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98714936995883e-05,
+      "loss": 0.0251,
+      "step": 26760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9871445659849827e-05,
+      "loss": 0.0204,
+      "step": 26770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9871397620111356e-05,
+      "loss": 0.0289,
+      "step": 26780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9871349580372886e-05,
+      "loss": 0.0239,
+      "step": 26790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9871301540634415e-05,
+      "loss": 0.0284,
+      "step": 26800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987125350089594e-05,
+      "loss": 0.0272,
+      "step": 26810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987120546115747e-05,
+      "loss": 0.0185,
+      "step": 26820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9871157421419e-05,
+      "loss": 0.0262,
+      "step": 26830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9871109381680525e-05,
+      "loss": 0.0309,
+      "step": 26840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9871061341942054e-05,
+      "loss": 0.0227,
+      "step": 26850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9871013302203584e-05,
+      "loss": 0.0183,
+      "step": 26860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870965262465113e-05,
+      "loss": 0.0197,
+      "step": 26870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987091722272664e-05,
+      "loss": 0.0268,
+      "step": 26880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870869182988168e-05,
+      "loss": 0.0269,
+      "step": 26890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870821143249697e-05,
+      "loss": 0.0237,
+      "step": 26900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870773103511227e-05,
+      "loss": 0.0252,
+      "step": 26910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870725063772753e-05,
+      "loss": 0.0199,
+      "step": 26920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870677024034285e-05,
+      "loss": 0.0222,
+      "step": 26930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987062898429581e-05,
+      "loss": 0.032,
+      "step": 26940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987058094455734e-05,
+      "loss": 0.0189,
+      "step": 26950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987053290481887e-05,
+      "loss": 0.0207,
+      "step": 26960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98704848650804e-05,
+      "loss": 0.026,
+      "step": 26970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870436825341925e-05,
+      "loss": 0.0262,
+      "step": 26980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870388785603454e-05,
+      "loss": 0.019,
+      "step": 26990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870340745864983e-05,
+      "loss": 0.0169,
+      "step": 27000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987029270612651e-05,
+      "loss": 0.0258,
+      "step": 27010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987024466638804e-05,
+      "loss": 0.0202,
+      "step": 27020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870196626649568e-05,
+      "loss": 0.0194,
+      "step": 27030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870148586911097e-05,
+      "loss": 0.0249,
+      "step": 27040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870100547172623e-05,
+      "loss": 0.0274,
+      "step": 27050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870052507434152e-05,
+      "loss": 0.0207,
+      "step": 27060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987000446769568e-05,
+      "loss": 0.0248,
+      "step": 27070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869956427957207e-05,
+      "loss": 0.015,
+      "step": 27080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869908388218737e-05,
+      "loss": 0.0187,
+      "step": 27090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869860348480266e-05,
+      "loss": 0.0294,
+      "step": 27100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869812308741795e-05,
+      "loss": 0.0209,
+      "step": 27110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.986976426900332e-05,
+      "loss": 0.0218,
+      "step": 27120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.986971622926485e-05,
+      "loss": 0.0219,
+      "step": 27130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.986966818952638e-05,
+      "loss": 0.0251,
+      "step": 27140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869620149787905e-05,
+      "loss": 0.021,
+      "step": 27150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869572110049435e-05,
+      "loss": 0.0223,
+      "step": 27160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869524070310964e-05,
+      "loss": 0.0239,
+      "step": 27170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869476030572493e-05,
+      "loss": 0.0305,
+      "step": 27180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.986942799083402e-05,
+      "loss": 0.0199,
+      "step": 27190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.986937995109555e-05,
+      "loss": 0.0222,
+      "step": 27200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869331911357078e-05,
+      "loss": 0.0253,
+      "step": 27210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869283871618607e-05,
+      "loss": 0.0254,
+      "step": 27220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869235831880133e-05,
+      "loss": 0.0249,
+      "step": 27230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869187792141662e-05,
+      "loss": 0.0257,
+      "step": 27240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.986913975240319e-05,
+      "loss": 0.0217,
+      "step": 27250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869091712664717e-05,
+      "loss": 0.0228,
+      "step": 27260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9869043672926246e-05,
+      "loss": 0.0218,
+      "step": 27270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9868995633187776e-05,
+      "loss": 0.0209,
+      "step": 27280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9868947593449305e-05,
+      "loss": 0.0303,
+      "step": 27290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.986889955371083e-05,
+      "loss": 0.0203,
+      "step": 27300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.986885151397236e-05,
+      "loss": 0.0215,
+      "step": 27310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.986880347423389e-05,
+      "loss": 0.0291,
+      "step": 27320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9868755434495415e-05,
+      "loss": 0.0269,
+      "step": 27330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9868707394756945e-05,
+      "loss": 0.0333,
+      "step": 27340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9868659355018474e-05,
+      "loss": 0.0215,
+      "step": 27350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9868611315280003e-05,
+      "loss": 0.0207,
+      "step": 27360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.986856327554153e-05,
+      "loss": 0.0172,
+      "step": 27370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9868515235803058e-05,
+      "loss": 0.0215,
+      "step": 27380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9868467196064588e-05,
+      "loss": 0.0337,
+      "step": 27390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9868419156326117e-05,
+      "loss": 0.0227,
+      "step": 27400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868371116587643e-05,
+      "loss": 0.0268,
+      "step": 27410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868323076849172e-05,
+      "loss": 0.0237,
+      "step": 27420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98682750371107e-05,
+      "loss": 0.0222,
+      "step": 27430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868226997372227e-05,
+      "loss": 0.0238,
+      "step": 27440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868178957633756e-05,
+      "loss": 0.0254,
+      "step": 27450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868130917895286e-05,
+      "loss": 0.0202,
+      "step": 27460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868082878156815e-05,
+      "loss": 0.026,
+      "step": 27470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986803483841834e-05,
+      "loss": 0.0219,
+      "step": 27480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986798679867987e-05,
+      "loss": 0.0259,
+      "step": 27490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98679387589414e-05,
+      "loss": 0.0247,
+      "step": 27500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867890719202925e-05,
+      "loss": 0.0234,
+      "step": 27510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867842679464455e-05,
+      "loss": 0.0267,
+      "step": 27520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867794639725984e-05,
+      "loss": 0.0255,
+      "step": 27530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867746599987513e-05,
+      "loss": 0.0273,
+      "step": 27540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986769856024904e-05,
+      "loss": 0.0184,
+      "step": 27550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867650520510568e-05,
+      "loss": 0.0195,
+      "step": 27560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867602480772097e-05,
+      "loss": 0.0222,
+      "step": 27570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867554441033627e-05,
+      "loss": 0.0227,
+      "step": 27580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867506401295153e-05,
+      "loss": 0.0215,
+      "step": 27590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867458361556682e-05,
+      "loss": 0.0298,
+      "step": 27600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986741032181821e-05,
+      "loss": 0.0253,
+      "step": 27610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867362282079737e-05,
+      "loss": 0.0229,
+      "step": 27620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867314242341266e-05,
+      "loss": 0.0294,
+      "step": 27630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867266202602796e-05,
+      "loss": 0.0244,
+      "step": 27640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867218162864325e-05,
+      "loss": 0.0346,
+      "step": 27650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986717012312585e-05,
+      "loss": 0.0258,
+      "step": 27660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986712208338738e-05,
+      "loss": 0.0242,
+      "step": 27670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986707404364891e-05,
+      "loss": 0.0212,
+      "step": 27680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867026003910435e-05,
+      "loss": 0.0229,
+      "step": 27690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866977964171964e-05,
+      "loss": 0.0252,
+      "step": 27700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866929924433494e-05,
+      "loss": 0.0204,
+      "step": 27710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866881884695023e-05,
+      "loss": 0.02,
+      "step": 27720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986683384495655e-05,
+      "loss": 0.0243,
+      "step": 27730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866785805218078e-05,
+      "loss": 0.026,
+      "step": 27740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866737765479607e-05,
+      "loss": 0.0273,
+      "step": 27750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866689725741137e-05,
+      "loss": 0.027,
+      "step": 27760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866641686002663e-05,
+      "loss": 0.0227,
+      "step": 27770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866593646264192e-05,
+      "loss": 0.0292,
+      "step": 27780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986654560652572e-05,
+      "loss": 0.0176,
+      "step": 27790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866497566787247e-05,
+      "loss": 0.0204,
+      "step": 27800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866449527048776e-05,
+      "loss": 0.0235,
+      "step": 27810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866401487310306e-05,
+      "loss": 0.0258,
+      "step": 27820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866353447571835e-05,
+      "loss": 0.0297,
+      "step": 27830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986630540783336e-05,
+      "loss": 0.0269,
+      "step": 27840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986625736809489e-05,
+      "loss": 0.0201,
+      "step": 27850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986620932835642e-05,
+      "loss": 0.03,
+      "step": 27860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866161288617945e-05,
+      "loss": 0.0215,
+      "step": 27870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866113248879474e-05,
+      "loss": 0.0227,
+      "step": 27880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866065209141004e-05,
+      "loss": 0.0259,
+      "step": 27890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866017169402533e-05,
+      "loss": 0.0219,
+      "step": 27900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986596912966406e-05,
+      "loss": 0.0234,
+      "step": 27910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865921089925588e-05,
+      "loss": 0.0242,
+      "step": 27920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865873050187117e-05,
+      "loss": 0.0234,
+      "step": 27930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865825010448647e-05,
+      "loss": 0.0309,
+      "step": 27940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865776970710172e-05,
+      "loss": 0.0259,
+      "step": 27950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865728930971702e-05,
+      "loss": 0.0225,
+      "step": 27960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986568089123323e-05,
+      "loss": 0.0211,
+      "step": 27970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865632851494757e-05,
+      "loss": 0.0231,
+      "step": 27980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865584811756286e-05,
+      "loss": 0.0192,
+      "step": 27990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865536772017815e-05,
+      "loss": 0.0263,
+      "step": 28000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865488732279345e-05,
+      "loss": 0.0322,
+      "step": 28010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986544069254087e-05,
+      "loss": 0.0226,
+      "step": 28020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98653926528024e-05,
+      "loss": 0.0296,
+      "step": 28030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986534461306393e-05,
+      "loss": 0.0212,
+      "step": 28040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865296573325455e-05,
+      "loss": 0.0265,
+      "step": 28050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865248533586984e-05,
+      "loss": 0.0243,
+      "step": 28060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865200493848514e-05,
+      "loss": 0.0234,
+      "step": 28070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865152454110043e-05,
+      "loss": 0.022,
+      "step": 28080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986510441437157e-05,
+      "loss": 0.0238,
+      "step": 28090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865056374633098e-05,
+      "loss": 0.0213,
+      "step": 28100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9865008334894627e-05,
+      "loss": 0.0203,
+      "step": 28110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864960295156156e-05,
+      "loss": 0.0292,
+      "step": 28120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864912255417682e-05,
+      "loss": 0.0325,
+      "step": 28130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986486421567921e-05,
+      "loss": 0.0199,
+      "step": 28140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986481617594074e-05,
+      "loss": 0.022,
+      "step": 28150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864768136202267e-05,
+      "loss": 0.0266,
+      "step": 28160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864720096463796e-05,
+      "loss": 0.0216,
+      "step": 28170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864672056725325e-05,
+      "loss": 0.0264,
+      "step": 28180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864624016986855e-05,
+      "loss": 0.0233,
+      "step": 28190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986457597724838e-05,
+      "loss": 0.0238,
+      "step": 28200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986452793750991e-05,
+      "loss": 0.0179,
+      "step": 28210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986447989777144e-05,
+      "loss": 0.0204,
+      "step": 28220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864431858032965e-05,
+      "loss": 0.0185,
+      "step": 28230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864383818294494e-05,
+      "loss": 0.0259,
+      "step": 28240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864335778556023e-05,
+      "loss": 0.0272,
+      "step": 28250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864287738817553e-05,
+      "loss": 0.0244,
+      "step": 28260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986423969907908e-05,
+      "loss": 0.025,
+      "step": 28270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864191659340608e-05,
+      "loss": 0.0193,
+      "step": 28280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864143619602137e-05,
+      "loss": 0.0257,
+      "step": 28290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864095579863666e-05,
+      "loss": 0.026,
+      "step": 28300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9864047540125192e-05,
+      "loss": 0.028,
+      "step": 28310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986399950038672e-05,
+      "loss": 0.0196,
+      "step": 28320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986395146064825e-05,
+      "loss": 0.0223,
+      "step": 28330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863903420909777e-05,
+      "loss": 0.0284,
+      "step": 28340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863855381171306e-05,
+      "loss": 0.0224,
+      "step": 28350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863807341432835e-05,
+      "loss": 0.024,
+      "step": 28360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863759301694365e-05,
+      "loss": 0.0241,
+      "step": 28370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986371126195589e-05,
+      "loss": 0.0248,
+      "step": 28380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986366322221742e-05,
+      "loss": 0.0236,
+      "step": 28390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986361518247895e-05,
+      "loss": 0.0274,
+      "step": 28400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863567142740475e-05,
+      "loss": 0.0211,
+      "step": 28410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863519103002004e-05,
+      "loss": 0.028,
+      "step": 28420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863471063263533e-05,
+      "loss": 0.0216,
+      "step": 28430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863423023525063e-05,
+      "loss": 0.0301,
+      "step": 28440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986337498378659e-05,
+      "loss": 0.0221,
+      "step": 28450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863326944048118e-05,
+      "loss": 0.0186,
+      "step": 28460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863278904309647e-05,
+      "loss": 0.0228,
+      "step": 28470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863230864571176e-05,
+      "loss": 0.0201,
+      "step": 28480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863182824832702e-05,
+      "loss": 0.0267,
+      "step": 28490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986313478509423e-05,
+      "loss": 0.0206,
+      "step": 28500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986308674535576e-05,
+      "loss": 0.0239,
+      "step": 28510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863038705617287e-05,
+      "loss": 0.0202,
+      "step": 28520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862990665878816e-05,
+      "loss": 0.0251,
+      "step": 28530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862942626140345e-05,
+      "loss": 0.0213,
+      "step": 28540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862894586401874e-05,
+      "loss": 0.0232,
+      "step": 28550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.98628465466634e-05,
+      "loss": 0.0208,
+      "step": 28560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986279850692493e-05,
+      "loss": 0.0237,
+      "step": 28570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986275046718646e-05,
+      "loss": 0.0216,
+      "step": 28580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862702427447985e-05,
+      "loss": 0.0256,
+      "step": 28590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862654387709514e-05,
+      "loss": 0.0269,
+      "step": 28600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862606347971043e-05,
+      "loss": 0.0207,
+      "step": 28610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862558308232573e-05,
+      "loss": 0.0219,
+      "step": 28620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.98625102684941e-05,
+      "loss": 0.0165,
+      "step": 28630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862462228755628e-05,
+      "loss": 0.023,
+      "step": 28640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862414189017157e-05,
+      "loss": 0.0257,
+      "step": 28650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862366149278686e-05,
+      "loss": 0.0201,
+      "step": 28660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862318109540212e-05,
+      "loss": 0.0198,
+      "step": 28670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986227006980174e-05,
+      "loss": 0.0258,
+      "step": 28680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986222203006327e-05,
+      "loss": 0.0226,
+      "step": 28690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862173990324797e-05,
+      "loss": 0.0179,
+      "step": 28700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862125950586326e-05,
+      "loss": 0.0231,
+      "step": 28710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862077910847855e-05,
+      "loss": 0.0328,
+      "step": 28720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862029871109384e-05,
+      "loss": 0.0226,
+      "step": 28730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986198183137091e-05,
+      "loss": 0.0299,
+      "step": 28740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986193379163244e-05,
+      "loss": 0.0207,
+      "step": 28750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986188575189397e-05,
+      "loss": 0.0234,
+      "step": 28760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861837712155495e-05,
+      "loss": 0.0228,
+      "step": 28770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861789672417024e-05,
+      "loss": 0.0224,
+      "step": 28780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861741632678553e-05,
+      "loss": 0.0229,
+      "step": 28790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9861693592940082e-05,
+      "loss": 0.0213,
+      "step": 28800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.986164555320161e-05,
+      "loss": 0.02,
+      "step": 28810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9861597513463138e-05,
+      "loss": 0.018,
+      "step": 28820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9861549473724667e-05,
+      "loss": 0.028,
+      "step": 28830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9861501433986196e-05,
+      "loss": 0.023,
+      "step": 28840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9861453394247722e-05,
+      "loss": 0.0287,
+      "step": 28850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.986140535450925e-05,
+      "loss": 0.023,
+      "step": 28860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.986135731477078e-05,
+      "loss": 0.023,
+      "step": 28870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9861309275032306e-05,
+      "loss": 0.023,
+      "step": 28880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9861261235293836e-05,
+      "loss": 0.026,
+      "step": 28890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9861213195555365e-05,
+      "loss": 0.0269,
+      "step": 28900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9861165155816894e-05,
+      "loss": 0.0202,
+      "step": 28910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.986111711607842e-05,
+      "loss": 0.0237,
+      "step": 28920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.986106907633995e-05,
+      "loss": 0.0212,
+      "step": 28930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.986102103660148e-05,
+      "loss": 0.0206,
+      "step": 28940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860972996863005e-05,
+      "loss": 0.0223,
+      "step": 28950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860924957124534e-05,
+      "loss": 0.0234,
+      "step": 28960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860876917386063e-05,
+      "loss": 0.0239,
+      "step": 28970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860828877647592e-05,
+      "loss": 0.02,
+      "step": 28980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860780837909118e-05,
+      "loss": 0.0272,
+      "step": 28990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860732798170648e-05,
+      "loss": 0.0218,
+      "step": 29000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860684758432177e-05,
+      "loss": 0.0276,
+      "step": 29010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860636718693703e-05,
+      "loss": 0.0211,
+      "step": 29020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860588678955232e-05,
+      "loss": 0.0234,
+      "step": 29030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.986054063921676e-05,
+      "loss": 0.0234,
+      "step": 29040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.986049259947829e-05,
+      "loss": 0.0272,
+      "step": 29050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860444559739816e-05,
+      "loss": 0.0243,
+      "step": 29060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860396520001346e-05,
+      "loss": 0.0272,
+      "step": 29070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860348480262875e-05,
+      "loss": 0.0226,
+      "step": 29080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860300440524404e-05,
+      "loss": 0.0201,
+      "step": 29090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.986025240078593e-05,
+      "loss": 0.0234,
+      "step": 29100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.986020436104746e-05,
+      "loss": 0.0178,
+      "step": 29110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.986015632130899e-05,
+      "loss": 0.0213,
+      "step": 29120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860108281570514e-05,
+      "loss": 0.019,
+      "step": 29130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860060241832044e-05,
+      "loss": 0.0194,
+      "step": 29140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9860012202093573e-05,
+      "loss": 0.0227,
+      "step": 29150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859964162355102e-05,
+      "loss": 0.0281,
+      "step": 29160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859916122616628e-05,
+      "loss": 0.0224,
+      "step": 29170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859868082878157e-05,
+      "loss": 0.0225,
+      "step": 29180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859820043139687e-05,
+      "loss": 0.0201,
+      "step": 29190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859772003401213e-05,
+      "loss": 0.0227,
+      "step": 29200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859723963662742e-05,
+      "loss": 0.0235,
+      "step": 29210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985967592392427e-05,
+      "loss": 0.0244,
+      "step": 29220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98596278841858e-05,
+      "loss": 0.0178,
+      "step": 29230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859579844447326e-05,
+      "loss": 0.0243,
+      "step": 29240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859531804708856e-05,
+      "loss": 0.0309,
+      "step": 29250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859483764970385e-05,
+      "loss": 0.0246,
+      "step": 29260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859435725231914e-05,
+      "loss": 0.0164,
+      "step": 29270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985938768549344e-05,
+      "loss": 0.0234,
+      "step": 29280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985933964575497e-05,
+      "loss": 0.0228,
+      "step": 29290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98592916060165e-05,
+      "loss": 0.0219,
+      "step": 29300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859243566278024e-05,
+      "loss": 0.0207,
+      "step": 29310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859195526539554e-05,
+      "loss": 0.0238,
+      "step": 29320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859147486801083e-05,
+      "loss": 0.0246,
+      "step": 29330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859099447062612e-05,
+      "loss": 0.0217,
+      "step": 29340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859051407324138e-05,
+      "loss": 0.0239,
+      "step": 29350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9859003367585667e-05,
+      "loss": 0.0205,
+      "step": 29360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9858955327847197e-05,
+      "loss": 0.0225,
+      "step": 29370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9858907288108723e-05,
+      "loss": 0.0278,
+      "step": 29380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9858859248370252e-05,
+      "loss": 0.0223,
+      "step": 29390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985881120863178e-05,
+      "loss": 0.0207,
+      "step": 29400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985876316889331e-05,
+      "loss": 0.0232,
+      "step": 29410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9858715129154836e-05,
+      "loss": 0.0295,
+      "step": 29420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9858667089416365e-05,
+      "loss": 0.0293,
+      "step": 29430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9858619049677895e-05,
+      "loss": 0.0243,
+      "step": 29440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9858571009939424e-05,
+      "loss": 0.0264,
+      "step": 29450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985852297020095e-05,
+      "loss": 0.0204,
+      "step": 29460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985847493046248e-05,
+      "loss": 0.0164,
+      "step": 29470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985842689072401e-05,
+      "loss": 0.0249,
+      "step": 29480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9858378850985534e-05,
+      "loss": 0.0185,
+      "step": 29490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9858330811247067e-05,
+      "loss": 0.0158,
+      "step": 29500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9858282771508596e-05,
+      "loss": 0.0222,
+      "step": 29510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9858234731770122e-05,
+      "loss": 0.0167,
+      "step": 29520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985818669203165e-05,
+      "loss": 0.022,
+      "step": 29530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985813865229318e-05,
+      "loss": 0.0244,
+      "step": 29540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9858090612554707e-05,
+      "loss": 0.0207,
+      "step": 29550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9858042572816236e-05,
+      "loss": 0.0216,
+      "step": 29560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857994533077765e-05,
+      "loss": 0.0252,
+      "step": 29570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857946493339294e-05,
+      "loss": 0.0242,
+      "step": 29580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985789845360082e-05,
+      "loss": 0.0294,
+      "step": 29590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985785041386235e-05,
+      "loss": 0.0228,
+      "step": 29600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985780237412388e-05,
+      "loss": 0.0307,
+      "step": 29610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857754334385405e-05,
+      "loss": 0.0214,
+      "step": 29620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857706294646934e-05,
+      "loss": 0.0215,
+      "step": 29630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857658254908463e-05,
+      "loss": 0.0198,
+      "step": 29640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857610215169992e-05,
+      "loss": 0.0329,
+      "step": 29650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985756217543152e-05,
+      "loss": 0.0223,
+      "step": 29660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857514135693048e-05,
+      "loss": 0.0231,
+      "step": 29670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857466095954577e-05,
+      "loss": 0.0245,
+      "step": 29680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857418056216106e-05,
+      "loss": 0.02,
+      "step": 29690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857370016477632e-05,
+      "loss": 0.0277,
+      "step": 29700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985732197673916e-05,
+      "loss": 0.0273,
+      "step": 29710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985727393700069e-05,
+      "loss": 0.0247,
+      "step": 29720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857225897262216e-05,
+      "loss": 0.0328,
+      "step": 29730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857177857523746e-05,
+      "loss": 0.0224,
+      "step": 29740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857129817785275e-05,
+      "loss": 0.0264,
+      "step": 29750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9857081778046804e-05,
+      "loss": 0.0218,
+      "step": 29760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985703373830833e-05,
+      "loss": 0.0165,
+      "step": 29770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985698569856986e-05,
+      "loss": 0.0226,
+      "step": 29780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985693765883139e-05,
+      "loss": 0.0234,
+      "step": 29790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856889619092915e-05,
+      "loss": 0.0234,
+      "step": 29800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856841579354444e-05,
+      "loss": 0.0255,
+      "step": 29810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856793539615973e-05,
+      "loss": 0.0247,
+      "step": 29820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856745499877502e-05,
+      "loss": 0.024,
+      "step": 29830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856697460139028e-05,
+      "loss": 0.0195,
+      "step": 29840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856649420400558e-05,
+      "loss": 0.0256,
+      "step": 29850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856601380662087e-05,
+      "loss": 0.0258,
+      "step": 29860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856553340923616e-05,
+      "loss": 0.0218,
+      "step": 29870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856505301185142e-05,
+      "loss": 0.0252,
+      "step": 29880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985645726144667e-05,
+      "loss": 0.02,
+      "step": 29890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.98564092217082e-05,
+      "loss": 0.0268,
+      "step": 29900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856361181969726e-05,
+      "loss": 0.021,
+      "step": 29910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856313142231256e-05,
+      "loss": 0.0207,
+      "step": 29920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856265102492785e-05,
+      "loss": 0.0215,
+      "step": 29930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856217062754314e-05,
+      "loss": 0.0167,
+      "step": 29940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985616902301584e-05,
+      "loss": 0.0221,
+      "step": 29950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985612098327737e-05,
+      "loss": 0.0207,
+      "step": 29960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.98560729435389e-05,
+      "loss": 0.0229,
+      "step": 29970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9856024903800424e-05,
+      "loss": 0.0269,
+      "step": 29980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855976864061954e-05,
+      "loss": 0.0212,
+      "step": 29990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855928824323483e-05,
+      "loss": 0.0227,
+      "step": 30000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855880784585012e-05,
+      "loss": 0.0252,
+      "step": 30010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855832744846538e-05,
+      "loss": 0.019,
+      "step": 30020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855784705108067e-05,
+      "loss": 0.0211,
+      "step": 30030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855736665369597e-05,
+      "loss": 0.0212,
+      "step": 30040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855688625631126e-05,
+      "loss": 0.0216,
+      "step": 30050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855640585892652e-05,
+      "loss": 0.0242,
+      "step": 30060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985559254615418e-05,
+      "loss": 0.022,
+      "step": 30070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985554450641571e-05,
+      "loss": 0.0164,
+      "step": 30080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855496466677236e-05,
+      "loss": 0.025,
+      "step": 30090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855448426938766e-05,
+      "loss": 0.0174,
+      "step": 30100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855400387200295e-05,
+      "loss": 0.0206,
+      "step": 30110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855352347461824e-05,
+      "loss": 0.0258,
+      "step": 30120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985530430772335e-05,
+      "loss": 0.0185,
+      "step": 30130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985525626798488e-05,
+      "loss": 0.024,
+      "step": 30140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.985520822824641e-05,
+      "loss": 0.023,
+      "step": 30150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855160188507934e-05,
+      "loss": 0.0263,
+      "step": 30160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855112148769464e-05,
+      "loss": 0.0221,
+      "step": 30170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9855064109030993e-05,
+      "loss": 0.0223,
+      "step": 30180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9855016069292522e-05,
+      "loss": 0.027,
+      "step": 30190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854968029554048e-05,
+      "loss": 0.0197,
+      "step": 30200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854919989815577e-05,
+      "loss": 0.019,
+      "step": 30210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854871950077107e-05,
+      "loss": 0.0212,
+      "step": 30220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854823910338636e-05,
+      "loss": 0.0278,
+      "step": 30230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854775870600162e-05,
+      "loss": 0.0322,
+      "step": 30240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985472783086169e-05,
+      "loss": 0.0251,
+      "step": 30250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985467979112322e-05,
+      "loss": 0.0346,
+      "step": 30260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854631751384746e-05,
+      "loss": 0.02,
+      "step": 30270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854583711646275e-05,
+      "loss": 0.0234,
+      "step": 30280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854535671907805e-05,
+      "loss": 0.0226,
+      "step": 30290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985449243614318e-05,
+      "loss": 0.025,
+      "step": 30300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985444439640471e-05,
+      "loss": 0.0251,
+      "step": 30310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854396356666238e-05,
+      "loss": 0.0219,
+      "step": 30320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854348316927764e-05,
+      "loss": 0.025,
+      "step": 30330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854300277189293e-05,
+      "loss": 0.0231,
+      "step": 30340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854252237450822e-05,
+      "loss": 0.0217,
+      "step": 30350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985420419771235e-05,
+      "loss": 0.0233,
+      "step": 30360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854156157973877e-05,
+      "loss": 0.0256,
+      "step": 30370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854108118235407e-05,
+      "loss": 0.0158,
+      "step": 30380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854060078496936e-05,
+      "loss": 0.0171,
+      "step": 30390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9854012038758462e-05,
+      "loss": 0.0234,
+      "step": 30400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985396399901999e-05,
+      "loss": 0.0259,
+      "step": 30410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985391595928152e-05,
+      "loss": 0.0256,
+      "step": 30420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985386791954305e-05,
+      "loss": 0.022,
+      "step": 30430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9853819879804576e-05,
+      "loss": 0.0192,
+      "step": 30440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9853771840066105e-05,
+      "loss": 0.0174,
+      "step": 30450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9853723800327634e-05,
+      "loss": 0.0181,
+      "step": 30460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985367576058916e-05,
+      "loss": 0.0228,
+      "step": 30470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985362772085069e-05,
+      "loss": 0.0227,
+      "step": 30480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985357968111222e-05,
+      "loss": 0.0308,
+      "step": 30490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9853531641373748e-05,
+      "loss": 0.0177,
+      "step": 30500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9853483601635274e-05,
+      "loss": 0.019,
+      "step": 30510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9853435561896803e-05,
+      "loss": 0.025,
+      "step": 30520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9853387522158332e-05,
+      "loss": 0.0241,
+      "step": 30530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985333948241986e-05,
+      "loss": 0.0272,
+      "step": 30540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9853291442681387e-05,
+      "loss": 0.0166,
+      "step": 30550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9853243402942917e-05,
+      "loss": 0.0194,
+      "step": 30560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9853195363204446e-05,
+      "loss": 0.0266,
+      "step": 30570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9853147323465972e-05,
+      "loss": 0.0181,
+      "step": 30580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98530992837275e-05,
+      "loss": 0.0214,
+      "step": 30590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985305124398903e-05,
+      "loss": 0.0233,
+      "step": 30600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985300320425056e-05,
+      "loss": 0.0227,
+      "step": 30610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852955164512086e-05,
+      "loss": 0.0245,
+      "step": 30620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852907124773615e-05,
+      "loss": 0.0196,
+      "step": 30630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852859085035144e-05,
+      "loss": 0.0165,
+      "step": 30640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985281104529667e-05,
+      "loss": 0.0237,
+      "step": 30650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98527630055582e-05,
+      "loss": 0.0218,
+      "step": 30660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985271496581973e-05,
+      "loss": 0.0257,
+      "step": 30670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852666926081258e-05,
+      "loss": 0.0267,
+      "step": 30680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852618886342784e-05,
+      "loss": 0.0222,
+      "step": 30690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852570846604313e-05,
+      "loss": 0.023,
+      "step": 30700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852522806865842e-05,
+      "loss": 0.0262,
+      "step": 30710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985247476712737e-05,
+      "loss": 0.0159,
+      "step": 30720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852426727388897e-05,
+      "loss": 0.0203,
+      "step": 30730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852378687650427e-05,
+      "loss": 0.0176,
+      "step": 30740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852330647911956e-05,
+      "loss": 0.0195,
+      "step": 30750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852282608173482e-05,
+      "loss": 0.0213,
+      "step": 30760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985223456843501e-05,
+      "loss": 0.0255,
+      "step": 30770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985218652869654e-05,
+      "loss": 0.0207,
+      "step": 30780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985213848895807e-05,
+      "loss": 0.0154,
+      "step": 30790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852090449219595e-05,
+      "loss": 0.0206,
+      "step": 30800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9852042409481125e-05,
+      "loss": 0.0259,
+      "step": 30810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9851994369742654e-05,
+      "loss": 0.0203,
+      "step": 30820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985194633000418e-05,
+      "loss": 0.0204,
+      "step": 30830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985189829026571e-05,
+      "loss": 0.0153,
+      "step": 30840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985185025052724e-05,
+      "loss": 0.0249,
+      "step": 30850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9851802210788768e-05,
+      "loss": 0.026,
+      "step": 30860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9851754171050294e-05,
+      "loss": 0.0242,
+      "step": 30870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9851706131311823e-05,
+      "loss": 0.0236,
+      "step": 30880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9851658091573352e-05,
+      "loss": 0.023,
+      "step": 30890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985161005183488e-05,
+      "loss": 0.0207,
+      "step": 30900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9851562012096407e-05,
+      "loss": 0.0266,
+      "step": 30910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9851513972357937e-05,
+      "loss": 0.0203,
+      "step": 30920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9851465932619466e-05,
+      "loss": 0.0258,
+      "step": 30930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985141789288099e-05,
+      "loss": 0.0205,
+      "step": 30940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985136985314252e-05,
+      "loss": 0.0248,
+      "step": 30950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985132181340405e-05,
+      "loss": 0.0227,
+      "step": 30960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985127377366558e-05,
+      "loss": 0.0226,
+      "step": 30970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9851225733927105e-05,
+      "loss": 0.0198,
+      "step": 30980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9851177694188635e-05,
+      "loss": 0.0281,
+      "step": 30990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9851129654450164e-05,
+      "loss": 0.0301,
+      "step": 31000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985108161471169e-05,
+      "loss": 0.029,
+      "step": 31010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985103357497322e-05,
+      "loss": 0.0291,
+      "step": 31020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985098553523475e-05,
+      "loss": 0.0251,
+      "step": 31030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850937495496278e-05,
+      "loss": 0.0163,
+      "step": 31040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850889455757803e-05,
+      "loss": 0.0262,
+      "step": 31050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850841416019333e-05,
+      "loss": 0.0139,
+      "step": 31060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850793376280862e-05,
+      "loss": 0.0235,
+      "step": 31070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985074533654239e-05,
+      "loss": 0.0333,
+      "step": 31080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850697296803917e-05,
+      "loss": 0.0264,
+      "step": 31090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850649257065446e-05,
+      "loss": 0.0224,
+      "step": 31100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850601217326976e-05,
+      "loss": 0.0201,
+      "step": 31110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98505531775885e-05,
+      "loss": 0.0227,
+      "step": 31120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985050513785003e-05,
+      "loss": 0.0227,
+      "step": 31130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985045709811156e-05,
+      "loss": 0.0161,
+      "step": 31140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985040905837309e-05,
+      "loss": 0.0219,
+      "step": 31150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850361018634615e-05,
+      "loss": 0.0334,
+      "step": 31160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850312978896145e-05,
+      "loss": 0.0206,
+      "step": 31170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850264939157674e-05,
+      "loss": 0.0214,
+      "step": 31180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98502168994192e-05,
+      "loss": 0.0218,
+      "step": 31190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985016885968073e-05,
+      "loss": 0.0195,
+      "step": 31200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850120819942258e-05,
+      "loss": 0.0207,
+      "step": 31210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850072780203787e-05,
+      "loss": 0.022,
+      "step": 31220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9850024740465313e-05,
+      "loss": 0.0171,
+      "step": 31230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849976700726843e-05,
+      "loss": 0.0194,
+      "step": 31240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849928660988372e-05,
+      "loss": 0.0229,
+      "step": 31250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98498806212499e-05,
+      "loss": 0.02,
+      "step": 31260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849832581511427e-05,
+      "loss": 0.0195,
+      "step": 31270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849784541772956e-05,
+      "loss": 0.0223,
+      "step": 31280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849736502034486e-05,
+      "loss": 0.0218,
+      "step": 31290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.984968846229601e-05,
+      "loss": 0.0226,
+      "step": 31300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.984964042255754e-05,
+      "loss": 0.0203,
+      "step": 31310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.984959238281907e-05,
+      "loss": 0.0251,
+      "step": 31320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98495443430806e-05,
+      "loss": 0.0216,
+      "step": 31330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849496303342125e-05,
+      "loss": 0.0259,
+      "step": 31340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849448263603654e-05,
+      "loss": 0.0186,
+      "step": 31350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849400223865184e-05,
+      "loss": 0.0191,
+      "step": 31360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.984935218412671e-05,
+      "loss": 0.0194,
+      "step": 31370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.984930414438824e-05,
+      "loss": 0.0235,
+      "step": 31380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849256104649768e-05,
+      "loss": 0.0194,
+      "step": 31390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849208064911297e-05,
+      "loss": 0.0238,
+      "step": 31400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849160025172823e-05,
+      "loss": 0.0238,
+      "step": 31410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849111985434353e-05,
+      "loss": 0.0226,
+      "step": 31420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849063945695882e-05,
+      "loss": 0.0226,
+      "step": 31430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.984901590595741e-05,
+      "loss": 0.0251,
+      "step": 31440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9848967866218937e-05,
+      "loss": 0.0299,
+      "step": 31450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9848919826480466e-05,
+      "loss": 0.0217,
+      "step": 31460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9848871786741996e-05,
+      "loss": 0.0246,
+      "step": 31470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.984882374700352e-05,
+      "loss": 0.0214,
+      "step": 31480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.984877570726505e-05,
+      "loss": 0.0207,
+      "step": 31490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.984872766752658e-05,
+      "loss": 0.022,
+      "step": 31500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.984867962778811e-05,
+      "loss": 0.0208,
+      "step": 31510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9848631588049635e-05,
+      "loss": 0.0223,
+      "step": 31520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9848583548311164e-05,
+      "loss": 0.0202,
+      "step": 31530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9848535508572694e-05,
+      "loss": 0.0225,
+      "step": 31540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.984848746883422e-05,
+      "loss": 0.0259,
+      "step": 31550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.984843942909575e-05,
+      "loss": 0.0213,
+      "step": 31560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9848391389357278e-05,
+      "loss": 0.0158,
+      "step": 31570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9848343349618807e-05,
+      "loss": 0.0249,
+      "step": 31580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9848295309880333e-05,
+      "loss": 0.0218,
+      "step": 31590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9848247270141862e-05,
+      "loss": 0.0216,
+      "step": 31600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9848199230403392e-05,
+      "loss": 0.0231,
+      "step": 31610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984815119066492e-05,
+      "loss": 0.0206,
+      "step": 31620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9848103150926447e-05,
+      "loss": 0.0232,
+      "step": 31630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9848055111187976e-05,
+      "loss": 0.0148,
+      "step": 31640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9848007071449505e-05,
+      "loss": 0.0214,
+      "step": 31650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984795903171103e-05,
+      "loss": 0.0177,
+      "step": 31660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984791099197256e-05,
+      "loss": 0.0257,
+      "step": 31670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984786295223409e-05,
+      "loss": 0.0219,
+      "step": 31680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984781491249562e-05,
+      "loss": 0.0178,
+      "step": 31690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9847766872757145e-05,
+      "loss": 0.0198,
+      "step": 31700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9847718833018674e-05,
+      "loss": 0.0214,
+      "step": 31710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9847670793280204e-05,
+      "loss": 0.0208,
+      "step": 31720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984762275354173e-05,
+      "loss": 0.0203,
+      "step": 31730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984757471380326e-05,
+      "loss": 0.0228,
+      "step": 31740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9847526674064788e-05,
+      "loss": 0.0188,
+      "step": 31750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9847478634326317e-05,
+      "loss": 0.0179,
+      "step": 31760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9847430594587843e-05,
+      "loss": 0.0234,
+      "step": 31770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9847382554849372e-05,
+      "loss": 0.0252,
+      "step": 31780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.98473345151109e-05,
+      "loss": 0.0235,
+      "step": 31790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984728647537243e-05,
+      "loss": 0.0237,
+      "step": 31800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9847238435633957e-05,
+      "loss": 0.018,
+      "step": 31810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9847190395895486e-05,
+      "loss": 0.0222,
+      "step": 31820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9847142356157015e-05,
+      "loss": 0.0264,
+      "step": 31830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984709431641854e-05,
+      "loss": 0.0233,
+      "step": 31840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984704627668007e-05,
+      "loss": 0.0232,
+      "step": 31850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.98469982369416e-05,
+      "loss": 0.0178,
+      "step": 31860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984695019720313e-05,
+      "loss": 0.0235,
+      "step": 31870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9846902157464655e-05,
+      "loss": 0.0216,
+      "step": 31880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9846854117726184e-05,
+      "loss": 0.0219,
+      "step": 31890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9846806077987713e-05,
+      "loss": 0.0251,
+      "step": 31900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984675803824924e-05,
+      "loss": 0.0283,
+      "step": 31910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984670999851077e-05,
+      "loss": 0.0205,
+      "step": 31920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9846661958772298e-05,
+      "loss": 0.019,
+      "step": 31930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9846613919033827e-05,
+      "loss": 0.0186,
+      "step": 31940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9846565879295353e-05,
+      "loss": 0.0256,
+      "step": 31950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9846517839556882e-05,
+      "loss": 0.0202,
+      "step": 31960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984646979981841e-05,
+      "loss": 0.0245,
+      "step": 31970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9846421760079937e-05,
+      "loss": 0.0204,
+      "step": 31980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9846373720341467e-05,
+      "loss": 0.0221,
+      "step": 31990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9846325680602996e-05,
+      "loss": 0.0196,
+      "step": 32000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9846277640864525e-05,
+      "loss": 0.0193,
+      "step": 32010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984622960112605e-05,
+      "loss": 0.0239,
+      "step": 32020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984618156138758e-05,
+      "loss": 0.0291,
+      "step": 32030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984613352164911e-05,
+      "loss": 0.025,
+      "step": 32040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984608548191064e-05,
+      "loss": 0.0243,
+      "step": 32050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9846037442172165e-05,
+      "loss": 0.0216,
+      "step": 32060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845989402433694e-05,
+      "loss": 0.0214,
+      "step": 32070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845941362695223e-05,
+      "loss": 0.0218,
+      "step": 32080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984589332295675e-05,
+      "loss": 0.0323,
+      "step": 32090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984584528321828e-05,
+      "loss": 0.0248,
+      "step": 32100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845797243479808e-05,
+      "loss": 0.0232,
+      "step": 32110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845749203741337e-05,
+      "loss": 0.0276,
+      "step": 32120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845701164002863e-05,
+      "loss": 0.0273,
+      "step": 32130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845653124264392e-05,
+      "loss": 0.0186,
+      "step": 32140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984560508452592e-05,
+      "loss": 0.0224,
+      "step": 32150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845557044787447e-05,
+      "loss": 0.027,
+      "step": 32160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845509005048977e-05,
+      "loss": 0.0191,
+      "step": 32170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845460965310506e-05,
+      "loss": 0.0218,
+      "step": 32180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845412925572035e-05,
+      "loss": 0.0231,
+      "step": 32190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984536488583356e-05,
+      "loss": 0.0251,
+      "step": 32200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984531684609509e-05,
+      "loss": 0.0264,
+      "step": 32210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984526880635662e-05,
+      "loss": 0.0245,
+      "step": 32220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.984522076661815e-05,
+      "loss": 0.0211,
+      "step": 32230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845172726879675e-05,
+      "loss": 0.0216,
+      "step": 32240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845124687141204e-05,
+      "loss": 0.0224,
+      "step": 32250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9845076647402733e-05,
+      "loss": 0.0208,
+      "step": 32260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984502860766426e-05,
+      "loss": 0.0173,
+      "step": 32270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984498056792579e-05,
+      "loss": 0.0251,
+      "step": 32280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844932528187318e-05,
+      "loss": 0.022,
+      "step": 32290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844884488448847e-05,
+      "loss": 0.0293,
+      "step": 32300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844836448710373e-05,
+      "loss": 0.0184,
+      "step": 32310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844788408971902e-05,
+      "loss": 0.0205,
+      "step": 32320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984474036923343e-05,
+      "loss": 0.0142,
+      "step": 32330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844692329494957e-05,
+      "loss": 0.0245,
+      "step": 32340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844644289756487e-05,
+      "loss": 0.0274,
+      "step": 32350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844596250018016e-05,
+      "loss": 0.0242,
+      "step": 32360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844548210279545e-05,
+      "loss": 0.0275,
+      "step": 32370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984450017054107e-05,
+      "loss": 0.0181,
+      "step": 32380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.98444521308026e-05,
+      "loss": 0.0206,
+      "step": 32390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984440409106413e-05,
+      "loss": 0.0259,
+      "step": 32400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984435605132566e-05,
+      "loss": 0.0195,
+      "step": 32410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844308011587185e-05,
+      "loss": 0.0221,
+      "step": 32420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844259971848714e-05,
+      "loss": 0.0219,
+      "step": 32430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844211932110243e-05,
+      "loss": 0.0194,
+      "step": 32440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984416389237177e-05,
+      "loss": 0.0228,
+      "step": 32450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.98441158526333e-05,
+      "loss": 0.0256,
+      "step": 32460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844067812894828e-05,
+      "loss": 0.028,
+      "step": 32470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9844019773156357e-05,
+      "loss": 0.0244,
+      "step": 32480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843971733417883e-05,
+      "loss": 0.0158,
+      "step": 32490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843923693679412e-05,
+      "loss": 0.0182,
+      "step": 32500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984387565394094e-05,
+      "loss": 0.0248,
+      "step": 32510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843827614202467e-05,
+      "loss": 0.0192,
+      "step": 32520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843779574463996e-05,
+      "loss": 0.0257,
+      "step": 32530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843736338699375e-05,
+      "loss": 0.0232,
+      "step": 32540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.98436882989609e-05,
+      "loss": 0.0218,
+      "step": 32550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984364025922243e-05,
+      "loss": 0.023,
+      "step": 32560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984359221948396e-05,
+      "loss": 0.0241,
+      "step": 32570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843544179745485e-05,
+      "loss": 0.0243,
+      "step": 32580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843496140007014e-05,
+      "loss": 0.0182,
+      "step": 32590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843448100268543e-05,
+      "loss": 0.0191,
+      "step": 32600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843400060530073e-05,
+      "loss": 0.0206,
+      "step": 32610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.98433520207916e-05,
+      "loss": 0.0229,
+      "step": 32620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843303981053128e-05,
+      "loss": 0.0204,
+      "step": 32630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843255941314657e-05,
+      "loss": 0.0218,
+      "step": 32640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843207901576186e-05,
+      "loss": 0.0274,
+      "step": 32650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843159861837712e-05,
+      "loss": 0.023,
+      "step": 32660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984311182209924e-05,
+      "loss": 0.024,
+      "step": 32670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984306378236077e-05,
+      "loss": 0.037,
+      "step": 32680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9843015742622297e-05,
+      "loss": 0.0188,
+      "step": 32690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842967702883826e-05,
+      "loss": 0.0245,
+      "step": 32700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842919663145355e-05,
+      "loss": 0.0239,
+      "step": 32710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842871623406884e-05,
+      "loss": 0.019,
+      "step": 32720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984282358366841e-05,
+      "loss": 0.0266,
+      "step": 32730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984277554392994e-05,
+      "loss": 0.0239,
+      "step": 32740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984272750419147e-05,
+      "loss": 0.0247,
+      "step": 32750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842679464452995e-05,
+      "loss": 0.0207,
+      "step": 32760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842631424714524e-05,
+      "loss": 0.0203,
+      "step": 32770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842583384976053e-05,
+      "loss": 0.0213,
+      "step": 32780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842535345237583e-05,
+      "loss": 0.0187,
+      "step": 32790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984248730549911e-05,
+      "loss": 0.0192,
+      "step": 32800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842439265760638e-05,
+      "loss": 0.0194,
+      "step": 32810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842391226022167e-05,
+      "loss": 0.0181,
+      "step": 32820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842343186283696e-05,
+      "loss": 0.0249,
+      "step": 32830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842295146545222e-05,
+      "loss": 0.0252,
+      "step": 32840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984224710680675e-05,
+      "loss": 0.0181,
+      "step": 32850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984219906706828e-05,
+      "loss": 0.0237,
+      "step": 32860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842151027329807e-05,
+      "loss": 0.0233,
+      "step": 32870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842102987591336e-05,
+      "loss": 0.022,
+      "step": 32880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842054947852865e-05,
+      "loss": 0.0235,
+      "step": 32890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9842006908114394e-05,
+      "loss": 0.0204,
+      "step": 32900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984195886837592e-05,
+      "loss": 0.0234,
+      "step": 32910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984191082863745e-05,
+      "loss": 0.0265,
+      "step": 32920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.984186278889898e-05,
+      "loss": 0.0179,
+      "step": 32930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9841814749160505e-05,
+      "loss": 0.0259,
+      "step": 32940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9841766709422034e-05,
+      "loss": 0.0207,
+      "step": 32950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9841718669683563e-05,
+      "loss": 0.0216,
+      "step": 32960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9841670629945092e-05,
+      "loss": 0.0201,
+      "step": 32970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9841627394180467e-05,
+      "loss": 0.0207,
+      "step": 32980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9841579354441996e-05,
+      "loss": 0.0177,
+      "step": 32990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9841531314703522e-05,
+      "loss": 0.0262,
+      "step": 33000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.984148327496505e-05,
+      "loss": 0.0189,
+      "step": 33010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.984143523522658e-05,
+      "loss": 0.02,
+      "step": 33020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.984138719548811e-05,
+      "loss": 0.0276,
+      "step": 33030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9841339155749636e-05,
+      "loss": 0.0192,
+      "step": 33040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9841291116011165e-05,
+      "loss": 0.0177,
+      "step": 33050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9841243076272694e-05,
+      "loss": 0.0209,
+      "step": 33060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9841195036534224e-05,
+      "loss": 0.0217,
+      "step": 33070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.984114699679575e-05,
+      "loss": 0.0208,
+      "step": 33080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.984109895705728e-05,
+      "loss": 0.0183,
+      "step": 33090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9841050917318808e-05,
+      "loss": 0.0259,
+      "step": 33100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9841002877580334e-05,
+      "loss": 0.0216,
+      "step": 33110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840954837841863e-05,
+      "loss": 0.0212,
+      "step": 33120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840906798103393e-05,
+      "loss": 0.0177,
+      "step": 33130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840858758364922e-05,
+      "loss": 0.0221,
+      "step": 33140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840810718626448e-05,
+      "loss": 0.0242,
+      "step": 33150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840762678887977e-05,
+      "loss": 0.0219,
+      "step": 33160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840714639149506e-05,
+      "loss": 0.0226,
+      "step": 33170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840666599411032e-05,
+      "loss": 0.0223,
+      "step": 33180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.984061855967256e-05,
+      "loss": 0.0219,
+      "step": 33190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.984057051993409e-05,
+      "loss": 0.0247,
+      "step": 33200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.984052248019562e-05,
+      "loss": 0.0188,
+      "step": 33210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840474440457146e-05,
+      "loss": 0.0263,
+      "step": 33220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840426400718675e-05,
+      "loss": 0.0231,
+      "step": 33230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840378360980204e-05,
+      "loss": 0.0262,
+      "step": 33240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.984033032124173e-05,
+      "loss": 0.0158,
+      "step": 33250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.984028228150326e-05,
+      "loss": 0.0218,
+      "step": 33260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.984023424176479e-05,
+      "loss": 0.0228,
+      "step": 33270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840186202026318e-05,
+      "loss": 0.0225,
+      "step": 33280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840138162287844e-05,
+      "loss": 0.02,
+      "step": 33290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840090122549373e-05,
+      "loss": 0.0215,
+      "step": 33300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9840042082810903e-05,
+      "loss": 0.0212,
+      "step": 33310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839994043072432e-05,
+      "loss": 0.0241,
+      "step": 33320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839946003333958e-05,
+      "loss": 0.0186,
+      "step": 33330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839897963595487e-05,
+      "loss": 0.0235,
+      "step": 33340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839849923857016e-05,
+      "loss": 0.0166,
+      "step": 33350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839801884118542e-05,
+      "loss": 0.0259,
+      "step": 33360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.983975384438007e-05,
+      "loss": 0.0246,
+      "step": 33370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.98397058046416e-05,
+      "loss": 0.0232,
+      "step": 33380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.983965776490313e-05,
+      "loss": 0.0259,
+      "step": 33390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839609725164656e-05,
+      "loss": 0.024,
+      "step": 33400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839561685426185e-05,
+      "loss": 0.0181,
+      "step": 33410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839513645687714e-05,
+      "loss": 0.0267,
+      "step": 33420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.983946560594924e-05,
+      "loss": 0.0218,
+      "step": 33430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.983941756621077e-05,
+      "loss": 0.0257,
+      "step": 33440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.98393695264723e-05,
+      "loss": 0.0251,
+      "step": 33450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839321486733828e-05,
+      "loss": 0.0238,
+      "step": 33460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839273446995354e-05,
+      "loss": 0.0183,
+      "step": 33470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839225407256883e-05,
+      "loss": 0.0245,
+      "step": 33480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839177367518412e-05,
+      "loss": 0.0239,
+      "step": 33490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839129327779942e-05,
+      "loss": 0.0248,
+      "step": 33500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839081288041468e-05,
+      "loss": 0.0174,
+      "step": 33510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9839033248302997e-05,
+      "loss": 0.0276,
+      "step": 33520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9838985208564526e-05,
+      "loss": 0.0141,
+      "step": 33530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9838937168826052e-05,
+      "loss": 0.0146,
+      "step": 33540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.983888912908758e-05,
+      "loss": 0.0207,
+      "step": 33550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.983884108934911e-05,
+      "loss": 0.0207,
+      "step": 33560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.983879304961064e-05,
+      "loss": 0.0187,
+      "step": 33570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9838745009872166e-05,
+      "loss": 0.0211,
+      "step": 33580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9838696970133695e-05,
+      "loss": 0.0186,
+      "step": 33590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9838648930395224e-05,
+      "loss": 0.0247,
+      "step": 33600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9838600890656753e-05,
+      "loss": 0.0221,
+      "step": 33610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9838552850918283e-05,
+      "loss": 0.0154,
+      "step": 33620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9838504811179812e-05,
+      "loss": 0.0254,
+      "step": 33630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9838456771441338e-05,
+      "loss": 0.0227,
+      "step": 33640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9838408731702867e-05,
+      "loss": 0.0196,
+      "step": 33650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9838360691964396e-05,
+      "loss": 0.0272,
+      "step": 33660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9838312652225922e-05,
+      "loss": 0.0226,
+      "step": 33670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983826461248745e-05,
+      "loss": 0.0205,
+      "step": 33680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983821657274898e-05,
+      "loss": 0.0196,
+      "step": 33690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983816853301051e-05,
+      "loss": 0.0259,
+      "step": 33700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9838120493272036e-05,
+      "loss": 0.0206,
+      "step": 33710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9838072453533565e-05,
+      "loss": 0.0199,
+      "step": 33720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9838024413795095e-05,
+      "loss": 0.0228,
+      "step": 33730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837976374056624e-05,
+      "loss": 0.0187,
+      "step": 33740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983792833431815e-05,
+      "loss": 0.0187,
+      "step": 33750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983788029457968e-05,
+      "loss": 0.0183,
+      "step": 33760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837832254841208e-05,
+      "loss": 0.0186,
+      "step": 33770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837784215102734e-05,
+      "loss": 0.0213,
+      "step": 33780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837736175364263e-05,
+      "loss": 0.0162,
+      "step": 33790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837688135625793e-05,
+      "loss": 0.0194,
+      "step": 33800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837640095887322e-05,
+      "loss": 0.0212,
+      "step": 33810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837592056148848e-05,
+      "loss": 0.0252,
+      "step": 33820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837544016410377e-05,
+      "loss": 0.0226,
+      "step": 33830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837495976671906e-05,
+      "loss": 0.0205,
+      "step": 33840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837447936933432e-05,
+      "loss": 0.0172,
+      "step": 33850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983739989719496e-05,
+      "loss": 0.0239,
+      "step": 33860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983735185745649e-05,
+      "loss": 0.0229,
+      "step": 33870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983730381771802e-05,
+      "loss": 0.0154,
+      "step": 33880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837255777979546e-05,
+      "loss": 0.0215,
+      "step": 33890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837207738241075e-05,
+      "loss": 0.0212,
+      "step": 33900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837159698502604e-05,
+      "loss": 0.0229,
+      "step": 33910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9837111658764134e-05,
+      "loss": 0.0242,
+      "step": 33920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983706361902566e-05,
+      "loss": 0.0237,
+      "step": 33930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983701557928719e-05,
+      "loss": 0.0187,
+      "step": 33940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836967539548718e-05,
+      "loss": 0.0174,
+      "step": 33950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836919499810244e-05,
+      "loss": 0.0242,
+      "step": 33960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836871460071773e-05,
+      "loss": 0.0165,
+      "step": 33970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836823420333303e-05,
+      "loss": 0.0201,
+      "step": 33980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836775380594832e-05,
+      "loss": 0.0204,
+      "step": 33990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836727340856358e-05,
+      "loss": 0.0248,
+      "step": 34000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836679301117887e-05,
+      "loss": 0.0208,
+      "step": 34010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836631261379416e-05,
+      "loss": 0.0226,
+      "step": 34020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836583221640942e-05,
+      "loss": 0.0184,
+      "step": 34030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983653518190247e-05,
+      "loss": 0.0212,
+      "step": 34040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836487142164e-05,
+      "loss": 0.0228,
+      "step": 34050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983643910242553e-05,
+      "loss": 0.0202,
+      "step": 34060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836391062687056e-05,
+      "loss": 0.0231,
+      "step": 34070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836343022948585e-05,
+      "loss": 0.0178,
+      "step": 34080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836294983210114e-05,
+      "loss": 0.0262,
+      "step": 34090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836246943471644e-05,
+      "loss": 0.0218,
+      "step": 34100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983619890373317e-05,
+      "loss": 0.0197,
+      "step": 34110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.98361508639947e-05,
+      "loss": 0.0212,
+      "step": 34120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836102824256228e-05,
+      "loss": 0.0219,
+      "step": 34130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836054784517754e-05,
+      "loss": 0.0211,
+      "step": 34140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836006744779283e-05,
+      "loss": 0.0239,
+      "step": 34150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835958705040813e-05,
+      "loss": 0.0237,
+      "step": 34160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835910665302342e-05,
+      "loss": 0.0231,
+      "step": 34170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835862625563868e-05,
+      "loss": 0.0225,
+      "step": 34180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835814585825397e-05,
+      "loss": 0.0176,
+      "step": 34190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835766546086926e-05,
+      "loss": 0.0157,
+      "step": 34200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835718506348452e-05,
+      "loss": 0.024,
+      "step": 34210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983567046660998e-05,
+      "loss": 0.0263,
+      "step": 34220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983562242687151e-05,
+      "loss": 0.0251,
+      "step": 34230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983557438713304e-05,
+      "loss": 0.0193,
+      "step": 34240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835526347394566e-05,
+      "loss": 0.0208,
+      "step": 34250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835478307656095e-05,
+      "loss": 0.0246,
+      "step": 34260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835430267917624e-05,
+      "loss": 0.0172,
+      "step": 34270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835382228179154e-05,
+      "loss": 0.0216,
+      "step": 34280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983533418844068e-05,
+      "loss": 0.0278,
+      "step": 34290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983528614870221e-05,
+      "loss": 0.0261,
+      "step": 34300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835238108963738e-05,
+      "loss": 0.018,
+      "step": 34310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835190069225264e-05,
+      "loss": 0.0245,
+      "step": 34320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835142029486793e-05,
+      "loss": 0.0201,
+      "step": 34330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835093989748322e-05,
+      "loss": 0.0201,
+      "step": 34340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9835045950009852e-05,
+      "loss": 0.0209,
+      "step": 34350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834997910271378e-05,
+      "loss": 0.0268,
+      "step": 34360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834949870532907e-05,
+      "loss": 0.0176,
+      "step": 34370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834901830794436e-05,
+      "loss": 0.0174,
+      "step": 34380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834853791055962e-05,
+      "loss": 0.0282,
+      "step": 34390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983480575131749e-05,
+      "loss": 0.028,
+      "step": 34400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983475771157902e-05,
+      "loss": 0.0199,
+      "step": 34410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983470967184055e-05,
+      "loss": 0.0188,
+      "step": 34420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834661632102076e-05,
+      "loss": 0.0187,
+      "step": 34430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834613592363605e-05,
+      "loss": 0.0224,
+      "step": 34440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834565552625134e-05,
+      "loss": 0.0206,
+      "step": 34450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834517512886664e-05,
+      "loss": 0.0191,
+      "step": 34460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983446947314819e-05,
+      "loss": 0.0217,
+      "step": 34470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983442143340972e-05,
+      "loss": 0.021,
+      "step": 34480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834373393671248e-05,
+      "loss": 0.0239,
+      "step": 34490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834325353932774e-05,
+      "loss": 0.021,
+      "step": 34500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834277314194303e-05,
+      "loss": 0.0235,
+      "step": 34510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834229274455832e-05,
+      "loss": 0.0197,
+      "step": 34520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983418123471736e-05,
+      "loss": 0.0216,
+      "step": 34530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834133194978887e-05,
+      "loss": 0.0199,
+      "step": 34540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834085155240417e-05,
+      "loss": 0.0191,
+      "step": 34550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9834037115501946e-05,
+      "loss": 0.0156,
+      "step": 34560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833989075763472e-05,
+      "loss": 0.0242,
+      "step": 34570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833941036025e-05,
+      "loss": 0.0173,
+      "step": 34580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983389299628653e-05,
+      "loss": 0.0191,
+      "step": 34590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983384495654806e-05,
+      "loss": 0.0222,
+      "step": 34600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833796916809586e-05,
+      "loss": 0.0217,
+      "step": 34610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833748877071115e-05,
+      "loss": 0.0182,
+      "step": 34620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833700837332644e-05,
+      "loss": 0.019,
+      "step": 34630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833652797594173e-05,
+      "loss": 0.0163,
+      "step": 34640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.98336047578557e-05,
+      "loss": 0.0174,
+      "step": 34650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983355671811723e-05,
+      "loss": 0.0227,
+      "step": 34660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833508678378758e-05,
+      "loss": 0.0249,
+      "step": 34670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833460638640284e-05,
+      "loss": 0.0162,
+      "step": 34680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833412598901813e-05,
+      "loss": 0.0231,
+      "step": 34690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833364559163342e-05,
+      "loss": 0.0191,
+      "step": 34700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983331651942487e-05,
+      "loss": 0.0205,
+      "step": 34710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833268479686397e-05,
+      "loss": 0.0197,
+      "step": 34720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833220439947927e-05,
+      "loss": 0.0213,
+      "step": 34730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833172400209456e-05,
+      "loss": 0.0208,
+      "step": 34740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9833124360470982e-05,
+      "loss": 0.0223,
+      "step": 34750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983307632073251e-05,
+      "loss": 0.0197,
+      "step": 34760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983302828099404e-05,
+      "loss": 0.0233,
+      "step": 34770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983298024125557e-05,
+      "loss": 0.0233,
+      "step": 34780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832932201517096e-05,
+      "loss": 0.0168,
+      "step": 34790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832884161778625e-05,
+      "loss": 0.0204,
+      "step": 34800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832836122040154e-05,
+      "loss": 0.0125,
+      "step": 34810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983278808230168e-05,
+      "loss": 0.0179,
+      "step": 34820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983274004256321e-05,
+      "loss": 0.0256,
+      "step": 34830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983269200282474e-05,
+      "loss": 0.024,
+      "step": 34840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832643963086268e-05,
+      "loss": 0.0167,
+      "step": 34850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832595923347794e-05,
+      "loss": 0.0259,
+      "step": 34860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832547883609323e-05,
+      "loss": 0.0199,
+      "step": 34870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832499843870852e-05,
+      "loss": 0.0215,
+      "step": 34880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983245180413238e-05,
+      "loss": 0.0252,
+      "step": 34890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832403764393907e-05,
+      "loss": 0.0254,
+      "step": 34900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832355724655437e-05,
+      "loss": 0.0229,
+      "step": 34910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832307684916966e-05,
+      "loss": 0.027,
+      "step": 34920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832259645178492e-05,
+      "loss": 0.0204,
+      "step": 34930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983221160544002e-05,
+      "loss": 0.0225,
+      "step": 34940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983216356570155e-05,
+      "loss": 0.0177,
+      "step": 34950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983211552596308e-05,
+      "loss": 0.0206,
+      "step": 34960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832067486224605e-05,
+      "loss": 0.0219,
+      "step": 34970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9832019446486135e-05,
+      "loss": 0.0259,
+      "step": 34980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9831971406747664e-05,
+      "loss": 0.0231,
+      "step": 34990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983192336700919e-05,
+      "loss": 0.0248,
+      "step": 35000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983187532727072e-05,
+      "loss": 0.0261,
+      "step": 35010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983182728753225e-05,
+      "loss": 0.028,
+      "step": 35020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9831779247793778e-05,
+      "loss": 0.021,
+      "step": 35030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9831731208055304e-05,
+      "loss": 0.0224,
+      "step": 35040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9831683168316833e-05,
+      "loss": 0.0219,
+      "step": 35050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9831635128578362e-05,
+      "loss": 0.0217,
+      "step": 35060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.983158708883989e-05,
+      "loss": 0.0189,
+      "step": 35070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9831539049101417e-05,
+      "loss": 0.0259,
+      "step": 35080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9831491009362947e-05,
+      "loss": 0.0184,
+      "step": 35090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9831442969624476e-05,
+      "loss": 0.0195,
+      "step": 35100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9831394929886e-05,
+      "loss": 0.0215,
+      "step": 35110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.983134689014753e-05,
+      "loss": 0.017,
+      "step": 35120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.983129885040906e-05,
+      "loss": 0.0187,
+      "step": 35130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.983125081067059e-05,
+      "loss": 0.0289,
+      "step": 35140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9831202770932115e-05,
+      "loss": 0.0199,
+      "step": 35150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9831154731193645e-05,
+      "loss": 0.0192,
+      "step": 35160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9831106691455174e-05,
+      "loss": 0.0217,
+      "step": 35170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.98310586517167e-05,
+      "loss": 0.0236,
+      "step": 35180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.983101061197823e-05,
+      "loss": 0.0235,
+      "step": 35190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.983096257223976e-05,
+      "loss": 0.0235,
+      "step": 35200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830914532501288e-05,
+      "loss": 0.0201,
+      "step": 35210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830866492762813e-05,
+      "loss": 0.0218,
+      "step": 35220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830818453024343e-05,
+      "loss": 0.0204,
+      "step": 35230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830770413285872e-05,
+      "loss": 0.0194,
+      "step": 35240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.98307223735474e-05,
+      "loss": 0.02,
+      "step": 35250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830674333808927e-05,
+      "loss": 0.021,
+      "step": 35260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830626294070456e-05,
+      "loss": 0.0203,
+      "step": 35270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830578254331986e-05,
+      "loss": 0.0196,
+      "step": 35280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.983053021459351e-05,
+      "loss": 0.019,
+      "step": 35290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.983048217485504e-05,
+      "loss": 0.0225,
+      "step": 35300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.983043413511657e-05,
+      "loss": 0.0198,
+      "step": 35310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.98303860953781e-05,
+      "loss": 0.0205,
+      "step": 35320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830338055639625e-05,
+      "loss": 0.0247,
+      "step": 35330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830290015901155e-05,
+      "loss": 0.0208,
+      "step": 35340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830241976162684e-05,
+      "loss": 0.0161,
+      "step": 35350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.983019393642421e-05,
+      "loss": 0.019,
+      "step": 35360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.983014589668574e-05,
+      "loss": 0.0212,
+      "step": 35370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830097856947268e-05,
+      "loss": 0.0268,
+      "step": 35380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830049817208797e-05,
+      "loss": 0.0216,
+      "step": 35390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9830001777470323e-05,
+      "loss": 0.0214,
+      "step": 35400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829953737731853e-05,
+      "loss": 0.0224,
+      "step": 35410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829905697993382e-05,
+      "loss": 0.0232,
+      "step": 35420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982985765825491e-05,
+      "loss": 0.0158,
+      "step": 35430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829809618516437e-05,
+      "loss": 0.0216,
+      "step": 35440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829761578777966e-05,
+      "loss": 0.0174,
+      "step": 35450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829713539039496e-05,
+      "loss": 0.019,
+      "step": 35460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982966549930102e-05,
+      "loss": 0.018,
+      "step": 35470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982961745956255e-05,
+      "loss": 0.0167,
+      "step": 35480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982956941982408e-05,
+      "loss": 0.0255,
+      "step": 35490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982952138008561e-05,
+      "loss": 0.0193,
+      "step": 35500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829473340347135e-05,
+      "loss": 0.0188,
+      "step": 35510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829425300608664e-05,
+      "loss": 0.0233,
+      "step": 35520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829377260870194e-05,
+      "loss": 0.0211,
+      "step": 35530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982932922113172e-05,
+      "loss": 0.0263,
+      "step": 35540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982928118139325e-05,
+      "loss": 0.0159,
+      "step": 35550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829233141654778e-05,
+      "loss": 0.0193,
+      "step": 35560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829185101916307e-05,
+      "loss": 0.0195,
+      "step": 35570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829137062177833e-05,
+      "loss": 0.0156,
+      "step": 35580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829089022439363e-05,
+      "loss": 0.0214,
+      "step": 35590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9829040982700892e-05,
+      "loss": 0.0227,
+      "step": 35600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982899294296242e-05,
+      "loss": 0.0176,
+      "step": 35610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9828944903223947e-05,
+      "loss": 0.0172,
+      "step": 35620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9828896863485476e-05,
+      "loss": 0.0185,
+      "step": 35630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9828848823747006e-05,
+      "loss": 0.0204,
+      "step": 35640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982880078400853e-05,
+      "loss": 0.0223,
+      "step": 35650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982875274427006e-05,
+      "loss": 0.0188,
+      "step": 35660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982870470453159e-05,
+      "loss": 0.0158,
+      "step": 35670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982865666479312e-05,
+      "loss": 0.0172,
+      "step": 35680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9828608625054645e-05,
+      "loss": 0.0197,
+      "step": 35690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9828560585316174e-05,
+      "loss": 0.024,
+      "step": 35700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9828512545577704e-05,
+      "loss": 0.0242,
+      "step": 35710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982846450583923e-05,
+      "loss": 0.0228,
+      "step": 35720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982841646610076e-05,
+      "loss": 0.0221,
+      "step": 35730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9828368426362288e-05,
+      "loss": 0.019,
+      "step": 35740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9828320386623817e-05,
+      "loss": 0.0202,
+      "step": 35750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9828272346885343e-05,
+      "loss": 0.0252,
+      "step": 35760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9828224307146872e-05,
+      "loss": 0.0196,
+      "step": 35770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9828176267408402e-05,
+      "loss": 0.0241,
+      "step": 35780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982812822766993e-05,
+      "loss": 0.0269,
+      "step": 35790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9828080187931457e-05,
+      "loss": 0.025,
+      "step": 35800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9828032148192986e-05,
+      "loss": 0.0208,
+      "step": 35810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9827984108454515e-05,
+      "loss": 0.0243,
+      "step": 35820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982793606871604e-05,
+      "loss": 0.0174,
+      "step": 35830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982788802897757e-05,
+      "loss": 0.0194,
+      "step": 35840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.98278399892391e-05,
+      "loss": 0.0243,
+      "step": 35850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982779194950063e-05,
+      "loss": 0.0261,
+      "step": 35860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9827743909762155e-05,
+      "loss": 0.0231,
+      "step": 35870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9827695870023684e-05,
+      "loss": 0.0228,
+      "step": 35880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9827647830285214e-05,
+      "loss": 0.031,
+      "step": 35890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982759979054674e-05,
+      "loss": 0.0237,
+      "step": 35900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982755175080827e-05,
+      "loss": 0.0159,
+      "step": 35910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9827503711069798e-05,
+      "loss": 0.0198,
+      "step": 35920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9827455671331327e-05,
+      "loss": 0.0172,
+      "step": 35930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9827407631592853e-05,
+      "loss": 0.0175,
+      "step": 35940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9827359591854382e-05,
+      "loss": 0.0209,
+      "step": 35950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982731155211591e-05,
+      "loss": 0.0223,
+      "step": 35960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982726351237744e-05,
+      "loss": 0.021,
+      "step": 35970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9827215472638967e-05,
+      "loss": 0.0172,
+      "step": 35980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9827167432900496e-05,
+      "loss": 0.0189,
+      "step": 35990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9827119393162025e-05,
+      "loss": 0.0179,
+      "step": 36000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982707135342355e-05,
+      "loss": 0.0204,
+      "step": 36010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982702331368508e-05,
+      "loss": 0.017,
+      "step": 36020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982697527394661e-05,
+      "loss": 0.023,
+      "step": 36030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982692723420814e-05,
+      "loss": 0.0225,
+      "step": 36040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9826879194469665e-05,
+      "loss": 0.0227,
+      "step": 36050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9826831154731194e-05,
+      "loss": 0.0252,
+      "step": 36060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9826783114992723e-05,
+      "loss": 0.0177,
+      "step": 36070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982673507525425e-05,
+      "loss": 0.0253,
+      "step": 36080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982668703551578e-05,
+      "loss": 0.0154,
+      "step": 36090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9826638995777308e-05,
+      "loss": 0.0187,
+      "step": 36100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9826590956038837e-05,
+      "loss": 0.0149,
+      "step": 36110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9826542916300363e-05,
+      "loss": 0.0162,
+      "step": 36120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9826494876561892e-05,
+      "loss": 0.0228,
+      "step": 36130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982644683682342e-05,
+      "loss": 0.0244,
+      "step": 36140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982639879708495e-05,
+      "loss": 0.0208,
+      "step": 36150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982635075734648e-05,
+      "loss": 0.0264,
+      "step": 36160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982630271760801e-05,
+      "loss": 0.0255,
+      "step": 36170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9826254677869535e-05,
+      "loss": 0.0198,
+      "step": 36180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9826206638131065e-05,
+      "loss": 0.0185,
+      "step": 36190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9826158598392594e-05,
+      "loss": 0.0167,
+      "step": 36200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982611055865412e-05,
+      "loss": 0.0161,
+      "step": 36210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982606251891565e-05,
+      "loss": 0.019,
+      "step": 36220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9826014479177178e-05,
+      "loss": 0.0173,
+      "step": 36230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9825966439438707e-05,
+      "loss": 0.0174,
+      "step": 36240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9825918399700233e-05,
+      "loss": 0.0209,
+      "step": 36250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9825870359961763e-05,
+      "loss": 0.0188,
+      "step": 36260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9825822320223292e-05,
+      "loss": 0.0184,
+      "step": 36270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982577428048482e-05,
+      "loss": 0.0246,
+      "step": 36280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9825726240746347e-05,
+      "loss": 0.0302,
+      "step": 36290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9825678201007876e-05,
+      "loss": 0.0224,
+      "step": 36300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9825630161269406e-05,
+      "loss": 0.0244,
+      "step": 36310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982558212153093e-05,
+      "loss": 0.0218,
+      "step": 36320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982553408179246e-05,
+      "loss": 0.0207,
+      "step": 36330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982548604205399e-05,
+      "loss": 0.0224,
+      "step": 36340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982543800231552e-05,
+      "loss": 0.0169,
+      "step": 36350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9825389962577045e-05,
+      "loss": 0.019,
+      "step": 36360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9825341922838574e-05,
+      "loss": 0.0219,
+      "step": 36370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9825293883100104e-05,
+      "loss": 0.0198,
+      "step": 36380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982524584336163e-05,
+      "loss": 0.0174,
+      "step": 36390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982519780362316e-05,
+      "loss": 0.025,
+      "step": 36400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9825149763884688e-05,
+      "loss": 0.0193,
+      "step": 36410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9825101724146217e-05,
+      "loss": 0.0222,
+      "step": 36420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9825053684407743e-05,
+      "loss": 0.0181,
+      "step": 36430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9825005644669273e-05,
+      "loss": 0.0232,
+      "step": 36440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824957604930802e-05,
+      "loss": 0.0251,
+      "step": 36450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982490956519233e-05,
+      "loss": 0.022,
+      "step": 36460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824861525453857e-05,
+      "loss": 0.0182,
+      "step": 36470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824813485715386e-05,
+      "loss": 0.024,
+      "step": 36480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824765445976916e-05,
+      "loss": 0.0202,
+      "step": 36490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982471740623844e-05,
+      "loss": 0.0236,
+      "step": 36500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982466936649997e-05,
+      "loss": 0.0179,
+      "step": 36510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.98246213267615e-05,
+      "loss": 0.0197,
+      "step": 36520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982457328702303e-05,
+      "loss": 0.0246,
+      "step": 36530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824525247284555e-05,
+      "loss": 0.025,
+      "step": 36540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824477207546084e-05,
+      "loss": 0.0251,
+      "step": 36550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824429167807614e-05,
+      "loss": 0.0181,
+      "step": 36560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982438112806914e-05,
+      "loss": 0.0176,
+      "step": 36570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982433308833067e-05,
+      "loss": 0.0248,
+      "step": 36580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824285048592198e-05,
+      "loss": 0.0234,
+      "step": 36590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824237008853727e-05,
+      "loss": 0.0215,
+      "step": 36600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824188969115253e-05,
+      "loss": 0.0188,
+      "step": 36610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824140929376782e-05,
+      "loss": 0.0188,
+      "step": 36620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9824092889638312e-05,
+      "loss": 0.02,
+      "step": 36630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982404484989984e-05,
+      "loss": 0.023,
+      "step": 36640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823996810161367e-05,
+      "loss": 0.0201,
+      "step": 36650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823948770422896e-05,
+      "loss": 0.0177,
+      "step": 36660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823900730684425e-05,
+      "loss": 0.0273,
+      "step": 36670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982385269094595e-05,
+      "loss": 0.0204,
+      "step": 36680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982380465120748e-05,
+      "loss": 0.0177,
+      "step": 36690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982375661146901e-05,
+      "loss": 0.0235,
+      "step": 36700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982370857173054e-05,
+      "loss": 0.0213,
+      "step": 36710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823660531992065e-05,
+      "loss": 0.0229,
+      "step": 36720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823612492253594e-05,
+      "loss": 0.0251,
+      "step": 36730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823564452515124e-05,
+      "loss": 0.0238,
+      "step": 36740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982351641277665e-05,
+      "loss": 0.0207,
+      "step": 36750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982346837303818e-05,
+      "loss": 0.0231,
+      "step": 36760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823420333299708e-05,
+      "loss": 0.0219,
+      "step": 36770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823372293561237e-05,
+      "loss": 0.0154,
+      "step": 36780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823324253822763e-05,
+      "loss": 0.0225,
+      "step": 36790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823276214084292e-05,
+      "loss": 0.0194,
+      "step": 36800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982322817434582e-05,
+      "loss": 0.0226,
+      "step": 36810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982318013460735e-05,
+      "loss": 0.0235,
+      "step": 36820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823132094868877e-05,
+      "loss": 0.0274,
+      "step": 36830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823084055130406e-05,
+      "loss": 0.0195,
+      "step": 36840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9823036015391935e-05,
+      "loss": 0.0189,
+      "step": 36850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982298797565346e-05,
+      "loss": 0.0212,
+      "step": 36860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982293993591499e-05,
+      "loss": 0.0191,
+      "step": 36870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982289189617652e-05,
+      "loss": 0.019,
+      "step": 36880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982284385643805e-05,
+      "loss": 0.0249,
+      "step": 36890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9822795816699575e-05,
+      "loss": 0.0146,
+      "step": 36900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9822747776961104e-05,
+      "loss": 0.0184,
+      "step": 36910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9822699737222633e-05,
+      "loss": 0.0204,
+      "step": 36920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982265169748416e-05,
+      "loss": 0.0213,
+      "step": 36930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982260365774569e-05,
+      "loss": 0.0214,
+      "step": 36940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9822555618007218e-05,
+      "loss": 0.0212,
+      "step": 36950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9822507578268747e-05,
+      "loss": 0.0207,
+      "step": 36960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9822459538530273e-05,
+      "loss": 0.0191,
+      "step": 36970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9822411498791802e-05,
+      "loss": 0.0199,
+      "step": 36980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982236345905333e-05,
+      "loss": 0.0296,
+      "step": 36990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982231541931486e-05,
+      "loss": 0.022,
+      "step": 37000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9822267379576387e-05,
+      "loss": 0.0216,
+      "step": 37010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9822219339837916e-05,
+      "loss": 0.017,
+      "step": 37020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9822171300099445e-05,
+      "loss": 0.0239,
+      "step": 37030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982212326036097e-05,
+      "loss": 0.0238,
+      "step": 37040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.98220752206225e-05,
+      "loss": 0.0197,
+      "step": 37050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982202718088403e-05,
+      "loss": 0.0207,
+      "step": 37060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982197914114556e-05,
+      "loss": 0.0225,
+      "step": 37070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9821931101407085e-05,
+      "loss": 0.0177,
+      "step": 37080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9821883061668614e-05,
+      "loss": 0.0194,
+      "step": 37090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9821835021930143e-05,
+      "loss": 0.0191,
+      "step": 37100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.982178698219167e-05,
+      "loss": 0.019,
+      "step": 37110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.98217389424532e-05,
+      "loss": 0.0234,
+      "step": 37120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9821690902714728e-05,
+      "loss": 0.0189,
+      "step": 37130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9821642862976257e-05,
+      "loss": 0.0221,
+      "step": 37140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9821594823237783e-05,
+      "loss": 0.0173,
+      "step": 37150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9821546783499312e-05,
+      "loss": 0.0156,
+      "step": 37160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982149874376084e-05,
+      "loss": 0.0174,
+      "step": 37170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982145070402237e-05,
+      "loss": 0.0186,
+      "step": 37180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9821402664283897e-05,
+      "loss": 0.0187,
+      "step": 37190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9821354624545426e-05,
+      "loss": 0.0204,
+      "step": 37200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9821306584806955e-05,
+      "loss": 0.022,
+      "step": 37210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982125854506848e-05,
+      "loss": 0.0179,
+      "step": 37220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982121050533001e-05,
+      "loss": 0.0157,
+      "step": 37230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982116246559154e-05,
+      "loss": 0.0167,
+      "step": 37240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982111442585307e-05,
+      "loss": 0.0186,
+      "step": 37250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9821066386114595e-05,
+      "loss": 0.0214,
+      "step": 37260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9821018346376124e-05,
+      "loss": 0.0247,
+      "step": 37270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9820970306637653e-05,
+      "loss": 0.0173,
+      "step": 37280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982092226689918e-05,
+      "loss": 0.0174,
+      "step": 37290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982087422716071e-05,
+      "loss": 0.018,
+      "step": 37300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9820826187422238e-05,
+      "loss": 0.02,
+      "step": 37310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9820778147683767e-05,
+      "loss": 0.0212,
+      "step": 37320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9820730107945293e-05,
+      "loss": 0.0255,
+      "step": 37330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9820682068206822e-05,
+      "loss": 0.0189,
+      "step": 37340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982063402846835e-05,
+      "loss": 0.0183,
+      "step": 37350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982058598872988e-05,
+      "loss": 0.0227,
+      "step": 37360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9820537948991407e-05,
+      "loss": 0.0188,
+      "step": 37370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9820489909252936e-05,
+      "loss": 0.0222,
+      "step": 37380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9820441869514465e-05,
+      "loss": 0.0184,
+      "step": 37390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982039382977599e-05,
+      "loss": 0.0197,
+      "step": 37400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982034579003752e-05,
+      "loss": 0.0243,
+      "step": 37410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982029775029905e-05,
+      "loss": 0.022,
+      "step": 37420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982024971056058e-05,
+      "loss": 0.0174,
+      "step": 37430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9820201670822105e-05,
+      "loss": 0.0252,
+      "step": 37440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9820153631083634e-05,
+      "loss": 0.0236,
+      "step": 37450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9820105591345163e-05,
+      "loss": 0.0207,
+      "step": 37460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982005755160669e-05,
+      "loss": 0.0257,
+      "step": 37470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.982000951186822e-05,
+      "loss": 0.0204,
+      "step": 37480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819961472129748e-05,
+      "loss": 0.0237,
+      "step": 37490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819913432391277e-05,
+      "loss": 0.0195,
+      "step": 37500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819865392652803e-05,
+      "loss": 0.0207,
+      "step": 37510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819817352914332e-05,
+      "loss": 0.0198,
+      "step": 37520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.981976931317586e-05,
+      "loss": 0.0157,
+      "step": 37530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.981972127343739e-05,
+      "loss": 0.0179,
+      "step": 37540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819673233698916e-05,
+      "loss": 0.0191,
+      "step": 37550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819625193960446e-05,
+      "loss": 0.0167,
+      "step": 37560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819577154221975e-05,
+      "loss": 0.0198,
+      "step": 37570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.98195291144835e-05,
+      "loss": 0.0153,
+      "step": 37580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.981948107474503e-05,
+      "loss": 0.0189,
+      "step": 37590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.981943303500656e-05,
+      "loss": 0.0178,
+      "step": 37600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.981938499526809e-05,
+      "loss": 0.0244,
+      "step": 37610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819336955529615e-05,
+      "loss": 0.0227,
+      "step": 37620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819288915791144e-05,
+      "loss": 0.0192,
+      "step": 37630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819240876052673e-05,
+      "loss": 0.0216,
+      "step": 37640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.98191928363142e-05,
+      "loss": 0.0214,
+      "step": 37650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819144796575728e-05,
+      "loss": 0.0148,
+      "step": 37660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819096756837258e-05,
+      "loss": 0.026,
+      "step": 37670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819048717098787e-05,
+      "loss": 0.0238,
+      "step": 37680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9819000677360313e-05,
+      "loss": 0.0233,
+      "step": 37690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9818952637621842e-05,
+      "loss": 0.0292,
+      "step": 37700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.981890459788337e-05,
+      "loss": 0.024,
+      "step": 37710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.98188565581449e-05,
+      "loss": 0.0156,
+      "step": 37720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9818808518406426e-05,
+      "loss": 0.0188,
+      "step": 37730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9818760478667956e-05,
+      "loss": 0.0218,
+      "step": 37740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9818712438929485e-05,
+      "loss": 0.0188,
+      "step": 37750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.981866439919101e-05,
+      "loss": 0.0217,
+      "step": 37760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.981861635945254e-05,
+      "loss": 0.0186,
+      "step": 37770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.981856831971407e-05,
+      "loss": 0.0179,
+      "step": 37780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.98185202799756e-05,
+      "loss": 0.0215,
+      "step": 37790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9818472240237125e-05,
+      "loss": 0.0264,
+      "step": 37800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9818424200498654e-05,
+      "loss": 0.0259,
+      "step": 37810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9818376160760183e-05,
+      "loss": 0.021,
+      "step": 37820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981832812102171e-05,
+      "loss": 0.0221,
+      "step": 37830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9818280081283238e-05,
+      "loss": 0.0207,
+      "step": 37840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9818232041544767e-05,
+      "loss": 0.0187,
+      "step": 37850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9818184001806297e-05,
+      "loss": 0.0178,
+      "step": 37860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9818135962067823e-05,
+      "loss": 0.0228,
+      "step": 37870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9818087922329352e-05,
+      "loss": 0.0137,
+      "step": 37880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981803988259088e-05,
+      "loss": 0.0199,
+      "step": 37890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981799184285241e-05,
+      "loss": 0.0197,
+      "step": 37900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817943803113936e-05,
+      "loss": 0.0244,
+      "step": 37910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817895763375466e-05,
+      "loss": 0.0222,
+      "step": 37920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817847723636995e-05,
+      "loss": 0.0183,
+      "step": 37930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981779968389852e-05,
+      "loss": 0.0212,
+      "step": 37940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981775164416005e-05,
+      "loss": 0.0247,
+      "step": 37950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981770360442158e-05,
+      "loss": 0.0203,
+      "step": 37960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981765556468311e-05,
+      "loss": 0.0224,
+      "step": 37970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817607524944634e-05,
+      "loss": 0.019,
+      "step": 37980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817559485206164e-05,
+      "loss": 0.0188,
+      "step": 37990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817511445467693e-05,
+      "loss": 0.02,
+      "step": 38000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981746340572922e-05,
+      "loss": 0.0202,
+      "step": 38010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817415365990748e-05,
+      "loss": 0.0191,
+      "step": 38020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817367326252277e-05,
+      "loss": 0.0238,
+      "step": 38030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817319286513807e-05,
+      "loss": 0.0195,
+      "step": 38040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817271246775333e-05,
+      "loss": 0.0172,
+      "step": 38050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817223207036862e-05,
+      "loss": 0.0241,
+      "step": 38060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981717516729839e-05,
+      "loss": 0.0213,
+      "step": 38070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817127127559917e-05,
+      "loss": 0.0187,
+      "step": 38080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817079087821446e-05,
+      "loss": 0.0189,
+      "step": 38090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9817031048082975e-05,
+      "loss": 0.017,
+      "step": 38100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816983008344505e-05,
+      "loss": 0.0173,
+      "step": 38110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981693496860603e-05,
+      "loss": 0.0233,
+      "step": 38120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981688692886756e-05,
+      "loss": 0.0241,
+      "step": 38130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981683888912909e-05,
+      "loss": 0.0199,
+      "step": 38140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981679084939062e-05,
+      "loss": 0.0192,
+      "step": 38150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816742809652144e-05,
+      "loss": 0.0179,
+      "step": 38160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816694769913674e-05,
+      "loss": 0.0184,
+      "step": 38170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816646730175203e-05,
+      "loss": 0.0198,
+      "step": 38180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981659869043673e-05,
+      "loss": 0.0228,
+      "step": 38190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816550650698258e-05,
+      "loss": 0.0223,
+      "step": 38200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816502610959787e-05,
+      "loss": 0.0189,
+      "step": 38210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816454571221317e-05,
+      "loss": 0.0149,
+      "step": 38220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816406531482842e-05,
+      "loss": 0.0193,
+      "step": 38230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816358491744372e-05,
+      "loss": 0.0204,
+      "step": 38240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.98163104520059e-05,
+      "loss": 0.0177,
+      "step": 38250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816262412267427e-05,
+      "loss": 0.0229,
+      "step": 38260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816214372528956e-05,
+      "loss": 0.0183,
+      "step": 38270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816166332790485e-05,
+      "loss": 0.0155,
+      "step": 38280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9816118293052015e-05,
+      "loss": 0.0217,
+      "step": 38290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981607025331354e-05,
+      "loss": 0.0256,
+      "step": 38300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981602221357507e-05,
+      "loss": 0.0201,
+      "step": 38310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.98159741738366e-05,
+      "loss": 0.0189,
+      "step": 38320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981592613409813e-05,
+      "loss": 0.0215,
+      "step": 38330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9815878094359654e-05,
+      "loss": 0.0192,
+      "step": 38340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9815830054621184e-05,
+      "loss": 0.0171,
+      "step": 38350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9815782014882713e-05,
+      "loss": 0.0227,
+      "step": 38360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981573397514424e-05,
+      "loss": 0.0199,
+      "step": 38370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9815685935405768e-05,
+      "loss": 0.0201,
+      "step": 38380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9815637895667297e-05,
+      "loss": 0.0185,
+      "step": 38390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9815589855928826e-05,
+      "loss": 0.0174,
+      "step": 38400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9815541816190352e-05,
+      "loss": 0.0194,
+      "step": 38410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981549377645188e-05,
+      "loss": 0.0214,
+      "step": 38420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981544573671341e-05,
+      "loss": 0.0218,
+      "step": 38430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9815397696974937e-05,
+      "loss": 0.0272,
+      "step": 38440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9815349657236466e-05,
+      "loss": 0.0235,
+      "step": 38450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9815301617497995e-05,
+      "loss": 0.0225,
+      "step": 38460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9815253577759525e-05,
+      "loss": 0.0167,
+      "step": 38470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981520553802105e-05,
+      "loss": 0.0226,
+      "step": 38480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981515749828258e-05,
+      "loss": 0.0216,
+      "step": 38490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.981510945854411e-05,
+      "loss": 0.0204,
+      "step": 38500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9815061418805638e-05,
+      "loss": 0.0152,
+      "step": 38510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9815013379067164e-05,
+      "loss": 0.0178,
+      "step": 38520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814965339328693e-05,
+      "loss": 0.0208,
+      "step": 38530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814917299590223e-05,
+      "loss": 0.0214,
+      "step": 38540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981486925985175e-05,
+      "loss": 0.0234,
+      "step": 38550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814821220113278e-05,
+      "loss": 0.0188,
+      "step": 38560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814773180374807e-05,
+      "loss": 0.0205,
+      "step": 38570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814725140636336e-05,
+      "loss": 0.0175,
+      "step": 38580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814677100897862e-05,
+      "loss": 0.0232,
+      "step": 38590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981462906115939e-05,
+      "loss": 0.02,
+      "step": 38600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981458102142092e-05,
+      "loss": 0.0162,
+      "step": 38610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814532981682447e-05,
+      "loss": 0.0179,
+      "step": 38620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814484941943976e-05,
+      "loss": 0.0209,
+      "step": 38630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814436902205505e-05,
+      "loss": 0.0218,
+      "step": 38640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814388862467035e-05,
+      "loss": 0.022,
+      "step": 38650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981434082272856e-05,
+      "loss": 0.0211,
+      "step": 38660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981429278299009e-05,
+      "loss": 0.0214,
+      "step": 38670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981424474325162e-05,
+      "loss": 0.0193,
+      "step": 38680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814196703513148e-05,
+      "loss": 0.0194,
+      "step": 38690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814148663774674e-05,
+      "loss": 0.0232,
+      "step": 38700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814100624036203e-05,
+      "loss": 0.0252,
+      "step": 38710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814052584297733e-05,
+      "loss": 0.0225,
+      "step": 38720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9814004544559262e-05,
+      "loss": 0.0215,
+      "step": 38730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981395650482079e-05,
+      "loss": 0.0204,
+      "step": 38740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981390846508232e-05,
+      "loss": 0.0216,
+      "step": 38750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9813860425343846e-05,
+      "loss": 0.0197,
+      "step": 38760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9813812385605376e-05,
+      "loss": 0.0227,
+      "step": 38770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9813764345866905e-05,
+      "loss": 0.022,
+      "step": 38780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981371630612843e-05,
+      "loss": 0.0203,
+      "step": 38790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981366826638996e-05,
+      "loss": 0.0199,
+      "step": 38800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981362022665149e-05,
+      "loss": 0.0168,
+      "step": 38810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981357218691302e-05,
+      "loss": 0.024,
+      "step": 38820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9813524147174544e-05,
+      "loss": 0.0196,
+      "step": 38830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9813476107436074e-05,
+      "loss": 0.0183,
+      "step": 38840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9813428067697603e-05,
+      "loss": 0.023,
+      "step": 38850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981338002795913e-05,
+      "loss": 0.0183,
+      "step": 38860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9813331988220658e-05,
+      "loss": 0.0179,
+      "step": 38870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9813283948482187e-05,
+      "loss": 0.0209,
+      "step": 38880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9813235908743717e-05,
+      "loss": 0.0198,
+      "step": 38890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9813187869005243e-05,
+      "loss": 0.0194,
+      "step": 38900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9813139829266772e-05,
+      "loss": 0.02,
+      "step": 38910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.98130917895283e-05,
+      "loss": 0.0219,
+      "step": 38920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981304374978983e-05,
+      "loss": 0.0182,
+      "step": 38930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812995710051356e-05,
+      "loss": 0.025,
+      "step": 38940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812947670312885e-05,
+      "loss": 0.0225,
+      "step": 38950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812899630574415e-05,
+      "loss": 0.0294,
+      "step": 38960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981285159083594e-05,
+      "loss": 0.0241,
+      "step": 38970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981280355109747e-05,
+      "loss": 0.0213,
+      "step": 38980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812755511359e-05,
+      "loss": 0.0186,
+      "step": 38990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981270747162053e-05,
+      "loss": 0.0308,
+      "step": 39000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812659431882054e-05,
+      "loss": 0.0158,
+      "step": 39010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812611392143584e-05,
+      "loss": 0.0215,
+      "step": 39020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812563352405113e-05,
+      "loss": 0.0193,
+      "step": 39030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981251531266664e-05,
+      "loss": 0.0152,
+      "step": 39040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812467272928168e-05,
+      "loss": 0.0188,
+      "step": 39050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812419233189697e-05,
+      "loss": 0.017,
+      "step": 39060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812371193451227e-05,
+      "loss": 0.0225,
+      "step": 39070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812323153712752e-05,
+      "loss": 0.0282,
+      "step": 39080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812275113974282e-05,
+      "loss": 0.0187,
+      "step": 39090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981222707423581e-05,
+      "loss": 0.0209,
+      "step": 39100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981217903449734e-05,
+      "loss": 0.0202,
+      "step": 39110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812130994758866e-05,
+      "loss": 0.0163,
+      "step": 39120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812082955020395e-05,
+      "loss": 0.0165,
+      "step": 39130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9812034915281925e-05,
+      "loss": 0.0196,
+      "step": 39140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981198687554345e-05,
+      "loss": 0.0177,
+      "step": 39150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981193883580498e-05,
+      "loss": 0.0231,
+      "step": 39160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981189079606651e-05,
+      "loss": 0.0163,
+      "step": 39170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.981184275632804e-05,
+      "loss": 0.0166,
+      "step": 39180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9811794716589564e-05,
+      "loss": 0.0214,
+      "step": 39190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9811746676851094e-05,
+      "loss": 0.0174,
+      "step": 39200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9811698637112623e-05,
+      "loss": 0.0206,
+      "step": 39210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.981165059737415e-05,
+      "loss": 0.0175,
+      "step": 39220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9811602557635678e-05,
+      "loss": 0.0216,
+      "step": 39230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9811554517897207e-05,
+      "loss": 0.0192,
+      "step": 39240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9811506478158736e-05,
+      "loss": 0.0298,
+      "step": 39250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9811458438420262e-05,
+      "loss": 0.017,
+      "step": 39260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.981141039868179e-05,
+      "loss": 0.0197,
+      "step": 39270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.981136235894332e-05,
+      "loss": 0.0274,
+      "step": 39280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.981131431920485e-05,
+      "loss": 0.0242,
+      "step": 39290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9811266279466376e-05,
+      "loss": 0.0236,
+      "step": 39300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9811218239727905e-05,
+      "loss": 0.0305,
+      "step": 39310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9811170199989435e-05,
+      "loss": 0.0296,
+      "step": 39320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.981112696422481e-05,
+      "loss": 0.0165,
+      "step": 39330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.981107892448634e-05,
+      "loss": 0.0278,
+      "step": 39340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9811030884747864e-05,
+      "loss": 0.0188,
+      "step": 39350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810982845009394e-05,
+      "loss": 0.0155,
+      "step": 39360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810934805270923e-05,
+      "loss": 0.0219,
+      "step": 39370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810886765532452e-05,
+      "loss": 0.0231,
+      "step": 39380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810838725793978e-05,
+      "loss": 0.0199,
+      "step": 39390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810790686055507e-05,
+      "loss": 0.0234,
+      "step": 39400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810742646317037e-05,
+      "loss": 0.0236,
+      "step": 39410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810694606578566e-05,
+      "loss": 0.0167,
+      "step": 39420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810646566840092e-05,
+      "loss": 0.0163,
+      "step": 39430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.981059852710162e-05,
+      "loss": 0.0221,
+      "step": 39440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.981055048736315e-05,
+      "loss": 0.0216,
+      "step": 39450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810502447624676e-05,
+      "loss": 0.0143,
+      "step": 39460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810454407886205e-05,
+      "loss": 0.0171,
+      "step": 39470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810406368147735e-05,
+      "loss": 0.0243,
+      "step": 39480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810358328409264e-05,
+      "loss": 0.0201,
+      "step": 39490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.981031028867079e-05,
+      "loss": 0.0202,
+      "step": 39500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.981026224893232e-05,
+      "loss": 0.0179,
+      "step": 39510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.981021420919385e-05,
+      "loss": 0.0158,
+      "step": 39520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810166169455374e-05,
+      "loss": 0.0182,
+      "step": 39530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810118129716904e-05,
+      "loss": 0.015,
+      "step": 39540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810070089978433e-05,
+      "loss": 0.0305,
+      "step": 39550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9810022050239962e-05,
+      "loss": 0.0301,
+      "step": 39560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809974010501488e-05,
+      "loss": 0.0253,
+      "step": 39570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809925970763017e-05,
+      "loss": 0.017,
+      "step": 39580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809877931024547e-05,
+      "loss": 0.0217,
+      "step": 39590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809829891286076e-05,
+      "loss": 0.0229,
+      "step": 39600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.98097818515476e-05,
+      "loss": 0.021,
+      "step": 39610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.980973381180913e-05,
+      "loss": 0.0178,
+      "step": 39620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.980968577207066e-05,
+      "loss": 0.022,
+      "step": 39630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809637732332186e-05,
+      "loss": 0.0186,
+      "step": 39640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809589692593715e-05,
+      "loss": 0.0176,
+      "step": 39650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809541652855245e-05,
+      "loss": 0.0203,
+      "step": 39660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809493613116774e-05,
+      "loss": 0.0204,
+      "step": 39670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.98094455733783e-05,
+      "loss": 0.0207,
+      "step": 39680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.980939753363983e-05,
+      "loss": 0.0227,
+      "step": 39690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.980934949390136e-05,
+      "loss": 0.0173,
+      "step": 39700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809301454162884e-05,
+      "loss": 0.0196,
+      "step": 39710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809253414424413e-05,
+      "loss": 0.0218,
+      "step": 39720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809205374685943e-05,
+      "loss": 0.0222,
+      "step": 39730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809157334947472e-05,
+      "loss": 0.0151,
+      "step": 39740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809109295208998e-05,
+      "loss": 0.017,
+      "step": 39750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809061255470527e-05,
+      "loss": 0.0257,
+      "step": 39760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809013215732056e-05,
+      "loss": 0.022,
+      "step": 39770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9808965175993586e-05,
+      "loss": 0.0208,
+      "step": 39780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.980891713625511e-05,
+      "loss": 0.0158,
+      "step": 39790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.980886909651664e-05,
+      "loss": 0.0183,
+      "step": 39800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.980882105677817e-05,
+      "loss": 0.0176,
+      "step": 39810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9808773017039696e-05,
+      "loss": 0.0219,
+      "step": 39820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9808724977301225e-05,
+      "loss": 0.0182,
+      "step": 39830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9808676937562755e-05,
+      "loss": 0.0162,
+      "step": 39840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9808628897824284e-05,
+      "loss": 0.0147,
+      "step": 39850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.980858085808581e-05,
+      "loss": 0.0172,
+      "step": 39860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.980853281834734e-05,
+      "loss": 0.0185,
+      "step": 39870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9808484778608868e-05,
+      "loss": 0.0219,
+      "step": 39880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9808436738870394e-05,
+      "loss": 0.0207,
+      "step": 39890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9808388699131923e-05,
+      "loss": 0.0184,
+      "step": 39900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9808340659393453e-05,
+      "loss": 0.0241,
+      "step": 39910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9808292619654982e-05,
+      "loss": 0.0224,
+      "step": 39920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9808244579916508e-05,
+      "loss": 0.0214,
+      "step": 39930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9808196540178037e-05,
+      "loss": 0.0235,
+      "step": 39940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9808148500439566e-05,
+      "loss": 0.0179,
+      "step": 39950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9808100460701096e-05,
+      "loss": 0.0277,
+      "step": 39960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980805242096262e-05,
+      "loss": 0.018,
+      "step": 39970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980800438122415e-05,
+      "loss": 0.0183,
+      "step": 39980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980795634148568e-05,
+      "loss": 0.0211,
+      "step": 39990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807908301747206e-05,
+      "loss": 0.0232,
+      "step": 40000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807860262008735e-05,
+      "loss": 0.0204,
+      "step": 40010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807812222270264e-05,
+      "loss": 0.0208,
+      "step": 40020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807764182531794e-05,
+      "loss": 0.0164,
+      "step": 40030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980771614279332e-05,
+      "loss": 0.0234,
+      "step": 40040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980766810305485e-05,
+      "loss": 0.0182,
+      "step": 40050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807620063316378e-05,
+      "loss": 0.0205,
+      "step": 40060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807572023577904e-05,
+      "loss": 0.0187,
+      "step": 40070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807523983839433e-05,
+      "loss": 0.0164,
+      "step": 40080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807475944100963e-05,
+      "loss": 0.0232,
+      "step": 40090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807427904362492e-05,
+      "loss": 0.0259,
+      "step": 40100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807379864624018e-05,
+      "loss": 0.0277,
+      "step": 40110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807331824885547e-05,
+      "loss": 0.0223,
+      "step": 40120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807283785147076e-05,
+      "loss": 0.023,
+      "step": 40130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807235745408606e-05,
+      "loss": 0.0232,
+      "step": 40140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980718770567013e-05,
+      "loss": 0.022,
+      "step": 40150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980713966593166e-05,
+      "loss": 0.0192,
+      "step": 40160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980709162619319e-05,
+      "loss": 0.0199,
+      "step": 40170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9807043586454716e-05,
+      "loss": 0.0148,
+      "step": 40180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806995546716245e-05,
+      "loss": 0.0216,
+      "step": 40190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806947506977774e-05,
+      "loss": 0.0195,
+      "step": 40200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806899467239304e-05,
+      "loss": 0.0208,
+      "step": 40210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980685142750083e-05,
+      "loss": 0.0191,
+      "step": 40220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980680338776236e-05,
+      "loss": 0.0239,
+      "step": 40230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806755348023888e-05,
+      "loss": 0.0188,
+      "step": 40240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806707308285414e-05,
+      "loss": 0.015,
+      "step": 40250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806659268546943e-05,
+      "loss": 0.0227,
+      "step": 40260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806611228808473e-05,
+      "loss": 0.0248,
+      "step": 40270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806563189070002e-05,
+      "loss": 0.0246,
+      "step": 40280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806515149331528e-05,
+      "loss": 0.0244,
+      "step": 40290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806467109593057e-05,
+      "loss": 0.0204,
+      "step": 40300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806419069854586e-05,
+      "loss": 0.0217,
+      "step": 40310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806371030116115e-05,
+      "loss": 0.024,
+      "step": 40320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980632299037764e-05,
+      "loss": 0.0146,
+      "step": 40330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980627495063917e-05,
+      "loss": 0.0213,
+      "step": 40340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.98062269109007e-05,
+      "loss": 0.0185,
+      "step": 40350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806178871162226e-05,
+      "loss": 0.0261,
+      "step": 40360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806130831423755e-05,
+      "loss": 0.02,
+      "step": 40370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806082791685284e-05,
+      "loss": 0.0198,
+      "step": 40380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9806034751946814e-05,
+      "loss": 0.0175,
+      "step": 40390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980598671220834e-05,
+      "loss": 0.0208,
+      "step": 40400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980593867246987e-05,
+      "loss": 0.024,
+      "step": 40410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805890632731398e-05,
+      "loss": 0.0137,
+      "step": 40420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805847396966773e-05,
+      "loss": 0.0209,
+      "step": 40430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805799357228302e-05,
+      "loss": 0.0219,
+      "step": 40440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980575131748983e-05,
+      "loss": 0.0271,
+      "step": 40450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805703277751357e-05,
+      "loss": 0.0203,
+      "step": 40460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805655238012886e-05,
+      "loss": 0.0211,
+      "step": 40470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805607198274416e-05,
+      "loss": 0.022,
+      "step": 40480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980555915853594e-05,
+      "loss": 0.0164,
+      "step": 40490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980551111879747e-05,
+      "loss": 0.0243,
+      "step": 40500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805463079059e-05,
+      "loss": 0.0205,
+      "step": 40510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980541503932053e-05,
+      "loss": 0.0177,
+      "step": 40520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805366999582055e-05,
+      "loss": 0.0178,
+      "step": 40530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805318959843584e-05,
+      "loss": 0.0136,
+      "step": 40540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805270920105114e-05,
+      "loss": 0.0153,
+      "step": 40550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805222880366643e-05,
+      "loss": 0.0214,
+      "step": 40560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.980517484062817e-05,
+      "loss": 0.0213,
+      "step": 40570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805126800889698e-05,
+      "loss": 0.0207,
+      "step": 40580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9805078761151227e-05,
+      "loss": 0.0198,
+      "step": 40590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9805030721412753e-05,
+      "loss": 0.0205,
+      "step": 40600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804982681674283e-05,
+      "loss": 0.0138,
+      "step": 40610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804934641935812e-05,
+      "loss": 0.02,
+      "step": 40620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980488660219734e-05,
+      "loss": 0.0217,
+      "step": 40630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804838562458867e-05,
+      "loss": 0.0181,
+      "step": 40640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804790522720396e-05,
+      "loss": 0.022,
+      "step": 40650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804742482981926e-05,
+      "loss": 0.0184,
+      "step": 40660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980469444324345e-05,
+      "loss": 0.0184,
+      "step": 40670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980464640350498e-05,
+      "loss": 0.0247,
+      "step": 40680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980459836376651e-05,
+      "loss": 0.0261,
+      "step": 40690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980455032402804e-05,
+      "loss": 0.0189,
+      "step": 40700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804502284289565e-05,
+      "loss": 0.0168,
+      "step": 40710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804454244551094e-05,
+      "loss": 0.0253,
+      "step": 40720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804406204812624e-05,
+      "loss": 0.0208,
+      "step": 40730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804358165074153e-05,
+      "loss": 0.0169,
+      "step": 40740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980431012533568e-05,
+      "loss": 0.0165,
+      "step": 40750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804262085597208e-05,
+      "loss": 0.0152,
+      "step": 40760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804214045858737e-05,
+      "loss": 0.0202,
+      "step": 40770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804166006120263e-05,
+      "loss": 0.0169,
+      "step": 40780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804117966381792e-05,
+      "loss": 0.0174,
+      "step": 40790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9804069926643322e-05,
+      "loss": 0.019,
+      "step": 40800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980402188690485e-05,
+      "loss": 0.0179,
+      "step": 40810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803973847166377e-05,
+      "loss": 0.024,
+      "step": 40820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803925807427906e-05,
+      "loss": 0.0202,
+      "step": 40830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803877767689435e-05,
+      "loss": 0.0224,
+      "step": 40840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980382972795096e-05,
+      "loss": 0.0183,
+      "step": 40850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980378168821249e-05,
+      "loss": 0.0252,
+      "step": 40860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980373364847402e-05,
+      "loss": 0.0189,
+      "step": 40870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980368560873555e-05,
+      "loss": 0.0163,
+      "step": 40880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803637568997075e-05,
+      "loss": 0.0158,
+      "step": 40890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803589529258604e-05,
+      "loss": 0.0277,
+      "step": 40900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803541489520134e-05,
+      "loss": 0.0187,
+      "step": 40910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980349344978166e-05,
+      "loss": 0.018,
+      "step": 40920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980344541004319e-05,
+      "loss": 0.0191,
+      "step": 40930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803397370304718e-05,
+      "loss": 0.0214,
+      "step": 40940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803349330566247e-05,
+      "loss": 0.0164,
+      "step": 40950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803301290827773e-05,
+      "loss": 0.0177,
+      "step": 40960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803253251089302e-05,
+      "loss": 0.0187,
+      "step": 40970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980320521135083e-05,
+      "loss": 0.0222,
+      "step": 40980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980315717161236e-05,
+      "loss": 0.0162,
+      "step": 40990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803109131873887e-05,
+      "loss": 0.0322,
+      "step": 41000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803061092135416e-05,
+      "loss": 0.0177,
+      "step": 41010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9803013052396945e-05,
+      "loss": 0.021,
+      "step": 41020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980296501265847e-05,
+      "loss": 0.022,
+      "step": 41030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980291697292e-05,
+      "loss": 0.0195,
+      "step": 41040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980286893318153e-05,
+      "loss": 0.0222,
+      "step": 41050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980282089344306e-05,
+      "loss": 0.0182,
+      "step": 41060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9802772853704585e-05,
+      "loss": 0.0221,
+      "step": 41070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9802724813966114e-05,
+      "loss": 0.0188,
+      "step": 41080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9802676774227643e-05,
+      "loss": 0.0185,
+      "step": 41090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980262873448917e-05,
+      "loss": 0.0215,
+      "step": 41100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.98025806947507e-05,
+      "loss": 0.017,
+      "step": 41110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9802532655012228e-05,
+      "loss": 0.0157,
+      "step": 41120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9802484615273757e-05,
+      "loss": 0.0177,
+      "step": 41130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9802436575535283e-05,
+      "loss": 0.0231,
+      "step": 41140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9802388535796812e-05,
+      "loss": 0.0132,
+      "step": 41150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980234049605834e-05,
+      "loss": 0.0175,
+      "step": 41160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980229245631987e-05,
+      "loss": 0.0135,
+      "step": 41170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9802244416581397e-05,
+      "loss": 0.0225,
+      "step": 41180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9802196376842926e-05,
+      "loss": 0.0165,
+      "step": 41190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9802148337104455e-05,
+      "loss": 0.0137,
+      "step": 41200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980210029736598e-05,
+      "loss": 0.0264,
+      "step": 41210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980205225762751e-05,
+      "loss": 0.0167,
+      "step": 41220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980200421788904e-05,
+      "loss": 0.0228,
+      "step": 41230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980195617815057e-05,
+      "loss": 0.0184,
+      "step": 41240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9801908138412095e-05,
+      "loss": 0.0214,
+      "step": 41250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9801860098673624e-05,
+      "loss": 0.0227,
+      "step": 41260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9801812058935153e-05,
+      "loss": 0.0219,
+      "step": 41270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.980176401919668e-05,
+      "loss": 0.0202,
+      "step": 41280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980171597945821e-05,
+      "loss": 0.0222,
+      "step": 41290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9801667939719738e-05,
+      "loss": 0.0193,
+      "step": 41300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9801619899981267e-05,
+      "loss": 0.0248,
+      "step": 41310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9801571860242793e-05,
+      "loss": 0.0192,
+      "step": 41320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9801523820504322e-05,
+      "loss": 0.0175,
+      "step": 41330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980147578076585e-05,
+      "loss": 0.0198,
+      "step": 41340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980142774102738e-05,
+      "loss": 0.0141,
+      "step": 41350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9801379701288907e-05,
+      "loss": 0.0176,
+      "step": 41360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9801331661550436e-05,
+      "loss": 0.0199,
+      "step": 41370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9801283621811965e-05,
+      "loss": 0.0231,
+      "step": 41380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980123558207349e-05,
+      "loss": 0.0226,
+      "step": 41390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980118754233502e-05,
+      "loss": 0.0198,
+      "step": 41400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980113950259655e-05,
+      "loss": 0.0221,
+      "step": 41410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980109146285808e-05,
+      "loss": 0.017,
+      "step": 41420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9801043423119605e-05,
+      "loss": 0.0206,
+      "step": 41430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800995383381134e-05,
+      "loss": 0.0164,
+      "step": 41440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800947343642663e-05,
+      "loss": 0.0205,
+      "step": 41450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980089930390419e-05,
+      "loss": 0.0244,
+      "step": 41460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980085126416572e-05,
+      "loss": 0.0221,
+      "step": 41470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800803224427248e-05,
+      "loss": 0.0205,
+      "step": 41480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800755184688777e-05,
+      "loss": 0.0176,
+      "step": 41490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800707144950303e-05,
+      "loss": 0.0195,
+      "step": 41500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800659105211832e-05,
+      "loss": 0.0155,
+      "step": 41510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980061106547336e-05,
+      "loss": 0.0192,
+      "step": 41520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980056302573489e-05,
+      "loss": 0.0217,
+      "step": 41530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800514985996417e-05,
+      "loss": 0.0196,
+      "step": 41540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800466946257946e-05,
+      "loss": 0.0198,
+      "step": 41550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800418906519475e-05,
+      "loss": 0.019,
+      "step": 41560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800370866781e-05,
+      "loss": 0.014,
+      "step": 41570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980032282704253e-05,
+      "loss": 0.0201,
+      "step": 41580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980027478730406e-05,
+      "loss": 0.0215,
+      "step": 41590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980022674756559e-05,
+      "loss": 0.0199,
+      "step": 41600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800178707827115e-05,
+      "loss": 0.0202,
+      "step": 41610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800130668088644e-05,
+      "loss": 0.0175,
+      "step": 41620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9800082628350173e-05,
+      "loss": 0.0218,
+      "step": 41630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.98000345886117e-05,
+      "loss": 0.0208,
+      "step": 41640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979998654887323e-05,
+      "loss": 0.0198,
+      "step": 41650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799938509134758e-05,
+      "loss": 0.022,
+      "step": 41660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799890469396287e-05,
+      "loss": 0.0159,
+      "step": 41670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799842429657813e-05,
+      "loss": 0.018,
+      "step": 41680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799794389919342e-05,
+      "loss": 0.0147,
+      "step": 41690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979974635018087e-05,
+      "loss": 0.0164,
+      "step": 41700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.97996983104424e-05,
+      "loss": 0.0189,
+      "step": 41710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799650270703926e-05,
+      "loss": 0.0252,
+      "step": 41720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799602230965456e-05,
+      "loss": 0.0175,
+      "step": 41730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799554191226985e-05,
+      "loss": 0.0156,
+      "step": 41740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979950615148851e-05,
+      "loss": 0.0157,
+      "step": 41750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979945811175004e-05,
+      "loss": 0.0183,
+      "step": 41760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979941007201157e-05,
+      "loss": 0.0198,
+      "step": 41770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.97993620322731e-05,
+      "loss": 0.0273,
+      "step": 41780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799313992534625e-05,
+      "loss": 0.0211,
+      "step": 41790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799265952796154e-05,
+      "loss": 0.0182,
+      "step": 41800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799217913057683e-05,
+      "loss": 0.0222,
+      "step": 41810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979916987331921e-05,
+      "loss": 0.0193,
+      "step": 41820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799121833580738e-05,
+      "loss": 0.0215,
+      "step": 41830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799073793842268e-05,
+      "loss": 0.0143,
+      "step": 41840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9799025754103797e-05,
+      "loss": 0.0221,
+      "step": 41850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9798977714365323e-05,
+      "loss": 0.0183,
+      "step": 41860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9798929674626852e-05,
+      "loss": 0.0196,
+      "step": 41870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979888163488838e-05,
+      "loss": 0.0207,
+      "step": 41880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979883359514991e-05,
+      "loss": 0.0142,
+      "step": 41890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9798785555411436e-05,
+      "loss": 0.0162,
+      "step": 41900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9798737515672966e-05,
+      "loss": 0.022,
+      "step": 41910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9798689475934495e-05,
+      "loss": 0.0222,
+      "step": 41920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979864143619602e-05,
+      "loss": 0.0169,
+      "step": 41930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979859339645755e-05,
+      "loss": 0.0152,
+      "step": 41940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979854535671908e-05,
+      "loss": 0.0158,
+      "step": 41950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979849731698061e-05,
+      "loss": 0.0169,
+      "step": 41960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9798449277242134e-05,
+      "loss": 0.0181,
+      "step": 41970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9798401237503664e-05,
+      "loss": 0.0166,
+      "step": 41980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9798353197765193e-05,
+      "loss": 0.0232,
+      "step": 41990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979830515802672e-05,
+      "loss": 0.0196,
+      "step": 42000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9798257118288248e-05,
+      "loss": 0.0211,
+      "step": 42010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9798209078549777e-05,
+      "loss": 0.0121,
+      "step": 42020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9798161038811307e-05,
+      "loss": 0.0212,
+      "step": 42030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9798112999072833e-05,
+      "loss": 0.0188,
+      "step": 42040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9798064959334362e-05,
+      "loss": 0.0133,
+      "step": 42050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979801691959589e-05,
+      "loss": 0.02,
+      "step": 42060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979796887985742e-05,
+      "loss": 0.0199,
+      "step": 42070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797920840118946e-05,
+      "loss": 0.0187,
+      "step": 42080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797872800380476e-05,
+      "loss": 0.0167,
+      "step": 42090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797824760642005e-05,
+      "loss": 0.0204,
+      "step": 42100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979777672090353e-05,
+      "loss": 0.0247,
+      "step": 42110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979772868116506e-05,
+      "loss": 0.0222,
+      "step": 42120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979768064142659e-05,
+      "loss": 0.0203,
+      "step": 42130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979763260168812e-05,
+      "loss": 0.0229,
+      "step": 42140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797584561949644e-05,
+      "loss": 0.0165,
+      "step": 42150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797536522211174e-05,
+      "loss": 0.0158,
+      "step": 42160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797488482472703e-05,
+      "loss": 0.0186,
+      "step": 42170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979744044273423e-05,
+      "loss": 0.0244,
+      "step": 42180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797392402995758e-05,
+      "loss": 0.0191,
+      "step": 42190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797344363257287e-05,
+      "loss": 0.0153,
+      "step": 42200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797296323518817e-05,
+      "loss": 0.0202,
+      "step": 42210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797248283780343e-05,
+      "loss": 0.0163,
+      "step": 42220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797200244041872e-05,
+      "loss": 0.0184,
+      "step": 42230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.97971522043034e-05,
+      "loss": 0.0336,
+      "step": 42240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979710416456493e-05,
+      "loss": 0.0205,
+      "step": 42250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797056124826456e-05,
+      "loss": 0.0174,
+      "step": 42260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9797008085087985e-05,
+      "loss": 0.0171,
+      "step": 42270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9796960045349515e-05,
+      "loss": 0.0181,
+      "step": 42280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979691200561104e-05,
+      "loss": 0.0238,
+      "step": 42290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979686396587257e-05,
+      "loss": 0.0173,
+      "step": 42300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.97968159261341e-05,
+      "loss": 0.015,
+      "step": 42310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979676788639563e-05,
+      "loss": 0.0207,
+      "step": 42320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9796719846657158e-05,
+      "loss": 0.0169,
+      "step": 42330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9796671806918687e-05,
+      "loss": 0.0186,
+      "step": 42340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9796623767180213e-05,
+      "loss": 0.0211,
+      "step": 42350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9796575727441742e-05,
+      "loss": 0.0201,
+      "step": 42360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979652768770327e-05,
+      "loss": 0.0214,
+      "step": 42370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.97964796479648e-05,
+      "loss": 0.0192,
+      "step": 42380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9796431608226327e-05,
+      "loss": 0.0215,
+      "step": 42390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9796383568487856e-05,
+      "loss": 0.0271,
+      "step": 42400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9796335528749385e-05,
+      "loss": 0.0208,
+      "step": 42410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979628748901091e-05,
+      "loss": 0.0161,
+      "step": 42420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979623944927244e-05,
+      "loss": 0.0163,
+      "step": 42430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979619140953397e-05,
+      "loss": 0.0197,
+      "step": 42440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.97961433697955e-05,
+      "loss": 0.0158,
+      "step": 42450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9796095330057025e-05,
+      "loss": 0.0181,
+      "step": 42460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9796047290318554e-05,
+      "loss": 0.0232,
+      "step": 42470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9795999250580083e-05,
+      "loss": 0.0229,
+      "step": 42480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979595121084161e-05,
+      "loss": 0.0261,
+      "step": 42490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979590317110314e-05,
+      "loss": 0.0197,
+      "step": 42500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9795855131364668e-05,
+      "loss": 0.0205,
+      "step": 42510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9795807091626197e-05,
+      "loss": 0.0168,
+      "step": 42520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9795759051887723e-05,
+      "loss": 0.0147,
+      "step": 42530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9795711012149252e-05,
+      "loss": 0.0209,
+      "step": 42540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979566297241078e-05,
+      "loss": 0.0183,
+      "step": 42550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979561493267231e-05,
+      "loss": 0.0184,
+      "step": 42560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9795566892933836e-05,
+      "loss": 0.0151,
+      "step": 42570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9795518853195366e-05,
+      "loss": 0.0175,
+      "step": 42580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9795470813456895e-05,
+      "loss": 0.0208,
+      "step": 42590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979542277371842e-05,
+      "loss": 0.0235,
+      "step": 42600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979537473397995e-05,
+      "loss": 0.0213,
+      "step": 42610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979532669424148e-05,
+      "loss": 0.024,
+      "step": 42620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979527865450301e-05,
+      "loss": 0.016,
+      "step": 42630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9795230614764535e-05,
+      "loss": 0.0202,
+      "step": 42640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9795182575026064e-05,
+      "loss": 0.018,
+      "step": 42650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9795134535287593e-05,
+      "loss": 0.019,
+      "step": 42660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.979508649554912e-05,
+      "loss": 0.0195,
+      "step": 42670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9795038455810648e-05,
+      "loss": 0.0217,
+      "step": 42680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794990416072178e-05,
+      "loss": 0.017,
+      "step": 42690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794942376333707e-05,
+      "loss": 0.0171,
+      "step": 42700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794894336595233e-05,
+      "loss": 0.0194,
+      "step": 42710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794846296856762e-05,
+      "loss": 0.0216,
+      "step": 42720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979479825711829e-05,
+      "loss": 0.0169,
+      "step": 42730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979475021737982e-05,
+      "loss": 0.0194,
+      "step": 42740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794702177641346e-05,
+      "loss": 0.0239,
+      "step": 42750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794654137902876e-05,
+      "loss": 0.0171,
+      "step": 42760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794606098164405e-05,
+      "loss": 0.0183,
+      "step": 42770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979455805842593e-05,
+      "loss": 0.0175,
+      "step": 42780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979451001868746e-05,
+      "loss": 0.0246,
+      "step": 42790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979446197894899e-05,
+      "loss": 0.0185,
+      "step": 42800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979441393921052e-05,
+      "loss": 0.0154,
+      "step": 42810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794365899472044e-05,
+      "loss": 0.0165,
+      "step": 42820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794317859733574e-05,
+      "loss": 0.0201,
+      "step": 42830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794269819995103e-05,
+      "loss": 0.026,
+      "step": 42840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979422178025663e-05,
+      "loss": 0.0241,
+      "step": 42850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794173740518158e-05,
+      "loss": 0.0144,
+      "step": 42860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794125700779687e-05,
+      "loss": 0.0198,
+      "step": 42870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794077661041217e-05,
+      "loss": 0.0284,
+      "step": 42880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9794029621302743e-05,
+      "loss": 0.0155,
+      "step": 42890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9793981581564272e-05,
+      "loss": 0.0166,
+      "step": 42900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.97939335418258e-05,
+      "loss": 0.0206,
+      "step": 42910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979388550208733e-05,
+      "loss": 0.019,
+      "step": 42920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9793837462348856e-05,
+      "loss": 0.017,
+      "step": 42930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9793789422610386e-05,
+      "loss": 0.0191,
+      "step": 42940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9793741382871915e-05,
+      "loss": 0.02,
+      "step": 42950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979369334313344e-05,
+      "loss": 0.0225,
+      "step": 42960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979364530339497e-05,
+      "loss": 0.0187,
+      "step": 42970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.97935972636565e-05,
+      "loss": 0.018,
+      "step": 42980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979354922391803e-05,
+      "loss": 0.019,
+      "step": 42990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9793501184179554e-05,
+      "loss": 0.021,
+      "step": 43000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9793457948414932e-05,
+      "loss": 0.0167,
+      "step": 43010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979340990867646e-05,
+      "loss": 0.018,
+      "step": 43020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9793361868937988e-05,
+      "loss": 0.0239,
+      "step": 43030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9793313829199517e-05,
+      "loss": 0.0261,
+      "step": 43040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9793265789461046e-05,
+      "loss": 0.022,
+      "step": 43050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9793217749722572e-05,
+      "loss": 0.0175,
+      "step": 43060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.97931697099841e-05,
+      "loss": 0.0189,
+      "step": 43070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979312167024563e-05,
+      "loss": 0.0199,
+      "step": 43080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9793073630507156e-05,
+      "loss": 0.0192,
+      "step": 43090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9793025590768686e-05,
+      "loss": 0.0201,
+      "step": 43100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9792977551030215e-05,
+      "loss": 0.0175,
+      "step": 43110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9792929511291744e-05,
+      "loss": 0.0246,
+      "step": 43120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979288147155327e-05,
+      "loss": 0.0197,
+      "step": 43130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.97928334318148e-05,
+      "loss": 0.0217,
+      "step": 43140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979278539207633e-05,
+      "loss": 0.0207,
+      "step": 43150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9792742156311703e-05,
+      "loss": 0.0216,
+      "step": 43160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979269411657323e-05,
+      "loss": 0.0206,
+      "step": 43170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979264607683476e-05,
+      "loss": 0.0235,
+      "step": 43180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9792598037096288e-05,
+      "loss": 0.0207,
+      "step": 43190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9792549997357814e-05,
+      "loss": 0.0233,
+      "step": 43200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9792501957619343e-05,
+      "loss": 0.0196,
+      "step": 43210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9792453917880872e-05,
+      "loss": 0.0176,
+      "step": 43220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.97924058781424e-05,
+      "loss": 0.0225,
+      "step": 43230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9792357838403927e-05,
+      "loss": 0.0221,
+      "step": 43240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9792309798665457e-05,
+      "loss": 0.0159,
+      "step": 43250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9792261758926986e-05,
+      "loss": 0.0178,
+      "step": 43260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9792213719188512e-05,
+      "loss": 0.0213,
+      "step": 43270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979216567945004e-05,
+      "loss": 0.0177,
+      "step": 43280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.979211763971157e-05,
+      "loss": 0.0244,
+      "step": 43290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.97920695999731e-05,
+      "loss": 0.019,
+      "step": 43300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9792021560234625e-05,
+      "loss": 0.0166,
+      "step": 43310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9791973520496155e-05,
+      "loss": 0.0207,
+      "step": 43320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9791925480757684e-05,
+      "loss": 0.024,
+      "step": 43330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9791877441019213e-05,
+      "loss": 0.0222,
+      "step": 43340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9791829401280743e-05,
+      "loss": 0.0186,
+      "step": 43350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9791781361542272e-05,
+      "loss": 0.0175,
+      "step": 43360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9791733321803798e-05,
+      "loss": 0.0149,
+      "step": 43370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9791685282065327e-05,
+      "loss": 0.0229,
+      "step": 43380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9791637242326856e-05,
+      "loss": 0.0152,
+      "step": 43390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9791589202588385e-05,
+      "loss": 0.0274,
+      "step": 43400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979154116284991e-05,
+      "loss": 0.0218,
+      "step": 43410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979149312311144e-05,
+      "loss": 0.0221,
+      "step": 43420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9791449887346815e-05,
+      "loss": 0.0217,
+      "step": 43430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979140184760834e-05,
+      "loss": 0.0183,
+      "step": 43440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979135380786987e-05,
+      "loss": 0.0188,
+      "step": 43450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.97913057681314e-05,
+      "loss": 0.0188,
+      "step": 43460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979125772839293e-05,
+      "loss": 0.019,
+      "step": 43470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9791209688654455e-05,
+      "loss": 0.0175,
+      "step": 43480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9791161648915984e-05,
+      "loss": 0.02,
+      "step": 43490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9791113609177513e-05,
+      "loss": 0.0175,
+      "step": 43500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979106556943904e-05,
+      "loss": 0.0205,
+      "step": 43510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979101752970057e-05,
+      "loss": 0.0201,
+      "step": 43520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790969489962098e-05,
+      "loss": 0.0147,
+      "step": 43530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790921450223627e-05,
+      "loss": 0.0244,
+      "step": 43540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790873410485153e-05,
+      "loss": 0.0148,
+      "step": 43550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790825370746682e-05,
+      "loss": 0.0177,
+      "step": 43560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979077733100821e-05,
+      "loss": 0.0185,
+      "step": 43570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979072929126974e-05,
+      "loss": 0.0209,
+      "step": 43580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790681251531267e-05,
+      "loss": 0.0177,
+      "step": 43590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790633211792796e-05,
+      "loss": 0.0163,
+      "step": 43600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790585172054325e-05,
+      "loss": 0.0144,
+      "step": 43610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979053713231585e-05,
+      "loss": 0.019,
+      "step": 43620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979048909257738e-05,
+      "loss": 0.0224,
+      "step": 43630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979044105283891e-05,
+      "loss": 0.0213,
+      "step": 43640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979039301310044e-05,
+      "loss": 0.0193,
+      "step": 43650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790344973361965e-05,
+      "loss": 0.022,
+      "step": 43660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790296933623494e-05,
+      "loss": 0.0194,
+      "step": 43670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790248893885023e-05,
+      "loss": 0.0208,
+      "step": 43680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979020085414655e-05,
+      "loss": 0.0238,
+      "step": 43690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.979015281440808e-05,
+      "loss": 0.0173,
+      "step": 43700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790104774669608e-05,
+      "loss": 0.0168,
+      "step": 43710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790056734931137e-05,
+      "loss": 0.0175,
+      "step": 43720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9790008695192663e-05,
+      "loss": 0.0142,
+      "step": 43730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789960655454192e-05,
+      "loss": 0.0228,
+      "step": 43740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.978991261571572e-05,
+      "loss": 0.0257,
+      "step": 43750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789864575977247e-05,
+      "loss": 0.0247,
+      "step": 43760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789816536238777e-05,
+      "loss": 0.022,
+      "step": 43770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789768496500306e-05,
+      "loss": 0.0203,
+      "step": 43780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789720456761835e-05,
+      "loss": 0.024,
+      "step": 43790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.978967241702336e-05,
+      "loss": 0.0227,
+      "step": 43800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.978962437728489e-05,
+      "loss": 0.0177,
+      "step": 43810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.978957633754642e-05,
+      "loss": 0.0128,
+      "step": 43820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.978952829780795e-05,
+      "loss": 0.0209,
+      "step": 43830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789480258069475e-05,
+      "loss": 0.0161,
+      "step": 43840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789432218331004e-05,
+      "loss": 0.0194,
+      "step": 43850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789384178592533e-05,
+      "loss": 0.0263,
+      "step": 43860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789336138854062e-05,
+      "loss": 0.0156,
+      "step": 43870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789288099115592e-05,
+      "loss": 0.0158,
+      "step": 43880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.978924005937712e-05,
+      "loss": 0.0167,
+      "step": 43890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789192019638647e-05,
+      "loss": 0.0229,
+      "step": 43900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789143979900176e-05,
+      "loss": 0.0216,
+      "step": 43910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9789095940161705e-05,
+      "loss": 0.0247,
+      "step": 43920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.978904790042323e-05,
+      "loss": 0.0189,
+      "step": 43930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.978899986068476e-05,
+      "loss": 0.0207,
+      "step": 43940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.978895182094629e-05,
+      "loss": 0.0188,
+      "step": 43950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.978890378120782e-05,
+      "loss": 0.0152,
+      "step": 43960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9788855741469345e-05,
+      "loss": 0.0219,
+      "step": 43970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9788807701730874e-05,
+      "loss": 0.0322,
+      "step": 43980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9788759661992404e-05,
+      "loss": 0.0173,
+      "step": 43990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.978871162225393e-05,
+      "loss": 0.0183,
+      "step": 44000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.978866358251546e-05,
+      "loss": 0.0215,
+      "step": 44010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9788615542776988e-05,
+      "loss": 0.015,
+      "step": 44020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9788567503038517e-05,
+      "loss": 0.0148,
+      "step": 44030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9788519463300043e-05,
+      "loss": 0.017,
+      "step": 44040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9788471423561572e-05,
+      "loss": 0.0196,
+      "step": 44050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.97884233838231e-05,
+      "loss": 0.0181,
+      "step": 44060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978837534408463e-05,
+      "loss": 0.0158,
+      "step": 44070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9788327304346157e-05,
+      "loss": 0.0165,
+      "step": 44080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9788279264607686e-05,
+      "loss": 0.0198,
+      "step": 44090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9788231224869215e-05,
+      "loss": 0.0131,
+      "step": 44100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978818318513074e-05,
+      "loss": 0.022,
+      "step": 44110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978813514539227e-05,
+      "loss": 0.022,
+      "step": 44120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.97880871056538e-05,
+      "loss": 0.0194,
+      "step": 44130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978803906591533e-05,
+      "loss": 0.0134,
+      "step": 44140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787991026176855e-05,
+      "loss": 0.0133,
+      "step": 44150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787942986438384e-05,
+      "loss": 0.019,
+      "step": 44160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787894946699913e-05,
+      "loss": 0.0218,
+      "step": 44170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978784690696144e-05,
+      "loss": 0.016,
+      "step": 44180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978779886722297e-05,
+      "loss": 0.0179,
+      "step": 44190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787750827484498e-05,
+      "loss": 0.0139,
+      "step": 44200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787702787746027e-05,
+      "loss": 0.0164,
+      "step": 44210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787654748007553e-05,
+      "loss": 0.0159,
+      "step": 44220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787606708269082e-05,
+      "loss": 0.0177,
+      "step": 44230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978755866853061e-05,
+      "loss": 0.018,
+      "step": 44240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978751062879214e-05,
+      "loss": 0.0182,
+      "step": 44250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787462589053667e-05,
+      "loss": 0.0189,
+      "step": 44260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787414549315196e-05,
+      "loss": 0.02,
+      "step": 44270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787366509576725e-05,
+      "loss": 0.0183,
+      "step": 44280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978731846983825e-05,
+      "loss": 0.0185,
+      "step": 44290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978727043009978e-05,
+      "loss": 0.0171,
+      "step": 44300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978722239036131e-05,
+      "loss": 0.0167,
+      "step": 44310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978717435062284e-05,
+      "loss": 0.0194,
+      "step": 44320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787126310884365e-05,
+      "loss": 0.0174,
+      "step": 44330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787078271145894e-05,
+      "loss": 0.0169,
+      "step": 44340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9787030231407423e-05,
+      "loss": 0.0155,
+      "step": 44350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978698219166895e-05,
+      "loss": 0.0163,
+      "step": 44360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978693415193048e-05,
+      "loss": 0.0194,
+      "step": 44370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9786886112192008e-05,
+      "loss": 0.0159,
+      "step": 44380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9786838072453537e-05,
+      "loss": 0.0161,
+      "step": 44390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9786790032715063e-05,
+      "loss": 0.0229,
+      "step": 44400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9786741992976592e-05,
+      "loss": 0.0275,
+      "step": 44410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978669395323812e-05,
+      "loss": 0.019,
+      "step": 44420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978664591349965e-05,
+      "loss": 0.0267,
+      "step": 44430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9786597873761177e-05,
+      "loss": 0.0236,
+      "step": 44440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9786549834022706e-05,
+      "loss": 0.02,
+      "step": 44450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9786501794284235e-05,
+      "loss": 0.0153,
+      "step": 44460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978645375454576e-05,
+      "loss": 0.0184,
+      "step": 44470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978640571480729e-05,
+      "loss": 0.0211,
+      "step": 44480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978635767506882e-05,
+      "loss": 0.0197,
+      "step": 44490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978630963533035e-05,
+      "loss": 0.0191,
+      "step": 44500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9786261595591875e-05,
+      "loss": 0.0165,
+      "step": 44510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9786213555853404e-05,
+      "loss": 0.0195,
+      "step": 44520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9786165516114933e-05,
+      "loss": 0.0202,
+      "step": 44530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978611747637646e-05,
+      "loss": 0.0182,
+      "step": 44540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978606943663799e-05,
+      "loss": 0.0126,
+      "step": 44550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9786021396899518e-05,
+      "loss": 0.0186,
+      "step": 44560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9785973357161047e-05,
+      "loss": 0.021,
+      "step": 44570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9785925317422573e-05,
+      "loss": 0.0137,
+      "step": 44580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9785877277684102e-05,
+      "loss": 0.0223,
+      "step": 44590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978582923794563e-05,
+      "loss": 0.016,
+      "step": 44600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978578119820716e-05,
+      "loss": 0.02,
+      "step": 44610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9785733158468687e-05,
+      "loss": 0.018,
+      "step": 44620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9785685118730216e-05,
+      "loss": 0.0159,
+      "step": 44630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9785637078991745e-05,
+      "loss": 0.0226,
+      "step": 44640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978558903925327e-05,
+      "loss": 0.0237,
+      "step": 44650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.97855409995148e-05,
+      "loss": 0.0184,
+      "step": 44660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978549295977633e-05,
+      "loss": 0.0148,
+      "step": 44670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978544492003786e-05,
+      "loss": 0.0162,
+      "step": 44680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9785396880299385e-05,
+      "loss": 0.0184,
+      "step": 44690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9785348840560914e-05,
+      "loss": 0.0161,
+      "step": 44700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9785300800822443e-05,
+      "loss": 0.0196,
+      "step": 44710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.978525276108397e-05,
+      "loss": 0.0226,
+      "step": 44720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.97852047213455e-05,
+      "loss": 0.0245,
+      "step": 44730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9785156681607028e-05,
+      "loss": 0.0235,
+      "step": 44740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9785108641868557e-05,
+      "loss": 0.0223,
+      "step": 44750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9785060602130083e-05,
+      "loss": 0.0157,
+      "step": 44760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9785012562391612e-05,
+      "loss": 0.0142,
+      "step": 44770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978496452265314e-05,
+      "loss": 0.0127,
+      "step": 44780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978491648291467e-05,
+      "loss": 0.0177,
+      "step": 44790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9784868443176196e-05,
+      "loss": 0.0156,
+      "step": 44800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9784820403437726e-05,
+      "loss": 0.0148,
+      "step": 44810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9784772363699255e-05,
+      "loss": 0.0207,
+      "step": 44820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978472432396078e-05,
+      "loss": 0.0145,
+      "step": 44830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978467628422231e-05,
+      "loss": 0.0173,
+      "step": 44840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978462824448384e-05,
+      "loss": 0.0163,
+      "step": 44850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978458020474537e-05,
+      "loss": 0.0205,
+      "step": 44860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9784532165006895e-05,
+      "loss": 0.0227,
+      "step": 44870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9784484125268424e-05,
+      "loss": 0.0218,
+      "step": 44880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9784436085529953e-05,
+      "loss": 0.0198,
+      "step": 44890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978438804579148e-05,
+      "loss": 0.0222,
+      "step": 44900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9784340006053008e-05,
+      "loss": 0.0209,
+      "step": 44910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9784291966314538e-05,
+      "loss": 0.0161,
+      "step": 44920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9784243926576067e-05,
+      "loss": 0.0215,
+      "step": 44930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9784195886837593e-05,
+      "loss": 0.0168,
+      "step": 44940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9784147847099122e-05,
+      "loss": 0.0234,
+      "step": 44950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978409980736065e-05,
+      "loss": 0.016,
+      "step": 44960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978405176762218e-05,
+      "loss": 0.018,
+      "step": 44970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9784003727883706e-05,
+      "loss": 0.0171,
+      "step": 44980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783955688145236e-05,
+      "loss": 0.0269,
+      "step": 44990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783907648406765e-05,
+      "loss": 0.0192,
+      "step": 45000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978385960866829e-05,
+      "loss": 0.0265,
+      "step": 45010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978381156892982e-05,
+      "loss": 0.0142,
+      "step": 45020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978376352919135e-05,
+      "loss": 0.0288,
+      "step": 45030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978371548945288e-05,
+      "loss": 0.0211,
+      "step": 45040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783667449714405e-05,
+      "loss": 0.0202,
+      "step": 45050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783619409975934e-05,
+      "loss": 0.0118,
+      "step": 45060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783571370237463e-05,
+      "loss": 0.0183,
+      "step": 45070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978352333049899e-05,
+      "loss": 0.0197,
+      "step": 45080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783475290760518e-05,
+      "loss": 0.0236,
+      "step": 45090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783427251022047e-05,
+      "loss": 0.0185,
+      "step": 45100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783379211283577e-05,
+      "loss": 0.0188,
+      "step": 45110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783331171545103e-05,
+      "loss": 0.0163,
+      "step": 45120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783283131806632e-05,
+      "loss": 0.0232,
+      "step": 45130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978323509206816e-05,
+      "loss": 0.0183,
+      "step": 45140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783187052329687e-05,
+      "loss": 0.017,
+      "step": 45150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783139012591216e-05,
+      "loss": 0.0215,
+      "step": 45160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783090972852746e-05,
+      "loss": 0.025,
+      "step": 45170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9783042933114275e-05,
+      "loss": 0.0182,
+      "step": 45180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.97829948933758e-05,
+      "loss": 0.0175,
+      "step": 45190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978294685363733e-05,
+      "loss": 0.0188,
+      "step": 45200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978289881389886e-05,
+      "loss": 0.0231,
+      "step": 45210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978285077416039e-05,
+      "loss": 0.0184,
+      "step": 45220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9782802734421914e-05,
+      "loss": 0.0126,
+      "step": 45230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9782754694683444e-05,
+      "loss": 0.0233,
+      "step": 45240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9782706654944973e-05,
+      "loss": 0.0183,
+      "step": 45250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.97826586152065e-05,
+      "loss": 0.0163,
+      "step": 45260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9782610575468028e-05,
+      "loss": 0.0173,
+      "step": 45270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9782562535729557e-05,
+      "loss": 0.0201,
+      "step": 45280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9782514495991087e-05,
+      "loss": 0.0162,
+      "step": 45290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9782466456252613e-05,
+      "loss": 0.0193,
+      "step": 45300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9782418416514142e-05,
+      "loss": 0.0187,
+      "step": 45310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978237037677567e-05,
+      "loss": 0.0215,
+      "step": 45320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9782322337037197e-05,
+      "loss": 0.0219,
+      "step": 45330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9782274297298726e-05,
+      "loss": 0.0186,
+      "step": 45340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9782226257560255e-05,
+      "loss": 0.018,
+      "step": 45350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9782178217821785e-05,
+      "loss": 0.021,
+      "step": 45360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978213017808331e-05,
+      "loss": 0.0187,
+      "step": 45370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978208213834484e-05,
+      "loss": 0.0145,
+      "step": 45380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978203409860637e-05,
+      "loss": 0.0184,
+      "step": 45390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.97819860588679e-05,
+      "loss": 0.024,
+      "step": 45400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9781938019129424e-05,
+      "loss": 0.0246,
+      "step": 45410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9781889979390954e-05,
+      "loss": 0.0245,
+      "step": 45420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9781841939652483e-05,
+      "loss": 0.0183,
+      "step": 45430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.978179389991401e-05,
+      "loss": 0.0156,
+      "step": 45440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9781745860175538e-05,
+      "loss": 0.0183,
+      "step": 45450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9781697820437067e-05,
+      "loss": 0.0202,
+      "step": 45460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9781649780698597e-05,
+      "loss": 0.0171,
+      "step": 45470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9781601740960122e-05,
+      "loss": 0.0119,
+      "step": 45480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9781553701221652e-05,
+      "loss": 0.0181,
+      "step": 45490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.978150566148318e-05,
+      "loss": 0.0152,
+      "step": 45500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9781457621744707e-05,
+      "loss": 0.0211,
+      "step": 45510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9781409582006236e-05,
+      "loss": 0.0228,
+      "step": 45520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9781361542267765e-05,
+      "loss": 0.0199,
+      "step": 45530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9781313502529295e-05,
+      "loss": 0.0194,
+      "step": 45540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.978126546279082e-05,
+      "loss": 0.024,
+      "step": 45550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.978121742305235e-05,
+      "loss": 0.0166,
+      "step": 45560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.978116938331388e-05,
+      "loss": 0.0158,
+      "step": 45570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.978112134357541e-05,
+      "loss": 0.0183,
+      "step": 45580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9781073303836934e-05,
+      "loss": 0.0174,
+      "step": 45590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9781025264098464e-05,
+      "loss": 0.0237,
+      "step": 45600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780977224359993e-05,
+      "loss": 0.0172,
+      "step": 45610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.978092918462152e-05,
+      "loss": 0.0193,
+      "step": 45620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780881144883048e-05,
+      "loss": 0.014,
+      "step": 45630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780833105144577e-05,
+      "loss": 0.0205,
+      "step": 45640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780785065406106e-05,
+      "loss": 0.0196,
+      "step": 45650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780737025667632e-05,
+      "loss": 0.0168,
+      "step": 45660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.978068898592916e-05,
+      "loss": 0.0175,
+      "step": 45670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.978064094619069e-05,
+      "loss": 0.016,
+      "step": 45680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780592906452217e-05,
+      "loss": 0.0207,
+      "step": 45690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780544866713746e-05,
+      "loss": 0.0184,
+      "step": 45700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780496826975275e-05,
+      "loss": 0.0169,
+      "step": 45710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780448787236805e-05,
+      "loss": 0.019,
+      "step": 45720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.978040074749833e-05,
+      "loss": 0.0179,
+      "step": 45730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.978035270775986e-05,
+      "loss": 0.0181,
+      "step": 45740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.978030466802139e-05,
+      "loss": 0.0174,
+      "step": 45750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780256628282918e-05,
+      "loss": 0.0176,
+      "step": 45760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780208588544444e-05,
+      "loss": 0.0174,
+      "step": 45770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780160548805973e-05,
+      "loss": 0.0164,
+      "step": 45780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780112509067503e-05,
+      "loss": 0.0214,
+      "step": 45790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.978006446932903e-05,
+      "loss": 0.025,
+      "step": 45800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9780016429590558e-05,
+      "loss": 0.0225,
+      "step": 45810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779968389852087e-05,
+      "loss": 0.0211,
+      "step": 45820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779920350113616e-05,
+      "loss": 0.0233,
+      "step": 45830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779872310375142e-05,
+      "loss": 0.0149,
+      "step": 45840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.977982427063667e-05,
+      "loss": 0.0186,
+      "step": 45850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.97797762308982e-05,
+      "loss": 0.0198,
+      "step": 45860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779728191159727e-05,
+      "loss": 0.0157,
+      "step": 45870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779680151421256e-05,
+      "loss": 0.0163,
+      "step": 45880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779632111682785e-05,
+      "loss": 0.0235,
+      "step": 45890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779584071944315e-05,
+      "loss": 0.0188,
+      "step": 45900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.977953603220584e-05,
+      "loss": 0.0156,
+      "step": 45910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.977948799246737e-05,
+      "loss": 0.0186,
+      "step": 45920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.97794399527289e-05,
+      "loss": 0.0229,
+      "step": 45930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779391912990428e-05,
+      "loss": 0.0158,
+      "step": 45940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779343873251954e-05,
+      "loss": 0.0208,
+      "step": 45950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779295833513483e-05,
+      "loss": 0.0176,
+      "step": 45960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779247793775013e-05,
+      "loss": 0.0193,
+      "step": 45970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.977919975403654e-05,
+      "loss": 0.024,
+      "step": 45980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779151714298068e-05,
+      "loss": 0.0182,
+      "step": 45990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779103674559597e-05,
+      "loss": 0.0203,
+      "step": 46000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779055634821126e-05,
+      "loss": 0.0231,
+      "step": 46010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9779007595082652e-05,
+      "loss": 0.0195,
+      "step": 46020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.977895955534418e-05,
+      "loss": 0.0178,
+      "step": 46030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.977891151560571e-05,
+      "loss": 0.0184,
+      "step": 46040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9778863475867237e-05,
+      "loss": 0.0196,
+      "step": 46050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9778815436128766e-05,
+      "loss": 0.0129,
+      "step": 46060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9778767396390295e-05,
+      "loss": 0.0189,
+      "step": 46070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9778719356651824e-05,
+      "loss": 0.0183,
+      "step": 46080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.977867131691335e-05,
+      "loss": 0.0174,
+      "step": 46090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.977862327717488e-05,
+      "loss": 0.0229,
+      "step": 46100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.977857523743641e-05,
+      "loss": 0.0157,
+      "step": 46110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9778527197697938e-05,
+      "loss": 0.0196,
+      "step": 46120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9778479157959464e-05,
+      "loss": 0.0231,
+      "step": 46130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9778431118220993e-05,
+      "loss": 0.0213,
+      "step": 46140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9778383078482523e-05,
+      "loss": 0.0163,
+      "step": 46150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977833503874405e-05,
+      "loss": 0.0259,
+      "step": 46160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9778286999005578e-05,
+      "loss": 0.0171,
+      "step": 46170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9778238959267107e-05,
+      "loss": 0.0145,
+      "step": 46180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9778190919528636e-05,
+      "loss": 0.0224,
+      "step": 46190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9778142879790162e-05,
+      "loss": 0.0152,
+      "step": 46200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977809484005169e-05,
+      "loss": 0.0206,
+      "step": 46210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977804680031322e-05,
+      "loss": 0.0183,
+      "step": 46220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777998760574747e-05,
+      "loss": 0.014,
+      "step": 46230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777950720836276e-05,
+      "loss": 0.0212,
+      "step": 46240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777902681097805e-05,
+      "loss": 0.0185,
+      "step": 46250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777854641359334e-05,
+      "loss": 0.0198,
+      "step": 46260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977780660162086e-05,
+      "loss": 0.0227,
+      "step": 46270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977775856188239e-05,
+      "loss": 0.0204,
+      "step": 46280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977771052214392e-05,
+      "loss": 0.0181,
+      "step": 46290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777662482405448e-05,
+      "loss": 0.0176,
+      "step": 46300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777614442666974e-05,
+      "loss": 0.0259,
+      "step": 46310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777566402928503e-05,
+      "loss": 0.0152,
+      "step": 46320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777518363190032e-05,
+      "loss": 0.0155,
+      "step": 46330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977747032345156e-05,
+      "loss": 0.0162,
+      "step": 46340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777422283713088e-05,
+      "loss": 0.0177,
+      "step": 46350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777374243974617e-05,
+      "loss": 0.0219,
+      "step": 46360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777326204236146e-05,
+      "loss": 0.0189,
+      "step": 46370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777278164497672e-05,
+      "loss": 0.016,
+      "step": 46380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.97772301247592e-05,
+      "loss": 0.019,
+      "step": 46390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977718208502073e-05,
+      "loss": 0.0242,
+      "step": 46400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9777134045282256e-05,
+      "loss": 0.0226,
+      "step": 46410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977708600554379e-05,
+      "loss": 0.022,
+      "step": 46420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977703796580532e-05,
+      "loss": 0.018,
+      "step": 46430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776989926066844e-05,
+      "loss": 0.0233,
+      "step": 46440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776941886328374e-05,
+      "loss": 0.0188,
+      "step": 46450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776893846589903e-05,
+      "loss": 0.0235,
+      "step": 46460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977684580685143e-05,
+      "loss": 0.0195,
+      "step": 46470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776797767112958e-05,
+      "loss": 0.0152,
+      "step": 46480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776749727374487e-05,
+      "loss": 0.0171,
+      "step": 46490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776701687636016e-05,
+      "loss": 0.0158,
+      "step": 46500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776653647897542e-05,
+      "loss": 0.0253,
+      "step": 46510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977660560815907e-05,
+      "loss": 0.0178,
+      "step": 46520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.97765575684206e-05,
+      "loss": 0.0168,
+      "step": 46530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977650952868213e-05,
+      "loss": 0.0199,
+      "step": 46540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776461488943656e-05,
+      "loss": 0.018,
+      "step": 46550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776413449205185e-05,
+      "loss": 0.02,
+      "step": 46560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776365409466715e-05,
+      "loss": 0.0186,
+      "step": 46570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977631736972824e-05,
+      "loss": 0.016,
+      "step": 46580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977626932998977e-05,
+      "loss": 0.0221,
+      "step": 46590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.97762212902513e-05,
+      "loss": 0.0139,
+      "step": 46600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776173250512828e-05,
+      "loss": 0.0186,
+      "step": 46610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776125210774354e-05,
+      "loss": 0.0154,
+      "step": 46620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776077171035883e-05,
+      "loss": 0.0181,
+      "step": 46630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9776029131297413e-05,
+      "loss": 0.0187,
+      "step": 46640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977598109155894e-05,
+      "loss": 0.0171,
+      "step": 46650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9775933051820468e-05,
+      "loss": 0.0237,
+      "step": 46660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9775885012081997e-05,
+      "loss": 0.0176,
+      "step": 46670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9775836972343526e-05,
+      "loss": 0.0173,
+      "step": 46680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9775788932605052e-05,
+      "loss": 0.0193,
+      "step": 46690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977574089286658e-05,
+      "loss": 0.0131,
+      "step": 46700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977569285312811e-05,
+      "loss": 0.0154,
+      "step": 46710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9775644813389637e-05,
+      "loss": 0.0201,
+      "step": 46720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9775596773651166e-05,
+      "loss": 0.0227,
+      "step": 46730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9775548733912695e-05,
+      "loss": 0.0142,
+      "step": 46740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9775500694174225e-05,
+      "loss": 0.0177,
+      "step": 46750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977545265443575e-05,
+      "loss": 0.0263,
+      "step": 46760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977540461469728e-05,
+      "loss": 0.0148,
+      "step": 46770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977535657495881e-05,
+      "loss": 0.0206,
+      "step": 46780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9775308535220338e-05,
+      "loss": 0.0161,
+      "step": 46790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9775260495481864e-05,
+      "loss": 0.0179,
+      "step": 46800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9775212455743393e-05,
+      "loss": 0.0232,
+      "step": 46810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9775164416004923e-05,
+      "loss": 0.0155,
+      "step": 46820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.977511637626645e-05,
+      "loss": 0.0183,
+      "step": 46830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9775068336527978e-05,
+      "loss": 0.0194,
+      "step": 46840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9775020296789507e-05,
+      "loss": 0.0241,
+      "step": 46850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774972257051036e-05,
+      "loss": 0.0172,
+      "step": 46860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774924217312562e-05,
+      "loss": 0.0151,
+      "step": 46870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977487617757409e-05,
+      "loss": 0.0228,
+      "step": 46880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977482813783562e-05,
+      "loss": 0.0161,
+      "step": 46890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774780098097147e-05,
+      "loss": 0.0163,
+      "step": 46900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774732058358676e-05,
+      "loss": 0.0194,
+      "step": 46910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774684018620205e-05,
+      "loss": 0.0157,
+      "step": 46920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774635978881734e-05,
+      "loss": 0.0188,
+      "step": 46930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977458793914326e-05,
+      "loss": 0.0147,
+      "step": 46940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977453989940479e-05,
+      "loss": 0.0182,
+      "step": 46950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977449185966632e-05,
+      "loss": 0.0189,
+      "step": 46960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774443819927848e-05,
+      "loss": 0.0167,
+      "step": 46970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774395780189374e-05,
+      "loss": 0.0185,
+      "step": 46980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774347740450903e-05,
+      "loss": 0.0199,
+      "step": 46990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774299700712433e-05,
+      "loss": 0.0258,
+      "step": 47000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977425166097396e-05,
+      "loss": 0.0234,
+      "step": 47010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774203621235488e-05,
+      "loss": 0.0196,
+      "step": 47020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774155581497017e-05,
+      "loss": 0.0175,
+      "step": 47030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774107541758546e-05,
+      "loss": 0.0198,
+      "step": 47040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9774059502020072e-05,
+      "loss": 0.0286,
+      "step": 47050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.97740114622816e-05,
+      "loss": 0.0151,
+      "step": 47060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977396342254313e-05,
+      "loss": 0.0226,
+      "step": 47070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773915382804657e-05,
+      "loss": 0.0183,
+      "step": 47080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773867343066186e-05,
+      "loss": 0.0193,
+      "step": 47090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773819303327715e-05,
+      "loss": 0.0171,
+      "step": 47100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773771263589244e-05,
+      "loss": 0.0244,
+      "step": 47110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977372322385077e-05,
+      "loss": 0.0191,
+      "step": 47120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.97736751841123e-05,
+      "loss": 0.0214,
+      "step": 47130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977362714437383e-05,
+      "loss": 0.0224,
+      "step": 47140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773579104635358e-05,
+      "loss": 0.024,
+      "step": 47150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773531064896884e-05,
+      "loss": 0.0146,
+      "step": 47160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773483025158413e-05,
+      "loss": 0.0198,
+      "step": 47170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773434985419942e-05,
+      "loss": 0.0221,
+      "step": 47180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977338694568147e-05,
+      "loss": 0.0148,
+      "step": 47190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773338905942998e-05,
+      "loss": 0.0169,
+      "step": 47200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773290866204527e-05,
+      "loss": 0.0231,
+      "step": 47210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773242826466056e-05,
+      "loss": 0.0211,
+      "step": 47220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773194786727582e-05,
+      "loss": 0.0155,
+      "step": 47230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977314674698911e-05,
+      "loss": 0.0203,
+      "step": 47240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977309870725064e-05,
+      "loss": 0.0179,
+      "step": 47250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773050667512166e-05,
+      "loss": 0.0244,
+      "step": 47260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9773002627773696e-05,
+      "loss": 0.0235,
+      "step": 47270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772954588035225e-05,
+      "loss": 0.0171,
+      "step": 47280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772906548296754e-05,
+      "loss": 0.0215,
+      "step": 47290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977285850855828e-05,
+      "loss": 0.0195,
+      "step": 47300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977281046881981e-05,
+      "loss": 0.0172,
+      "step": 47310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977276242908134e-05,
+      "loss": 0.0129,
+      "step": 47320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772714389342868e-05,
+      "loss": 0.021,
+      "step": 47330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772666349604394e-05,
+      "loss": 0.021,
+      "step": 47340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772618309865923e-05,
+      "loss": 0.019,
+      "step": 47350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772570270127452e-05,
+      "loss": 0.021,
+      "step": 47360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772522230388978e-05,
+      "loss": 0.0153,
+      "step": 47370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772474190650508e-05,
+      "loss": 0.0225,
+      "step": 47380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772426150912037e-05,
+      "loss": 0.0192,
+      "step": 47390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772378111173566e-05,
+      "loss": 0.0152,
+      "step": 47400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772330071435092e-05,
+      "loss": 0.0165,
+      "step": 47410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977228203169662e-05,
+      "loss": 0.0394,
+      "step": 47420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977223399195815e-05,
+      "loss": 0.0194,
+      "step": 47430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772185952219676e-05,
+      "loss": 0.0145,
+      "step": 47440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772137912481206e-05,
+      "loss": 0.0191,
+      "step": 47450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772089872742735e-05,
+      "loss": 0.0153,
+      "step": 47460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9772041833004264e-05,
+      "loss": 0.0236,
+      "step": 47470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977199379326579e-05,
+      "loss": 0.0189,
+      "step": 47480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977194575352732e-05,
+      "loss": 0.0191,
+      "step": 47490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977189771378885e-05,
+      "loss": 0.0193,
+      "step": 47500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9771849674050378e-05,
+      "loss": 0.0225,
+      "step": 47510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9771801634311904e-05,
+      "loss": 0.0183,
+      "step": 47520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9771753594573433e-05,
+      "loss": 0.0194,
+      "step": 47530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9771705554834962e-05,
+      "loss": 0.0179,
+      "step": 47540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9771657515096488e-05,
+      "loss": 0.0202,
+      "step": 47550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9771609475358017e-05,
+      "loss": 0.017,
+      "step": 47560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9771561435619547e-05,
+      "loss": 0.0162,
+      "step": 47570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9771513395881076e-05,
+      "loss": 0.0204,
+      "step": 47580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9771465356142602e-05,
+      "loss": 0.0176,
+      "step": 47590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.977141731640413e-05,
+      "loss": 0.0128,
+      "step": 47600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.977136927666566e-05,
+      "loss": 0.019,
+      "step": 47610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9771321236927186e-05,
+      "loss": 0.0208,
+      "step": 47620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9771273197188716e-05,
+      "loss": 0.0198,
+      "step": 47630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9771225157450245e-05,
+      "loss": 0.0163,
+      "step": 47640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9771177117711774e-05,
+      "loss": 0.018,
+      "step": 47650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.97711290779733e-05,
+      "loss": 0.0208,
+      "step": 47660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.977108103823483e-05,
+      "loss": 0.0157,
+      "step": 47670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.977103299849636e-05,
+      "loss": 0.017,
+      "step": 47680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770984958757888e-05,
+      "loss": 0.019,
+      "step": 47690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770936919019414e-05,
+      "loss": 0.0204,
+      "step": 47700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770888879280943e-05,
+      "loss": 0.0168,
+      "step": 47710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770840839542472e-05,
+      "loss": 0.015,
+      "step": 47720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770792799803998e-05,
+      "loss": 0.0121,
+      "step": 47730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770744760065527e-05,
+      "loss": 0.0201,
+      "step": 47740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770696720327057e-05,
+      "loss": 0.0166,
+      "step": 47750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770648680588586e-05,
+      "loss": 0.0176,
+      "step": 47760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770600640850112e-05,
+      "loss": 0.0205,
+      "step": 47770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.977055260111164e-05,
+      "loss": 0.0151,
+      "step": 47780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.977050456137317e-05,
+      "loss": 0.0154,
+      "step": 47790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770456521634696e-05,
+      "loss": 0.0174,
+      "step": 47800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770408481896225e-05,
+      "loss": 0.0151,
+      "step": 47810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770360442157755e-05,
+      "loss": 0.0188,
+      "step": 47820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770312402419284e-05,
+      "loss": 0.012,
+      "step": 47830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.977026436268081e-05,
+      "loss": 0.0284,
+      "step": 47840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.977021632294234e-05,
+      "loss": 0.0227,
+      "step": 47850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.977016828320387e-05,
+      "loss": 0.02,
+      "step": 47860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770120243465398e-05,
+      "loss": 0.0203,
+      "step": 47870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770072203726924e-05,
+      "loss": 0.0181,
+      "step": 47880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9770024163988453e-05,
+      "loss": 0.0239,
+      "step": 47890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769976124249982e-05,
+      "loss": 0.0207,
+      "step": 47900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769928084511508e-05,
+      "loss": 0.0173,
+      "step": 47910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769880044773037e-05,
+      "loss": 0.0179,
+      "step": 47920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769832005034567e-05,
+      "loss": 0.0181,
+      "step": 47930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769783965296096e-05,
+      "loss": 0.0217,
+      "step": 47940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.976973592555762e-05,
+      "loss": 0.0238,
+      "step": 47950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.976968788581915e-05,
+      "loss": 0.0196,
+      "step": 47960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.976963984608068e-05,
+      "loss": 0.0161,
+      "step": 47970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769591806342206e-05,
+      "loss": 0.0138,
+      "step": 47980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769543766603735e-05,
+      "loss": 0.0233,
+      "step": 47990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769495726865265e-05,
+      "loss": 0.0204,
+      "step": 48000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769447687126794e-05,
+      "loss": 0.0212,
+      "step": 48010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.976939964738832e-05,
+      "loss": 0.0218,
+      "step": 48020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.976935160764985e-05,
+      "loss": 0.0189,
+      "step": 48030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.976930356791138e-05,
+      "loss": 0.0178,
+      "step": 48040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769255528172908e-05,
+      "loss": 0.0142,
+      "step": 48050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769207488434433e-05,
+      "loss": 0.0163,
+      "step": 48060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769159448695963e-05,
+      "loss": 0.0229,
+      "step": 48070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769111408957492e-05,
+      "loss": 0.0177,
+      "step": 48080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769063369219018e-05,
+      "loss": 0.0155,
+      "step": 48090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9769015329480547e-05,
+      "loss": 0.0177,
+      "step": 48100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9768967289742076e-05,
+      "loss": 0.0196,
+      "step": 48110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9768919250003606e-05,
+      "loss": 0.0273,
+      "step": 48120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.976887121026513e-05,
+      "loss": 0.0178,
+      "step": 48130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.976882317052666e-05,
+      "loss": 0.0127,
+      "step": 48140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.976877513078819e-05,
+      "loss": 0.0239,
+      "step": 48150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9768727091049716e-05,
+      "loss": 0.0149,
+      "step": 48160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9768679051311245e-05,
+      "loss": 0.0181,
+      "step": 48170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9768631011572775e-05,
+      "loss": 0.0171,
+      "step": 48180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9768582971834304e-05,
+      "loss": 0.0198,
+      "step": 48190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.976853493209583e-05,
+      "loss": 0.015,
+      "step": 48200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.976848689235736e-05,
+      "loss": 0.0205,
+      "step": 48210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9768438852618888e-05,
+      "loss": 0.021,
+      "step": 48220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9768390812880418e-05,
+      "loss": 0.0196,
+      "step": 48230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9768342773141943e-05,
+      "loss": 0.0203,
+      "step": 48240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9768294733403473e-05,
+      "loss": 0.013,
+      "step": 48250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9768246693665002e-05,
+      "loss": 0.0262,
+      "step": 48260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9768198653926528e-05,
+      "loss": 0.0195,
+      "step": 48270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9768150614188057e-05,
+      "loss": 0.0168,
+      "step": 48280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9768102574449586e-05,
+      "loss": 0.0183,
+      "step": 48290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9768054534711116e-05,
+      "loss": 0.0202,
+      "step": 48300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976800649497264e-05,
+      "loss": 0.0188,
+      "step": 48310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976795845523417e-05,
+      "loss": 0.0192,
+      "step": 48320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.97679104154957e-05,
+      "loss": 0.02,
+      "step": 48330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767862375757226e-05,
+      "loss": 0.0172,
+      "step": 48340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767814336018755e-05,
+      "loss": 0.0192,
+      "step": 48350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767766296280284e-05,
+      "loss": 0.0158,
+      "step": 48360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767718256541814e-05,
+      "loss": 0.0172,
+      "step": 48370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976767021680334e-05,
+      "loss": 0.0138,
+      "step": 48380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976762217706487e-05,
+      "loss": 0.0151,
+      "step": 48390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767574137326398e-05,
+      "loss": 0.0162,
+      "step": 48400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767526097587924e-05,
+      "loss": 0.02,
+      "step": 48410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767478057849453e-05,
+      "loss": 0.0188,
+      "step": 48420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767430018110983e-05,
+      "loss": 0.0155,
+      "step": 48430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767381978372512e-05,
+      "loss": 0.0178,
+      "step": 48440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767333938634038e-05,
+      "loss": 0.0229,
+      "step": 48450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767285898895567e-05,
+      "loss": 0.016,
+      "step": 48460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767237859157096e-05,
+      "loss": 0.0164,
+      "step": 48470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9767189819418626e-05,
+      "loss": 0.0197,
+      "step": 48480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976714177968015e-05,
+      "loss": 0.0183,
+      "step": 48490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976709373994168e-05,
+      "loss": 0.0155,
+      "step": 48500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976704570020321e-05,
+      "loss": 0.0181,
+      "step": 48510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766997660464736e-05,
+      "loss": 0.0196,
+      "step": 48520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766949620726265e-05,
+      "loss": 0.0138,
+      "step": 48530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766901580987794e-05,
+      "loss": 0.0157,
+      "step": 48540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766853541249324e-05,
+      "loss": 0.021,
+      "step": 48550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976680550151085e-05,
+      "loss": 0.0148,
+      "step": 48560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976675746177238e-05,
+      "loss": 0.0198,
+      "step": 48570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766709422033908e-05,
+      "loss": 0.0179,
+      "step": 48580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766661382295434e-05,
+      "loss": 0.0186,
+      "step": 48590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766613342556963e-05,
+      "loss": 0.0243,
+      "step": 48600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766565302818492e-05,
+      "loss": 0.0198,
+      "step": 48610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766517263080022e-05,
+      "loss": 0.0183,
+      "step": 48620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766469223341548e-05,
+      "loss": 0.021,
+      "step": 48630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766421183603077e-05,
+      "loss": 0.0182,
+      "step": 48640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766373143864606e-05,
+      "loss": 0.0202,
+      "step": 48650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766325104126135e-05,
+      "loss": 0.0156,
+      "step": 48660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976627706438766e-05,
+      "loss": 0.0197,
+      "step": 48670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976622902464919e-05,
+      "loss": 0.0208,
+      "step": 48680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976618098491072e-05,
+      "loss": 0.0132,
+      "step": 48690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766132945172246e-05,
+      "loss": 0.0205,
+      "step": 48700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766084905433775e-05,
+      "loss": 0.0141,
+      "step": 48710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9766036865695304e-05,
+      "loss": 0.0178,
+      "step": 48720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765988825956834e-05,
+      "loss": 0.0196,
+      "step": 48730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976594078621836e-05,
+      "loss": 0.0191,
+      "step": 48740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976589274647989e-05,
+      "loss": 0.0196,
+      "step": 48750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765844706741418e-05,
+      "loss": 0.0125,
+      "step": 48760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765796667002944e-05,
+      "loss": 0.0125,
+      "step": 48770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765748627264473e-05,
+      "loss": 0.0181,
+      "step": 48780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765700587526002e-05,
+      "loss": 0.0199,
+      "step": 48790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976565254778753e-05,
+      "loss": 0.023,
+      "step": 48800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765604508049058e-05,
+      "loss": 0.0174,
+      "step": 48810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765556468310587e-05,
+      "loss": 0.0145,
+      "step": 48820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765508428572116e-05,
+      "loss": 0.0157,
+      "step": 48830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765460388833645e-05,
+      "loss": 0.0186,
+      "step": 48840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976541234909517e-05,
+      "loss": 0.0183,
+      "step": 48850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.97653643093567e-05,
+      "loss": 0.0184,
+      "step": 48860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.976531626961823e-05,
+      "loss": 0.0201,
+      "step": 48870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765268229879756e-05,
+      "loss": 0.0206,
+      "step": 48880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765220190141285e-05,
+      "loss": 0.0158,
+      "step": 48890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765172150402814e-05,
+      "loss": 0.018,
+      "step": 48900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9765124110664343e-05,
+      "loss": 0.0146,
+      "step": 48910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976507607092587e-05,
+      "loss": 0.0191,
+      "step": 48920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.97650280311874e-05,
+      "loss": 0.0193,
+      "step": 48930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764979991448928e-05,
+      "loss": 0.0186,
+      "step": 48940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764931951710454e-05,
+      "loss": 0.0168,
+      "step": 48950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764883911971983e-05,
+      "loss": 0.0157,
+      "step": 48960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764835872233512e-05,
+      "loss": 0.018,
+      "step": 48970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976478783249504e-05,
+      "loss": 0.0325,
+      "step": 48980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976473979275657e-05,
+      "loss": 0.0209,
+      "step": 48990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.97646917530181e-05,
+      "loss": 0.0174,
+      "step": 49000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764643713279626e-05,
+      "loss": 0.0184,
+      "step": 49010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764595673541155e-05,
+      "loss": 0.018,
+      "step": 49020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764547633802685e-05,
+      "loss": 0.0207,
+      "step": 49030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764499594064214e-05,
+      "loss": 0.02,
+      "step": 49040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976445155432574e-05,
+      "loss": 0.0233,
+      "step": 49050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976440351458727e-05,
+      "loss": 0.0198,
+      "step": 49060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764355474848798e-05,
+      "loss": 0.0239,
+      "step": 49070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764307435110328e-05,
+      "loss": 0.0168,
+      "step": 49080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764259395371853e-05,
+      "loss": 0.0245,
+      "step": 49090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764211355633383e-05,
+      "loss": 0.0203,
+      "step": 49100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764163315894912e-05,
+      "loss": 0.0173,
+      "step": 49110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764115276156438e-05,
+      "loss": 0.0189,
+      "step": 49120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764067236417967e-05,
+      "loss": 0.0166,
+      "step": 49130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9764019196679496e-05,
+      "loss": 0.0144,
+      "step": 49140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763971156941026e-05,
+      "loss": 0.0184,
+      "step": 49150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976392311720255e-05,
+      "loss": 0.0222,
+      "step": 49160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976387507746408e-05,
+      "loss": 0.0173,
+      "step": 49170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976382703772561e-05,
+      "loss": 0.0239,
+      "step": 49180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763778997987136e-05,
+      "loss": 0.0183,
+      "step": 49190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763730958248665e-05,
+      "loss": 0.0185,
+      "step": 49200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763682918510194e-05,
+      "loss": 0.0177,
+      "step": 49210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763634878771724e-05,
+      "loss": 0.0163,
+      "step": 49220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976358683903325e-05,
+      "loss": 0.0166,
+      "step": 49230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976353879929478e-05,
+      "loss": 0.0154,
+      "step": 49240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763490759556308e-05,
+      "loss": 0.0236,
+      "step": 49250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763442719817837e-05,
+      "loss": 0.0159,
+      "step": 49260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763394680079363e-05,
+      "loss": 0.0178,
+      "step": 49270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763346640340893e-05,
+      "loss": 0.0217,
+      "step": 49280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763298600602422e-05,
+      "loss": 0.0188,
+      "step": 49290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763250560863948e-05,
+      "loss": 0.0204,
+      "step": 49300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763202521125477e-05,
+      "loss": 0.0191,
+      "step": 49310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763154481387006e-05,
+      "loss": 0.0156,
+      "step": 49320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9763106441648536e-05,
+      "loss": 0.0188,
+      "step": 49330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976305840191006e-05,
+      "loss": 0.0138,
+      "step": 49340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976301036217159e-05,
+      "loss": 0.0178,
+      "step": 49350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976296232243312e-05,
+      "loss": 0.0175,
+      "step": 49360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762914282694646e-05,
+      "loss": 0.0128,
+      "step": 49370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762866242956175e-05,
+      "loss": 0.0188,
+      "step": 49380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762818203217704e-05,
+      "loss": 0.0217,
+      "step": 49390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762770163479234e-05,
+      "loss": 0.0175,
+      "step": 49400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976272212374076e-05,
+      "loss": 0.0214,
+      "step": 49410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976267408400229e-05,
+      "loss": 0.0187,
+      "step": 49420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762626044263818e-05,
+      "loss": 0.0153,
+      "step": 49430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762578004525347e-05,
+      "loss": 0.0192,
+      "step": 49440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762529964786873e-05,
+      "loss": 0.0189,
+      "step": 49450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762481925048402e-05,
+      "loss": 0.0157,
+      "step": 49460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762433885309932e-05,
+      "loss": 0.0176,
+      "step": 49470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762385845571458e-05,
+      "loss": 0.0191,
+      "step": 49480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762337805832987e-05,
+      "loss": 0.0156,
+      "step": 49490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762289766094516e-05,
+      "loss": 0.0167,
+      "step": 49500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762241726356045e-05,
+      "loss": 0.0171,
+      "step": 49510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976219368661757e-05,
+      "loss": 0.0169,
+      "step": 49520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.97621456468791e-05,
+      "loss": 0.0131,
+      "step": 49530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976209760714063e-05,
+      "loss": 0.0215,
+      "step": 49540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762054371376005e-05,
+      "loss": 0.0205,
+      "step": 49550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9762006331637534e-05,
+      "loss": 0.019,
+      "step": 49560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9761958291899063e-05,
+      "loss": 0.0164,
+      "step": 49570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.976191025216059e-05,
+      "loss": 0.0175,
+      "step": 49580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9761862212422118e-05,
+      "loss": 0.0162,
+      "step": 49590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9761814172683647e-05,
+      "loss": 0.0169,
+      "step": 49600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9761766132945173e-05,
+      "loss": 0.0154,
+      "step": 49610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9761718093206703e-05,
+      "loss": 0.0178,
+      "step": 49620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9761670053468232e-05,
+      "loss": 0.0151,
+      "step": 49630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976162201372976e-05,
+      "loss": 0.0181,
+      "step": 49640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9761573973991287e-05,
+      "loss": 0.0137,
+      "step": 49650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9761525934252816e-05,
+      "loss": 0.0179,
+      "step": 49660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9761477894514346e-05,
+      "loss": 0.0172,
+      "step": 49670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976142985477587e-05,
+      "loss": 0.0119,
+      "step": 49680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.97613818150374e-05,
+      "loss": 0.0168,
+      "step": 49690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976133377529893e-05,
+      "loss": 0.0179,
+      "step": 49700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976128573556046e-05,
+      "loss": 0.014,
+      "step": 49710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9761237695821985e-05,
+      "loss": 0.0182,
+      "step": 49720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9761189656083514e-05,
+      "loss": 0.0206,
+      "step": 49730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9761141616345044e-05,
+      "loss": 0.0178,
+      "step": 49740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9761093576606573e-05,
+      "loss": 0.0121,
+      "step": 49750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.97610455368681e-05,
+      "loss": 0.0201,
+      "step": 49760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760997497129628e-05,
+      "loss": 0.0236,
+      "step": 49770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760949457391157e-05,
+      "loss": 0.0174,
+      "step": 49780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760901417652683e-05,
+      "loss": 0.0218,
+      "step": 49790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760853377914213e-05,
+      "loss": 0.0199,
+      "step": 49800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760805338175742e-05,
+      "loss": 0.0127,
+      "step": 49810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976075729843727e-05,
+      "loss": 0.0217,
+      "step": 49820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760709258698797e-05,
+      "loss": 0.0159,
+      "step": 49830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760661218960326e-05,
+      "loss": 0.014,
+      "step": 49840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760613179221856e-05,
+      "loss": 0.0148,
+      "step": 49850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976056513948338e-05,
+      "loss": 0.022,
+      "step": 49860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976051709974491e-05,
+      "loss": 0.0201,
+      "step": 49870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976046906000644e-05,
+      "loss": 0.0194,
+      "step": 49880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976042102026797e-05,
+      "loss": 0.0238,
+      "step": 49890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760372980529495e-05,
+      "loss": 0.0129,
+      "step": 49900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760324940791024e-05,
+      "loss": 0.0133,
+      "step": 49910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760276901052554e-05,
+      "loss": 0.0184,
+      "step": 49920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760228861314083e-05,
+      "loss": 0.0163,
+      "step": 49930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976018082157561e-05,
+      "loss": 0.0114,
+      "step": 49940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760132781837138e-05,
+      "loss": 0.0166,
+      "step": 49950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760084742098667e-05,
+      "loss": 0.0201,
+      "step": 49960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9760036702360193e-05,
+      "loss": 0.021,
+      "step": 49970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759988662621722e-05,
+      "loss": 0.0218,
+      "step": 49980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759940622883252e-05,
+      "loss": 0.0165,
+      "step": 49990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975989258314478e-05,
+      "loss": 0.0235,
+      "step": 50000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759844543406307e-05,
+      "loss": 0.0189,
+      "step": 50010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759796503667836e-05,
+      "loss": 0.0217,
+      "step": 50020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759748463929365e-05,
+      "loss": 0.0179,
+      "step": 50030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975970042419089e-05,
+      "loss": 0.0191,
+      "step": 50040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975965238445242e-05,
+      "loss": 0.0177,
+      "step": 50050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975960434471395e-05,
+      "loss": 0.0151,
+      "step": 50060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975955630497548e-05,
+      "loss": 0.022,
+      "step": 50070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759508265237005e-05,
+      "loss": 0.0181,
+      "step": 50080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759460225498534e-05,
+      "loss": 0.0203,
+      "step": 50090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759412185760064e-05,
+      "loss": 0.0171,
+      "step": 50100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759364146021593e-05,
+      "loss": 0.0187,
+      "step": 50110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975931610628312e-05,
+      "loss": 0.0194,
+      "step": 50120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759268066544648e-05,
+      "loss": 0.0165,
+      "step": 50130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759220026806177e-05,
+      "loss": 0.0152,
+      "step": 50140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759171987067703e-05,
+      "loss": 0.0132,
+      "step": 50150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759123947329232e-05,
+      "loss": 0.0223,
+      "step": 50160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975907590759076e-05,
+      "loss": 0.0162,
+      "step": 50170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975902786785229e-05,
+      "loss": 0.019,
+      "step": 50180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9758979828113817e-05,
+      "loss": 0.0206,
+      "step": 50190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9758931788375346e-05,
+      "loss": 0.0168,
+      "step": 50200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9758883748636875e-05,
+      "loss": 0.0199,
+      "step": 50210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.97588357088984e-05,
+      "loss": 0.0212,
+      "step": 50220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975878766915993e-05,
+      "loss": 0.0224,
+      "step": 50230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975873962942146e-05,
+      "loss": 0.0208,
+      "step": 50240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975869158968299e-05,
+      "loss": 0.0192,
+      "step": 50250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9758643549944515e-05,
+      "loss": 0.0157,
+      "step": 50260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9758595510206044e-05,
+      "loss": 0.0195,
+      "step": 50270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9758547470467573e-05,
+      "loss": 0.0154,
+      "step": 50280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9758499430729103e-05,
+      "loss": 0.0181,
+      "step": 50290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975845139099063e-05,
+      "loss": 0.0139,
+      "step": 50300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9758403351252158e-05,
+      "loss": 0.0166,
+      "step": 50310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9758355311513687e-05,
+      "loss": 0.0188,
+      "step": 50320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9758307271775213e-05,
+      "loss": 0.0123,
+      "step": 50330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9758259232036742e-05,
+      "loss": 0.016,
+      "step": 50340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975821119229827e-05,
+      "loss": 0.0171,
+      "step": 50350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.97581631525598e-05,
+      "loss": 0.0242,
+      "step": 50360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9758115112821327e-05,
+      "loss": 0.0161,
+      "step": 50370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9758067073082856e-05,
+      "loss": 0.016,
+      "step": 50380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9758019033344385e-05,
+      "loss": 0.0131,
+      "step": 50390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975797099360591e-05,
+      "loss": 0.0169,
+      "step": 50400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975792295386744e-05,
+      "loss": 0.0159,
+      "step": 50410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975787491412897e-05,
+      "loss": 0.0156,
+      "step": 50420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.97578268743905e-05,
+      "loss": 0.0187,
+      "step": 50430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9757778834652025e-05,
+      "loss": 0.0183,
+      "step": 50440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9757730794913554e-05,
+      "loss": 0.0153,
+      "step": 50450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9757682755175083e-05,
+      "loss": 0.0175,
+      "step": 50460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9757634715436613e-05,
+      "loss": 0.0187,
+      "step": 50470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975758667569814e-05,
+      "loss": 0.0156,
+      "step": 50480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9757538635959668e-05,
+      "loss": 0.0168,
+      "step": 50490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9757490596221197e-05,
+      "loss": 0.015,
+      "step": 50500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9757442556482723e-05,
+      "loss": 0.017,
+      "step": 50510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9757394516744252e-05,
+      "loss": 0.0144,
+      "step": 50520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975734647700578e-05,
+      "loss": 0.0195,
+      "step": 50530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975729843726731e-05,
+      "loss": 0.0134,
+      "step": 50540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9757250397528837e-05,
+      "loss": 0.019,
+      "step": 50550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9757202357790366e-05,
+      "loss": 0.0189,
+      "step": 50560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9757154318051895e-05,
+      "loss": 0.0186,
+      "step": 50570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975710627831342e-05,
+      "loss": 0.0182,
+      "step": 50580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975705823857495e-05,
+      "loss": 0.016,
+      "step": 50590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975701019883648e-05,
+      "loss": 0.019,
+      "step": 50600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975696215909801e-05,
+      "loss": 0.0186,
+      "step": 50610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756914119359535e-05,
+      "loss": 0.0166,
+      "step": 50620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756866079621064e-05,
+      "loss": 0.019,
+      "step": 50630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756818039882593e-05,
+      "loss": 0.0125,
+      "step": 50640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756770000144123e-05,
+      "loss": 0.0133,
+      "step": 50650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975672196040565e-05,
+      "loss": 0.0234,
+      "step": 50660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756673920667178e-05,
+      "loss": 0.0203,
+      "step": 50670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756625880928707e-05,
+      "loss": 0.0173,
+      "step": 50680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756577841190233e-05,
+      "loss": 0.0173,
+      "step": 50690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756529801451762e-05,
+      "loss": 0.024,
+      "step": 50700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975648176171329e-05,
+      "loss": 0.0179,
+      "step": 50710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975643372197482e-05,
+      "loss": 0.016,
+      "step": 50720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756385682236347e-05,
+      "loss": 0.0212,
+      "step": 50730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756337642497876e-05,
+      "loss": 0.0155,
+      "step": 50740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756289602759405e-05,
+      "loss": 0.0199,
+      "step": 50750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975624156302093e-05,
+      "loss": 0.0192,
+      "step": 50760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975619352328246e-05,
+      "loss": 0.0229,
+      "step": 50770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975614548354399e-05,
+      "loss": 0.0177,
+      "step": 50780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975609744380552e-05,
+      "loss": 0.0197,
+      "step": 50790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756049404067045e-05,
+      "loss": 0.0159,
+      "step": 50800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9756001364328574e-05,
+      "loss": 0.0166,
+      "step": 50810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9755953324590103e-05,
+      "loss": 0.0183,
+      "step": 50820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9755905284851632e-05,
+      "loss": 0.0166,
+      "step": 50830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975585724511316e-05,
+      "loss": 0.0134,
+      "step": 50840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9755809205374688e-05,
+      "loss": 0.0251,
+      "step": 50850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9755761165636217e-05,
+      "loss": 0.0142,
+      "step": 50860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9755713125897743e-05,
+      "loss": 0.0173,
+      "step": 50870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9755665086159272e-05,
+      "loss": 0.0215,
+      "step": 50880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.97556170464208e-05,
+      "loss": 0.0167,
+      "step": 50890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975556900668233e-05,
+      "loss": 0.0204,
+      "step": 50900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9755520966943856e-05,
+      "loss": 0.0172,
+      "step": 50910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9755472927205386e-05,
+      "loss": 0.0195,
+      "step": 50920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9755424887466915e-05,
+      "loss": 0.0162,
+      "step": 50930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975537684772844e-05,
+      "loss": 0.0171,
+      "step": 50940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975532880798997e-05,
+      "loss": 0.0207,
+      "step": 50950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.97552807682515e-05,
+      "loss": 0.0147,
+      "step": 50960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.975523272851303e-05,
+      "loss": 0.0159,
+      "step": 50970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9755184688774555e-05,
+      "loss": 0.0154,
+      "step": 50980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9755136649036084e-05,
+      "loss": 0.0192,
+      "step": 50990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9755088609297613e-05,
+      "loss": 0.0145,
+      "step": 51000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9755040569559142e-05,
+      "loss": 0.0187,
+      "step": 51010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754992529820668e-05,
+      "loss": 0.0151,
+      "step": 51020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754944490082198e-05,
+      "loss": 0.0276,
+      "step": 51030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754896450343727e-05,
+      "loss": 0.0169,
+      "step": 51040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754848410605253e-05,
+      "loss": 0.0188,
+      "step": 51050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754800370866782e-05,
+      "loss": 0.0143,
+      "step": 51060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975475233112831e-05,
+      "loss": 0.0121,
+      "step": 51070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975470429138984e-05,
+      "loss": 0.0193,
+      "step": 51080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754656251651366e-05,
+      "loss": 0.0138,
+      "step": 51090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754608211912896e-05,
+      "loss": 0.0174,
+      "step": 51100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754560172174425e-05,
+      "loss": 0.0141,
+      "step": 51110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975451213243595e-05,
+      "loss": 0.017,
+      "step": 51120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975446409269748e-05,
+      "loss": 0.0163,
+      "step": 51130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975441605295901e-05,
+      "loss": 0.0183,
+      "step": 51140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975436801322054e-05,
+      "loss": 0.0182,
+      "step": 51150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754319973482064e-05,
+      "loss": 0.0168,
+      "step": 51160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754271933743594e-05,
+      "loss": 0.0188,
+      "step": 51170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754223894005123e-05,
+      "loss": 0.0124,
+      "step": 51180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754175854266652e-05,
+      "loss": 0.0188,
+      "step": 51190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754127814528178e-05,
+      "loss": 0.0234,
+      "step": 51200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754079774789707e-05,
+      "loss": 0.0178,
+      "step": 51210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9754031735051237e-05,
+      "loss": 0.0165,
+      "step": 51220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753983695312763e-05,
+      "loss": 0.0187,
+      "step": 51230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753935655574292e-05,
+      "loss": 0.0134,
+      "step": 51240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975388761583582e-05,
+      "loss": 0.0218,
+      "step": 51250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975383957609735e-05,
+      "loss": 0.0178,
+      "step": 51260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753791536358876e-05,
+      "loss": 0.0159,
+      "step": 51270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753743496620406e-05,
+      "loss": 0.0217,
+      "step": 51280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753695456881935e-05,
+      "loss": 0.0235,
+      "step": 51290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975364741714346e-05,
+      "loss": 0.0201,
+      "step": 51300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975359937740499e-05,
+      "loss": 0.0151,
+      "step": 51310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975355133766652e-05,
+      "loss": 0.0222,
+      "step": 51320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975350329792805e-05,
+      "loss": 0.0195,
+      "step": 51330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753455258189574e-05,
+      "loss": 0.0193,
+      "step": 51340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753407218451104e-05,
+      "loss": 0.0252,
+      "step": 51350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753359178712633e-05,
+      "loss": 0.0147,
+      "step": 51360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753311138974162e-05,
+      "loss": 0.0202,
+      "step": 51370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753263099235688e-05,
+      "loss": 0.0159,
+      "step": 51380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753215059497217e-05,
+      "loss": 0.0164,
+      "step": 51390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753167019758747e-05,
+      "loss": 0.0193,
+      "step": 51400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753118980020273e-05,
+      "loss": 0.0183,
+      "step": 51410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9753070940281802e-05,
+      "loss": 0.0183,
+      "step": 51420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975302290054333e-05,
+      "loss": 0.0174,
+      "step": 51430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975297486080486e-05,
+      "loss": 0.0162,
+      "step": 51440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752926821066386e-05,
+      "loss": 0.0153,
+      "step": 51450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752878781327915e-05,
+      "loss": 0.0175,
+      "step": 51460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752830741589445e-05,
+      "loss": 0.0191,
+      "step": 51470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975278270185097e-05,
+      "loss": 0.0204,
+      "step": 51480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.97527346621125e-05,
+      "loss": 0.021,
+      "step": 51490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975268662237403e-05,
+      "loss": 0.0148,
+      "step": 51500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975263858263556e-05,
+      "loss": 0.0161,
+      "step": 51510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752590542897084e-05,
+      "loss": 0.0191,
+      "step": 51520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752542503158614e-05,
+      "loss": 0.0188,
+      "step": 51530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752494463420143e-05,
+      "loss": 0.0189,
+      "step": 51540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975244642368167e-05,
+      "loss": 0.0199,
+      "step": 51550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752398383943198e-05,
+      "loss": 0.0189,
+      "step": 51560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752350344204727e-05,
+      "loss": 0.0192,
+      "step": 51570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752302304466257e-05,
+      "loss": 0.0158,
+      "step": 51580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752254264727782e-05,
+      "loss": 0.0177,
+      "step": 51590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752206224989312e-05,
+      "loss": 0.0181,
+      "step": 51600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975215818525084e-05,
+      "loss": 0.0142,
+      "step": 51610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975211014551237e-05,
+      "loss": 0.0125,
+      "step": 51620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752062105773896e-05,
+      "loss": 0.0208,
+      "step": 51630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9752014066035425e-05,
+      "loss": 0.014,
+      "step": 51640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9751966026296955e-05,
+      "loss": 0.0185,
+      "step": 51650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975191798655848e-05,
+      "loss": 0.0202,
+      "step": 51660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975186994682001e-05,
+      "loss": 0.0167,
+      "step": 51670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975182190708154e-05,
+      "loss": 0.0192,
+      "step": 51680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975177386734307e-05,
+      "loss": 0.0163,
+      "step": 51690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9751725827604594e-05,
+      "loss": 0.0143,
+      "step": 51700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9751677787866124e-05,
+      "loss": 0.0293,
+      "step": 51710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9751629748127653e-05,
+      "loss": 0.0179,
+      "step": 51720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975158170838918e-05,
+      "loss": 0.0193,
+      "step": 51730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9751533668650708e-05,
+      "loss": 0.0213,
+      "step": 51740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9751485628912237e-05,
+      "loss": 0.0158,
+      "step": 51750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9751437589173766e-05,
+      "loss": 0.0166,
+      "step": 51760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9751389549435292e-05,
+      "loss": 0.0181,
+      "step": 51770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975134150969682e-05,
+      "loss": 0.0212,
+      "step": 51780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975129346995835e-05,
+      "loss": 0.0128,
+      "step": 51790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975124543021988e-05,
+      "loss": 0.0152,
+      "step": 51800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9751197390481406e-05,
+      "loss": 0.0156,
+      "step": 51810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9751149350742935e-05,
+      "loss": 0.0223,
+      "step": 51820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9751101311004465e-05,
+      "loss": 0.0134,
+      "step": 51830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975105327126599e-05,
+      "loss": 0.0206,
+      "step": 51840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975100523152752e-05,
+      "loss": 0.0126,
+      "step": 51850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975095719178905e-05,
+      "loss": 0.017,
+      "step": 51860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750913956024424e-05,
+      "loss": 0.0358,
+      "step": 51870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750865916285953e-05,
+      "loss": 0.0227,
+      "step": 51880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750817876547482e-05,
+      "loss": 0.0158,
+      "step": 51890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750769836809008e-05,
+      "loss": 0.0179,
+      "step": 51900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750721797070537e-05,
+      "loss": 0.015,
+      "step": 51910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750673757332067e-05,
+      "loss": 0.0179,
+      "step": 51920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750625717593596e-05,
+      "loss": 0.0151,
+      "step": 51930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750577677855122e-05,
+      "loss": 0.0199,
+      "step": 51940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975052963811665e-05,
+      "loss": 0.0142,
+      "step": 51950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975048159837818e-05,
+      "loss": 0.0188,
+      "step": 51960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750433558639706e-05,
+      "loss": 0.0159,
+      "step": 51970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750385518901235e-05,
+      "loss": 0.0172,
+      "step": 51980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750337479162765e-05,
+      "loss": 0.0158,
+      "step": 51990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750289439424294e-05,
+      "loss": 0.0218,
+      "step": 52000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975024139968582e-05,
+      "loss": 0.0151,
+      "step": 52010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975019335994735e-05,
+      "loss": 0.0184,
+      "step": 52020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975014532020888e-05,
+      "loss": 0.014,
+      "step": 52030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750097280470408e-05,
+      "loss": 0.0204,
+      "step": 52040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750049240731934e-05,
+      "loss": 0.0145,
+      "step": 52050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750001200993463e-05,
+      "loss": 0.0236,
+      "step": 52060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749953161254992e-05,
+      "loss": 0.0162,
+      "step": 52070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749905121516518e-05,
+      "loss": 0.0173,
+      "step": 52080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749857081778047e-05,
+      "loss": 0.0197,
+      "step": 52090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749809042039577e-05,
+      "loss": 0.0206,
+      "step": 52100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749761002301106e-05,
+      "loss": 0.0206,
+      "step": 52110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.974971296256263e-05,
+      "loss": 0.0141,
+      "step": 52120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.974966492282416e-05,
+      "loss": 0.0147,
+      "step": 52130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.974961688308569e-05,
+      "loss": 0.0172,
+      "step": 52140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749568843347216e-05,
+      "loss": 0.0173,
+      "step": 52150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749520803608745e-05,
+      "loss": 0.0188,
+      "step": 52160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749472763870275e-05,
+      "loss": 0.0165,
+      "step": 52170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749424724131804e-05,
+      "loss": 0.0138,
+      "step": 52180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.974937668439333e-05,
+      "loss": 0.0258,
+      "step": 52190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.974932864465486e-05,
+      "loss": 0.0145,
+      "step": 52200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.974928060491639e-05,
+      "loss": 0.0213,
+      "step": 52210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749232565177918e-05,
+      "loss": 0.0173,
+      "step": 52220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749184525439443e-05,
+      "loss": 0.0157,
+      "step": 52230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749136485700973e-05,
+      "loss": 0.0223,
+      "step": 52240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749088445962502e-05,
+      "loss": 0.0162,
+      "step": 52250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9749040406224028e-05,
+      "loss": 0.0193,
+      "step": 52260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9748992366485557e-05,
+      "loss": 0.013,
+      "step": 52270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9748944326747086e-05,
+      "loss": 0.0141,
+      "step": 52280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9748896287008616e-05,
+      "loss": 0.0178,
+      "step": 52290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.974884824727014e-05,
+      "loss": 0.0169,
+      "step": 52300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.974880020753167e-05,
+      "loss": 0.022,
+      "step": 52310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.97487521677932e-05,
+      "loss": 0.015,
+      "step": 52320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9748704128054726e-05,
+      "loss": 0.0152,
+      "step": 52330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9748656088316255e-05,
+      "loss": 0.0224,
+      "step": 52340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9748608048577785e-05,
+      "loss": 0.0219,
+      "step": 52350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9748560008839314e-05,
+      "loss": 0.0201,
+      "step": 52360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.974851196910084e-05,
+      "loss": 0.0143,
+      "step": 52370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.974846392936237e-05,
+      "loss": 0.014,
+      "step": 52380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9748415889623898e-05,
+      "loss": 0.0172,
+      "step": 52390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9748367849885428e-05,
+      "loss": 0.0146,
+      "step": 52400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9748319810146953e-05,
+      "loss": 0.018,
+      "step": 52410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9748271770408483e-05,
+      "loss": 0.0218,
+      "step": 52420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9748223730670012e-05,
+      "loss": 0.0265,
+      "step": 52430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9748175690931538e-05,
+      "loss": 0.0148,
+      "step": 52440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9748127651193067e-05,
+      "loss": 0.0177,
+      "step": 52450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9748079611454596e-05,
+      "loss": 0.0199,
+      "step": 52460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9748031571716126e-05,
+      "loss": 0.0172,
+      "step": 52470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974798353197765e-05,
+      "loss": 0.0193,
+      "step": 52480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974793549223918e-05,
+      "loss": 0.0128,
+      "step": 52490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974788745250071e-05,
+      "loss": 0.0149,
+      "step": 52500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747839412762236e-05,
+      "loss": 0.0151,
+      "step": 52510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747791373023765e-05,
+      "loss": 0.0183,
+      "step": 52520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747743333285294e-05,
+      "loss": 0.0202,
+      "step": 52530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747695293546824e-05,
+      "loss": 0.013,
+      "step": 52540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974764725380835e-05,
+      "loss": 0.0198,
+      "step": 52550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974759921406988e-05,
+      "loss": 0.0212,
+      "step": 52560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747551174331408e-05,
+      "loss": 0.0201,
+      "step": 52570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747503134592937e-05,
+      "loss": 0.0185,
+      "step": 52580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747455094854467e-05,
+      "loss": 0.0208,
+      "step": 52590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747407055115996e-05,
+      "loss": 0.0091,
+      "step": 52600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747359015377522e-05,
+      "loss": 0.0237,
+      "step": 52610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974731097563905e-05,
+      "loss": 0.016,
+      "step": 52620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974726293590058e-05,
+      "loss": 0.0169,
+      "step": 52630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747214896162106e-05,
+      "loss": 0.0187,
+      "step": 52640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747166856423636e-05,
+      "loss": 0.0153,
+      "step": 52650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747118816685165e-05,
+      "loss": 0.0196,
+      "step": 52660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9747070776946694e-05,
+      "loss": 0.0167,
+      "step": 52670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974702273720822e-05,
+      "loss": 0.0208,
+      "step": 52680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974697469746975e-05,
+      "loss": 0.0205,
+      "step": 52690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974692665773128e-05,
+      "loss": 0.0168,
+      "step": 52700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746878617992808e-05,
+      "loss": 0.0196,
+      "step": 52710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746830578254334e-05,
+      "loss": 0.02,
+      "step": 52720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746782538515863e-05,
+      "loss": 0.0165,
+      "step": 52730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746734498777392e-05,
+      "loss": 0.0149,
+      "step": 52740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746686459038918e-05,
+      "loss": 0.0145,
+      "step": 52750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746638419300447e-05,
+      "loss": 0.0199,
+      "step": 52760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746590379561977e-05,
+      "loss": 0.0101,
+      "step": 52770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746542339823506e-05,
+      "loss": 0.0185,
+      "step": 52780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746494300085032e-05,
+      "loss": 0.0145,
+      "step": 52790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974644626034656e-05,
+      "loss": 0.0179,
+      "step": 52800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974639822060809e-05,
+      "loss": 0.0166,
+      "step": 52810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746350180869616e-05,
+      "loss": 0.024,
+      "step": 52820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746302141131145e-05,
+      "loss": 0.0152,
+      "step": 52830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746254101392675e-05,
+      "loss": 0.0162,
+      "step": 52840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746206061654204e-05,
+      "loss": 0.0144,
+      "step": 52850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974615802191573e-05,
+      "loss": 0.0172,
+      "step": 52860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974610998217726e-05,
+      "loss": 0.0167,
+      "step": 52870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974606194243879e-05,
+      "loss": 0.0259,
+      "step": 52880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9746013902700318e-05,
+      "loss": 0.018,
+      "step": 52890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745965862961844e-05,
+      "loss": 0.0155,
+      "step": 52900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745917823223373e-05,
+      "loss": 0.0152,
+      "step": 52910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745869783484902e-05,
+      "loss": 0.0143,
+      "step": 52920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745821743746428e-05,
+      "loss": 0.0201,
+      "step": 52930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745773704007957e-05,
+      "loss": 0.0161,
+      "step": 52940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745725664269487e-05,
+      "loss": 0.02,
+      "step": 52950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745677624531016e-05,
+      "loss": 0.0176,
+      "step": 52960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974562958479254e-05,
+      "loss": 0.0166,
+      "step": 52970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974558154505407e-05,
+      "loss": 0.0144,
+      "step": 52980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.97455335053156e-05,
+      "loss": 0.0116,
+      "step": 52990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745485465577126e-05,
+      "loss": 0.0217,
+      "step": 53000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745437425838655e-05,
+      "loss": 0.0139,
+      "step": 53010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745389386100185e-05,
+      "loss": 0.0175,
+      "step": 53020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745341346361714e-05,
+      "loss": 0.0171,
+      "step": 53030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974529330662324e-05,
+      "loss": 0.0199,
+      "step": 53040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.974524526688477e-05,
+      "loss": 0.0124,
+      "step": 53050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.97451972271463e-05,
+      "loss": 0.0183,
+      "step": 53060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745149187407828e-05,
+      "loss": 0.0224,
+      "step": 53070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9745101147669353e-05,
+      "loss": 0.0188,
+      "step": 53080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9745053107930883e-05,
+      "loss": 0.0188,
+      "step": 53090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9745005068192412e-05,
+      "loss": 0.0229,
+      "step": 53100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744957028453938e-05,
+      "loss": 0.0196,
+      "step": 53110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744908988715467e-05,
+      "loss": 0.0151,
+      "step": 53120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744860948976996e-05,
+      "loss": 0.0171,
+      "step": 53130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744812909238526e-05,
+      "loss": 0.0204,
+      "step": 53140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974476486950005e-05,
+      "loss": 0.0223,
+      "step": 53150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974471682976158e-05,
+      "loss": 0.0171,
+      "step": 53160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974466879002311e-05,
+      "loss": 0.0158,
+      "step": 53170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744620750284636e-05,
+      "loss": 0.0148,
+      "step": 53180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744572710546165e-05,
+      "loss": 0.0187,
+      "step": 53190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744524670807695e-05,
+      "loss": 0.018,
+      "step": 53200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744476631069224e-05,
+      "loss": 0.0178,
+      "step": 53210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974442859133075e-05,
+      "loss": 0.0173,
+      "step": 53220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974438055159228e-05,
+      "loss": 0.02,
+      "step": 53230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744332511853808e-05,
+      "loss": 0.0199,
+      "step": 53240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744284472115338e-05,
+      "loss": 0.0211,
+      "step": 53250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744236432376863e-05,
+      "loss": 0.0279,
+      "step": 53260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744188392638393e-05,
+      "loss": 0.0167,
+      "step": 53270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744140352899922e-05,
+      "loss": 0.0156,
+      "step": 53280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744092313161448e-05,
+      "loss": 0.0172,
+      "step": 53290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9744044273422977e-05,
+      "loss": 0.0171,
+      "step": 53300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743996233684506e-05,
+      "loss": 0.0125,
+      "step": 53310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743948193946036e-05,
+      "loss": 0.0167,
+      "step": 53320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974390015420756e-05,
+      "loss": 0.0261,
+      "step": 53330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974385211446909e-05,
+      "loss": 0.0135,
+      "step": 53340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974380407473062e-05,
+      "loss": 0.0175,
+      "step": 53350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743756034992146e-05,
+      "loss": 0.0209,
+      "step": 53360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743707995253675e-05,
+      "loss": 0.019,
+      "step": 53370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743659955515204e-05,
+      "loss": 0.0145,
+      "step": 53380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743611915776734e-05,
+      "loss": 0.0156,
+      "step": 53390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974356387603826e-05,
+      "loss": 0.0189,
+      "step": 53400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974351583629979e-05,
+      "loss": 0.0192,
+      "step": 53410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743467796561318e-05,
+      "loss": 0.0167,
+      "step": 53420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743419756822847e-05,
+      "loss": 0.0193,
+      "step": 53430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743371717084373e-05,
+      "loss": 0.0134,
+      "step": 53440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743323677345903e-05,
+      "loss": 0.0135,
+      "step": 53450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743275637607432e-05,
+      "loss": 0.0202,
+      "step": 53460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743227597868958e-05,
+      "loss": 0.0177,
+      "step": 53470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743179558130487e-05,
+      "loss": 0.0198,
+      "step": 53480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743131518392016e-05,
+      "loss": 0.0123,
+      "step": 53490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9743083478653546e-05,
+      "loss": 0.0133,
+      "step": 53500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974303543891507e-05,
+      "loss": 0.0205,
+      "step": 53510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.97429873991766e-05,
+      "loss": 0.0177,
+      "step": 53520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974293935943813e-05,
+      "loss": 0.017,
+      "step": 53530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742891319699656e-05,
+      "loss": 0.0148,
+      "step": 53540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742843279961185e-05,
+      "loss": 0.0171,
+      "step": 53550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742795240222714e-05,
+      "loss": 0.0237,
+      "step": 53560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742747200484244e-05,
+      "loss": 0.0204,
+      "step": 53570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974269916074577e-05,
+      "loss": 0.0183,
+      "step": 53580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.97426511210073e-05,
+      "loss": 0.0263,
+      "step": 53590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742603081268828e-05,
+      "loss": 0.0196,
+      "step": 53600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742555041530357e-05,
+      "loss": 0.0212,
+      "step": 53610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742507001791883e-05,
+      "loss": 0.0177,
+      "step": 53620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742458962053412e-05,
+      "loss": 0.0195,
+      "step": 53630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742410922314942e-05,
+      "loss": 0.0153,
+      "step": 53640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742362882576468e-05,
+      "loss": 0.0188,
+      "step": 53650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742314842837997e-05,
+      "loss": 0.0183,
+      "step": 53660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742266803099526e-05,
+      "loss": 0.0209,
+      "step": 53670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742218763361055e-05,
+      "loss": 0.0175,
+      "step": 53680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974217072362258e-05,
+      "loss": 0.0185,
+      "step": 53690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974212268388411e-05,
+      "loss": 0.0096,
+      "step": 53700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974207464414564e-05,
+      "loss": 0.0172,
+      "step": 53710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9742026604407166e-05,
+      "loss": 0.0175,
+      "step": 53720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9741978564668695e-05,
+      "loss": 0.0216,
+      "step": 53730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9741930524930224e-05,
+      "loss": 0.0179,
+      "step": 53740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9741882485191754e-05,
+      "loss": 0.0199,
+      "step": 53750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974183444545328e-05,
+      "loss": 0.0232,
+      "step": 53760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.974178640571481e-05,
+      "loss": 0.0122,
+      "step": 53770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9741738365976338e-05,
+      "loss": 0.0151,
+      "step": 53780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9741690326237867e-05,
+      "loss": 0.0145,
+      "step": 53790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9741642286499393e-05,
+      "loss": 0.0198,
+      "step": 53800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9741594246760922e-05,
+      "loss": 0.0119,
+      "step": 53810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.974154620702245e-05,
+      "loss": 0.0165,
+      "step": 53820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9741498167283978e-05,
+      "loss": 0.0195,
+      "step": 53830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9741450127545507e-05,
+      "loss": 0.0144,
+      "step": 53840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9741402087807036e-05,
+      "loss": 0.0154,
+      "step": 53850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9741354048068565e-05,
+      "loss": 0.0188,
+      "step": 53860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.974130600833009e-05,
+      "loss": 0.0219,
+      "step": 53870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.974125796859162e-05,
+      "loss": 0.0144,
+      "step": 53880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.974120992885315e-05,
+      "loss": 0.0188,
+      "step": 53890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9741161889114676e-05,
+      "loss": 0.0182,
+      "step": 53900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9741113849376205e-05,
+      "loss": 0.0191,
+      "step": 53910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9741065809637734e-05,
+      "loss": 0.0126,
+      "step": 53920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9741017769899263e-05,
+      "loss": 0.0217,
+      "step": 53930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.974096973016079e-05,
+      "loss": 0.0188,
+      "step": 53940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.974092169042232e-05,
+      "loss": 0.0132,
+      "step": 53950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740873650683848e-05,
+      "loss": 0.0231,
+      "step": 53960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740825610945377e-05,
+      "loss": 0.0251,
+      "step": 53970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740777571206903e-05,
+      "loss": 0.0171,
+      "step": 53980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740729531468432e-05,
+      "loss": 0.0186,
+      "step": 53990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.974068149172996e-05,
+      "loss": 0.0199,
+      "step": 54000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740633451991487e-05,
+      "loss": 0.0304,
+      "step": 54010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740585412253017e-05,
+      "loss": 0.0172,
+      "step": 54020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740537372514546e-05,
+      "loss": 0.0188,
+      "step": 54030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740489332776075e-05,
+      "loss": 0.0197,
+      "step": 54040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.97404412930376e-05,
+      "loss": 0.0153,
+      "step": 54050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.974039325329913e-05,
+      "loss": 0.0138,
+      "step": 54060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.974034521356066e-05,
+      "loss": 0.0206,
+      "step": 54070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740297173822186e-05,
+      "loss": 0.021,
+      "step": 54080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740249134083715e-05,
+      "loss": 0.0205,
+      "step": 54090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740201094345244e-05,
+      "loss": 0.0169,
+      "step": 54100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740153054606773e-05,
+      "loss": 0.0138,
+      "step": 54110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.97401050148683e-05,
+      "loss": 0.0216,
+      "step": 54120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.974005697512983e-05,
+      "loss": 0.0189,
+      "step": 54130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9740008935391358e-05,
+      "loss": 0.0162,
+      "step": 54140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739960895652887e-05,
+      "loss": 0.0124,
+      "step": 54150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739912855914413e-05,
+      "loss": 0.0177,
+      "step": 54160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739864816175942e-05,
+      "loss": 0.0154,
+      "step": 54170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.973981677643747e-05,
+      "loss": 0.0226,
+      "step": 54180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739768736698997e-05,
+      "loss": 0.0168,
+      "step": 54190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739720696960527e-05,
+      "loss": 0.0177,
+      "step": 54200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739672657222056e-05,
+      "loss": 0.0182,
+      "step": 54210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739624617483585e-05,
+      "loss": 0.0178,
+      "step": 54220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.973957657774511e-05,
+      "loss": 0.0174,
+      "step": 54230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.973952853800664e-05,
+      "loss": 0.0147,
+      "step": 54240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.973948049826817e-05,
+      "loss": 0.0147,
+      "step": 54250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739432458529696e-05,
+      "loss": 0.0133,
+      "step": 54260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739384418791225e-05,
+      "loss": 0.0176,
+      "step": 54270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739336379052754e-05,
+      "loss": 0.0135,
+      "step": 54280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739288339314283e-05,
+      "loss": 0.0171,
+      "step": 54290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.973924029957581e-05,
+      "loss": 0.0158,
+      "step": 54300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.973919225983734e-05,
+      "loss": 0.0179,
+      "step": 54310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739144220098868e-05,
+      "loss": 0.0158,
+      "step": 54320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739096180360397e-05,
+      "loss": 0.0134,
+      "step": 54330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739048140621923e-05,
+      "loss": 0.015,
+      "step": 54340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9739000100883452e-05,
+      "loss": 0.0134,
+      "step": 54350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.973895206114498e-05,
+      "loss": 0.0132,
+      "step": 54360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9738904021406507e-05,
+      "loss": 0.0193,
+      "step": 54370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9738855981668037e-05,
+      "loss": 0.0142,
+      "step": 54380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9738807941929566e-05,
+      "loss": 0.0159,
+      "step": 54390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9738759902191095e-05,
+      "loss": 0.0159,
+      "step": 54400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.973871186245262e-05,
+      "loss": 0.0134,
+      "step": 54410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.973866382271415e-05,
+      "loss": 0.0135,
+      "step": 54420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.973861578297568e-05,
+      "loss": 0.0122,
+      "step": 54430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9738567743237205e-05,
+      "loss": 0.0221,
+      "step": 54440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9738519703498735e-05,
+      "loss": 0.014,
+      "step": 54450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9738471663760264e-05,
+      "loss": 0.0187,
+      "step": 54460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9738423624021793e-05,
+      "loss": 0.0164,
+      "step": 54470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973837558428332e-05,
+      "loss": 0.0177,
+      "step": 54480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973832754454485e-05,
+      "loss": 0.0231,
+      "step": 54490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9738279504806378e-05,
+      "loss": 0.0164,
+      "step": 54500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9738231465067904e-05,
+      "loss": 0.0174,
+      "step": 54510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9738183425329433e-05,
+      "loss": 0.0161,
+      "step": 54520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9738135385590962e-05,
+      "loss": 0.0191,
+      "step": 54530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973808734585249e-05,
+      "loss": 0.0211,
+      "step": 54540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9738039306114017e-05,
+      "loss": 0.0189,
+      "step": 54550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737991266375546e-05,
+      "loss": 0.0174,
+      "step": 54560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737943226637076e-05,
+      "loss": 0.0139,
+      "step": 54570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737895186898605e-05,
+      "loss": 0.022,
+      "step": 54580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973784714716013e-05,
+      "loss": 0.0151,
+      "step": 54590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973779910742166e-05,
+      "loss": 0.0132,
+      "step": 54600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973775106768319e-05,
+      "loss": 0.0212,
+      "step": 54610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737703027944715e-05,
+      "loss": 0.012,
+      "step": 54620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737654988206245e-05,
+      "loss": 0.0169,
+      "step": 54630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737606948467774e-05,
+      "loss": 0.0154,
+      "step": 54640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737558908729303e-05,
+      "loss": 0.0134,
+      "step": 54650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973751086899083e-05,
+      "loss": 0.0159,
+      "step": 54660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737462829252358e-05,
+      "loss": 0.0172,
+      "step": 54670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737414789513888e-05,
+      "loss": 0.0181,
+      "step": 54680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737366749775413e-05,
+      "loss": 0.0218,
+      "step": 54690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737318710036943e-05,
+      "loss": 0.0166,
+      "step": 54700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737270670298472e-05,
+      "loss": 0.0173,
+      "step": 54710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973722263056e-05,
+      "loss": 0.0148,
+      "step": 54720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737174590821527e-05,
+      "loss": 0.0204,
+      "step": 54730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737126551083056e-05,
+      "loss": 0.0153,
+      "step": 54740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737078511344586e-05,
+      "loss": 0.0162,
+      "step": 54750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9737030471606115e-05,
+      "loss": 0.0151,
+      "step": 54760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973698243186764e-05,
+      "loss": 0.018,
+      "step": 54770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973693439212917e-05,
+      "loss": 0.0185,
+      "step": 54780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.97368863523907e-05,
+      "loss": 0.0168,
+      "step": 54790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736838312652225e-05,
+      "loss": 0.0152,
+      "step": 54800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736790272913755e-05,
+      "loss": 0.014,
+      "step": 54810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736742233175284e-05,
+      "loss": 0.0159,
+      "step": 54820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736694193436813e-05,
+      "loss": 0.0165,
+      "step": 54830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973664615369834e-05,
+      "loss": 0.0184,
+      "step": 54840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736598113959868e-05,
+      "loss": 0.0151,
+      "step": 54850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736550074221397e-05,
+      "loss": 0.0164,
+      "step": 54860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736502034482923e-05,
+      "loss": 0.018,
+      "step": 54870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736453994744453e-05,
+      "loss": 0.0218,
+      "step": 54880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736405955005982e-05,
+      "loss": 0.0135,
+      "step": 54890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973635791526751e-05,
+      "loss": 0.0185,
+      "step": 54900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736309875529037e-05,
+      "loss": 0.0214,
+      "step": 54910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736261835790566e-05,
+      "loss": 0.0122,
+      "step": 54920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973621860002594e-05,
+      "loss": 0.0168,
+      "step": 54930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973617056028747e-05,
+      "loss": 0.0155,
+      "step": 54940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736122520549e-05,
+      "loss": 0.0163,
+      "step": 54950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973607448081053e-05,
+      "loss": 0.0199,
+      "step": 54960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9736026441072055e-05,
+      "loss": 0.0189,
+      "step": 54970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9735978401333584e-05,
+      "loss": 0.0161,
+      "step": 54980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9735930361595113e-05,
+      "loss": 0.0099,
+      "step": 54990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9735882321856642e-05,
+      "loss": 0.0153,
+      "step": 55000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973583428211817e-05,
+      "loss": 0.019,
+      "step": 55010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9735786242379698e-05,
+      "loss": 0.0229,
+      "step": 55020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9735738202641227e-05,
+      "loss": 0.0131,
+      "step": 55030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9735690162902753e-05,
+      "loss": 0.0203,
+      "step": 55040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9735642123164282e-05,
+      "loss": 0.0162,
+      "step": 55050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973559408342581e-05,
+      "loss": 0.0187,
+      "step": 55060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973554604368734e-05,
+      "loss": 0.0147,
+      "step": 55070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9735498003948866e-05,
+      "loss": 0.0204,
+      "step": 55080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9735449964210396e-05,
+      "loss": 0.0191,
+      "step": 55090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9735401924471925e-05,
+      "loss": 0.0169,
+      "step": 55100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973535388473345e-05,
+      "loss": 0.0191,
+      "step": 55110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973530584499498e-05,
+      "loss": 0.0211,
+      "step": 55120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973525780525651e-05,
+      "loss": 0.0178,
+      "step": 55130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.973520976551804e-05,
+      "loss": 0.0186,
+      "step": 55140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9735161725779565e-05,
+      "loss": 0.013,
+      "step": 55150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9735113686041094e-05,
+      "loss": 0.0192,
+      "step": 55160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9735065646302623e-05,
+      "loss": 0.0179,
+      "step": 55170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9735017606564152e-05,
+      "loss": 0.0123,
+      "step": 55180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734969566825678e-05,
+      "loss": 0.0232,
+      "step": 55190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734921527087208e-05,
+      "loss": 0.017,
+      "step": 55200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734873487348737e-05,
+      "loss": 0.0177,
+      "step": 55210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734825447610263e-05,
+      "loss": 0.0152,
+      "step": 55220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734777407871792e-05,
+      "loss": 0.0141,
+      "step": 55230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973472936813332e-05,
+      "loss": 0.0184,
+      "step": 55240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973468132839485e-05,
+      "loss": 0.0206,
+      "step": 55250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734633288656376e-05,
+      "loss": 0.0154,
+      "step": 55260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734585248917906e-05,
+      "loss": 0.0236,
+      "step": 55270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734537209179435e-05,
+      "loss": 0.0185,
+      "step": 55280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973448916944096e-05,
+      "loss": 0.0173,
+      "step": 55290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973444112970249e-05,
+      "loss": 0.0168,
+      "step": 55300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973439308996402e-05,
+      "loss": 0.0182,
+      "step": 55310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973434505022555e-05,
+      "loss": 0.022,
+      "step": 55320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734297010487074e-05,
+      "loss": 0.0194,
+      "step": 55330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734248970748604e-05,
+      "loss": 0.018,
+      "step": 55340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734200931010133e-05,
+      "loss": 0.0155,
+      "step": 55350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734152891271662e-05,
+      "loss": 0.0195,
+      "step": 55360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734104851533188e-05,
+      "loss": 0.0146,
+      "step": 55370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734056811794717e-05,
+      "loss": 0.0157,
+      "step": 55380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9734008772056247e-05,
+      "loss": 0.0187,
+      "step": 55390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733960732317773e-05,
+      "loss": 0.0154,
+      "step": 55400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733912692579302e-05,
+      "loss": 0.0136,
+      "step": 55410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973386465284083e-05,
+      "loss": 0.017,
+      "step": 55420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973381661310236e-05,
+      "loss": 0.0181,
+      "step": 55430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733768573363886e-05,
+      "loss": 0.0168,
+      "step": 55440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733720533625416e-05,
+      "loss": 0.0172,
+      "step": 55450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733672493886945e-05,
+      "loss": 0.0179,
+      "step": 55460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973362445414847e-05,
+      "loss": 0.0176,
+      "step": 55470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973357641441e-05,
+      "loss": 0.0203,
+      "step": 55480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973352837467153e-05,
+      "loss": 0.0152,
+      "step": 55490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973348033493306e-05,
+      "loss": 0.016,
+      "step": 55500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733432295194584e-05,
+      "loss": 0.0153,
+      "step": 55510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733384255456114e-05,
+      "loss": 0.0119,
+      "step": 55520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733336215717643e-05,
+      "loss": 0.0145,
+      "step": 55530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733288175979172e-05,
+      "loss": 0.0165,
+      "step": 55540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733240136240698e-05,
+      "loss": 0.0135,
+      "step": 55550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733192096502227e-05,
+      "loss": 0.0186,
+      "step": 55560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733144056763757e-05,
+      "loss": 0.019,
+      "step": 55570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733096017025283e-05,
+      "loss": 0.0153,
+      "step": 55580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733047977286812e-05,
+      "loss": 0.0157,
+      "step": 55590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973299993754834e-05,
+      "loss": 0.0137,
+      "step": 55600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973295189780987e-05,
+      "loss": 0.0177,
+      "step": 55610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732903858071396e-05,
+      "loss": 0.0129,
+      "step": 55620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732855818332925e-05,
+      "loss": 0.0175,
+      "step": 55630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732807778594455e-05,
+      "loss": 0.0221,
+      "step": 55640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973275973885598e-05,
+      "loss": 0.0204,
+      "step": 55650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732711699117513e-05,
+      "loss": 0.0201,
+      "step": 55660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732663659379043e-05,
+      "loss": 0.0165,
+      "step": 55670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973261561964057e-05,
+      "loss": 0.0135,
+      "step": 55680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732567579902098e-05,
+      "loss": 0.0158,
+      "step": 55690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732519540163627e-05,
+      "loss": 0.0179,
+      "step": 55700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732471500425153e-05,
+      "loss": 0.0194,
+      "step": 55710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732423460686682e-05,
+      "loss": 0.0196,
+      "step": 55720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973237542094821e-05,
+      "loss": 0.0179,
+      "step": 55730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973232738120974e-05,
+      "loss": 0.0163,
+      "step": 55740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732279341471267e-05,
+      "loss": 0.0151,
+      "step": 55750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732231301732796e-05,
+      "loss": 0.0164,
+      "step": 55760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9732183261994325e-05,
+      "loss": 0.0197,
+      "step": 55770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973213522225585e-05,
+      "loss": 0.0153,
+      "step": 55780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973208718251738e-05,
+      "loss": 0.0134,
+      "step": 55790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973203914277891e-05,
+      "loss": 0.0147,
+      "step": 55800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973199110304044e-05,
+      "loss": 0.0165,
+      "step": 55810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9731943063301965e-05,
+      "loss": 0.0191,
+      "step": 55820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9731895023563494e-05,
+      "loss": 0.0149,
+      "step": 55830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9731846983825023e-05,
+      "loss": 0.0168,
+      "step": 55840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9731798944086552e-05,
+      "loss": 0.018,
+      "step": 55850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973175090434808e-05,
+      "loss": 0.0141,
+      "step": 55860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9731702864609608e-05,
+      "loss": 0.0188,
+      "step": 55870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9731654824871137e-05,
+      "loss": 0.0171,
+      "step": 55880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9731606785132663e-05,
+      "loss": 0.016,
+      "step": 55890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9731558745394192e-05,
+      "loss": 0.0159,
+      "step": 55900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973151070565572e-05,
+      "loss": 0.0175,
+      "step": 55910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973146266591725e-05,
+      "loss": 0.0165,
+      "step": 55920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9731414626178776e-05,
+      "loss": 0.0159,
+      "step": 55930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9731366586440306e-05,
+      "loss": 0.0143,
+      "step": 55940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9731318546701835e-05,
+      "loss": 0.0168,
+      "step": 55950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973127050696336e-05,
+      "loss": 0.0174,
+      "step": 55960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973122246722489e-05,
+      "loss": 0.018,
+      "step": 55970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973117442748642e-05,
+      "loss": 0.0168,
+      "step": 55980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973112638774795e-05,
+      "loss": 0.0135,
+      "step": 55990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9731078348009475e-05,
+      "loss": 0.0221,
+      "step": 56000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9731030308271004e-05,
+      "loss": 0.0183,
+      "step": 56010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730982268532533e-05,
+      "loss": 0.0151,
+      "step": 56020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730934228794062e-05,
+      "loss": 0.0151,
+      "step": 56030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730886189055588e-05,
+      "loss": 0.0154,
+      "step": 56040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730838149317118e-05,
+      "loss": 0.0153,
+      "step": 56050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730790109578647e-05,
+      "loss": 0.0174,
+      "step": 56060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730742069840173e-05,
+      "loss": 0.0215,
+      "step": 56070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730694030101702e-05,
+      "loss": 0.0198,
+      "step": 56080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973064599036323e-05,
+      "loss": 0.0135,
+      "step": 56090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973059795062476e-05,
+      "loss": 0.018,
+      "step": 56100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730549910886286e-05,
+      "loss": 0.013,
+      "step": 56110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730501871147816e-05,
+      "loss": 0.0249,
+      "step": 56120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730453831409345e-05,
+      "loss": 0.0186,
+      "step": 56130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973040579167087e-05,
+      "loss": 0.0159,
+      "step": 56140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.97303577519324e-05,
+      "loss": 0.0147,
+      "step": 56150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973030971219393e-05,
+      "loss": 0.015,
+      "step": 56160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.973026167245546e-05,
+      "loss": 0.0188,
+      "step": 56170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730213632716984e-05,
+      "loss": 0.0182,
+      "step": 56180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730165592978514e-05,
+      "loss": 0.0182,
+      "step": 56190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730117553240043e-05,
+      "loss": 0.0186,
+      "step": 56200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730069513501572e-05,
+      "loss": 0.0246,
+      "step": 56210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9730021473763098e-05,
+      "loss": 0.0177,
+      "step": 56220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729973434024627e-05,
+      "loss": 0.014,
+      "step": 56230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729925394286157e-05,
+      "loss": 0.0173,
+      "step": 56240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729877354547683e-05,
+      "loss": 0.0172,
+      "step": 56250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729829314809212e-05,
+      "loss": 0.02,
+      "step": 56260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972978127507074e-05,
+      "loss": 0.0134,
+      "step": 56270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972973323533227e-05,
+      "loss": 0.0156,
+      "step": 56280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729685195593796e-05,
+      "loss": 0.0184,
+      "step": 56290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729637155855326e-05,
+      "loss": 0.0171,
+      "step": 56300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729589116116855e-05,
+      "loss": 0.0124,
+      "step": 56310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972954107637838e-05,
+      "loss": 0.0163,
+      "step": 56320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972949303663991e-05,
+      "loss": 0.0153,
+      "step": 56330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972944499690144e-05,
+      "loss": 0.0146,
+      "step": 56340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972939695716297e-05,
+      "loss": 0.0174,
+      "step": 56350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729348917424494e-05,
+      "loss": 0.0226,
+      "step": 56360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729300877686024e-05,
+      "loss": 0.0151,
+      "step": 56370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729252837947553e-05,
+      "loss": 0.0169,
+      "step": 56380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729204798209082e-05,
+      "loss": 0.018,
+      "step": 56390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729156758470608e-05,
+      "loss": 0.0168,
+      "step": 56400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729108718732137e-05,
+      "loss": 0.0129,
+      "step": 56410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729060678993667e-05,
+      "loss": 0.0161,
+      "step": 56420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9729012639255193e-05,
+      "loss": 0.0155,
+      "step": 56430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9728964599516722e-05,
+      "loss": 0.0188,
+      "step": 56440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972891655977825e-05,
+      "loss": 0.0217,
+      "step": 56450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972886852003978e-05,
+      "loss": 0.0135,
+      "step": 56460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9728820480301306e-05,
+      "loss": 0.0159,
+      "step": 56470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9728772440562835e-05,
+      "loss": 0.0229,
+      "step": 56480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9728724400824365e-05,
+      "loss": 0.0206,
+      "step": 56490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972867636108589e-05,
+      "loss": 0.016,
+      "step": 56500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972862832134742e-05,
+      "loss": 0.0164,
+      "step": 56510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972858028160895e-05,
+      "loss": 0.0162,
+      "step": 56520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972853224187048e-05,
+      "loss": 0.0164,
+      "step": 56530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9728484202132004e-05,
+      "loss": 0.016,
+      "step": 56540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9728436162393534e-05,
+      "loss": 0.0135,
+      "step": 56550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9728388122655063e-05,
+      "loss": 0.0175,
+      "step": 56560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9728340082916592e-05,
+      "loss": 0.0192,
+      "step": 56570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9728292043178118e-05,
+      "loss": 0.0204,
+      "step": 56580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9728244003439647e-05,
+      "loss": 0.0206,
+      "step": 56590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9728195963701177e-05,
+      "loss": 0.027,
+      "step": 56600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9728147923962702e-05,
+      "loss": 0.0153,
+      "step": 56610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9728099884224232e-05,
+      "loss": 0.0184,
+      "step": 56620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972805184448576e-05,
+      "loss": 0.0195,
+      "step": 56630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972800380474729e-05,
+      "loss": 0.0194,
+      "step": 56640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727955765008816e-05,
+      "loss": 0.019,
+      "step": 56650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727907725270345e-05,
+      "loss": 0.0171,
+      "step": 56660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727859685531875e-05,
+      "loss": 0.0256,
+      "step": 56670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.97278116457934e-05,
+      "loss": 0.0171,
+      "step": 56680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972776360605493e-05,
+      "loss": 0.0173,
+      "step": 56690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972771556631646e-05,
+      "loss": 0.0188,
+      "step": 56700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972766752657799e-05,
+      "loss": 0.0163,
+      "step": 56710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727619486839514e-05,
+      "loss": 0.0139,
+      "step": 56720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727571447101044e-05,
+      "loss": 0.0124,
+      "step": 56730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727523407362573e-05,
+      "loss": 0.0172,
+      "step": 56740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727475367624102e-05,
+      "loss": 0.0143,
+      "step": 56750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727427327885628e-05,
+      "loss": 0.0162,
+      "step": 56760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727379288147157e-05,
+      "loss": 0.0168,
+      "step": 56770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727331248408686e-05,
+      "loss": 0.0248,
+      "step": 56780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727283208670212e-05,
+      "loss": 0.0172,
+      "step": 56790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972723516893174e-05,
+      "loss": 0.0124,
+      "step": 56800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972718712919327e-05,
+      "loss": 0.0191,
+      "step": 56810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.97271390894548e-05,
+      "loss": 0.0182,
+      "step": 56820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727091049716326e-05,
+      "loss": 0.0139,
+      "step": 56830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9727043009977855e-05,
+      "loss": 0.0135,
+      "step": 56840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726994970239385e-05,
+      "loss": 0.0189,
+      "step": 56850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972694693050091e-05,
+      "loss": 0.018,
+      "step": 56860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972689889076244e-05,
+      "loss": 0.0132,
+      "step": 56870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972685085102397e-05,
+      "loss": 0.0161,
+      "step": 56880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726802811285498e-05,
+      "loss": 0.016,
+      "step": 56890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726754771547024e-05,
+      "loss": 0.0166,
+      "step": 56900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726706731808553e-05,
+      "loss": 0.0183,
+      "step": 56910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726658692070083e-05,
+      "loss": 0.016,
+      "step": 56920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726610652331612e-05,
+      "loss": 0.0189,
+      "step": 56930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726562612593138e-05,
+      "loss": 0.0152,
+      "step": 56940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726514572854667e-05,
+      "loss": 0.0167,
+      "step": 56950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726466533116196e-05,
+      "loss": 0.0152,
+      "step": 56960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726418493377722e-05,
+      "loss": 0.0227,
+      "step": 56970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972637045363925e-05,
+      "loss": 0.0143,
+      "step": 56980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972632241390078e-05,
+      "loss": 0.0159,
+      "step": 56990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972627437416231e-05,
+      "loss": 0.0197,
+      "step": 57000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726226334423836e-05,
+      "loss": 0.0156,
+      "step": 57010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726178294685365e-05,
+      "loss": 0.0161,
+      "step": 57020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9726130254946894e-05,
+      "loss": 0.0169,
+      "step": 57030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972608221520842e-05,
+      "loss": 0.0175,
+      "step": 57040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972603417546995e-05,
+      "loss": 0.0159,
+      "step": 57050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972598613573148e-05,
+      "loss": 0.0194,
+      "step": 57060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9725938095993008e-05,
+      "loss": 0.0165,
+      "step": 57070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9725890056254534e-05,
+      "loss": 0.023,
+      "step": 57080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9725842016516063e-05,
+      "loss": 0.0138,
+      "step": 57090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9725798780751438e-05,
+      "loss": 0.0176,
+      "step": 57100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9725750741012967e-05,
+      "loss": 0.0136,
+      "step": 57110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9725702701274497e-05,
+      "loss": 0.02,
+      "step": 57120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9725654661536026e-05,
+      "loss": 0.0201,
+      "step": 57130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972560662179755e-05,
+      "loss": 0.0156,
+      "step": 57140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972555858205908e-05,
+      "loss": 0.0172,
+      "step": 57150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972551054232061e-05,
+      "loss": 0.0199,
+      "step": 57160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972546250258214e-05,
+      "loss": 0.0148,
+      "step": 57170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9725414462843665e-05,
+      "loss": 0.0132,
+      "step": 57180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9725366423105195e-05,
+      "loss": 0.0143,
+      "step": 57190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9725318383366724e-05,
+      "loss": 0.0153,
+      "step": 57200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972527034362825e-05,
+      "loss": 0.0147,
+      "step": 57210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972522230388978e-05,
+      "loss": 0.0195,
+      "step": 57220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.972517426415131e-05,
+      "loss": 0.0154,
+      "step": 57230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9725126224412838e-05,
+      "loss": 0.0186,
+      "step": 57240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9725078184674363e-05,
+      "loss": 0.0148,
+      "step": 57250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9725030144935893e-05,
+      "loss": 0.0166,
+      "step": 57260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724982105197422e-05,
+      "loss": 0.0155,
+      "step": 57270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724934065458948e-05,
+      "loss": 0.022,
+      "step": 57280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724886025720477e-05,
+      "loss": 0.0157,
+      "step": 57290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724837985982006e-05,
+      "loss": 0.0175,
+      "step": 57300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724789946243536e-05,
+      "loss": 0.0169,
+      "step": 57310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972474190650506e-05,
+      "loss": 0.0169,
+      "step": 57320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972469386676659e-05,
+      "loss": 0.0114,
+      "step": 57330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972464582702812e-05,
+      "loss": 0.0122,
+      "step": 57340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724597787289646e-05,
+      "loss": 0.0163,
+      "step": 57350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724549747551175e-05,
+      "loss": 0.0164,
+      "step": 57360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724501707812705e-05,
+      "loss": 0.0172,
+      "step": 57370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724453668074234e-05,
+      "loss": 0.0124,
+      "step": 57380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972440562833576e-05,
+      "loss": 0.0155,
+      "step": 57390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972435758859729e-05,
+      "loss": 0.0176,
+      "step": 57400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724309548858818e-05,
+      "loss": 0.0152,
+      "step": 57410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724261509120348e-05,
+      "loss": 0.0112,
+      "step": 57420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724213469381873e-05,
+      "loss": 0.0191,
+      "step": 57430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724165429643403e-05,
+      "loss": 0.0116,
+      "step": 57440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724117389904932e-05,
+      "loss": 0.018,
+      "step": 57450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724069350166458e-05,
+      "loss": 0.0159,
+      "step": 57460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9724021310427987e-05,
+      "loss": 0.0191,
+      "step": 57470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723973270689516e-05,
+      "loss": 0.0139,
+      "step": 57480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723925230951046e-05,
+      "loss": 0.0202,
+      "step": 57490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972387719121257e-05,
+      "loss": 0.0119,
+      "step": 57500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.97238291514741e-05,
+      "loss": 0.0173,
+      "step": 57510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972378111173563e-05,
+      "loss": 0.0197,
+      "step": 57520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723733071997156e-05,
+      "loss": 0.0171,
+      "step": 57530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723685032258685e-05,
+      "loss": 0.0178,
+      "step": 57540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723636992520214e-05,
+      "loss": 0.0194,
+      "step": 57550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723588952781744e-05,
+      "loss": 0.0154,
+      "step": 57560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972354091304327e-05,
+      "loss": 0.0129,
+      "step": 57570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.97234928733048e-05,
+      "loss": 0.0103,
+      "step": 57580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723444833566328e-05,
+      "loss": 0.015,
+      "step": 57590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723396793827857e-05,
+      "loss": 0.0119,
+      "step": 57600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723348754089383e-05,
+      "loss": 0.0141,
+      "step": 57610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723300714350913e-05,
+      "loss": 0.0183,
+      "step": 57620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723252674612442e-05,
+      "loss": 0.0143,
+      "step": 57630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723204634873968e-05,
+      "loss": 0.0149,
+      "step": 57640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723156595135497e-05,
+      "loss": 0.0157,
+      "step": 57650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723108555397026e-05,
+      "loss": 0.0138,
+      "step": 57660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723060515658556e-05,
+      "loss": 0.0196,
+      "step": 57670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972301247592008e-05,
+      "loss": 0.0204,
+      "step": 57680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972296443618161e-05,
+      "loss": 0.019,
+      "step": 57690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972291639644314e-05,
+      "loss": 0.0165,
+      "step": 57700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722868356704666e-05,
+      "loss": 0.0142,
+      "step": 57710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722820316966195e-05,
+      "loss": 0.016,
+      "step": 57720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722772277227724e-05,
+      "loss": 0.013,
+      "step": 57730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722724237489254e-05,
+      "loss": 0.0195,
+      "step": 57740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972267619775078e-05,
+      "loss": 0.0125,
+      "step": 57750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972262815801231e-05,
+      "loss": 0.0217,
+      "step": 57760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722580118273838e-05,
+      "loss": 0.0155,
+      "step": 57770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722532078535367e-05,
+      "loss": 0.0165,
+      "step": 57780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722484038796893e-05,
+      "loss": 0.0203,
+      "step": 57790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722435999058422e-05,
+      "loss": 0.0156,
+      "step": 57800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722387959319952e-05,
+      "loss": 0.0176,
+      "step": 57810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722339919581478e-05,
+      "loss": 0.0137,
+      "step": 57820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722291879843007e-05,
+      "loss": 0.0147,
+      "step": 57830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722243840104536e-05,
+      "loss": 0.0194,
+      "step": 57840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722195800366065e-05,
+      "loss": 0.0137,
+      "step": 57850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972214776062759e-05,
+      "loss": 0.0155,
+      "step": 57860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972209972088912e-05,
+      "loss": 0.0152,
+      "step": 57870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972205168115065e-05,
+      "loss": 0.0268,
+      "step": 57880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722003641412176e-05,
+      "loss": 0.023,
+      "step": 57890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9721955601673705e-05,
+      "loss": 0.0142,
+      "step": 57900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9721907561935234e-05,
+      "loss": 0.0125,
+      "step": 57910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9721859522196764e-05,
+      "loss": 0.0165,
+      "step": 57920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972181148245829e-05,
+      "loss": 0.0168,
+      "step": 57930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972176344271982e-05,
+      "loss": 0.0188,
+      "step": 57940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9721715402981348e-05,
+      "loss": 0.0167,
+      "step": 57950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9721667363242877e-05,
+      "loss": 0.0208,
+      "step": 57960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9721619323504403e-05,
+      "loss": 0.0168,
+      "step": 57970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9721571283765932e-05,
+      "loss": 0.0137,
+      "step": 57980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972152324402746e-05,
+      "loss": 0.0193,
+      "step": 57990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9721475204288988e-05,
+      "loss": 0.0146,
+      "step": 58000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9721427164550517e-05,
+      "loss": 0.0208,
+      "step": 58010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9721379124812046e-05,
+      "loss": 0.0153,
+      "step": 58020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9721331085073575e-05,
+      "loss": 0.0128,
+      "step": 58030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.97212830453351e-05,
+      "loss": 0.0179,
+      "step": 58040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972123500559663e-05,
+      "loss": 0.0184,
+      "step": 58050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972118696585816e-05,
+      "loss": 0.0153,
+      "step": 58060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9721138926119686e-05,
+      "loss": 0.0154,
+      "step": 58070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9721090886381215e-05,
+      "loss": 0.0202,
+      "step": 58080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9721042846642744e-05,
+      "loss": 0.0236,
+      "step": 58090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720994806904273e-05,
+      "loss": 0.0163,
+      "step": 58100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.97209467671658e-05,
+      "loss": 0.0229,
+      "step": 58110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972089872742733e-05,
+      "loss": 0.0191,
+      "step": 58120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720850687688858e-05,
+      "loss": 0.0171,
+      "step": 58130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720802647950387e-05,
+      "loss": 0.0207,
+      "step": 58140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720754608211913e-05,
+      "loss": 0.0171,
+      "step": 58150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720706568473442e-05,
+      "loss": 0.0152,
+      "step": 58160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972065852873497e-05,
+      "loss": 0.0194,
+      "step": 58170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720610488996497e-05,
+      "loss": 0.0159,
+      "step": 58180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720562449258027e-05,
+      "loss": 0.0139,
+      "step": 58190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720514409519556e-05,
+      "loss": 0.0136,
+      "step": 58200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720466369781085e-05,
+      "loss": 0.0127,
+      "step": 58210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972041833004261e-05,
+      "loss": 0.0141,
+      "step": 58220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972037029030414e-05,
+      "loss": 0.0183,
+      "step": 58230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972032225056567e-05,
+      "loss": 0.0195,
+      "step": 58240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720274210827196e-05,
+      "loss": 0.0171,
+      "step": 58250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720226171088725e-05,
+      "loss": 0.0159,
+      "step": 58260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720178131350254e-05,
+      "loss": 0.0164,
+      "step": 58270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9720130091611783e-05,
+      "loss": 0.0177,
+      "step": 58280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972008205187331e-05,
+      "loss": 0.0146,
+      "step": 58290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.972003401213484e-05,
+      "loss": 0.0151,
+      "step": 58300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719985972396368e-05,
+      "loss": 0.0139,
+      "step": 58310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719937932657897e-05,
+      "loss": 0.0239,
+      "step": 58320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719889892919423e-05,
+      "loss": 0.0157,
+      "step": 58330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719841853180952e-05,
+      "loss": 0.023,
+      "step": 58340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971979381344248e-05,
+      "loss": 0.0171,
+      "step": 58350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719745773704007e-05,
+      "loss": 0.0146,
+      "step": 58360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719697733965537e-05,
+      "loss": 0.0216,
+      "step": 58370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719649694227066e-05,
+      "loss": 0.0157,
+      "step": 58380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719601654488595e-05,
+      "loss": 0.0154,
+      "step": 58390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971955361475012e-05,
+      "loss": 0.0195,
+      "step": 58400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971950557501165e-05,
+      "loss": 0.0141,
+      "step": 58410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971945753527318e-05,
+      "loss": 0.0192,
+      "step": 58420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719409495534705e-05,
+      "loss": 0.0149,
+      "step": 58430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719361455796235e-05,
+      "loss": 0.0117,
+      "step": 58440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719313416057764e-05,
+      "loss": 0.0126,
+      "step": 58450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719265376319293e-05,
+      "loss": 0.0193,
+      "step": 58460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971921733658082e-05,
+      "loss": 0.0138,
+      "step": 58470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971916929684235e-05,
+      "loss": 0.0176,
+      "step": 58480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719121257103878e-05,
+      "loss": 0.0161,
+      "step": 58490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719073217365407e-05,
+      "loss": 0.0161,
+      "step": 58500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9719025177626933e-05,
+      "loss": 0.0181,
+      "step": 58510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9718977137888462e-05,
+      "loss": 0.0151,
+      "step": 58520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971892909814999e-05,
+      "loss": 0.0124,
+      "step": 58530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9718881058411517e-05,
+      "loss": 0.0159,
+      "step": 58540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9718833018673047e-05,
+      "loss": 0.0214,
+      "step": 58550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9718784978934576e-05,
+      "loss": 0.0164,
+      "step": 58560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9718736939196105e-05,
+      "loss": 0.0143,
+      "step": 58570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971868889945763e-05,
+      "loss": 0.0117,
+      "step": 58580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971864085971916e-05,
+      "loss": 0.0144,
+      "step": 58590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.971859281998069e-05,
+      "loss": 0.017,
+      "step": 58600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9718544780242215e-05,
+      "loss": 0.0138,
+      "step": 58610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9718496740503745e-05,
+      "loss": 0.0178,
+      "step": 58620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9718448700765274e-05,
+      "loss": 0.0156,
+      "step": 58630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9718400661026803e-05,
+      "loss": 0.0143,
+      "step": 58640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971835262128833e-05,
+      "loss": 0.0192,
+      "step": 58650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971830458154986e-05,
+      "loss": 0.0148,
+      "step": 58660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9718256541811388e-05,
+      "loss": 0.0177,
+      "step": 58670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9718208502072917e-05,
+      "loss": 0.0176,
+      "step": 58680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9718160462334443e-05,
+      "loss": 0.014,
+      "step": 58690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9718112422595972e-05,
+      "loss": 0.0157,
+      "step": 58700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.97180643828575e-05,
+      "loss": 0.0187,
+      "step": 58710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9718016343119027e-05,
+      "loss": 0.0171,
+      "step": 58720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971796830338056e-05,
+      "loss": 0.0174,
+      "step": 58730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717920263642086e-05,
+      "loss": 0.0176,
+      "step": 58740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717872223903615e-05,
+      "loss": 0.0159,
+      "step": 58750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717824184165144e-05,
+      "loss": 0.0137,
+      "step": 58760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717776144426674e-05,
+      "loss": 0.0181,
+      "step": 58770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.97177281046882e-05,
+      "loss": 0.0226,
+      "step": 58780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971768006494973e-05,
+      "loss": 0.0138,
+      "step": 58790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717632025211258e-05,
+      "loss": 0.0137,
+      "step": 58800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717583985472787e-05,
+      "loss": 0.0146,
+      "step": 58810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717535945734313e-05,
+      "loss": 0.0156,
+      "step": 58820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717487905995842e-05,
+      "loss": 0.0171,
+      "step": 58830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971743986625737e-05,
+      "loss": 0.0137,
+      "step": 58840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717391826518898e-05,
+      "loss": 0.0117,
+      "step": 58850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717343786780427e-05,
+      "loss": 0.0148,
+      "step": 58860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717295747041956e-05,
+      "loss": 0.0176,
+      "step": 58870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717247707303485e-05,
+      "loss": 0.0119,
+      "step": 58880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971719966756501e-05,
+      "loss": 0.0106,
+      "step": 58890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971715162782654e-05,
+      "loss": 0.0152,
+      "step": 58900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971710358808807e-05,
+      "loss": 0.0154,
+      "step": 58910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717055548349596e-05,
+      "loss": 0.0142,
+      "step": 58920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9717007508611125e-05,
+      "loss": 0.0228,
+      "step": 58930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716959468872654e-05,
+      "loss": 0.0176,
+      "step": 58940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716911429134183e-05,
+      "loss": 0.0165,
+      "step": 58950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971686338939571e-05,
+      "loss": 0.0132,
+      "step": 58960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971681534965724e-05,
+      "loss": 0.0175,
+      "step": 58970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716767309918768e-05,
+      "loss": 0.0145,
+      "step": 58980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716719270180297e-05,
+      "loss": 0.013,
+      "step": 58990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716671230441823e-05,
+      "loss": 0.017,
+      "step": 59000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716623190703352e-05,
+      "loss": 0.0107,
+      "step": 59010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971657515096488e-05,
+      "loss": 0.0165,
+      "step": 59020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716527111226407e-05,
+      "loss": 0.016,
+      "step": 59030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716479071487937e-05,
+      "loss": 0.0157,
+      "step": 59040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716431031749466e-05,
+      "loss": 0.0179,
+      "step": 59050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716382992010995e-05,
+      "loss": 0.0212,
+      "step": 59060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971633495227252e-05,
+      "loss": 0.0141,
+      "step": 59070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971628691253405e-05,
+      "loss": 0.017,
+      "step": 59080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971623887279558e-05,
+      "loss": 0.0151,
+      "step": 59090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716190833057106e-05,
+      "loss": 0.017,
+      "step": 59100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716142793318635e-05,
+      "loss": 0.02,
+      "step": 59110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716094753580164e-05,
+      "loss": 0.0211,
+      "step": 59120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9716046713841693e-05,
+      "loss": 0.0177,
+      "step": 59130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971599867410322e-05,
+      "loss": 0.0179,
+      "step": 59140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971595063436475e-05,
+      "loss": 0.0115,
+      "step": 59150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9715902594626278e-05,
+      "loss": 0.0121,
+      "step": 59160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9715854554887807e-05,
+      "loss": 0.0216,
+      "step": 59170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9715806515149333e-05,
+      "loss": 0.0189,
+      "step": 59180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9715758475410862e-05,
+      "loss": 0.0189,
+      "step": 59190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971571043567239e-05,
+      "loss": 0.0169,
+      "step": 59200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9715662395933917e-05,
+      "loss": 0.0145,
+      "step": 59210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9715614356195447e-05,
+      "loss": 0.0149,
+      "step": 59220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9715566316456976e-05,
+      "loss": 0.0141,
+      "step": 59230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9715518276718505e-05,
+      "loss": 0.0165,
+      "step": 59240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971547023698003e-05,
+      "loss": 0.0216,
+      "step": 59250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971542219724156e-05,
+      "loss": 0.0144,
+      "step": 59260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971537415750309e-05,
+      "loss": 0.0152,
+      "step": 59270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9715326117764616e-05,
+      "loss": 0.0131,
+      "step": 59280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9715278078026145e-05,
+      "loss": 0.0193,
+      "step": 59290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9715230038287674e-05,
+      "loss": 0.0209,
+      "step": 59300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9715181998549203e-05,
+      "loss": 0.0169,
+      "step": 59310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.971513395881073e-05,
+      "loss": 0.0149,
+      "step": 59320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971508591907226e-05,
+      "loss": 0.0157,
+      "step": 59330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9715037879333788e-05,
+      "loss": 0.0194,
+      "step": 59340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714989839595317e-05,
+      "loss": 0.012,
+      "step": 59350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714941799856843e-05,
+      "loss": 0.0261,
+      "step": 59360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714893760118372e-05,
+      "loss": 0.0122,
+      "step": 59370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.97148457203799e-05,
+      "loss": 0.0136,
+      "step": 59380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714797680641427e-05,
+      "loss": 0.0134,
+      "step": 59390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714749640902957e-05,
+      "loss": 0.0142,
+      "step": 59400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714701601164486e-05,
+      "loss": 0.0258,
+      "step": 59410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714653561426015e-05,
+      "loss": 0.0171,
+      "step": 59420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971460552168754e-05,
+      "loss": 0.015,
+      "step": 59430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971455748194907e-05,
+      "loss": 0.0176,
+      "step": 59440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.97145094422106e-05,
+      "loss": 0.0109,
+      "step": 59450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714461402472125e-05,
+      "loss": 0.0158,
+      "step": 59460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714413362733655e-05,
+      "loss": 0.0222,
+      "step": 59470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714365322995184e-05,
+      "loss": 0.0135,
+      "step": 59480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714317283256713e-05,
+      "loss": 0.0175,
+      "step": 59490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971426924351824e-05,
+      "loss": 0.0207,
+      "step": 59500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971422120377977e-05,
+      "loss": 0.0131,
+      "step": 59510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714173164041298e-05,
+      "loss": 0.0166,
+      "step": 59520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714125124302827e-05,
+      "loss": 0.0138,
+      "step": 59530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714077084564353e-05,
+      "loss": 0.0167,
+      "step": 59540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9714029044825882e-05,
+      "loss": 0.0166,
+      "step": 59550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971398100508741e-05,
+      "loss": 0.0167,
+      "step": 59560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713937769322786e-05,
+      "loss": 0.0166,
+      "step": 59570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713889729584315e-05,
+      "loss": 0.0199,
+      "step": 59580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713841689845845e-05,
+      "loss": 0.0148,
+      "step": 59590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971379365010737e-05,
+      "loss": 0.0218,
+      "step": 59600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.97137456103689e-05,
+      "loss": 0.0146,
+      "step": 59610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971369757063043e-05,
+      "loss": 0.0162,
+      "step": 59620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713649530891955e-05,
+      "loss": 0.0193,
+      "step": 59630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713601491153484e-05,
+      "loss": 0.022,
+      "step": 59640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713553451415013e-05,
+      "loss": 0.0216,
+      "step": 59650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713505411676543e-05,
+      "loss": 0.0157,
+      "step": 59660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971345737193807e-05,
+      "loss": 0.013,
+      "step": 59670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713409332199598e-05,
+      "loss": 0.0166,
+      "step": 59680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713361292461127e-05,
+      "loss": 0.0172,
+      "step": 59690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713313252722653e-05,
+      "loss": 0.0163,
+      "step": 59700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713265212984182e-05,
+      "loss": 0.0126,
+      "step": 59710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971321717324571e-05,
+      "loss": 0.0168,
+      "step": 59720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971316913350724e-05,
+      "loss": 0.0166,
+      "step": 59730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713121093768767e-05,
+      "loss": 0.0234,
+      "step": 59740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713073054030296e-05,
+      "loss": 0.0135,
+      "step": 59750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9713025014291825e-05,
+      "loss": 0.013,
+      "step": 59760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712976974553354e-05,
+      "loss": 0.0166,
+      "step": 59770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971292893481488e-05,
+      "loss": 0.0111,
+      "step": 59780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971288089507641e-05,
+      "loss": 0.0196,
+      "step": 59790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971283285533794e-05,
+      "loss": 0.0123,
+      "step": 59800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712784815599465e-05,
+      "loss": 0.014,
+      "step": 59810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712736775860994e-05,
+      "loss": 0.0237,
+      "step": 59820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712688736122523e-05,
+      "loss": 0.0157,
+      "step": 59830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712640696384053e-05,
+      "loss": 0.0186,
+      "step": 59840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971259265664558e-05,
+      "loss": 0.0173,
+      "step": 59850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712544616907108e-05,
+      "loss": 0.0172,
+      "step": 59860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712496577168637e-05,
+      "loss": 0.0241,
+      "step": 59870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712448537430163e-05,
+      "loss": 0.0122,
+      "step": 59880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712400497691692e-05,
+      "loss": 0.0151,
+      "step": 59890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971235245795322e-05,
+      "loss": 0.0161,
+      "step": 59900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971230441821475e-05,
+      "loss": 0.0159,
+      "step": 59910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712256378476277e-05,
+      "loss": 0.0133,
+      "step": 59920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712208338737806e-05,
+      "loss": 0.0184,
+      "step": 59930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712160298999335e-05,
+      "loss": 0.0127,
+      "step": 59940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9712112259260864e-05,
+      "loss": 0.0185,
+      "step": 59950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971206421952239e-05,
+      "loss": 0.0201,
+      "step": 59960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971201617978392e-05,
+      "loss": 0.0124,
+      "step": 59970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971196814004545e-05,
+      "loss": 0.0144,
+      "step": 59980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9711920100306975e-05,
+      "loss": 0.0147,
+      "step": 59990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9711872060568504e-05,
+      "loss": 0.0164,
+      "step": 60000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9711824020830033e-05,
+      "loss": 0.0174,
+      "step": 60010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9711775981091562e-05,
+      "loss": 0.0137,
+      "step": 60020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.971172794135309e-05,
+      "loss": 0.0159,
+      "step": 60030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9711679901614618e-05,
+      "loss": 0.014,
+      "step": 60040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9711631861876147e-05,
+      "loss": 0.0147,
+      "step": 60050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9711583822137673e-05,
+      "loss": 0.0209,
+      "step": 60060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9711535782399202e-05,
+      "loss": 0.0149,
+      "step": 60070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.971148774266073e-05,
+      "loss": 0.0105,
+      "step": 60080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.971143970292226e-05,
+      "loss": 0.0123,
+      "step": 60090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9711391663183786e-05,
+      "loss": 0.016,
+      "step": 60100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9711343623445316e-05,
+      "loss": 0.0124,
+      "step": 60110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9711295583706845e-05,
+      "loss": 0.0203,
+      "step": 60120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9711247543968374e-05,
+      "loss": 0.0175,
+      "step": 60130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.97111995042299e-05,
+      "loss": 0.0194,
+      "step": 60140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.971115146449143e-05,
+      "loss": 0.0142,
+      "step": 60150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.971110342475296e-05,
+      "loss": 0.0156,
+      "step": 60160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9711055385014485e-05,
+      "loss": 0.0135,
+      "step": 60170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9711007345276014e-05,
+      "loss": 0.0147,
+      "step": 60180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710959305537543e-05,
+      "loss": 0.016,
+      "step": 60190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710911265799072e-05,
+      "loss": 0.0164,
+      "step": 60200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710863226060598e-05,
+      "loss": 0.0221,
+      "step": 60210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710815186322128e-05,
+      "loss": 0.0148,
+      "step": 60220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710767146583657e-05,
+      "loss": 0.0146,
+      "step": 60230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710719106845183e-05,
+      "loss": 0.0141,
+      "step": 60240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710671067106712e-05,
+      "loss": 0.0159,
+      "step": 60250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.971062302736824e-05,
+      "loss": 0.0166,
+      "step": 60260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.971057498762977e-05,
+      "loss": 0.0208,
+      "step": 60270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710526947891296e-05,
+      "loss": 0.0157,
+      "step": 60280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710478908152826e-05,
+      "loss": 0.0178,
+      "step": 60290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710430868414355e-05,
+      "loss": 0.0172,
+      "step": 60300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.971038282867588e-05,
+      "loss": 0.0154,
+      "step": 60310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.971033478893741e-05,
+      "loss": 0.0185,
+      "step": 60320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.971028674919894e-05,
+      "loss": 0.0224,
+      "step": 60330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.971023870946047e-05,
+      "loss": 0.0179,
+      "step": 60340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710190669721994e-05,
+      "loss": 0.0144,
+      "step": 60350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710142629983524e-05,
+      "loss": 0.0165,
+      "step": 60360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710094590245053e-05,
+      "loss": 0.0201,
+      "step": 60370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9710046550506582e-05,
+      "loss": 0.0185,
+      "step": 60380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709998510768108e-05,
+      "loss": 0.0143,
+      "step": 60390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709950471029637e-05,
+      "loss": 0.0156,
+      "step": 60400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709902431291167e-05,
+      "loss": 0.0206,
+      "step": 60410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709854391552693e-05,
+      "loss": 0.0153,
+      "step": 60420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709806351814222e-05,
+      "loss": 0.0146,
+      "step": 60430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.970975831207575e-05,
+      "loss": 0.0199,
+      "step": 60440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.970971027233728e-05,
+      "loss": 0.0139,
+      "step": 60450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709662232598806e-05,
+      "loss": 0.0163,
+      "step": 60460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709614192860336e-05,
+      "loss": 0.0163,
+      "step": 60470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709566153121865e-05,
+      "loss": 0.0175,
+      "step": 60480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.970951811338339e-05,
+      "loss": 0.0141,
+      "step": 60490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.970947007364492e-05,
+      "loss": 0.0169,
+      "step": 60500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.970942203390645e-05,
+      "loss": 0.0194,
+      "step": 60510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.970937399416798e-05,
+      "loss": 0.0152,
+      "step": 60520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709325954429504e-05,
+      "loss": 0.0158,
+      "step": 60530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709277914691034e-05,
+      "loss": 0.0162,
+      "step": 60540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709229874952563e-05,
+      "loss": 0.0145,
+      "step": 60550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709181835214092e-05,
+      "loss": 0.0127,
+      "step": 60560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709133795475618e-05,
+      "loss": 0.0172,
+      "step": 60570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709085755737147e-05,
+      "loss": 0.0142,
+      "step": 60580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9709037715998677e-05,
+      "loss": 0.0129,
+      "step": 60590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9708989676260203e-05,
+      "loss": 0.0175,
+      "step": 60600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9708941636521732e-05,
+      "loss": 0.0191,
+      "step": 60610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.970889359678326e-05,
+      "loss": 0.0159,
+      "step": 60620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.970884555704479e-05,
+      "loss": 0.0198,
+      "step": 60630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9708797517306316e-05,
+      "loss": 0.0172,
+      "step": 60640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9708749477567845e-05,
+      "loss": 0.0172,
+      "step": 60650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9708701437829375e-05,
+      "loss": 0.0193,
+      "step": 60660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.97086533980909e-05,
+      "loss": 0.0157,
+      "step": 60670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.970860535835243e-05,
+      "loss": 0.0162,
+      "step": 60680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.970855731861396e-05,
+      "loss": 0.0171,
+      "step": 60690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.970850927887549e-05,
+      "loss": 0.0128,
+      "step": 60700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9708461239137014e-05,
+      "loss": 0.0194,
+      "step": 60710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9708413199398544e-05,
+      "loss": 0.0202,
+      "step": 60720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9708365159660073e-05,
+      "loss": 0.0121,
+      "step": 60730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9708317119921602e-05,
+      "loss": 0.0145,
+      "step": 60740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9708269080183128e-05,
+      "loss": 0.0174,
+      "step": 60750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9708221040444657e-05,
+      "loss": 0.0137,
+      "step": 60760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9708173000706187e-05,
+      "loss": 0.0153,
+      "step": 60770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9708124960967712e-05,
+      "loss": 0.0224,
+      "step": 60780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9708076921229242e-05,
+      "loss": 0.0135,
+      "step": 60790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970802888149077e-05,
+      "loss": 0.015,
+      "step": 60800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.97079808417523e-05,
+      "loss": 0.0146,
+      "step": 60810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707932802013826e-05,
+      "loss": 0.0205,
+      "step": 60820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707884762275355e-05,
+      "loss": 0.0135,
+      "step": 60830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707836722536885e-05,
+      "loss": 0.0172,
+      "step": 60840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970778868279841e-05,
+      "loss": 0.0156,
+      "step": 60850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970774064305994e-05,
+      "loss": 0.0132,
+      "step": 60860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970769260332147e-05,
+      "loss": 0.0143,
+      "step": 60870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707644563583e-05,
+      "loss": 0.0121,
+      "step": 60880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707596523844524e-05,
+      "loss": 0.0166,
+      "step": 60890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707548484106054e-05,
+      "loss": 0.019,
+      "step": 60900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707500444367583e-05,
+      "loss": 0.0168,
+      "step": 60910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707452404629112e-05,
+      "loss": 0.0146,
+      "step": 60920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707404364890638e-05,
+      "loss": 0.0151,
+      "step": 60930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707356325152167e-05,
+      "loss": 0.0162,
+      "step": 60940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707308285413696e-05,
+      "loss": 0.0172,
+      "step": 60950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707260245675222e-05,
+      "loss": 0.0261,
+      "step": 60960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970721220593675e-05,
+      "loss": 0.0157,
+      "step": 60970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970716416619828e-05,
+      "loss": 0.0182,
+      "step": 60980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970711612645981e-05,
+      "loss": 0.014,
+      "step": 60990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707068086721336e-05,
+      "loss": 0.0171,
+      "step": 61000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9707020046982865e-05,
+      "loss": 0.0181,
+      "step": 61010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706972007244395e-05,
+      "loss": 0.0141,
+      "step": 61020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970692396750592e-05,
+      "loss": 0.0142,
+      "step": 61030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970687592776745e-05,
+      "loss": 0.0116,
+      "step": 61040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970682788802898e-05,
+      "loss": 0.0189,
+      "step": 61050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706779848290508e-05,
+      "loss": 0.0189,
+      "step": 61060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706731808552034e-05,
+      "loss": 0.0158,
+      "step": 61070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706683768813563e-05,
+      "loss": 0.0137,
+      "step": 61080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706635729075093e-05,
+      "loss": 0.0222,
+      "step": 61090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706587689336622e-05,
+      "loss": 0.0203,
+      "step": 61100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706539649598148e-05,
+      "loss": 0.021,
+      "step": 61110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706491609859677e-05,
+      "loss": 0.0149,
+      "step": 61120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706443570121206e-05,
+      "loss": 0.0197,
+      "step": 61130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706395530382732e-05,
+      "loss": 0.0177,
+      "step": 61140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970634749064426e-05,
+      "loss": 0.0116,
+      "step": 61150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970629945090579e-05,
+      "loss": 0.0162,
+      "step": 61160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970625141116732e-05,
+      "loss": 0.0133,
+      "step": 61170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706203371428846e-05,
+      "loss": 0.0129,
+      "step": 61180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706155331690375e-05,
+      "loss": 0.0187,
+      "step": 61190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9706107291951904e-05,
+      "loss": 0.0135,
+      "step": 61200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970605925221343e-05,
+      "loss": 0.0143,
+      "step": 61210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970601121247496e-05,
+      "loss": 0.0115,
+      "step": 61220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970596317273649e-05,
+      "loss": 0.017,
+      "step": 61230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9705915132998018e-05,
+      "loss": 0.0181,
+      "step": 61240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9705867093259544e-05,
+      "loss": 0.0171,
+      "step": 61250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9705819053521073e-05,
+      "loss": 0.0155,
+      "step": 61260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9705771013782603e-05,
+      "loss": 0.0142,
+      "step": 61270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9705722974044132e-05,
+      "loss": 0.0165,
+      "step": 61280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9705674934305658e-05,
+      "loss": 0.0154,
+      "step": 61290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9705626894567187e-05,
+      "loss": 0.0187,
+      "step": 61300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9705578854828716e-05,
+      "loss": 0.0184,
+      "step": 61310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9705530815090242e-05,
+      "loss": 0.0167,
+      "step": 61320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970548277535177e-05,
+      "loss": 0.0166,
+      "step": 61330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.97054347356133e-05,
+      "loss": 0.0138,
+      "step": 61340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970538669587483e-05,
+      "loss": 0.0108,
+      "step": 61350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9705338656136356e-05,
+      "loss": 0.0189,
+      "step": 61360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9705290616397885e-05,
+      "loss": 0.0175,
+      "step": 61370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9705242576659414e-05,
+      "loss": 0.0197,
+      "step": 61380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970519453692094e-05,
+      "loss": 0.0128,
+      "step": 61390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.970514649718247e-05,
+      "loss": 0.0163,
+      "step": 61400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9705098457444e-05,
+      "loss": 0.0142,
+      "step": 61410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9705050417705528e-05,
+      "loss": 0.0212,
+      "step": 61420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9705002377967054e-05,
+      "loss": 0.0144,
+      "step": 61430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704954338228583e-05,
+      "loss": 0.0132,
+      "step": 61440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704906298490113e-05,
+      "loss": 0.017,
+      "step": 61450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704858258751642e-05,
+      "loss": 0.0162,
+      "step": 61460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704810219013168e-05,
+      "loss": 0.0126,
+      "step": 61470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704762179274697e-05,
+      "loss": 0.0153,
+      "step": 61480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704714139536226e-05,
+      "loss": 0.0146,
+      "step": 61490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704666099797752e-05,
+      "loss": 0.0189,
+      "step": 61500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970461806005928e-05,
+      "loss": 0.0165,
+      "step": 61510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970457002032081e-05,
+      "loss": 0.0212,
+      "step": 61520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970452198058234e-05,
+      "loss": 0.0157,
+      "step": 61530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704473940843866e-05,
+      "loss": 0.0126,
+      "step": 61540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704425901105395e-05,
+      "loss": 0.017,
+      "step": 61550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704377861366924e-05,
+      "loss": 0.0173,
+      "step": 61560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970432982162845e-05,
+      "loss": 0.0225,
+      "step": 61570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970428178188998e-05,
+      "loss": 0.0162,
+      "step": 61580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970423374215151e-05,
+      "loss": 0.0137,
+      "step": 61590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704185702413038e-05,
+      "loss": 0.0139,
+      "step": 61600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704137662674564e-05,
+      "loss": 0.0155,
+      "step": 61610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704089622936093e-05,
+      "loss": 0.0179,
+      "step": 61620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9704041583197622e-05,
+      "loss": 0.0166,
+      "step": 61630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9703993543459152e-05,
+      "loss": 0.0173,
+      "step": 61640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9703945503720678e-05,
+      "loss": 0.0147,
+      "step": 61650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9703897463982207e-05,
+      "loss": 0.0152,
+      "step": 61660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9703849424243736e-05,
+      "loss": 0.019,
+      "step": 61670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9703801384505262e-05,
+      "loss": 0.0119,
+      "step": 61680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970375334476679e-05,
+      "loss": 0.0129,
+      "step": 61690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970370530502832e-05,
+      "loss": 0.0184,
+      "step": 61700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970365726528985e-05,
+      "loss": 0.016,
+      "step": 61710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9703609225551376e-05,
+      "loss": 0.0152,
+      "step": 61720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9703561185812905e-05,
+      "loss": 0.0157,
+      "step": 61730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9703513146074434e-05,
+      "loss": 0.0174,
+      "step": 61740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970346510633596e-05,
+      "loss": 0.0133,
+      "step": 61750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970341706659749e-05,
+      "loss": 0.0143,
+      "step": 61760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970336902685902e-05,
+      "loss": 0.0181,
+      "step": 61770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9703320987120548e-05,
+      "loss": 0.0174,
+      "step": 61780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9703272947382074e-05,
+      "loss": 0.0166,
+      "step": 61790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9703224907643603e-05,
+      "loss": 0.0148,
+      "step": 61800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9703176867905132e-05,
+      "loss": 0.0172,
+      "step": 61810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970312882816666e-05,
+      "loss": 0.0136,
+      "step": 61820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970308078842819e-05,
+      "loss": 0.0186,
+      "step": 61830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970303274868972e-05,
+      "loss": 0.0174,
+      "step": 61840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702984708951246e-05,
+      "loss": 0.0208,
+      "step": 61850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702936669212775e-05,
+      "loss": 0.0155,
+      "step": 61860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702888629474305e-05,
+      "loss": 0.0132,
+      "step": 61870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970284058973583e-05,
+      "loss": 0.0137,
+      "step": 61880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970279254999736e-05,
+      "loss": 0.0158,
+      "step": 61890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970274451025889e-05,
+      "loss": 0.0177,
+      "step": 61900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702696470520418e-05,
+      "loss": 0.017,
+      "step": 61910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702648430781944e-05,
+      "loss": 0.0167,
+      "step": 61920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702600391043473e-05,
+      "loss": 0.0206,
+      "step": 61930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702552351305003e-05,
+      "loss": 0.0187,
+      "step": 61940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702504311566532e-05,
+      "loss": 0.017,
+      "step": 61950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702456271828058e-05,
+      "loss": 0.0154,
+      "step": 61960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702408232089587e-05,
+      "loss": 0.013,
+      "step": 61970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702360192351116e-05,
+      "loss": 0.0148,
+      "step": 61980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702312152612642e-05,
+      "loss": 0.015,
+      "step": 61990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970226411287417e-05,
+      "loss": 0.0183,
+      "step": 62000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.97022160731357e-05,
+      "loss": 0.0145,
+      "step": 62010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970216803339723e-05,
+      "loss": 0.0169,
+      "step": 62020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702119993658756e-05,
+      "loss": 0.0188,
+      "step": 62030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702071953920285e-05,
+      "loss": 0.0168,
+      "step": 62040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9702023914181814e-05,
+      "loss": 0.0181,
+      "step": 62050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970197587444334e-05,
+      "loss": 0.017,
+      "step": 62060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.970192783470487e-05,
+      "loss": 0.0127,
+      "step": 62070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.97018797949664e-05,
+      "loss": 0.0164,
+      "step": 62080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9701831755227928e-05,
+      "loss": 0.0183,
+      "step": 62090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9701783715489454e-05,
+      "loss": 0.0119,
+      "step": 62100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9701735675750983e-05,
+      "loss": 0.0152,
+      "step": 62110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9701687636012513e-05,
+      "loss": 0.0147,
+      "step": 62120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9701639596274042e-05,
+      "loss": 0.0176,
+      "step": 62130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9701591556535568e-05,
+      "loss": 0.0172,
+      "step": 62140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9701543516797097e-05,
+      "loss": 0.0112,
+      "step": 62150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9701495477058626e-05,
+      "loss": 0.0124,
+      "step": 62160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9701447437320152e-05,
+      "loss": 0.0171,
+      "step": 62170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.970139939758168e-05,
+      "loss": 0.0172,
+      "step": 62180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.970135135784321e-05,
+      "loss": 0.0146,
+      "step": 62190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.970130331810474e-05,
+      "loss": 0.0162,
+      "step": 62200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9701255278366266e-05,
+      "loss": 0.017,
+      "step": 62210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9701207238627795e-05,
+      "loss": 0.016,
+      "step": 62220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9701159198889324e-05,
+      "loss": 0.0177,
+      "step": 62230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.970111115915085e-05,
+      "loss": 0.0175,
+      "step": 62240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.970106311941238e-05,
+      "loss": 0.0146,
+      "step": 62250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.970101507967391e-05,
+      "loss": 0.0167,
+      "step": 62260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700967039935438e-05,
+      "loss": 0.0134,
+      "step": 62270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700919000196964e-05,
+      "loss": 0.0182,
+      "step": 62280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700870960458493e-05,
+      "loss": 0.0245,
+      "step": 62290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700822920720023e-05,
+      "loss": 0.0204,
+      "step": 62300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700774880981552e-05,
+      "loss": 0.0181,
+      "step": 62310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700726841243078e-05,
+      "loss": 0.0157,
+      "step": 62320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700678801504607e-05,
+      "loss": 0.018,
+      "step": 62330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700630761766136e-05,
+      "loss": 0.0163,
+      "step": 62340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700582722027662e-05,
+      "loss": 0.0196,
+      "step": 62350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.970053468228919e-05,
+      "loss": 0.0158,
+      "step": 62360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.970048664255072e-05,
+      "loss": 0.0116,
+      "step": 62370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.970043860281225e-05,
+      "loss": 0.0183,
+      "step": 62380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700390563073776e-05,
+      "loss": 0.0128,
+      "step": 62390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700342523335305e-05,
+      "loss": 0.0155,
+      "step": 62400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700294483596834e-05,
+      "loss": 0.0125,
+      "step": 62410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.970024644385836e-05,
+      "loss": 0.0111,
+      "step": 62420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.970019840411989e-05,
+      "loss": 0.0162,
+      "step": 62430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.970015036438142e-05,
+      "loss": 0.0168,
+      "step": 62440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700102324642948e-05,
+      "loss": 0.0181,
+      "step": 62450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700054284904474e-05,
+      "loss": 0.0186,
+      "step": 62460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9700006245166003e-05,
+      "loss": 0.0167,
+      "step": 62470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699958205427532e-05,
+      "loss": 0.0162,
+      "step": 62480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699910165689062e-05,
+      "loss": 0.0128,
+      "step": 62490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699862125950588e-05,
+      "loss": 0.0133,
+      "step": 62500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699814086212117e-05,
+      "loss": 0.0129,
+      "step": 62510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699766046473646e-05,
+      "loss": 0.0179,
+      "step": 62520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699718006735172e-05,
+      "loss": 0.0188,
+      "step": 62530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.96996699669967e-05,
+      "loss": 0.018,
+      "step": 62540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.969962192725823e-05,
+      "loss": 0.014,
+      "step": 62550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.969957388751976e-05,
+      "loss": 0.0198,
+      "step": 62560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699525847781286e-05,
+      "loss": 0.0144,
+      "step": 62570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699477808042815e-05,
+      "loss": 0.0184,
+      "step": 62580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699429768304344e-05,
+      "loss": 0.0151,
+      "step": 62590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.969938172856587e-05,
+      "loss": 0.0174,
+      "step": 62600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.96993336888274e-05,
+      "loss": 0.017,
+      "step": 62610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.969928564908893e-05,
+      "loss": 0.0114,
+      "step": 62620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699237609350458e-05,
+      "loss": 0.0153,
+      "step": 62630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699189569611984e-05,
+      "loss": 0.0149,
+      "step": 62640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699141529873513e-05,
+      "loss": 0.0156,
+      "step": 62650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9699093490135042e-05,
+      "loss": 0.0133,
+      "step": 62660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.969904545039657e-05,
+      "loss": 0.016,
+      "step": 62670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9698997410658097e-05,
+      "loss": 0.0125,
+      "step": 62680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9698949370919627e-05,
+      "loss": 0.0147,
+      "step": 62690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9698901331181156e-05,
+      "loss": 0.0143,
+      "step": 62700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9698853291442682e-05,
+      "loss": 0.0163,
+      "step": 62710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.969880525170421e-05,
+      "loss": 0.0158,
+      "step": 62720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.969875721196574e-05,
+      "loss": 0.0179,
+      "step": 62730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.969870917222727e-05,
+      "loss": 0.0132,
+      "step": 62740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9698661132488796e-05,
+      "loss": 0.0127,
+      "step": 62750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9698613092750325e-05,
+      "loss": 0.0185,
+      "step": 62760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9698565053011854e-05,
+      "loss": 0.0127,
+      "step": 62770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.969851701327338e-05,
+      "loss": 0.0175,
+      "step": 62780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.969846897353491e-05,
+      "loss": 0.0187,
+      "step": 62790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969842093379644e-05,
+      "loss": 0.0113,
+      "step": 62800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9698372894057968e-05,
+      "loss": 0.0179,
+      "step": 62810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9698324854319494e-05,
+      "loss": 0.0136,
+      "step": 62820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9698276814581023e-05,
+      "loss": 0.0183,
+      "step": 62830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9698228774842552e-05,
+      "loss": 0.0155,
+      "step": 62840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969818073510408e-05,
+      "loss": 0.0168,
+      "step": 62850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9698132695365607e-05,
+      "loss": 0.0186,
+      "step": 62860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9698084655627137e-05,
+      "loss": 0.0178,
+      "step": 62870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9698036615888666e-05,
+      "loss": 0.0224,
+      "step": 62880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9697988576150192e-05,
+      "loss": 0.0159,
+      "step": 62890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969794053641172e-05,
+      "loss": 0.0144,
+      "step": 62900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969789249667325e-05,
+      "loss": 0.0167,
+      "step": 62910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969784445693478e-05,
+      "loss": 0.0161,
+      "step": 62920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9697796417196306e-05,
+      "loss": 0.021,
+      "step": 62930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9697748377457835e-05,
+      "loss": 0.0216,
+      "step": 62940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9697700337719364e-05,
+      "loss": 0.0171,
+      "step": 62950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969765229798089e-05,
+      "loss": 0.0133,
+      "step": 62960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969760425824242e-05,
+      "loss": 0.0181,
+      "step": 62970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969755621850395e-05,
+      "loss": 0.0135,
+      "step": 62980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9697508178765478e-05,
+      "loss": 0.017,
+      "step": 62990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9697460139027004e-05,
+      "loss": 0.0133,
+      "step": 63000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9697412099288533e-05,
+      "loss": 0.0122,
+      "step": 63010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9697364059550062e-05,
+      "loss": 0.017,
+      "step": 63020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969731601981159e-05,
+      "loss": 0.014,
+      "step": 63030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9697267980073117e-05,
+      "loss": 0.0218,
+      "step": 63040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9697219940334647e-05,
+      "loss": 0.02,
+      "step": 63050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9697171900596176e-05,
+      "loss": 0.0131,
+      "step": 63060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9697123860857702e-05,
+      "loss": 0.0111,
+      "step": 63070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969707582111923e-05,
+      "loss": 0.0185,
+      "step": 63080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969702778138076e-05,
+      "loss": 0.0137,
+      "step": 63090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969697974164229e-05,
+      "loss": 0.0174,
+      "step": 63100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9696931701903815e-05,
+      "loss": 0.0206,
+      "step": 63110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9696883662165345e-05,
+      "loss": 0.0139,
+      "step": 63120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9696835622426874e-05,
+      "loss": 0.0136,
+      "step": 63130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.96967875826884e-05,
+      "loss": 0.0113,
+      "step": 63140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969673954294993e-05,
+      "loss": 0.0152,
+      "step": 63150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969669150321146e-05,
+      "loss": 0.0141,
+      "step": 63160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9696643463472988e-05,
+      "loss": 0.0196,
+      "step": 63170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9696595423734514e-05,
+      "loss": 0.0119,
+      "step": 63180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9696547383996043e-05,
+      "loss": 0.0231,
+      "step": 63190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9696499344257572e-05,
+      "loss": 0.019,
+      "step": 63200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.96964513045191e-05,
+      "loss": 0.012,
+      "step": 63210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9696403264780627e-05,
+      "loss": 0.02,
+      "step": 63220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9696355225042157e-05,
+      "loss": 0.0149,
+      "step": 63230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9696307185303686e-05,
+      "loss": 0.0129,
+      "step": 63240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969625914556521e-05,
+      "loss": 0.0126,
+      "step": 63250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969621110582674e-05,
+      "loss": 0.0127,
+      "step": 63260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969616306608827e-05,
+      "loss": 0.0213,
+      "step": 63270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.96961150263498e-05,
+      "loss": 0.0128,
+      "step": 63280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9696066986611325e-05,
+      "loss": 0.0147,
+      "step": 63290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9696018946872855e-05,
+      "loss": 0.0148,
+      "step": 63300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9695970907134384e-05,
+      "loss": 0.0162,
+      "step": 63310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969592286739591e-05,
+      "loss": 0.0138,
+      "step": 63320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969587482765744e-05,
+      "loss": 0.0117,
+      "step": 63330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969582678791897e-05,
+      "loss": 0.0161,
+      "step": 63340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9695778748180498e-05,
+      "loss": 0.0168,
+      "step": 63350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9695730708442023e-05,
+      "loss": 0.0154,
+      "step": 63360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9695682668703553e-05,
+      "loss": 0.0127,
+      "step": 63370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9695634628965082e-05,
+      "loss": 0.0185,
+      "step": 63380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969558658922661e-05,
+      "loss": 0.0147,
+      "step": 63390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9695538549488137e-05,
+      "loss": 0.0141,
+      "step": 63400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9695490509749666e-05,
+      "loss": 0.0164,
+      "step": 63410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969544727398504e-05,
+      "loss": 0.0134,
+      "step": 63420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.969539923424657e-05,
+      "loss": 0.0181,
+      "step": 63430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.96953511945081e-05,
+      "loss": 0.0111,
+      "step": 63440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9695303154769625e-05,
+      "loss": 0.0151,
+      "step": 63450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9695255115031155e-05,
+      "loss": 0.0142,
+      "step": 63460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9695207075292684e-05,
+      "loss": 0.0183,
+      "step": 63470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9695159035554213e-05,
+      "loss": 0.0183,
+      "step": 63480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969511099581574e-05,
+      "loss": 0.0188,
+      "step": 63490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969506295607727e-05,
+      "loss": 0.0161,
+      "step": 63500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9695014916338798e-05,
+      "loss": 0.0146,
+      "step": 63510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694966876600327e-05,
+      "loss": 0.0164,
+      "step": 63520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694918836861853e-05,
+      "loss": 0.0163,
+      "step": 63530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694870797123382e-05,
+      "loss": 0.0156,
+      "step": 63540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969482275738491e-05,
+      "loss": 0.0163,
+      "step": 63550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694774717646437e-05,
+      "loss": 0.017,
+      "step": 63560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694726677907967e-05,
+      "loss": 0.012,
+      "step": 63570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694678638169496e-05,
+      "loss": 0.0179,
+      "step": 63580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694630598431025e-05,
+      "loss": 0.0125,
+      "step": 63590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969458255869255e-05,
+      "loss": 0.0181,
+      "step": 63600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969453451895408e-05,
+      "loss": 0.0153,
+      "step": 63610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969448647921561e-05,
+      "loss": 0.0154,
+      "step": 63620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694438439477135e-05,
+      "loss": 0.0145,
+      "step": 63630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694390399738665e-05,
+      "loss": 0.0171,
+      "step": 63640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694342360000194e-05,
+      "loss": 0.0193,
+      "step": 63650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694294320261723e-05,
+      "loss": 0.0153,
+      "step": 63660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969424628052325e-05,
+      "loss": 0.0134,
+      "step": 63670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969419824078478e-05,
+      "loss": 0.0183,
+      "step": 63680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694150201046308e-05,
+      "loss": 0.0184,
+      "step": 63690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694102161307837e-05,
+      "loss": 0.0148,
+      "step": 63700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694054121569363e-05,
+      "loss": 0.0164,
+      "step": 63710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9694006081830892e-05,
+      "loss": 0.0159,
+      "step": 63720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969395804209242e-05,
+      "loss": 0.0186,
+      "step": 63730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693910002353947e-05,
+      "loss": 0.0173,
+      "step": 63740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693861962615476e-05,
+      "loss": 0.0152,
+      "step": 63750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693813922877006e-05,
+      "loss": 0.014,
+      "step": 63760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693765883138535e-05,
+      "loss": 0.0159,
+      "step": 63770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969371784340006e-05,
+      "loss": 0.0137,
+      "step": 63780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969366980366159e-05,
+      "loss": 0.015,
+      "step": 63790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969362176392312e-05,
+      "loss": 0.0196,
+      "step": 63800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693573724184645e-05,
+      "loss": 0.0173,
+      "step": 63810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693525684446175e-05,
+      "loss": 0.0132,
+      "step": 63820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693477644707704e-05,
+      "loss": 0.0153,
+      "step": 63830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693429604969233e-05,
+      "loss": 0.018,
+      "step": 63840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969338156523076e-05,
+      "loss": 0.0136,
+      "step": 63850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693333525492288e-05,
+      "loss": 0.0197,
+      "step": 63860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693285485753818e-05,
+      "loss": 0.0123,
+      "step": 63870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693237446015347e-05,
+      "loss": 0.0171,
+      "step": 63880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693189406276873e-05,
+      "loss": 0.014,
+      "step": 63890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693141366538402e-05,
+      "loss": 0.0178,
+      "step": 63900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969309332679993e-05,
+      "loss": 0.0139,
+      "step": 63910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9693045287061457e-05,
+      "loss": 0.0137,
+      "step": 63920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692997247322986e-05,
+      "loss": 0.0101,
+      "step": 63930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692949207584516e-05,
+      "loss": 0.0171,
+      "step": 63940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692901167846045e-05,
+      "loss": 0.0141,
+      "step": 63950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969285312810757e-05,
+      "loss": 0.0148,
+      "step": 63960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.96928050883691e-05,
+      "loss": 0.0143,
+      "step": 63970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969275704863063e-05,
+      "loss": 0.0117,
+      "step": 63980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692709008892155e-05,
+      "loss": 0.0163,
+      "step": 63990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692660969153685e-05,
+      "loss": 0.0107,
+      "step": 64000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692612929415214e-05,
+      "loss": 0.0171,
+      "step": 64010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692564889676743e-05,
+      "loss": 0.0219,
+      "step": 64020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969251684993827e-05,
+      "loss": 0.0154,
+      "step": 64030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692468810199798e-05,
+      "loss": 0.0128,
+      "step": 64040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692420770461327e-05,
+      "loss": 0.0141,
+      "step": 64050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692372730722857e-05,
+      "loss": 0.0153,
+      "step": 64060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692324690984383e-05,
+      "loss": 0.0157,
+      "step": 64070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692276651245912e-05,
+      "loss": 0.0153,
+      "step": 64080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969222861150744e-05,
+      "loss": 0.013,
+      "step": 64090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692180571768967e-05,
+      "loss": 0.0156,
+      "step": 64100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692132532030496e-05,
+      "loss": 0.0165,
+      "step": 64110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692084492292026e-05,
+      "loss": 0.0138,
+      "step": 64120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9692036452553555e-05,
+      "loss": 0.0163,
+      "step": 64130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969198841281508e-05,
+      "loss": 0.0169,
+      "step": 64140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969194037307661e-05,
+      "loss": 0.0172,
+      "step": 64150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969189233333814e-05,
+      "loss": 0.0174,
+      "step": 64160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9691844293599665e-05,
+      "loss": 0.0142,
+      "step": 64170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9691796253861194e-05,
+      "loss": 0.0159,
+      "step": 64180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9691748214122724e-05,
+      "loss": 0.0225,
+      "step": 64190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9691700174384253e-05,
+      "loss": 0.014,
+      "step": 64200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.969165213464578e-05,
+      "loss": 0.0101,
+      "step": 64210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9691604094907308e-05,
+      "loss": 0.0161,
+      "step": 64220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9691556055168837e-05,
+      "loss": 0.0141,
+      "step": 64230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9691508015430367e-05,
+      "loss": 0.0123,
+      "step": 64240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9691459975691893e-05,
+      "loss": 0.0154,
+      "step": 64250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9691411935953422e-05,
+      "loss": 0.0144,
+      "step": 64260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.969136389621495e-05,
+      "loss": 0.02,
+      "step": 64270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9691315856476477e-05,
+      "loss": 0.0185,
+      "step": 64280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9691267816738006e-05,
+      "loss": 0.018,
+      "step": 64290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9691219776999535e-05,
+      "loss": 0.0179,
+      "step": 64300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9691171737261065e-05,
+      "loss": 0.0157,
+      "step": 64310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.969112369752259e-05,
+      "loss": 0.0136,
+      "step": 64320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.969107565778412e-05,
+      "loss": 0.0147,
+      "step": 64330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.969102761804565e-05,
+      "loss": 0.0152,
+      "step": 64340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690979578307175e-05,
+      "loss": 0.0171,
+      "step": 64350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690931538568704e-05,
+      "loss": 0.0206,
+      "step": 64360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690883498830234e-05,
+      "loss": 0.0186,
+      "step": 64370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690835459091763e-05,
+      "loss": 0.0129,
+      "step": 64380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.969078741935329e-05,
+      "loss": 0.0198,
+      "step": 64390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690739379614818e-05,
+      "loss": 0.0115,
+      "step": 64400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690691339876347e-05,
+      "loss": 0.0175,
+      "step": 64410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690643300137877e-05,
+      "loss": 0.0143,
+      "step": 64420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690595260399402e-05,
+      "loss": 0.0169,
+      "step": 64430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690547220660932e-05,
+      "loss": 0.0189,
+      "step": 64440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.969049918092246e-05,
+      "loss": 0.0168,
+      "step": 64450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690451141183987e-05,
+      "loss": 0.0142,
+      "step": 64460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690403101445516e-05,
+      "loss": 0.0168,
+      "step": 64470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690355061707045e-05,
+      "loss": 0.017,
+      "step": 64480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690307021968575e-05,
+      "loss": 0.0147,
+      "step": 64490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.96902589822301e-05,
+      "loss": 0.0187,
+      "step": 64500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.969021094249163e-05,
+      "loss": 0.0185,
+      "step": 64510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.969016290275316e-05,
+      "loss": 0.0225,
+      "step": 64520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690114863014685e-05,
+      "loss": 0.0133,
+      "step": 64530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690066823276214e-05,
+      "loss": 0.0186,
+      "step": 64540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9690018783537744e-05,
+      "loss": 0.0208,
+      "step": 64550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689970743799273e-05,
+      "loss": 0.016,
+      "step": 64560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.96899227040608e-05,
+      "loss": 0.0168,
+      "step": 64570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689874664322328e-05,
+      "loss": 0.0118,
+      "step": 64580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689826624583857e-05,
+      "loss": 0.0107,
+      "step": 64590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689778584845386e-05,
+      "loss": 0.0108,
+      "step": 64600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689730545106912e-05,
+      "loss": 0.0129,
+      "step": 64610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.968968250536844e-05,
+      "loss": 0.0185,
+      "step": 64620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.968963446562997e-05,
+      "loss": 0.0156,
+      "step": 64630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689586425891497e-05,
+      "loss": 0.0172,
+      "step": 64640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689538386153026e-05,
+      "loss": 0.0169,
+      "step": 64650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689490346414555e-05,
+      "loss": 0.0156,
+      "step": 64660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689442306676085e-05,
+      "loss": 0.0167,
+      "step": 64670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.968939426693761e-05,
+      "loss": 0.0147,
+      "step": 64680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.968934622719914e-05,
+      "loss": 0.0171,
+      "step": 64690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.968929818746067e-05,
+      "loss": 0.0139,
+      "step": 64700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689250147722195e-05,
+      "loss": 0.0128,
+      "step": 64710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689202107983724e-05,
+      "loss": 0.0148,
+      "step": 64720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689154068245253e-05,
+      "loss": 0.0148,
+      "step": 64730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689106028506783e-05,
+      "loss": 0.0187,
+      "step": 64740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.968905798876831e-05,
+      "loss": 0.0143,
+      "step": 64750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9689009949029838e-05,
+      "loss": 0.0128,
+      "step": 64760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9688961909291367e-05,
+      "loss": 0.0176,
+      "step": 64770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9688913869552896e-05,
+      "loss": 0.0167,
+      "step": 64780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9688865829814422e-05,
+      "loss": 0.0134,
+      "step": 64790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.968881779007595e-05,
+      "loss": 0.0153,
+      "step": 64800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.968876975033748e-05,
+      "loss": 0.0146,
+      "step": 64810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9688721710599007e-05,
+      "loss": 0.0149,
+      "step": 64820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9688673670860536e-05,
+      "loss": 0.0134,
+      "step": 64830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9688625631122065e-05,
+      "loss": 0.0171,
+      "step": 64840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9688577591383595e-05,
+      "loss": 0.0121,
+      "step": 64850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.968852955164512e-05,
+      "loss": 0.0149,
+      "step": 64860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.968848151190665e-05,
+      "loss": 0.0123,
+      "step": 64870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968843347216818e-05,
+      "loss": 0.0185,
+      "step": 64880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9688385432429708e-05,
+      "loss": 0.02,
+      "step": 64890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9688337392691237e-05,
+      "loss": 0.0138,
+      "step": 64900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9688289352952767e-05,
+      "loss": 0.0204,
+      "step": 64910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9688241313214293e-05,
+      "loss": 0.0155,
+      "step": 64920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9688193273475822e-05,
+      "loss": 0.0198,
+      "step": 64930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968814523373735e-05,
+      "loss": 0.0236,
+      "step": 64940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9688097193998877e-05,
+      "loss": 0.0114,
+      "step": 64950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9688049154260406e-05,
+      "loss": 0.0164,
+      "step": 64960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9688001114521936e-05,
+      "loss": 0.0149,
+      "step": 64970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687953074783465e-05,
+      "loss": 0.017,
+      "step": 64980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968790503504499e-05,
+      "loss": 0.0115,
+      "step": 64990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968785699530652e-05,
+      "loss": 0.0201,
+      "step": 65000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968780895556805e-05,
+      "loss": 0.0169,
+      "step": 65010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687760915829575e-05,
+      "loss": 0.015,
+      "step": 65020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687712876091104e-05,
+      "loss": 0.0149,
+      "step": 65030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687664836352634e-05,
+      "loss": 0.012,
+      "step": 65040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687616796614163e-05,
+      "loss": 0.0149,
+      "step": 65050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968756875687569e-05,
+      "loss": 0.017,
+      "step": 65060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687520717137218e-05,
+      "loss": 0.0186,
+      "step": 65070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687472677398747e-05,
+      "loss": 0.0129,
+      "step": 65080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687424637660277e-05,
+      "loss": 0.0165,
+      "step": 65090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687376597921803e-05,
+      "loss": 0.0118,
+      "step": 65100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687328558183332e-05,
+      "loss": 0.0191,
+      "step": 65110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968728051844486e-05,
+      "loss": 0.0122,
+      "step": 65120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687232478706387e-05,
+      "loss": 0.0184,
+      "step": 65130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687184438967916e-05,
+      "loss": 0.0173,
+      "step": 65140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687136399229446e-05,
+      "loss": 0.0183,
+      "step": 65150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687088359490975e-05,
+      "loss": 0.0129,
+      "step": 65160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.96870403197525e-05,
+      "loss": 0.0218,
+      "step": 65170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968699228001403e-05,
+      "loss": 0.0185,
+      "step": 65180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968694424027556e-05,
+      "loss": 0.0151,
+      "step": 65190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686896200537085e-05,
+      "loss": 0.016,
+      "step": 65200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686848160798614e-05,
+      "loss": 0.0153,
+      "step": 65210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686800121060144e-05,
+      "loss": 0.0145,
+      "step": 65220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686752081321673e-05,
+      "loss": 0.0137,
+      "step": 65230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.96867040415832e-05,
+      "loss": 0.0134,
+      "step": 65240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686656001844728e-05,
+      "loss": 0.0173,
+      "step": 65250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686607962106257e-05,
+      "loss": 0.0137,
+      "step": 65260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686559922367787e-05,
+      "loss": 0.0152,
+      "step": 65270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686511882629312e-05,
+      "loss": 0.0144,
+      "step": 65280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686463842890842e-05,
+      "loss": 0.0188,
+      "step": 65290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968641580315237e-05,
+      "loss": 0.0196,
+      "step": 65300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686367763413897e-05,
+      "loss": 0.0175,
+      "step": 65310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686319723675426e-05,
+      "loss": 0.0148,
+      "step": 65320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686271683936955e-05,
+      "loss": 0.0122,
+      "step": 65330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686223644198485e-05,
+      "loss": 0.0094,
+      "step": 65340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968617560446001e-05,
+      "loss": 0.0135,
+      "step": 65350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968612756472154e-05,
+      "loss": 0.0156,
+      "step": 65360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968607952498307e-05,
+      "loss": 0.0132,
+      "step": 65370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9686031485244595e-05,
+      "loss": 0.0134,
+      "step": 65380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685983445506124e-05,
+      "loss": 0.0148,
+      "step": 65390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685935405767654e-05,
+      "loss": 0.0144,
+      "step": 65400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685887366029183e-05,
+      "loss": 0.014,
+      "step": 65410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968583932629071e-05,
+      "loss": 0.0212,
+      "step": 65420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685791286552238e-05,
+      "loss": 0.0124,
+      "step": 65430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685743246813767e-05,
+      "loss": 0.0167,
+      "step": 65440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685695207075296e-05,
+      "loss": 0.0155,
+      "step": 65450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685647167336822e-05,
+      "loss": 0.0146,
+      "step": 65460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968559912759835e-05,
+      "loss": 0.0146,
+      "step": 65470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968555108785988e-05,
+      "loss": 0.0143,
+      "step": 65480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685503048121407e-05,
+      "loss": 0.0162,
+      "step": 65490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685455008382936e-05,
+      "loss": 0.0143,
+      "step": 65500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685406968644465e-05,
+      "loss": 0.0139,
+      "step": 65510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685358928905995e-05,
+      "loss": 0.0141,
+      "step": 65520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968531088916752e-05,
+      "loss": 0.0203,
+      "step": 65530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968526284942905e-05,
+      "loss": 0.015,
+      "step": 65540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.968521480969058e-05,
+      "loss": 0.0121,
+      "step": 65550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685166769952105e-05,
+      "loss": 0.019,
+      "step": 65560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9685118730213634e-05,
+      "loss": 0.0115,
+      "step": 65570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9685070690475163e-05,
+      "loss": 0.0154,
+      "step": 65580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9685022650736693e-05,
+      "loss": 0.0193,
+      "step": 65590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968497461099822e-05,
+      "loss": 0.0135,
+      "step": 65600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684926571259748e-05,
+      "loss": 0.015,
+      "step": 65610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684878531521277e-05,
+      "loss": 0.0191,
+      "step": 65620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684830491782806e-05,
+      "loss": 0.0206,
+      "step": 65630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684782452044332e-05,
+      "loss": 0.014,
+      "step": 65640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968473441230586e-05,
+      "loss": 0.0152,
+      "step": 65650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968468637256739e-05,
+      "loss": 0.0137,
+      "step": 65660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684638332828917e-05,
+      "loss": 0.0125,
+      "step": 65670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684590293090446e-05,
+      "loss": 0.0133,
+      "step": 65680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684542253351975e-05,
+      "loss": 0.0154,
+      "step": 65690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684494213613505e-05,
+      "loss": 0.0137,
+      "step": 65700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968444617387503e-05,
+      "loss": 0.0119,
+      "step": 65710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968439813413656e-05,
+      "loss": 0.0163,
+      "step": 65720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968435009439809e-05,
+      "loss": 0.0121,
+      "step": 65730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684302054659615e-05,
+      "loss": 0.016,
+      "step": 65740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684254014921144e-05,
+      "loss": 0.012,
+      "step": 65750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684205975182673e-05,
+      "loss": 0.0168,
+      "step": 65760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684157935444203e-05,
+      "loss": 0.015,
+      "step": 65770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968410989570573e-05,
+      "loss": 0.015,
+      "step": 65780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684061855967258e-05,
+      "loss": 0.0166,
+      "step": 65790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9684013816228787e-05,
+      "loss": 0.0119,
+      "step": 65800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683965776490316e-05,
+      "loss": 0.0166,
+      "step": 65810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683917736751842e-05,
+      "loss": 0.0145,
+      "step": 65820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968386969701337e-05,
+      "loss": 0.0135,
+      "step": 65830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.96838216572749e-05,
+      "loss": 0.0197,
+      "step": 65840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683773617536427e-05,
+      "loss": 0.0172,
+      "step": 65850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683725577797956e-05,
+      "loss": 0.0151,
+      "step": 65860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683677538059485e-05,
+      "loss": 0.0128,
+      "step": 65870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683629498321014e-05,
+      "loss": 0.0172,
+      "step": 65880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968358145858254e-05,
+      "loss": 0.0141,
+      "step": 65890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968353341884407e-05,
+      "loss": 0.0116,
+      "step": 65900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.96834853791056e-05,
+      "loss": 0.0199,
+      "step": 65910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683437339367125e-05,
+      "loss": 0.0179,
+      "step": 65920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683389299628654e-05,
+      "loss": 0.0122,
+      "step": 65930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683341259890183e-05,
+      "loss": 0.0139,
+      "step": 65940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683293220151713e-05,
+      "loss": 0.0164,
+      "step": 65950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968324518041324e-05,
+      "loss": 0.0189,
+      "step": 65960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683197140674768e-05,
+      "loss": 0.0173,
+      "step": 65970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683149100936297e-05,
+      "loss": 0.0176,
+      "step": 65980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683101061197826e-05,
+      "loss": 0.0228,
+      "step": 65990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9683053021459352e-05,
+      "loss": 0.0215,
+      "step": 66000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968300498172088e-05,
+      "loss": 0.013,
+      "step": 66010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968295694198241e-05,
+      "loss": 0.0181,
+      "step": 66020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682908902243937e-05,
+      "loss": 0.0146,
+      "step": 66030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682860862505466e-05,
+      "loss": 0.0173,
+      "step": 66040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682812822766995e-05,
+      "loss": 0.0137,
+      "step": 66050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682764783028524e-05,
+      "loss": 0.0149,
+      "step": 66060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968271674329005e-05,
+      "loss": 0.017,
+      "step": 66070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968266870355158e-05,
+      "loss": 0.0174,
+      "step": 66080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968262066381311e-05,
+      "loss": 0.0174,
+      "step": 66090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682572624074635e-05,
+      "loss": 0.0202,
+      "step": 66100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682524584336164e-05,
+      "loss": 0.0132,
+      "step": 66110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682481348571542e-05,
+      "loss": 0.0144,
+      "step": 66120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682433308833068e-05,
+      "loss": 0.0131,
+      "step": 66130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682385269094597e-05,
+      "loss": 0.0155,
+      "step": 66140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682337229356126e-05,
+      "loss": 0.0151,
+      "step": 66150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682289189617652e-05,
+      "loss": 0.0176,
+      "step": 66160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968224114987918e-05,
+      "loss": 0.0196,
+      "step": 66170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968219311014071e-05,
+      "loss": 0.0159,
+      "step": 66180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968214507040224e-05,
+      "loss": 0.0162,
+      "step": 66190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682097030663766e-05,
+      "loss": 0.0192,
+      "step": 66200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682048990925295e-05,
+      "loss": 0.012,
+      "step": 66210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9682000951186824e-05,
+      "loss": 0.0134,
+      "step": 66220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9681952911448354e-05,
+      "loss": 0.0163,
+      "step": 66230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968190487170988e-05,
+      "loss": 0.0137,
+      "step": 66240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.968185683197141e-05,
+      "loss": 0.0137,
+      "step": 66250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9681808792232938e-05,
+      "loss": 0.0116,
+      "step": 66260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9681760752494464e-05,
+      "loss": 0.0158,
+      "step": 66270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9681712712755993e-05,
+      "loss": 0.0199,
+      "step": 66280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9681664673017523e-05,
+      "loss": 0.0149,
+      "step": 66290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9681616633279052e-05,
+      "loss": 0.017,
+      "step": 66300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9681568593540578e-05,
+      "loss": 0.0159,
+      "step": 66310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9681520553802107e-05,
+      "loss": 0.0145,
+      "step": 66320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9681472514063636e-05,
+      "loss": 0.0136,
+      "step": 66330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9681424474325162e-05,
+      "loss": 0.0142,
+      "step": 66340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.968137643458669e-05,
+      "loss": 0.0173,
+      "step": 66350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.968132839484822e-05,
+      "loss": 0.0271,
+      "step": 66360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.968128035510975e-05,
+      "loss": 0.018,
+      "step": 66370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9681232315371276e-05,
+      "loss": 0.0166,
+      "step": 66380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9681184275632805e-05,
+      "loss": 0.0166,
+      "step": 66390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9681136235894334e-05,
+      "loss": 0.0149,
+      "step": 66400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.968108819615586e-05,
+      "loss": 0.0161,
+      "step": 66410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.968104015641739e-05,
+      "loss": 0.0193,
+      "step": 66420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.968099211667892e-05,
+      "loss": 0.012,
+      "step": 66430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680944076940448e-05,
+      "loss": 0.0181,
+      "step": 66440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680896037201974e-05,
+      "loss": 0.0236,
+      "step": 66450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680847997463503e-05,
+      "loss": 0.0132,
+      "step": 66460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680799957725033e-05,
+      "loss": 0.0176,
+      "step": 66470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680751917986562e-05,
+      "loss": 0.0144,
+      "step": 66480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680703878248088e-05,
+      "loss": 0.0174,
+      "step": 66490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680655838509617e-05,
+      "loss": 0.0163,
+      "step": 66500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680607798771146e-05,
+      "loss": 0.0107,
+      "step": 66510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680559759032672e-05,
+      "loss": 0.0196,
+      "step": 66520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.96805117192942e-05,
+      "loss": 0.0148,
+      "step": 66530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.968046367955573e-05,
+      "loss": 0.015,
+      "step": 66540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.968041563981726e-05,
+      "loss": 0.0143,
+      "step": 66550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680367600078786e-05,
+      "loss": 0.014,
+      "step": 66560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680319560340315e-05,
+      "loss": 0.0145,
+      "step": 66570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680271520601844e-05,
+      "loss": 0.0181,
+      "step": 66580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.968022348086337e-05,
+      "loss": 0.0133,
+      "step": 66590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.96801754411249e-05,
+      "loss": 0.0178,
+      "step": 66600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.968012740138643e-05,
+      "loss": 0.0122,
+      "step": 66610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680079361647958e-05,
+      "loss": 0.0149,
+      "step": 66620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9680031321909484e-05,
+      "loss": 0.0222,
+      "step": 66630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679983282171013e-05,
+      "loss": 0.0134,
+      "step": 66640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679935242432542e-05,
+      "loss": 0.0194,
+      "step": 66650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679887202694072e-05,
+      "loss": 0.0146,
+      "step": 66660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679839162955598e-05,
+      "loss": 0.0139,
+      "step": 66670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679791123217127e-05,
+      "loss": 0.0161,
+      "step": 66680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679743083478656e-05,
+      "loss": 0.0178,
+      "step": 66690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679695043740182e-05,
+      "loss": 0.0161,
+      "step": 66700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.967964700400171e-05,
+      "loss": 0.0138,
+      "step": 66710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.967959896426324e-05,
+      "loss": 0.0118,
+      "step": 66720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.967955092452477e-05,
+      "loss": 0.0134,
+      "step": 66730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679502884786296e-05,
+      "loss": 0.014,
+      "step": 66740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679454845047825e-05,
+      "loss": 0.015,
+      "step": 66750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679406805309354e-05,
+      "loss": 0.0167,
+      "step": 66760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.967935876557088e-05,
+      "loss": 0.0124,
+      "step": 66770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.967931072583241e-05,
+      "loss": 0.0195,
+      "step": 66780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.967926268609394e-05,
+      "loss": 0.016,
+      "step": 66790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679214646355468e-05,
+      "loss": 0.0118,
+      "step": 66800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679166606616994e-05,
+      "loss": 0.0132,
+      "step": 66810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679118566878523e-05,
+      "loss": 0.0155,
+      "step": 66820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9679070527140052e-05,
+      "loss": 0.0137,
+      "step": 66830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.967902248740158e-05,
+      "loss": 0.0143,
+      "step": 66840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9678974447663107e-05,
+      "loss": 0.0129,
+      "step": 66850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9678926407924637e-05,
+      "loss": 0.0156,
+      "step": 66860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9678878368186166e-05,
+      "loss": 0.016,
+      "step": 66870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9678830328447692e-05,
+      "loss": 0.0121,
+      "step": 66880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.967878228870922e-05,
+      "loss": 0.0147,
+      "step": 66890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.967873424897075e-05,
+      "loss": 0.0149,
+      "step": 66900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.967868620923228e-05,
+      "loss": 0.013,
+      "step": 66910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9678638169493806e-05,
+      "loss": 0.0121,
+      "step": 66920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9678590129755335e-05,
+      "loss": 0.012,
+      "step": 66930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9678542090016864e-05,
+      "loss": 0.0132,
+      "step": 66940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.967849405027839e-05,
+      "loss": 0.0148,
+      "step": 66950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967844601053992e-05,
+      "loss": 0.0186,
+      "step": 66960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967839797080145e-05,
+      "loss": 0.017,
+      "step": 66970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9678349931062978e-05,
+      "loss": 0.0132,
+      "step": 66980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9678301891324504e-05,
+      "loss": 0.0138,
+      "step": 66990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9678253851586033e-05,
+      "loss": 0.0186,
+      "step": 67000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9678205811847562e-05,
+      "loss": 0.0123,
+      "step": 67010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967815777210909e-05,
+      "loss": 0.0144,
+      "step": 67020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9678109732370617e-05,
+      "loss": 0.0159,
+      "step": 67030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9678061692632147e-05,
+      "loss": 0.0199,
+      "step": 67040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9678013652893676e-05,
+      "loss": 0.0125,
+      "step": 67050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9677965613155202e-05,
+      "loss": 0.0151,
+      "step": 67060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967791757341673e-05,
+      "loss": 0.0138,
+      "step": 67070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967786953367826e-05,
+      "loss": 0.0132,
+      "step": 67080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967782149393979e-05,
+      "loss": 0.0165,
+      "step": 67090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9677773454201316e-05,
+      "loss": 0.0141,
+      "step": 67100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9677725414462845e-05,
+      "loss": 0.0142,
+      "step": 67110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9677677374724374e-05,
+      "loss": 0.0142,
+      "step": 67120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.96776293349859e-05,
+      "loss": 0.0171,
+      "step": 67130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967758129524743e-05,
+      "loss": 0.0167,
+      "step": 67140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967753325550896e-05,
+      "loss": 0.0171,
+      "step": 67150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9677485215770488e-05,
+      "loss": 0.016,
+      "step": 67160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9677437176032014e-05,
+      "loss": 0.0145,
+      "step": 67170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9677389136293543e-05,
+      "loss": 0.0116,
+      "step": 67180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9677341096555072e-05,
+      "loss": 0.0183,
+      "step": 67190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.96772930568166e-05,
+      "loss": 0.0121,
+      "step": 67200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9677245017078127e-05,
+      "loss": 0.0173,
+      "step": 67210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9677196977339657e-05,
+      "loss": 0.0161,
+      "step": 67220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9677148937601186e-05,
+      "loss": 0.0104,
+      "step": 67230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9677100897862712e-05,
+      "loss": 0.0136,
+      "step": 67240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967705285812424e-05,
+      "loss": 0.0117,
+      "step": 67250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967700481838577e-05,
+      "loss": 0.0121,
+      "step": 67260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.96769567786473e-05,
+      "loss": 0.0155,
+      "step": 67270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9676908738908825e-05,
+      "loss": 0.0191,
+      "step": 67280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9676860699170355e-05,
+      "loss": 0.0154,
+      "step": 67290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9676812659431884e-05,
+      "loss": 0.0153,
+      "step": 67300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967676461969341e-05,
+      "loss": 0.0127,
+      "step": 67310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967671657995494e-05,
+      "loss": 0.0142,
+      "step": 67320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967666854021647e-05,
+      "loss": 0.0151,
+      "step": 67330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9676620500477998e-05,
+      "loss": 0.0139,
+      "step": 67340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9676572460739524e-05,
+      "loss": 0.0129,
+      "step": 67350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9676524421001053e-05,
+      "loss": 0.0158,
+      "step": 67360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9676476381262582e-05,
+      "loss": 0.0189,
+      "step": 67370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967642834152411e-05,
+      "loss": 0.0154,
+      "step": 67380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9676380301785637e-05,
+      "loss": 0.0143,
+      "step": 67390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9676332262047167e-05,
+      "loss": 0.0197,
+      "step": 67400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9676284222308696e-05,
+      "loss": 0.0124,
+      "step": 67410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967623618257022e-05,
+      "loss": 0.0175,
+      "step": 67420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967618814283175e-05,
+      "loss": 0.0148,
+      "step": 67430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967614010309328e-05,
+      "loss": 0.0147,
+      "step": 67440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967609206335481e-05,
+      "loss": 0.0152,
+      "step": 67450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9676044023616335e-05,
+      "loss": 0.0139,
+      "step": 67460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9675995983877865e-05,
+      "loss": 0.0155,
+      "step": 67470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9675947944139394e-05,
+      "loss": 0.0191,
+      "step": 67480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967589990440092e-05,
+      "loss": 0.0165,
+      "step": 67490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967585186466245e-05,
+      "loss": 0.016,
+      "step": 67500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967580382492398e-05,
+      "loss": 0.0127,
+      "step": 67510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9675755785185508e-05,
+      "loss": 0.0238,
+      "step": 67520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9675707745447033e-05,
+      "loss": 0.0126,
+      "step": 67530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9675659705708563e-05,
+      "loss": 0.0141,
+      "step": 67540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9675611665970092e-05,
+      "loss": 0.0173,
+      "step": 67550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967556362623162e-05,
+      "loss": 0.0179,
+      "step": 67560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9675515586493147e-05,
+      "loss": 0.0148,
+      "step": 67570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9675467546754676e-05,
+      "loss": 0.0191,
+      "step": 67580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9675419507016206e-05,
+      "loss": 0.0131,
+      "step": 67590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967537146727773e-05,
+      "loss": 0.0133,
+      "step": 67600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967532342753926e-05,
+      "loss": 0.0151,
+      "step": 67610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967527538780079e-05,
+      "loss": 0.0184,
+      "step": 67620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967522734806232e-05,
+      "loss": 0.0155,
+      "step": 67630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9675179308323845e-05,
+      "loss": 0.0125,
+      "step": 67640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9675131268585375e-05,
+      "loss": 0.0162,
+      "step": 67650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9675083228846904e-05,
+      "loss": 0.0147,
+      "step": 67660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967503518910843e-05,
+      "loss": 0.0224,
+      "step": 67670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967498714936996e-05,
+      "loss": 0.0195,
+      "step": 67680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674939109631488e-05,
+      "loss": 0.0142,
+      "step": 67690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674891069893017e-05,
+      "loss": 0.0122,
+      "step": 67700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674843030154543e-05,
+      "loss": 0.014,
+      "step": 67710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674794990416073e-05,
+      "loss": 0.0178,
+      "step": 67720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674746950677602e-05,
+      "loss": 0.0127,
+      "step": 67730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967469891093913e-05,
+      "loss": 0.0167,
+      "step": 67740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674650871200657e-05,
+      "loss": 0.0142,
+      "step": 67750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674602831462186e-05,
+      "loss": 0.0144,
+      "step": 67760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674554791723716e-05,
+      "loss": 0.0125,
+      "step": 67770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967450675198524e-05,
+      "loss": 0.0116,
+      "step": 67780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967445871224677e-05,
+      "loss": 0.0152,
+      "step": 67790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.96744106725083e-05,
+      "loss": 0.017,
+      "step": 67800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967436263276983e-05,
+      "loss": 0.0164,
+      "step": 67810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674314593031355e-05,
+      "loss": 0.0103,
+      "step": 67820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674266553292884e-05,
+      "loss": 0.0112,
+      "step": 67830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674218513554414e-05,
+      "loss": 0.0159,
+      "step": 67840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967417047381594e-05,
+      "loss": 0.0168,
+      "step": 67850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967412243407747e-05,
+      "loss": 0.0185,
+      "step": 67860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674074394338998e-05,
+      "loss": 0.0143,
+      "step": 67870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9674026354600527e-05,
+      "loss": 0.0131,
+      "step": 67880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673978314862053e-05,
+      "loss": 0.0118,
+      "step": 67890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673930275123583e-05,
+      "loss": 0.0165,
+      "step": 67900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673882235385112e-05,
+      "loss": 0.0167,
+      "step": 67910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967383419564664e-05,
+      "loss": 0.0141,
+      "step": 67920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673786155908167e-05,
+      "loss": 0.013,
+      "step": 67930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673738116169696e-05,
+      "loss": 0.0167,
+      "step": 67940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673690076431226e-05,
+      "loss": 0.014,
+      "step": 67950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673642036692755e-05,
+      "loss": 0.0123,
+      "step": 67960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673593996954284e-05,
+      "loss": 0.0158,
+      "step": 67970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967354595721581e-05,
+      "loss": 0.0098,
+      "step": 67980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967349791747734e-05,
+      "loss": 0.0164,
+      "step": 67990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967344987773887e-05,
+      "loss": 0.0124,
+      "step": 68000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673401838000398e-05,
+      "loss": 0.0126,
+      "step": 68010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673353798261924e-05,
+      "loss": 0.0162,
+      "step": 68020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673305758523453e-05,
+      "loss": 0.0159,
+      "step": 68030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673257718784982e-05,
+      "loss": 0.0164,
+      "step": 68040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967320967904651e-05,
+      "loss": 0.0169,
+      "step": 68050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673161639308037e-05,
+      "loss": 0.0148,
+      "step": 68060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673113599569567e-05,
+      "loss": 0.0133,
+      "step": 68070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673065559831096e-05,
+      "loss": 0.017,
+      "step": 68080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9673017520092622e-05,
+      "loss": 0.0159,
+      "step": 68090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967296948035415e-05,
+      "loss": 0.0214,
+      "step": 68100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967292144061568e-05,
+      "loss": 0.0193,
+      "step": 68110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967287340087721e-05,
+      "loss": 0.0159,
+      "step": 68120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9672825361138735e-05,
+      "loss": 0.0112,
+      "step": 68130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9672777321400265e-05,
+      "loss": 0.0169,
+      "step": 68140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9672729281661794e-05,
+      "loss": 0.0154,
+      "step": 68150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967268124192332e-05,
+      "loss": 0.0156,
+      "step": 68160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967263320218485e-05,
+      "loss": 0.019,
+      "step": 68170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967258516244638e-05,
+      "loss": 0.0149,
+      "step": 68180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9672537122707908e-05,
+      "loss": 0.0113,
+      "step": 68190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9672489082969434e-05,
+      "loss": 0.0161,
+      "step": 68200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9672441043230963e-05,
+      "loss": 0.0165,
+      "step": 68210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9672393003492492e-05,
+      "loss": 0.0203,
+      "step": 68220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967234496375402e-05,
+      "loss": 0.0145,
+      "step": 68230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9672296924015547e-05,
+      "loss": 0.0143,
+      "step": 68240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9672248884277077e-05,
+      "loss": 0.0151,
+      "step": 68250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9672200844538606e-05,
+      "loss": 0.0134,
+      "step": 68260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967215280480013e-05,
+      "loss": 0.0134,
+      "step": 68270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967210476506166e-05,
+      "loss": 0.0169,
+      "step": 68280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967205672532319e-05,
+      "loss": 0.0204,
+      "step": 68290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967200868558472e-05,
+      "loss": 0.0123,
+      "step": 68300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9671960645846245e-05,
+      "loss": 0.0147,
+      "step": 68310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9671912606107775e-05,
+      "loss": 0.0168,
+      "step": 68320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9671864566369304e-05,
+      "loss": 0.0173,
+      "step": 68330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.967181652663083e-05,
+      "loss": 0.0169,
+      "step": 68340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967176848689236e-05,
+      "loss": 0.0121,
+      "step": 68350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967172044715389e-05,
+      "loss": 0.0246,
+      "step": 68360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9671672407415418e-05,
+      "loss": 0.0135,
+      "step": 68370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9671624367676943e-05,
+      "loss": 0.0128,
+      "step": 68380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9671576327938473e-05,
+      "loss": 0.0175,
+      "step": 68390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9671528288200002e-05,
+      "loss": 0.0136,
+      "step": 68400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967148024846153e-05,
+      "loss": 0.016,
+      "step": 68410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9671432208723057e-05,
+      "loss": 0.0132,
+      "step": 68420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9671384168984586e-05,
+      "loss": 0.0215,
+      "step": 68430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9671336129246116e-05,
+      "loss": 0.0208,
+      "step": 68440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967128808950764e-05,
+      "loss": 0.0163,
+      "step": 68450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967124004976917e-05,
+      "loss": 0.0198,
+      "step": 68460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.96711920100307e-05,
+      "loss": 0.0146,
+      "step": 68470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967114397029223e-05,
+      "loss": 0.0136,
+      "step": 68480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9671095930553755e-05,
+      "loss": 0.0119,
+      "step": 68490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9671047890815285e-05,
+      "loss": 0.0163,
+      "step": 68500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670999851076814e-05,
+      "loss": 0.0119,
+      "step": 68510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967095181133834e-05,
+      "loss": 0.0114,
+      "step": 68520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967090377159987e-05,
+      "loss": 0.0144,
+      "step": 68530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670855731861398e-05,
+      "loss": 0.0135,
+      "step": 68540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670807692122927e-05,
+      "loss": 0.0215,
+      "step": 68550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670759652384453e-05,
+      "loss": 0.0207,
+      "step": 68560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670711612645983e-05,
+      "loss": 0.0203,
+      "step": 68570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670663572907512e-05,
+      "loss": 0.0127,
+      "step": 68580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967061553316904e-05,
+      "loss": 0.0135,
+      "step": 68590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670567493430567e-05,
+      "loss": 0.0169,
+      "step": 68600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670519453692096e-05,
+      "loss": 0.0161,
+      "step": 68610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670471413953626e-05,
+      "loss": 0.0243,
+      "step": 68620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967042337421515e-05,
+      "loss": 0.0175,
+      "step": 68630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967037533447668e-05,
+      "loss": 0.0115,
+      "step": 68640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967032729473821e-05,
+      "loss": 0.0148,
+      "step": 68650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.967027925499974e-05,
+      "loss": 0.0174,
+      "step": 68660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670231215261265e-05,
+      "loss": 0.0135,
+      "step": 68670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670183175522794e-05,
+      "loss": 0.0115,
+      "step": 68680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670135135784324e-05,
+      "loss": 0.0136,
+      "step": 68690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.96700919000197e-05,
+      "loss": 0.0131,
+      "step": 68700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9670043860281228e-05,
+      "loss": 0.0128,
+      "step": 68710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669995820542757e-05,
+      "loss": 0.0166,
+      "step": 68720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669947780804283e-05,
+      "loss": 0.0207,
+      "step": 68730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669899741065812e-05,
+      "loss": 0.0122,
+      "step": 68740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.966985170132734e-05,
+      "loss": 0.0172,
+      "step": 68750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669803661588867e-05,
+      "loss": 0.0138,
+      "step": 68760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669755621850396e-05,
+      "loss": 0.0107,
+      "step": 68770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669707582111926e-05,
+      "loss": 0.0184,
+      "step": 68780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669659542373455e-05,
+      "loss": 0.0123,
+      "step": 68790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.966961150263498e-05,
+      "loss": 0.0159,
+      "step": 68800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.966956346289651e-05,
+      "loss": 0.0129,
+      "step": 68810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.966951542315804e-05,
+      "loss": 0.0133,
+      "step": 68820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.966946738341957e-05,
+      "loss": 0.014,
+      "step": 68830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669419343681095e-05,
+      "loss": 0.019,
+      "step": 68840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669371303942624e-05,
+      "loss": 0.0176,
+      "step": 68850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669323264204153e-05,
+      "loss": 0.0131,
+      "step": 68860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.966927522446568e-05,
+      "loss": 0.0135,
+      "step": 68870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669227184727208e-05,
+      "loss": 0.0154,
+      "step": 68880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669179144988738e-05,
+      "loss": 0.0118,
+      "step": 68890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669131105250267e-05,
+      "loss": 0.0103,
+      "step": 68900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669083065511793e-05,
+      "loss": 0.0151,
+      "step": 68910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9669035025773322e-05,
+      "loss": 0.0137,
+      "step": 68920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.966898698603485e-05,
+      "loss": 0.0147,
+      "step": 68930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9668938946296377e-05,
+      "loss": 0.015,
+      "step": 68940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9668890906557906e-05,
+      "loss": 0.0182,
+      "step": 68950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9668842866819436e-05,
+      "loss": 0.0149,
+      "step": 68960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9668794827080965e-05,
+      "loss": 0.0131,
+      "step": 68970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.966874678734249e-05,
+      "loss": 0.0137,
+      "step": 68980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.966869874760402e-05,
+      "loss": 0.0137,
+      "step": 68990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.966865070786555e-05,
+      "loss": 0.0101,
+      "step": 69000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.966860266812708e-05,
+      "loss": 0.016,
+      "step": 69010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9668554628388605e-05,
+      "loss": 0.0133,
+      "step": 69020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9668506588650134e-05,
+      "loss": 0.0178,
+      "step": 69030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9668458548911663e-05,
+      "loss": 0.0131,
+      "step": 69040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966841050917319e-05,
+      "loss": 0.0142,
+      "step": 69050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9668362469434718e-05,
+      "loss": 0.0147,
+      "step": 69060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9668314429696247e-05,
+      "loss": 0.0164,
+      "step": 69070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9668266389957777e-05,
+      "loss": 0.0123,
+      "step": 69080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9668218350219303e-05,
+      "loss": 0.0152,
+      "step": 69090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9668170310480832e-05,
+      "loss": 0.0183,
+      "step": 69100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966812227074236e-05,
+      "loss": 0.0405,
+      "step": 69110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9668074231003887e-05,
+      "loss": 0.0136,
+      "step": 69120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9668026191265416e-05,
+      "loss": 0.0182,
+      "step": 69130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667978151526946e-05,
+      "loss": 0.0161,
+      "step": 69140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667930111788475e-05,
+      "loss": 0.0142,
+      "step": 69150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966788207205e-05,
+      "loss": 0.0145,
+      "step": 69160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966783403231153e-05,
+      "loss": 0.012,
+      "step": 69170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966778599257306e-05,
+      "loss": 0.0207,
+      "step": 69180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966773795283459e-05,
+      "loss": 0.0206,
+      "step": 69190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667689913096114e-05,
+      "loss": 0.016,
+      "step": 69200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667641873357644e-05,
+      "loss": 0.0153,
+      "step": 69210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667593833619173e-05,
+      "loss": 0.0157,
+      "step": 69220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.96675457938807e-05,
+      "loss": 0.0207,
+      "step": 69230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667497754142228e-05,
+      "loss": 0.0171,
+      "step": 69240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667449714403757e-05,
+      "loss": 0.011,
+      "step": 69250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667401674665287e-05,
+      "loss": 0.0116,
+      "step": 69260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667353634926813e-05,
+      "loss": 0.0186,
+      "step": 69270
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667305595188342e-05,
+      "loss": 0.0174,
+      "step": 69280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966725755544987e-05,
+      "loss": 0.0138,
+      "step": 69290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667209515711397e-05,
+      "loss": 0.0106,
+      "step": 69300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667161475972926e-05,
+      "loss": 0.0142,
+      "step": 69310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667113436234455e-05,
+      "loss": 0.0143,
+      "step": 69320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9667065396495985e-05,
+      "loss": 0.0184,
+      "step": 69330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966701735675751e-05,
+      "loss": 0.0146,
+      "step": 69340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966696931701904e-05,
+      "loss": 0.0187,
+      "step": 69350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966692127728057e-05,
+      "loss": 0.0133,
+      "step": 69360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666873237542095e-05,
+      "loss": 0.0186,
+      "step": 69370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666825197803624e-05,
+      "loss": 0.0138,
+      "step": 69380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666777158065154e-05,
+      "loss": 0.0134,
+      "step": 69390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666729118326683e-05,
+      "loss": 0.009,
+      "step": 69400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966668107858821e-05,
+      "loss": 0.0172,
+      "step": 69410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666633038849738e-05,
+      "loss": 0.0123,
+      "step": 69420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666584999111267e-05,
+      "loss": 0.0111,
+      "step": 69430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666536959372797e-05,
+      "loss": 0.009,
+      "step": 69440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666488919634322e-05,
+      "loss": 0.0124,
+      "step": 69450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666440879895852e-05,
+      "loss": 0.0171,
+      "step": 69460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966639284015738e-05,
+      "loss": 0.0131,
+      "step": 69470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666344800418907e-05,
+      "loss": 0.0123,
+      "step": 69480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666296760680436e-05,
+      "loss": 0.0189,
+      "step": 69490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666248720941965e-05,
+      "loss": 0.0158,
+      "step": 69500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666200681203495e-05,
+      "loss": 0.0109,
+      "step": 69510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966615264146502e-05,
+      "loss": 0.0082,
+      "step": 69520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966610460172655e-05,
+      "loss": 0.0164,
+      "step": 69530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966605656198808e-05,
+      "loss": 0.0139,
+      "step": 69540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9666008522249605e-05,
+      "loss": 0.0157,
+      "step": 69550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9665960482511134e-05,
+      "loss": 0.0146,
+      "step": 69560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9665912442772664e-05,
+      "loss": 0.0176,
+      "step": 69570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9665864403034193e-05,
+      "loss": 0.0128,
+      "step": 69580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966581636329572e-05,
+      "loss": 0.0111,
+      "step": 69590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9665768323557248e-05,
+      "loss": 0.0172,
+      "step": 69600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9665720283818777e-05,
+      "loss": 0.0111,
+      "step": 69610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9665672244080306e-05,
+      "loss": 0.0107,
+      "step": 69620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9665624204341832e-05,
+      "loss": 0.0121,
+      "step": 69630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966557616460336e-05,
+      "loss": 0.0152,
+      "step": 69640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966552812486489e-05,
+      "loss": 0.0103,
+      "step": 69650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9665480085126417e-05,
+      "loss": 0.0128,
+      "step": 69660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9665432045387946e-05,
+      "loss": 0.0117,
+      "step": 69670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9665384005649475e-05,
+      "loss": 0.0131,
+      "step": 69680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9665335965911005e-05,
+      "loss": 0.0139,
+      "step": 69690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966528792617253e-05,
+      "loss": 0.009,
+      "step": 69700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966523988643406e-05,
+      "loss": 0.0154,
+      "step": 69710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966519184669559e-05,
+      "loss": 0.0135,
+      "step": 69720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9665143806957115e-05,
+      "loss": 0.0119,
+      "step": 69730
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9665095767218644e-05,
+      "loss": 0.0107,
+      "step": 69740
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9665047727480173e-05,
+      "loss": 0.0113,
+      "step": 69750
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664999687741703e-05,
+      "loss": 0.0118,
+      "step": 69760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966495164800323e-05,
+      "loss": 0.012,
+      "step": 69770
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664903608264758e-05,
+      "loss": 0.0155,
+      "step": 69780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664855568526287e-05,
+      "loss": 0.0131,
+      "step": 69790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664807528787816e-05,
+      "loss": 0.0117,
+      "step": 69800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664759489049342e-05,
+      "loss": 0.0102,
+      "step": 69810
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966471144931087e-05,
+      "loss": 0.0117,
+      "step": 69820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.96646634095724e-05,
+      "loss": 0.0126,
+      "step": 69830
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664615369833927e-05,
+      "loss": 0.0158,
+      "step": 69840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664567330095456e-05,
+      "loss": 0.0104,
+      "step": 69850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664519290356985e-05,
+      "loss": 0.0179,
+      "step": 69860
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664471250618515e-05,
+      "loss": 0.0153,
+      "step": 69870
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966442321088004e-05,
+      "loss": 0.0215,
+      "step": 69880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966437517114157e-05,
+      "loss": 0.0121,
+      "step": 69890
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.96643271314031e-05,
+      "loss": 0.0123,
+      "step": 69900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664279091664625e-05,
+      "loss": 0.0141,
+      "step": 69910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664231051926154e-05,
+      "loss": 0.0129,
+      "step": 69920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664183012187683e-05,
+      "loss": 0.0139,
+      "step": 69930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664134972449213e-05,
+      "loss": 0.013,
+      "step": 69940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966408693271074e-05,
+      "loss": 0.0123,
+      "step": 69950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9664038892972268e-05,
+      "loss": 0.0191,
+      "step": 69960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663990853233797e-05,
+      "loss": 0.0119,
+      "step": 69970
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663942813495326e-05,
+      "loss": 0.0115,
+      "step": 69980
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663894773756852e-05,
+      "loss": 0.0092,
+      "step": 69990
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966384673401838e-05,
+      "loss": 0.0204,
+      "step": 70000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966379869427991e-05,
+      "loss": 0.0124,
+      "step": 70010
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663750654541437e-05,
+      "loss": 0.0161,
+      "step": 70020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663702614802966e-05,
+      "loss": 0.0087,
+      "step": 70030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663654575064495e-05,
+      "loss": 0.0148,
+      "step": 70040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663606535326024e-05,
+      "loss": 0.0199,
+      "step": 70050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966355849558755e-05,
+      "loss": 0.0108,
+      "step": 70060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966351045584908e-05,
+      "loss": 0.0174,
+      "step": 70070
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966346241611061e-05,
+      "loss": 0.0137,
+      "step": 70080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663414376372135e-05,
+      "loss": 0.0118,
+      "step": 70090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663366336633664e-05,
+      "loss": 0.0096,
+      "step": 70100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663318296895193e-05,
+      "loss": 0.0103,
+      "step": 70110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663270257156723e-05,
+      "loss": 0.014,
+      "step": 70120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966322221741825e-05,
+      "loss": 0.0155,
+      "step": 70130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663174177679778e-05,
+      "loss": 0.0124,
+      "step": 70140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663126137941307e-05,
+      "loss": 0.011,
+      "step": 70150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663078098202836e-05,
+      "loss": 0.0116,
+      "step": 70160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9663030058464362e-05,
+      "loss": 0.0155,
+      "step": 70170
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966298201872589e-05,
+      "loss": 0.0143,
+      "step": 70180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966293397898742e-05,
+      "loss": 0.0093,
+      "step": 70190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662885939248947e-05,
+      "loss": 0.0114,
+      "step": 70200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662837899510476e-05,
+      "loss": 0.0132,
+      "step": 70210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662789859772005e-05,
+      "loss": 0.0174,
+      "step": 70220
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662741820033534e-05,
+      "loss": 0.0112,
+      "step": 70230
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966269378029506e-05,
+      "loss": 0.0128,
+      "step": 70240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966264574055659e-05,
+      "loss": 0.0155,
+      "step": 70250
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966259770081812e-05,
+      "loss": 0.0117,
+      "step": 70260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662549661079645e-05,
+      "loss": 0.0198,
+      "step": 70270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662501621341174e-05,
+      "loss": 0.0178,
+      "step": 70280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662453581602703e-05,
+      "loss": 0.0108,
+      "step": 70290
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662405541864232e-05,
+      "loss": 0.0144,
+      "step": 70300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966235750212576e-05,
+      "loss": 0.0088,
+      "step": 70310
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662309462387288e-05,
+      "loss": 0.0125,
+      "step": 70320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662261422648817e-05,
+      "loss": 0.016,
+      "step": 70330
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662213382910346e-05,
+      "loss": 0.0137,
+      "step": 70340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662165343171872e-05,
+      "loss": 0.0125,
+      "step": 70350
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.96621173034334e-05,
+      "loss": 0.0152,
+      "step": 70360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966206926369493e-05,
+      "loss": 0.0124,
+      "step": 70370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9662021223956456e-05,
+      "loss": 0.0127,
+      "step": 70380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9661973184217986e-05,
+      "loss": 0.0132,
+      "step": 70390
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9661925144479515e-05,
+      "loss": 0.0176,
+      "step": 70400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9661877104741044e-05,
+      "loss": 0.0162,
+      "step": 70410
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966182906500257e-05,
+      "loss": 0.0139,
+      "step": 70420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.96617810252641e-05,
+      "loss": 0.0173,
+      "step": 70430
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.966173298552563e-05,
+      "loss": 0.0139,
+      "step": 70440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661684945787155e-05,
+      "loss": 0.0123,
+      "step": 70450
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661636906048684e-05,
+      "loss": 0.0188,
+      "step": 70460
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661588866310213e-05,
+      "loss": 0.0197,
+      "step": 70470
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661540826571742e-05,
+      "loss": 0.0177,
+      "step": 70480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661492786833268e-05,
+      "loss": 0.0115,
+      "step": 70490
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661444747094798e-05,
+      "loss": 0.0201,
+      "step": 70500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661396707356327e-05,
+      "loss": 0.0138,
+      "step": 70510
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661348667617856e-05,
+      "loss": 0.0134,
+      "step": 70520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661300627879382e-05,
+      "loss": 0.0142,
+      "step": 70530
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.966125258814091e-05,
+      "loss": 0.0139,
+      "step": 70540
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.966120454840244e-05,
+      "loss": 0.0125,
+      "step": 70550
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661156508663966e-05,
+      "loss": 0.0165,
+      "step": 70560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661108468925496e-05,
+      "loss": 0.0152,
+      "step": 70570
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661060429187025e-05,
+      "loss": 0.0152,
+      "step": 70580
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9661012389448554e-05,
+      "loss": 0.0122,
+      "step": 70590
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.966096434971008e-05,
+      "loss": 0.0083,
+      "step": 70600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.966091630997161e-05,
+      "loss": 0.0107,
+      "step": 70610
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.966086827023314e-05,
+      "loss": 0.0127,
+      "step": 70620
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660820230494664e-05,
+      "loss": 0.0137,
+      "step": 70630
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660772190756194e-05,
+      "loss": 0.0102,
+      "step": 70640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660724151017723e-05,
+      "loss": 0.0119,
+      "step": 70650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660676111279252e-05,
+      "loss": 0.0135,
+      "step": 70660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660628071540778e-05,
+      "loss": 0.0114,
+      "step": 70670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660580031802307e-05,
+      "loss": 0.0136,
+      "step": 70680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660531992063837e-05,
+      "loss": 0.0115,
+      "step": 70690
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660483952325366e-05,
+      "loss": 0.0145,
+      "step": 70700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660435912586892e-05,
+      "loss": 0.0124,
+      "step": 70710
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.966038787284842e-05,
+      "loss": 0.0126,
+      "step": 70720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.966033983310995e-05,
+      "loss": 0.0159,
+      "step": 70730
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660291793371476e-05,
+      "loss": 0.0142,
+      "step": 70740
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660243753633006e-05,
+      "loss": 0.0123,
+      "step": 70750
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660195713894535e-05,
+      "loss": 0.016,
+      "step": 70760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9660147674156064e-05,
+      "loss": 0.0159,
+      "step": 70770
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.966009963441759e-05,
+      "loss": 0.0129,
+      "step": 70780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.966005159467912e-05,
+      "loss": 0.0121,
+      "step": 70790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.966000355494065e-05,
+      "loss": 0.014,
+      "step": 70800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659955515202174e-05,
+      "loss": 0.0104,
+      "step": 70810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659907475463704e-05,
+      "loss": 0.0087,
+      "step": 70820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659859435725233e-05,
+      "loss": 0.0143,
+      "step": 70830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659811395986762e-05,
+      "loss": 0.0098,
+      "step": 70840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659763356248288e-05,
+      "loss": 0.0156,
+      "step": 70850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659715316509817e-05,
+      "loss": 0.015,
+      "step": 70860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659667276771347e-05,
+      "loss": 0.0088,
+      "step": 70870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659619237032876e-05,
+      "loss": 0.0104,
+      "step": 70880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659571197294402e-05,
+      "loss": 0.0106,
+      "step": 70890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.965952315755593e-05,
+      "loss": 0.0122,
+      "step": 70900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.965947511781746e-05,
+      "loss": 0.0101,
+      "step": 70910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659427078078986e-05,
+      "loss": 0.0104,
+      "step": 70920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659379038340515e-05,
+      "loss": 0.0132,
+      "step": 70930
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659330998602045e-05,
+      "loss": 0.0117,
+      "step": 70940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659282958863574e-05,
+      "loss": 0.0118,
+      "step": 70950
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.96592349191251e-05,
+      "loss": 0.0179,
+      "step": 70960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.965918687938663e-05,
+      "loss": 0.0081,
+      "step": 70970
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.965913883964816e-05,
+      "loss": 0.0163,
+      "step": 70980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659090799909684e-05,
+      "loss": 0.0134,
+      "step": 70990
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9659042760171214e-05,
+      "loss": 0.0131,
+      "step": 71000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9658994720432743e-05,
+      "loss": 0.0097,
+      "step": 71010
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9658946680694272e-05,
+      "loss": 0.0192,
+      "step": 71020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9658898640955798e-05,
+      "loss": 0.0157,
+      "step": 71030
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.965885060121733e-05,
+      "loss": 0.0125,
+      "step": 71040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9658802561478857e-05,
+      "loss": 0.0135,
+      "step": 71050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9658754521740386e-05,
+      "loss": 0.0127,
+      "step": 71060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9658706482001915e-05,
+      "loss": 0.0134,
+      "step": 71070
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9658658442263444e-05,
+      "loss": 0.016,
+      "step": 71080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.965861040252497e-05,
+      "loss": 0.0117,
+      "step": 71090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.96585623627865e-05,
+      "loss": 0.0156,
+      "step": 71100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.965851432304803e-05,
+      "loss": 0.0096,
+      "step": 71110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9658466283309555e-05,
+      "loss": 0.0135,
+      "step": 71120
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9658418243571084e-05,
+      "loss": 0.0118,
+      "step": 71130
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9658370203832613e-05,
+      "loss": 0.0119,
+      "step": 71140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9658322164094142e-05,
+      "loss": 0.0148,
+      "step": 71150
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965827412435567e-05,
+      "loss": 0.0151,
+      "step": 71160
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9658226084617198e-05,
+      "loss": 0.0121,
+      "step": 71170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9658178044878727e-05,
+      "loss": 0.0097,
+      "step": 71180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9658130005140256e-05,
+      "loss": 0.0138,
+      "step": 71190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9658081965401782e-05,
+      "loss": 0.0128,
+      "step": 71200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965803392566331e-05,
+      "loss": 0.0115,
+      "step": 71210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965798588592484e-05,
+      "loss": 0.0108,
+      "step": 71220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657937846186366e-05,
+      "loss": 0.0088,
+      "step": 71230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657889806447896e-05,
+      "loss": 0.0141,
+      "step": 71240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657841766709425e-05,
+      "loss": 0.0105,
+      "step": 71250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657793726970954e-05,
+      "loss": 0.0099,
+      "step": 71260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965774568723248e-05,
+      "loss": 0.0137,
+      "step": 71270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657702451467855e-05,
+      "loss": 0.0085,
+      "step": 71280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657654411729384e-05,
+      "loss": 0.0105,
+      "step": 71290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965760637199091e-05,
+      "loss": 0.0145,
+      "step": 71300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965755833225244e-05,
+      "loss": 0.0131,
+      "step": 71310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965751029251397e-05,
+      "loss": 0.0124,
+      "step": 71320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657462252775498e-05,
+      "loss": 0.0114,
+      "step": 71330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657414213037024e-05,
+      "loss": 0.0144,
+      "step": 71340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657366173298553e-05,
+      "loss": 0.0097,
+      "step": 71350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657318133560082e-05,
+      "loss": 0.0178,
+      "step": 71360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965727009382161e-05,
+      "loss": 0.0165,
+      "step": 71370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657222054083137e-05,
+      "loss": 0.0168,
+      "step": 71380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657174014344667e-05,
+      "loss": 0.0124,
+      "step": 71390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657125974606196e-05,
+      "loss": 0.0133,
+      "step": 71400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9657077934867722e-05,
+      "loss": 0.017,
+      "step": 71410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965702989512925e-05,
+      "loss": 0.016,
+      "step": 71420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965698185539078e-05,
+      "loss": 0.0126,
+      "step": 71430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965693381565231e-05,
+      "loss": 0.0137,
+      "step": 71440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9656885775913835e-05,
+      "loss": 0.013,
+      "step": 71450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9656837736175365e-05,
+      "loss": 0.0163,
+      "step": 71460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9656789696436894e-05,
+      "loss": 0.0118,
+      "step": 71470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965674165669842e-05,
+      "loss": 0.0131,
+      "step": 71480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965669361695995e-05,
+      "loss": 0.014,
+      "step": 71490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965664557722148e-05,
+      "loss": 0.0116,
+      "step": 71500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9656597537483008e-05,
+      "loss": 0.0117,
+      "step": 71510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9656549497744534e-05,
+      "loss": 0.0118,
+      "step": 71520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9656501458006063e-05,
+      "loss": 0.0201,
+      "step": 71530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9656453418267592e-05,
+      "loss": 0.016,
+      "step": 71540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965640537852912e-05,
+      "loss": 0.0114,
+      "step": 71550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965635733879065e-05,
+      "loss": 0.0135,
+      "step": 71560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965630929905218e-05,
+      "loss": 0.0097,
+      "step": 71570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9656261259313706e-05,
+      "loss": 0.0117,
+      "step": 71580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9656213219575235e-05,
+      "loss": 0.0119,
+      "step": 71590
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9656165179836764e-05,
+      "loss": 0.0134,
+      "step": 71600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9656117140098294e-05,
+      "loss": 0.0132,
+      "step": 71610
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965606910035982e-05,
+      "loss": 0.0117,
+      "step": 71620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965602106062135e-05,
+      "loss": 0.0146,
+      "step": 71630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655973020882878e-05,
+      "loss": 0.0134,
+      "step": 71640
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655924981144404e-05,
+      "loss": 0.0124,
+      "step": 71650
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655876941405933e-05,
+      "loss": 0.0135,
+      "step": 71660
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655828901667462e-05,
+      "loss": 0.0148,
+      "step": 71670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655780861928992e-05,
+      "loss": 0.0219,
+      "step": 71680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655732822190518e-05,
+      "loss": 0.0119,
+      "step": 71690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655684782452047e-05,
+      "loss": 0.0146,
+      "step": 71700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655636742713576e-05,
+      "loss": 0.0147,
+      "step": 71710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655588702975102e-05,
+      "loss": 0.0164,
+      "step": 71720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965554066323663e-05,
+      "loss": 0.0127,
+      "step": 71730
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965549262349816e-05,
+      "loss": 0.0124,
+      "step": 71740
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965544458375969e-05,
+      "loss": 0.0193,
+      "step": 71750
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655396544021216e-05,
+      "loss": 0.0127,
+      "step": 71760
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655348504282745e-05,
+      "loss": 0.0137,
+      "step": 71770
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655300464544274e-05,
+      "loss": 0.013,
+      "step": 71780
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9655252424805803e-05,
+      "loss": 0.0126,
+      "step": 71790
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965520438506733e-05,
+      "loss": 0.0125,
+      "step": 71800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965515634532886e-05,
+      "loss": 0.0118,
+      "step": 71810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9655108305590388e-05,
+      "loss": 0.0143,
+      "step": 71820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9655060265851914e-05,
+      "loss": 0.012,
+      "step": 71830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9655012226113443e-05,
+      "loss": 0.0154,
+      "step": 71840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654964186374972e-05,
+      "loss": 0.0122,
+      "step": 71850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.96549161466365e-05,
+      "loss": 0.0127,
+      "step": 71860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654868106898027e-05,
+      "loss": 0.013,
+      "step": 71870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654820067159557e-05,
+      "loss": 0.0139,
+      "step": 71880
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654772027421086e-05,
+      "loss": 0.0134,
+      "step": 71890
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654723987682612e-05,
+      "loss": 0.0157,
+      "step": 71900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965467594794414e-05,
+      "loss": 0.0114,
+      "step": 71910
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965462790820567e-05,
+      "loss": 0.0117,
+      "step": 71920
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.96545798684672e-05,
+      "loss": 0.0122,
+      "step": 71930
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654531828728726e-05,
+      "loss": 0.0192,
+      "step": 71940
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654483788990255e-05,
+      "loss": 0.0132,
+      "step": 71950
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654435749251784e-05,
+      "loss": 0.0111,
+      "step": 71960
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654387709513313e-05,
+      "loss": 0.0143,
+      "step": 71970
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965433966977484e-05,
+      "loss": 0.014,
+      "step": 71980
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965429163003637e-05,
+      "loss": 0.0123,
+      "step": 71990
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654243590297898e-05,
+      "loss": 0.0104,
+      "step": 72000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654195550559424e-05,
+      "loss": 0.0107,
+      "step": 72010
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654147510820953e-05,
+      "loss": 0.0178,
+      "step": 72020
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654099471082482e-05,
+      "loss": 0.0142,
+      "step": 72030
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965405143134401e-05,
+      "loss": 0.0154,
+      "step": 72040
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9654003391605537e-05,
+      "loss": 0.0092,
+      "step": 72050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653955351867067e-05,
+      "loss": 0.0125,
+      "step": 72060
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653907312128596e-05,
+      "loss": 0.0161,
+      "step": 72070
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653859272390122e-05,
+      "loss": 0.013,
+      "step": 72080
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965381123265165e-05,
+      "loss": 0.011,
+      "step": 72090
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965376319291318e-05,
+      "loss": 0.014,
+      "step": 72100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965371515317471e-05,
+      "loss": 0.0181,
+      "step": 72110
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653667113436236e-05,
+      "loss": 0.0136,
+      "step": 72120
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653619073697765e-05,
+      "loss": 0.0169,
+      "step": 72130
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653571033959294e-05,
+      "loss": 0.0144,
+      "step": 72140
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653522994220823e-05,
+      "loss": 0.0116,
+      "step": 72150
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965347495448235e-05,
+      "loss": 0.0164,
+      "step": 72160
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965342691474388e-05,
+      "loss": 0.0112,
+      "step": 72170
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653378875005408e-05,
+      "loss": 0.0115,
+      "step": 72180
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653330835266934e-05,
+      "loss": 0.0153,
+      "step": 72190
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653282795528463e-05,
+      "loss": 0.0188,
+      "step": 72200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653234755789992e-05,
+      "loss": 0.0173,
+      "step": 72210
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965318671605152e-05,
+      "loss": 0.0122,
+      "step": 72220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653138676313047e-05,
+      "loss": 0.0122,
+      "step": 72230
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653090636574577e-05,
+      "loss": 0.0119,
+      "step": 72240
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9653042596836106e-05,
+      "loss": 0.0185,
+      "step": 72250
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9652994557097632e-05,
+      "loss": 0.016,
+      "step": 72260
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965294651735916e-05,
+      "loss": 0.0111,
+      "step": 72270
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965289847762069e-05,
+      "loss": 0.0155,
+      "step": 72280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965285043788222e-05,
+      "loss": 0.0138,
+      "step": 72290
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9652802398143745e-05,
+      "loss": 0.0121,
+      "step": 72300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9652754358405275e-05,
+      "loss": 0.0139,
+      "step": 72310
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9652706318666804e-05,
+      "loss": 0.0187,
+      "step": 72320
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9652658278928333e-05,
+      "loss": 0.0161,
+      "step": 72330
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965261023918986e-05,
+      "loss": 0.017,
+      "step": 72340
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965256219945139e-05,
+      "loss": 0.0165,
+      "step": 72350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9652514159712918e-05,
+      "loss": 0.011,
+      "step": 72360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9652466119974444e-05,
+      "loss": 0.0159,
+      "step": 72370
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9652418080235973e-05,
+      "loss": 0.0157,
+      "step": 72380
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9652370040497502e-05,
+      "loss": 0.0155,
+      "step": 72390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965232200075903e-05,
+      "loss": 0.011,
+      "step": 72400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9652273961020557e-05,
+      "loss": 0.0129,
+      "step": 72410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9652225921282087e-05,
+      "loss": 0.0121,
+      "step": 72420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9652177881543616e-05,
+      "loss": 0.0158,
+      "step": 72430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965212984180514e-05,
+      "loss": 0.0111,
+      "step": 72440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965208180206667e-05,
+      "loss": 0.0163,
+      "step": 72450
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.96520337623282e-05,
+      "loss": 0.0151,
+      "step": 72460
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.965198572258973e-05,
+      "loss": 0.0188,
+      "step": 72470
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9651937682851255e-05,
+      "loss": 0.0132,
+      "step": 72480
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9651889643112785e-05,
+      "loss": 0.0136,
+      "step": 72490
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9651841603374314e-05,
+      "loss": 0.0092,
+      "step": 72500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965179356363584e-05,
+      "loss": 0.0137,
+      "step": 72510
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965174552389737e-05,
+      "loss": 0.0187,
+      "step": 72520
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.96516974841589e-05,
+      "loss": 0.0147,
+      "step": 72530
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9651649444420428e-05,
+      "loss": 0.0092,
+      "step": 72540
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9651601404681953e-05,
+      "loss": 0.0096,
+      "step": 72550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9651553364943483e-05,
+      "loss": 0.0146,
+      "step": 72560
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9651505325205012e-05,
+      "loss": 0.018,
+      "step": 72570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965145728546654e-05,
+      "loss": 0.0115,
+      "step": 72580
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9651409245728067e-05,
+      "loss": 0.0158,
+      "step": 72590
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9651361205989596e-05,
+      "loss": 0.0091,
+      "step": 72600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9651313166251126e-05,
+      "loss": 0.0153,
+      "step": 72610
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965126512651265e-05,
+      "loss": 0.0143,
+      "step": 72620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965121708677418e-05,
+      "loss": 0.011,
+      "step": 72630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965116904703571e-05,
+      "loss": 0.0103,
+      "step": 72640
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965112100729724e-05,
+      "loss": 0.0108,
+      "step": 72650
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9651072967558765e-05,
+      "loss": 0.0153,
+      "step": 72660
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9651024927820295e-05,
+      "loss": 0.0126,
+      "step": 72670
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9650976888081824e-05,
+      "loss": 0.0094,
+      "step": 72680
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965092884834335e-05,
+      "loss": 0.0113,
+      "step": 72690
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965088080860488e-05,
+      "loss": 0.0132,
+      "step": 72700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9650832768866408e-05,
+      "loss": 0.0133,
+      "step": 72710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9650784729127937e-05,
+      "loss": 0.0198,
+      "step": 72720
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9650736689389463e-05,
+      "loss": 0.0114,
+      "step": 72730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9650688649650993e-05,
+      "loss": 0.0138,
+      "step": 72740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9650640609912522e-05,
+      "loss": 0.0116,
+      "step": 72750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965059257017405e-05,
+      "loss": 0.0121,
+      "step": 72760
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9650544530435577e-05,
+      "loss": 0.0134,
+      "step": 72770
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9650496490697106e-05,
+      "loss": 0.0117,
+      "step": 72780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9650448450958636e-05,
+      "loss": 0.0121,
+      "step": 72790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965040041122016e-05,
+      "loss": 0.0131,
+      "step": 72800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965035237148169e-05,
+      "loss": 0.0113,
+      "step": 72810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965030433174322e-05,
+      "loss": 0.0126,
+      "step": 72820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965025629200475e-05,
+      "loss": 0.0128,
+      "step": 72830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9650208252266275e-05,
+      "loss": 0.0162,
+      "step": 72840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9650160212527804e-05,
+      "loss": 0.0138,
+      "step": 72850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9650112172789334e-05,
+      "loss": 0.0178,
+      "step": 72860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965006413305086e-05,
+      "loss": 0.0108,
+      "step": 72870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.965001609331239e-05,
+      "loss": 0.0135,
+      "step": 72880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649968053573918e-05,
+      "loss": 0.0151,
+      "step": 72890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649920013835447e-05,
+      "loss": 0.0156,
+      "step": 72900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649871974096973e-05,
+      "loss": 0.0129,
+      "step": 72910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649823934358503e-05,
+      "loss": 0.0213,
+      "step": 72920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649775894620032e-05,
+      "loss": 0.0117,
+      "step": 72930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.964972785488156e-05,
+      "loss": 0.0119,
+      "step": 72940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649679815143087e-05,
+      "loss": 0.0108,
+      "step": 72950
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649631775404616e-05,
+      "loss": 0.0134,
+      "step": 72960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649583735666146e-05,
+      "loss": 0.0135,
+      "step": 72970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.964953569592767e-05,
+      "loss": 0.0128,
+      "step": 72980
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.96494876561892e-05,
+      "loss": 0.0115,
+      "step": 72990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.964943961645073e-05,
+      "loss": 0.0118,
+      "step": 73000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.964939157671226e-05,
+      "loss": 0.0126,
+      "step": 73010
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649343536973785e-05,
+      "loss": 0.0128,
+      "step": 73020
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649295497235314e-05,
+      "loss": 0.0138,
+      "step": 73030
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649247457496844e-05,
+      "loss": 0.0111,
+      "step": 73040
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.964919941775837e-05,
+      "loss": 0.0108,
+      "step": 73050
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.96491513780199e-05,
+      "loss": 0.0081,
+      "step": 73060
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649103338281428e-05,
+      "loss": 0.0119,
+      "step": 73070
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649055298542957e-05,
+      "loss": 0.0122,
+      "step": 73080
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9649007258804483e-05,
+      "loss": 0.0158,
+      "step": 73090
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9648959219066012e-05,
+      "loss": 0.0102,
+      "step": 73100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9648911179327542e-05,
+      "loss": 0.011,
+      "step": 73110
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.964886313958907e-05,
+      "loss": 0.0161,
+      "step": 73120
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9648815099850597e-05,
+      "loss": 0.0143,
+      "step": 73130
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9648767060112126e-05,
+      "loss": 0.0131,
+      "step": 73140
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9648719020373655e-05,
+      "loss": 0.0103,
+      "step": 73150
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.964867098063518e-05,
+      "loss": 0.0142,
+      "step": 73160
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.964862294089671e-05,
+      "loss": 0.0122,
+      "step": 73170
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.964857490115824e-05,
+      "loss": 0.0127,
+      "step": 73180
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.964852686141977e-05,
+      "loss": 0.0143,
+      "step": 73190
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9648478821681295e-05,
+      "loss": 0.0169,
+      "step": 73200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9648430781942824e-05,
+      "loss": 0.0121,
+      "step": 73210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9648382742204354e-05,
+      "loss": 0.0128,
+      "step": 73220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964833470246588e-05,
+      "loss": 0.0155,
+      "step": 73230
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964828666272741e-05,
+      "loss": 0.016,
+      "step": 73240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9648238622988938e-05,
+      "loss": 0.0105,
+      "step": 73250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9648190583250467e-05,
+      "loss": 0.0136,
+      "step": 73260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9648142543511993e-05,
+      "loss": 0.0117,
+      "step": 73270
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9648094503773522e-05,
+      "loss": 0.0116,
+      "step": 73280
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964804646403505e-05,
+      "loss": 0.0135,
+      "step": 73290
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964799842429658e-05,
+      "loss": 0.0156,
+      "step": 73300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9647950384558107e-05,
+      "loss": 0.0145,
+      "step": 73310
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9647902344819636e-05,
+      "loss": 0.016,
+      "step": 73320
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9647854305081165e-05,
+      "loss": 0.0106,
+      "step": 73330
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964780626534269e-05,
+      "loss": 0.0131,
+      "step": 73340
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964775822560422e-05,
+      "loss": 0.0153,
+      "step": 73350
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964771018586575e-05,
+      "loss": 0.0159,
+      "step": 73360
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964766214612728e-05,
+      "loss": 0.0157,
+      "step": 73370
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9647614106388805e-05,
+      "loss": 0.0113,
+      "step": 73380
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9647566066650334e-05,
+      "loss": 0.0106,
+      "step": 73390
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9647518026911863e-05,
+      "loss": 0.0163,
+      "step": 73400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964746998717339e-05,
+      "loss": 0.0115,
+      "step": 73410
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964742194743492e-05,
+      "loss": 0.0119,
+      "step": 73420
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9647373907696448e-05,
+      "loss": 0.0149,
+      "step": 73430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9647325867957977e-05,
+      "loss": 0.01,
+      "step": 73440
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9647277828219503e-05,
+      "loss": 0.0097,
+      "step": 73450
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9647229788481032e-05,
+      "loss": 0.0156,
+      "step": 73460
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964718174874256e-05,
+      "loss": 0.0099,
+      "step": 73470
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964713370900409e-05,
+      "loss": 0.0159,
+      "step": 73480
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9647085669265617e-05,
+      "loss": 0.0128,
+      "step": 73490
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9647037629527146e-05,
+      "loss": 0.0157,
+      "step": 73500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9646989589788675e-05,
+      "loss": 0.0118,
+      "step": 73510
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.96469415500502e-05,
+      "loss": 0.0119,
+      "step": 73520
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964689351031173e-05,
+      "loss": 0.0107,
+      "step": 73530
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964684547057326e-05,
+      "loss": 0.0136,
+      "step": 73540
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964679743083479e-05,
+      "loss": 0.0152,
+      "step": 73550
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9646749391096315e-05,
+      "loss": 0.0135,
+      "step": 73560
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9646701351357844e-05,
+      "loss": 0.0156,
+      "step": 73570
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9646653311619373e-05,
+      "loss": 0.0149,
+      "step": 73580
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.96466052718809e-05,
+      "loss": 0.015,
+      "step": 73590
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964655723214243e-05,
+      "loss": 0.0134,
+      "step": 73600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9646509192403958e-05,
+      "loss": 0.0121,
+      "step": 73610
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9646461152665487e-05,
+      "loss": 0.0154,
+      "step": 73620
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9646413112927013e-05,
+      "loss": 0.0136,
+      "step": 73630
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9646365073188542e-05,
+      "loss": 0.0128,
+      "step": 73640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964631703345007e-05,
+      "loss": 0.0144,
+      "step": 73650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.96462689937116e-05,
+      "loss": 0.0091,
+      "step": 73660
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9646220953973127e-05,
+      "loss": 0.0123,
+      "step": 73670
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9646172914234656e-05,
+      "loss": 0.0112,
+      "step": 73680
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9646124874496185e-05,
+      "loss": 0.0168,
+      "step": 73690
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964607683475771e-05,
+      "loss": 0.0143,
+      "step": 73700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964602879501924e-05,
+      "loss": 0.0112,
+      "step": 73710
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964598075528077e-05,
+      "loss": 0.0194,
+      "step": 73720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.96459327155423e-05,
+      "loss": 0.0092,
+      "step": 73730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9645884675803825e-05,
+      "loss": 0.0111,
+      "step": 73740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9645836636065354e-05,
+      "loss": 0.0157,
+      "step": 73750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9645788596326883e-05,
+      "loss": 0.0122,
+      "step": 73760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964574055658841e-05,
+      "loss": 0.0147,
+      "step": 73770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964569251684994e-05,
+      "loss": 0.012,
+      "step": 73780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9645644477111468e-05,
+      "loss": 0.0174,
+      "step": 73790
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9645596437372997e-05,
+      "loss": 0.0153,
+      "step": 73800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9645548397634523e-05,
+      "loss": 0.0152,
+      "step": 73810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9645500357896052e-05,
+      "loss": 0.0104,
+      "step": 73820
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964545231815758e-05,
+      "loss": 0.0111,
+      "step": 73830
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964540427841911e-05,
+      "loss": 0.0182,
+      "step": 73840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9645356238680637e-05,
+      "loss": 0.0083,
+      "step": 73850
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9645308198942166e-05,
+      "loss": 0.0152,
+      "step": 73860
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9645260159203695e-05,
+      "loss": 0.0137,
+      "step": 73870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964521211946522e-05,
+      "loss": 0.0172,
+      "step": 73880
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.964516407972675e-05,
+      "loss": 0.0105,
+      "step": 73890
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964511603998828e-05,
+      "loss": 0.0121,
+      "step": 73900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964506800024981e-05,
+      "loss": 0.0114,
+      "step": 73910
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9645019960511335e-05,
+      "loss": 0.0134,
+      "step": 73920
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644971920772864e-05,
+      "loss": 0.0152,
+      "step": 73930
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644923881034393e-05,
+      "loss": 0.0116,
+      "step": 73940
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964487584129592e-05,
+      "loss": 0.0151,
+      "step": 73950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964482780155745e-05,
+      "loss": 0.0131,
+      "step": 73960
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644779761818978e-05,
+      "loss": 0.0088,
+      "step": 73970
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644731722080507e-05,
+      "loss": 0.0125,
+      "step": 73980
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644683682342033e-05,
+      "loss": 0.0152,
+      "step": 73990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644635642603562e-05,
+      "loss": 0.0134,
+      "step": 74000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964458760286509e-05,
+      "loss": 0.0175,
+      "step": 74010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964453956312662e-05,
+      "loss": 0.0132,
+      "step": 74020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644491523388146e-05,
+      "loss": 0.0089,
+      "step": 74030
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644443483649676e-05,
+      "loss": 0.0123,
+      "step": 74040
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644395443911205e-05,
+      "loss": 0.0132,
+      "step": 74050
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964434740417273e-05,
+      "loss": 0.0127,
+      "step": 74060
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964429936443426e-05,
+      "loss": 0.0086,
+      "step": 74070
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964425132469579e-05,
+      "loss": 0.0097,
+      "step": 74080
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964420328495732e-05,
+      "loss": 0.009,
+      "step": 74090
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644155245218845e-05,
+      "loss": 0.0177,
+      "step": 74100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644107205480374e-05,
+      "loss": 0.0122,
+      "step": 74110
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644059165741903e-05,
+      "loss": 0.0183,
+      "step": 74120
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9644011126003432e-05,
+      "loss": 0.013,
+      "step": 74130
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964396308626496e-05,
+      "loss": 0.0124,
+      "step": 74140
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964391504652649e-05,
+      "loss": 0.0118,
+      "step": 74150
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9643867006788017e-05,
+      "loss": 0.0182,
+      "step": 74160
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9643818967049546e-05,
+      "loss": 0.0132,
+      "step": 74170
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9643770927311075e-05,
+      "loss": 0.01,
+      "step": 74180
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.96437228875726e-05,
+      "loss": 0.0143,
+      "step": 74190
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964367484783413e-05,
+      "loss": 0.0125,
+      "step": 74200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964362680809566e-05,
+      "loss": 0.0122,
+      "step": 74210
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964357876835719e-05,
+      "loss": 0.0128,
+      "step": 74220
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9643530728618715e-05,
+      "loss": 0.0119,
+      "step": 74230
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9643482688880244e-05,
+      "loss": 0.0162,
+      "step": 74240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9643434649141773e-05,
+      "loss": 0.0144,
+      "step": 74250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.96433866094033e-05,
+      "loss": 0.0135,
+      "step": 74260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964333856966483e-05,
+      "loss": 0.0128,
+      "step": 74270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9643290529926358e-05,
+      "loss": 0.013,
+      "step": 74280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9643242490187887e-05,
+      "loss": 0.0138,
+      "step": 74290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9643194450449413e-05,
+      "loss": 0.0096,
+      "step": 74300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9643146410710942e-05,
+      "loss": 0.0148,
+      "step": 74310
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964309837097247e-05,
+      "loss": 0.0077,
+      "step": 74320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9643050331234e-05,
+      "loss": 0.0129,
+      "step": 74330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9643002291495527e-05,
+      "loss": 0.0159,
+      "step": 74340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9642954251757056e-05,
+      "loss": 0.0163,
+      "step": 74350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9642906212018585e-05,
+      "loss": 0.01,
+      "step": 74360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964285817228011e-05,
+      "loss": 0.0184,
+      "step": 74370
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964281013254164e-05,
+      "loss": 0.0127,
+      "step": 74380
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964276209280317e-05,
+      "loss": 0.0098,
+      "step": 74390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.96427140530647e-05,
+      "loss": 0.0144,
+      "step": 74400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9642666013326225e-05,
+      "loss": 0.0123,
+      "step": 74410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9642617973587754e-05,
+      "loss": 0.0118,
+      "step": 74420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9642569933849283e-05,
+      "loss": 0.0128,
+      "step": 74430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964252189411081e-05,
+      "loss": 0.0098,
+      "step": 74440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964247385437234e-05,
+      "loss": 0.0216,
+      "step": 74450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9642425814633868e-05,
+      "loss": 0.0121,
+      "step": 74460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9642377774895397e-05,
+      "loss": 0.0128,
+      "step": 74470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9642329735156923e-05,
+      "loss": 0.0099,
+      "step": 74480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9642281695418452e-05,
+      "loss": 0.0124,
+      "step": 74490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964223365567998e-05,
+      "loss": 0.0103,
+      "step": 74500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964218561594151e-05,
+      "loss": 0.011,
+      "step": 74510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9642137576203037e-05,
+      "loss": 0.0142,
+      "step": 74520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9642089536464566e-05,
+      "loss": 0.0093,
+      "step": 74530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9642041496726095e-05,
+      "loss": 0.0124,
+      "step": 74540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964199345698762e-05,
+      "loss": 0.0148,
+      "step": 74550
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964194541724915e-05,
+      "loss": 0.0104,
+      "step": 74560
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964189737751068e-05,
+      "loss": 0.0134,
+      "step": 74570
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.964184933777221e-05,
+      "loss": 0.0193,
+      "step": 74580
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9641801298033735e-05,
+      "loss": 0.0144,
+      "step": 74590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9641753258295264e-05,
+      "loss": 0.0111,
+      "step": 74600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9641705218556793e-05,
+      "loss": 0.0119,
+      "step": 74610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964165717881832e-05,
+      "loss": 0.0099,
+      "step": 74620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964160913907985e-05,
+      "loss": 0.0137,
+      "step": 74630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9641561099341378e-05,
+      "loss": 0.0103,
+      "step": 74640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9641513059602907e-05,
+      "loss": 0.0136,
+      "step": 74650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9641465019864433e-05,
+      "loss": 0.014,
+      "step": 74660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9641416980125962e-05,
+      "loss": 0.0166,
+      "step": 74670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964136894038749e-05,
+      "loss": 0.0154,
+      "step": 74680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964132090064902e-05,
+      "loss": 0.0153,
+      "step": 74690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9641272860910547e-05,
+      "loss": 0.0166,
+      "step": 74700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9641224821172076e-05,
+      "loss": 0.0141,
+      "step": 74710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9641176781433605e-05,
+      "loss": 0.0099,
+      "step": 74720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964112874169513e-05,
+      "loss": 0.0118,
+      "step": 74730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964108070195666e-05,
+      "loss": 0.0135,
+      "step": 74740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964103266221819e-05,
+      "loss": 0.0118,
+      "step": 74750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964098462247972e-05,
+      "loss": 0.0112,
+      "step": 74760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640936582741245e-05,
+      "loss": 0.0161,
+      "step": 74770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640888543002774e-05,
+      "loss": 0.0134,
+      "step": 74780
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640840503264303e-05,
+      "loss": 0.0146,
+      "step": 74790
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964079246352583e-05,
+      "loss": 0.0136,
+      "step": 74800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964074442378736e-05,
+      "loss": 0.0073,
+      "step": 74810
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640696384048888e-05,
+      "loss": 0.0112,
+      "step": 74820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640648344310417e-05,
+      "loss": 0.0133,
+      "step": 74830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640600304571943e-05,
+      "loss": 0.01,
+      "step": 74840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640552264833472e-05,
+      "loss": 0.0174,
+      "step": 74850
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640504225095e-05,
+      "loss": 0.0161,
+      "step": 74860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964045618535653e-05,
+      "loss": 0.0112,
+      "step": 74870
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640408145618056e-05,
+      "loss": 0.0126,
+      "step": 74880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640360105879586e-05,
+      "loss": 0.0129,
+      "step": 74890
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640312066141115e-05,
+      "loss": 0.0117,
+      "step": 74900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964026402640264e-05,
+      "loss": 0.0168,
+      "step": 74910
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964021598666417e-05,
+      "loss": 0.0141,
+      "step": 74920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.96401679469257e-05,
+      "loss": 0.0106,
+      "step": 74930
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.964011990718723e-05,
+      "loss": 0.0097,
+      "step": 74940
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640071867448755e-05,
+      "loss": 0.0135,
+      "step": 74950
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9640023827710284e-05,
+      "loss": 0.0144,
+      "step": 74960
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639975787971813e-05,
+      "loss": 0.0131,
+      "step": 74970
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.963992774823334e-05,
+      "loss": 0.0112,
+      "step": 74980
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639879708494868e-05,
+      "loss": 0.011,
+      "step": 74990
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639831668756398e-05,
+      "loss": 0.0116,
+      "step": 75000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639783629017927e-05,
+      "loss": 0.0115,
+      "step": 75010
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639735589279453e-05,
+      "loss": 0.0144,
+      "step": 75020
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639687549540982e-05,
+      "loss": 0.0132,
+      "step": 75030
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.963963950980251e-05,
+      "loss": 0.0144,
+      "step": 75040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.963959147006404e-05,
+      "loss": 0.0135,
+      "step": 75050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639543430325566e-05,
+      "loss": 0.0122,
+      "step": 75060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639495390587096e-05,
+      "loss": 0.0115,
+      "step": 75070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639447350848625e-05,
+      "loss": 0.0097,
+      "step": 75080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.963939931111015e-05,
+      "loss": 0.0092,
+      "step": 75090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.963935127137168e-05,
+      "loss": 0.0095,
+      "step": 75100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.963930323163321e-05,
+      "loss": 0.0111,
+      "step": 75110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.963925519189474e-05,
+      "loss": 0.0135,
+      "step": 75120
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639207152156264e-05,
+      "loss": 0.0131,
+      "step": 75130
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639159112417794e-05,
+      "loss": 0.0153,
+      "step": 75140
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639111072679323e-05,
+      "loss": 0.0141,
+      "step": 75150
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.963906303294085e-05,
+      "loss": 0.0142,
+      "step": 75160
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9639014993202378e-05,
+      "loss": 0.0116,
+      "step": 75170
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9638966953463907e-05,
+      "loss": 0.0156,
+      "step": 75180
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9638918913725437e-05,
+      "loss": 0.0158,
+      "step": 75190
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9638870873986963e-05,
+      "loss": 0.0115,
+      "step": 75200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9638822834248492e-05,
+      "loss": 0.0264,
+      "step": 75210
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.963877479451002e-05,
+      "loss": 0.0134,
+      "step": 75220
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.963872675477155e-05,
+      "loss": 0.0123,
+      "step": 75230
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9638678715033076e-05,
+      "loss": 0.0166,
+      "step": 75240
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9638630675294606e-05,
+      "loss": 0.0124,
+      "step": 75250
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9638582635556135e-05,
+      "loss": 0.0129,
+      "step": 75260
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.963853459581766e-05,
+      "loss": 0.0169,
+      "step": 75270
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.963848655607919e-05,
+      "loss": 0.0132,
+      "step": 75280
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963843851634072e-05,
+      "loss": 0.012,
+      "step": 75290
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963839047660225e-05,
+      "loss": 0.0121,
+      "step": 75300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9638342436863774e-05,
+      "loss": 0.0132,
+      "step": 75310
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9638294397125304e-05,
+      "loss": 0.0145,
+      "step": 75320
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9638246357386833e-05,
+      "loss": 0.0128,
+      "step": 75330
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963819831764836e-05,
+      "loss": 0.0109,
+      "step": 75340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9638150277909888e-05,
+      "loss": 0.0084,
+      "step": 75350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9638102238171417e-05,
+      "loss": 0.0101,
+      "step": 75360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9638054198432947e-05,
+      "loss": 0.0134,
+      "step": 75370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9638006158694473e-05,
+      "loss": 0.0126,
+      "step": 75380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637958118956002e-05,
+      "loss": 0.0162,
+      "step": 75390
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963791007921753e-05,
+      "loss": 0.0114,
+      "step": 75400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963786203947906e-05,
+      "loss": 0.013,
+      "step": 75410
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637813999740586e-05,
+      "loss": 0.0132,
+      "step": 75420
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637770763975964e-05,
+      "loss": 0.0152,
+      "step": 75430
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963772272423749e-05,
+      "loss": 0.0137,
+      "step": 75440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963767468449902e-05,
+      "loss": 0.0105,
+      "step": 75450
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963762664476055e-05,
+      "loss": 0.012,
+      "step": 75460
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637578605022075e-05,
+      "loss": 0.0125,
+      "step": 75470
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637530565283604e-05,
+      "loss": 0.0118,
+      "step": 75480
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637482525545133e-05,
+      "loss": 0.0113,
+      "step": 75490
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637434485806662e-05,
+      "loss": 0.0126,
+      "step": 75500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637386446068188e-05,
+      "loss": 0.012,
+      "step": 75510
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637338406329718e-05,
+      "loss": 0.012,
+      "step": 75520
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637290366591247e-05,
+      "loss": 0.0105,
+      "step": 75530
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637242326852776e-05,
+      "loss": 0.0199,
+      "step": 75540
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637194287114302e-05,
+      "loss": 0.0115,
+      "step": 75550
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963714624737583e-05,
+      "loss": 0.0138,
+      "step": 75560
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963709820763736e-05,
+      "loss": 0.0116,
+      "step": 75570
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637050167898886e-05,
+      "loss": 0.0155,
+      "step": 75580
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9637002128160416e-05,
+      "loss": 0.014,
+      "step": 75590
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636954088421945e-05,
+      "loss": 0.0121,
+      "step": 75600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636906048683474e-05,
+      "loss": 0.0099,
+      "step": 75610
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636858008945e-05,
+      "loss": 0.0144,
+      "step": 75620
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963680996920653e-05,
+      "loss": 0.0142,
+      "step": 75630
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963676192946806e-05,
+      "loss": 0.0199,
+      "step": 75640
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636713889729584e-05,
+      "loss": 0.0125,
+      "step": 75650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636665849991114e-05,
+      "loss": 0.0119,
+      "step": 75660
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636617810252643e-05,
+      "loss": 0.0125,
+      "step": 75670
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636569770514172e-05,
+      "loss": 0.0072,
+      "step": 75680
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636521730775698e-05,
+      "loss": 0.011,
+      "step": 75690
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636473691037227e-05,
+      "loss": 0.0121,
+      "step": 75700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636425651298757e-05,
+      "loss": 0.0112,
+      "step": 75710
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636377611560286e-05,
+      "loss": 0.0118,
+      "step": 75720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636329571821812e-05,
+      "loss": 0.015,
+      "step": 75730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963628153208334e-05,
+      "loss": 0.0113,
+      "step": 75740
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963623349234487e-05,
+      "loss": 0.0174,
+      "step": 75750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636185452606396e-05,
+      "loss": 0.0124,
+      "step": 75760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636137412867926e-05,
+      "loss": 0.0208,
+      "step": 75770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636089373129455e-05,
+      "loss": 0.0123,
+      "step": 75780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9636041333390984e-05,
+      "loss": 0.0181,
+      "step": 75790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963599329365251e-05,
+      "loss": 0.0133,
+      "step": 75800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963594525391404e-05,
+      "loss": 0.0091,
+      "step": 75810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963589721417557e-05,
+      "loss": 0.012,
+      "step": 75820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635849174437094e-05,
+      "loss": 0.0128,
+      "step": 75830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635801134698624e-05,
+      "loss": 0.0121,
+      "step": 75840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635753094960153e-05,
+      "loss": 0.0128,
+      "step": 75850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635705055221682e-05,
+      "loss": 0.0131,
+      "step": 75860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635657015483208e-05,
+      "loss": 0.0114,
+      "step": 75870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635608975744737e-05,
+      "loss": 0.012,
+      "step": 75880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635560936006267e-05,
+      "loss": 0.0159,
+      "step": 75890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635512896267796e-05,
+      "loss": 0.0098,
+      "step": 75900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635464856529322e-05,
+      "loss": 0.0137,
+      "step": 75910
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963541681679085e-05,
+      "loss": 0.0111,
+      "step": 75920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.963536877705238e-05,
+      "loss": 0.0136,
+      "step": 75930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635320737313906e-05,
+      "loss": 0.0127,
+      "step": 75940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635272697575435e-05,
+      "loss": 0.0116,
+      "step": 75950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635224657836965e-05,
+      "loss": 0.0099,
+      "step": 75960
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9635176618098494e-05,
+      "loss": 0.0133,
+      "step": 75970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963512857836002e-05,
+      "loss": 0.0168,
+      "step": 75980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963508053862155e-05,
+      "loss": 0.0119,
+      "step": 75990
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963503249888308e-05,
+      "loss": 0.0106,
+      "step": 76000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634984459144604e-05,
+      "loss": 0.0132,
+      "step": 76010
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634936419406134e-05,
+      "loss": 0.0084,
+      "step": 76020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634888379667663e-05,
+      "loss": 0.012,
+      "step": 76030
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634840339929192e-05,
+      "loss": 0.0117,
+      "step": 76040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634792300190718e-05,
+      "loss": 0.0107,
+      "step": 76050
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634744260452247e-05,
+      "loss": 0.0122,
+      "step": 76060
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634696220713777e-05,
+      "loss": 0.014,
+      "step": 76070
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634648180975306e-05,
+      "loss": 0.0141,
+      "step": 76080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963460014123683e-05,
+      "loss": 0.0129,
+      "step": 76090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963455210149836e-05,
+      "loss": 0.012,
+      "step": 76100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963450406175989e-05,
+      "loss": 0.0124,
+      "step": 76110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634456022021416e-05,
+      "loss": 0.0147,
+      "step": 76120
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634407982282945e-05,
+      "loss": 0.0127,
+      "step": 76130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634359942544475e-05,
+      "loss": 0.0123,
+      "step": 76140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634311902806004e-05,
+      "loss": 0.0121,
+      "step": 76150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963426386306753e-05,
+      "loss": 0.0166,
+      "step": 76160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963421582332906e-05,
+      "loss": 0.0165,
+      "step": 76170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963416778359059e-05,
+      "loss": 0.0151,
+      "step": 76180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634119743852114e-05,
+      "loss": 0.0141,
+      "step": 76190
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634071704113643e-05,
+      "loss": 0.0143,
+      "step": 76200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9634023664375173e-05,
+      "loss": 0.0094,
+      "step": 76210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633975624636702e-05,
+      "loss": 0.0115,
+      "step": 76220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633927584898228e-05,
+      "loss": 0.0181,
+      "step": 76230
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633879545159757e-05,
+      "loss": 0.0144,
+      "step": 76240
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633831505421286e-05,
+      "loss": 0.0157,
+      "step": 76250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633783465682816e-05,
+      "loss": 0.0109,
+      "step": 76260
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963373542594434e-05,
+      "loss": 0.0104,
+      "step": 76270
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963368738620587e-05,
+      "loss": 0.014,
+      "step": 76280
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.96336393464674e-05,
+      "loss": 0.0162,
+      "step": 76290
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633591306728926e-05,
+      "loss": 0.0093,
+      "step": 76300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633543266990455e-05,
+      "loss": 0.0132,
+      "step": 76310
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633495227251985e-05,
+      "loss": 0.0137,
+      "step": 76320
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633447187513514e-05,
+      "loss": 0.0129,
+      "step": 76330
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963339914777504e-05,
+      "loss": 0.0127,
+      "step": 76340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963335110803657e-05,
+      "loss": 0.0092,
+      "step": 76350
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633303068298098e-05,
+      "loss": 0.0117,
+      "step": 76360
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633255028559624e-05,
+      "loss": 0.0117,
+      "step": 76370
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633206988821153e-05,
+      "loss": 0.0143,
+      "step": 76380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633158949082683e-05,
+      "loss": 0.0095,
+      "step": 76390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633110909344212e-05,
+      "loss": 0.0129,
+      "step": 76400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633062869605738e-05,
+      "loss": 0.0115,
+      "step": 76410
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9633014829867267e-05,
+      "loss": 0.0131,
+      "step": 76420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632966790128796e-05,
+      "loss": 0.0089,
+      "step": 76430
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632918750390326e-05,
+      "loss": 0.0139,
+      "step": 76440
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963287071065185e-05,
+      "loss": 0.0146,
+      "step": 76450
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963282267091338e-05,
+      "loss": 0.0105,
+      "step": 76460
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963277463117491e-05,
+      "loss": 0.0083,
+      "step": 76470
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632726591436436e-05,
+      "loss": 0.0105,
+      "step": 76480
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632678551697965e-05,
+      "loss": 0.0181,
+      "step": 76490
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632630511959494e-05,
+      "loss": 0.0149,
+      "step": 76500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632582472221024e-05,
+      "loss": 0.0105,
+      "step": 76510
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963253443248255e-05,
+      "loss": 0.0111,
+      "step": 76520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963248639274408e-05,
+      "loss": 0.0108,
+      "step": 76530
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632438353005608e-05,
+      "loss": 0.0175,
+      "step": 76540
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632390313267134e-05,
+      "loss": 0.015,
+      "step": 76550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632342273528663e-05,
+      "loss": 0.014,
+      "step": 76560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632294233790193e-05,
+      "loss": 0.0111,
+      "step": 76570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632246194051722e-05,
+      "loss": 0.013,
+      "step": 76580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632198154313248e-05,
+      "loss": 0.013,
+      "step": 76590
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632150114574777e-05,
+      "loss": 0.0134,
+      "step": 76600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632102074836306e-05,
+      "loss": 0.0177,
+      "step": 76610
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9632054035097836e-05,
+      "loss": 0.0132,
+      "step": 76620
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963200599535936e-05,
+      "loss": 0.0103,
+      "step": 76630
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963195795562089e-05,
+      "loss": 0.0132,
+      "step": 76640
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.963190991588242e-05,
+      "loss": 0.0137,
+      "step": 76650
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9631861876143946e-05,
+      "loss": 0.0152,
+      "step": 76660
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9631813836405475e-05,
+      "loss": 0.0115,
+      "step": 76670
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9631765796667004e-05,
+      "loss": 0.0137,
+      "step": 76680
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9631717756928534e-05,
+      "loss": 0.0152,
+      "step": 76690
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.963166971719006e-05,
+      "loss": 0.0151,
+      "step": 76700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.963162167745159e-05,
+      "loss": 0.0122,
+      "step": 76710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9631573637713118e-05,
+      "loss": 0.0115,
+      "step": 76720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9631525597974644e-05,
+      "loss": 0.0122,
+      "step": 76730
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9631477558236173e-05,
+      "loss": 0.0145,
+      "step": 76740
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9631429518497703e-05,
+      "loss": 0.0133,
+      "step": 76750
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9631381478759232e-05,
+      "loss": 0.0114,
+      "step": 76760
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9631333439020758e-05,
+      "loss": 0.0108,
+      "step": 76770
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9631285399282287e-05,
+      "loss": 0.0175,
+      "step": 76780
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9631237359543816e-05,
+      "loss": 0.0153,
+      "step": 76790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9631189319805345e-05,
+      "loss": 0.0125,
+      "step": 76800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.963114128006687e-05,
+      "loss": 0.0093,
+      "step": 76810
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.96310932403284e-05,
+      "loss": 0.0159,
+      "step": 76820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.963104520058993e-05,
+      "loss": 0.0172,
+      "step": 76830
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630997160851456e-05,
+      "loss": 0.0139,
+      "step": 76840
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630949121112985e-05,
+      "loss": 0.0098,
+      "step": 76850
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630901081374514e-05,
+      "loss": 0.0151,
+      "step": 76860
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630853041636044e-05,
+      "loss": 0.0115,
+      "step": 76870
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.963080500189757e-05,
+      "loss": 0.0136,
+      "step": 76880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.96307569621591e-05,
+      "loss": 0.0124,
+      "step": 76890
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630708922420628e-05,
+      "loss": 0.0136,
+      "step": 76900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630660882682154e-05,
+      "loss": 0.0115,
+      "step": 76910
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630612842943683e-05,
+      "loss": 0.0108,
+      "step": 76920
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630564803205212e-05,
+      "loss": 0.0136,
+      "step": 76930
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.963051676346674e-05,
+      "loss": 0.0129,
+      "step": 76940
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630468723728268e-05,
+      "loss": 0.0111,
+      "step": 76950
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630420683989797e-05,
+      "loss": 0.0109,
+      "step": 76960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630372644251326e-05,
+      "loss": 0.0111,
+      "step": 76970
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630324604512855e-05,
+      "loss": 0.0136,
+      "step": 76980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.963027656477438e-05,
+      "loss": 0.0112,
+      "step": 76990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.963022852503591e-05,
+      "loss": 0.0096,
+      "step": 77000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.963018048529744e-05,
+      "loss": 0.0157,
+      "step": 77010
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630132445558966e-05,
+      "loss": 0.0127,
+      "step": 77020
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630084405820495e-05,
+      "loss": 0.0159,
+      "step": 77030
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9630036366082024e-05,
+      "loss": 0.0138,
+      "step": 77040
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629988326343553e-05,
+      "loss": 0.013,
+      "step": 77050
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.962994028660508e-05,
+      "loss": 0.0139,
+      "step": 77060
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.962989224686661e-05,
+      "loss": 0.0131,
+      "step": 77070
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629844207128138e-05,
+      "loss": 0.0147,
+      "step": 77080
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629796167389664e-05,
+      "loss": 0.0146,
+      "step": 77090
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629748127651193e-05,
+      "loss": 0.0109,
+      "step": 77100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629700087912722e-05,
+      "loss": 0.0143,
+      "step": 77110
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.962965204817425e-05,
+      "loss": 0.0116,
+      "step": 77120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629604008435777e-05,
+      "loss": 0.0121,
+      "step": 77130
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629555968697307e-05,
+      "loss": 0.0143,
+      "step": 77140
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629507928958836e-05,
+      "loss": 0.0169,
+      "step": 77150
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629459889220365e-05,
+      "loss": 0.0119,
+      "step": 77160
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.962941184948189e-05,
+      "loss": 0.0105,
+      "step": 77170
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.962936380974342e-05,
+      "loss": 0.0093,
+      "step": 77180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.962931577000495e-05,
+      "loss": 0.0114,
+      "step": 77190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.962926773026648e-05,
+      "loss": 0.0114,
+      "step": 77200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629219690528008e-05,
+      "loss": 0.0104,
+      "step": 77210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629171650789534e-05,
+      "loss": 0.0112,
+      "step": 77220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629123611051063e-05,
+      "loss": 0.0121,
+      "step": 77230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629075571312593e-05,
+      "loss": 0.0196,
+      "step": 77240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9629027531574122e-05,
+      "loss": 0.0127,
+      "step": 77250
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9628979491835648e-05,
+      "loss": 0.0135,
+      "step": 77260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9628931452097177e-05,
+      "loss": 0.0125,
+      "step": 77270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9628883412358706e-05,
+      "loss": 0.0144,
+      "step": 77280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9628835372620236e-05,
+      "loss": 0.0111,
+      "step": 77290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.962878733288176e-05,
+      "loss": 0.0102,
+      "step": 77300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.962873929314329e-05,
+      "loss": 0.0151,
+      "step": 77310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.962869125340482e-05,
+      "loss": 0.0106,
+      "step": 77320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9628643213666346e-05,
+      "loss": 0.0117,
+      "step": 77330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9628595173927875e-05,
+      "loss": 0.0109,
+      "step": 77340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9628547134189404e-05,
+      "loss": 0.0157,
+      "step": 77350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9628499094450934e-05,
+      "loss": 0.0134,
+      "step": 77360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962845105471246e-05,
+      "loss": 0.0139,
+      "step": 77370
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962840301497399e-05,
+      "loss": 0.0144,
+      "step": 77380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9628354975235518e-05,
+      "loss": 0.0117,
+      "step": 77390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9628306935497044e-05,
+      "loss": 0.0103,
+      "step": 77400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9628258895758573e-05,
+      "loss": 0.0131,
+      "step": 77410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9628210856020103e-05,
+      "loss": 0.0129,
+      "step": 77420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9628162816281632e-05,
+      "loss": 0.0135,
+      "step": 77430
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9628114776543158e-05,
+      "loss": 0.0139,
+      "step": 77440
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9628066736804687e-05,
+      "loss": 0.0097,
+      "step": 77450
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9628018697066216e-05,
+      "loss": 0.0129,
+      "step": 77460
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627970657327746e-05,
+      "loss": 0.0101,
+      "step": 77470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962792261758927e-05,
+      "loss": 0.0184,
+      "step": 77480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.96278745778508e-05,
+      "loss": 0.012,
+      "step": 77490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962782653811233e-05,
+      "loss": 0.0127,
+      "step": 77500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627778498373856e-05,
+      "loss": 0.0134,
+      "step": 77510
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627730458635385e-05,
+      "loss": 0.0146,
+      "step": 77520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627682418896914e-05,
+      "loss": 0.0135,
+      "step": 77530
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627634379158444e-05,
+      "loss": 0.0174,
+      "step": 77540
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962758633941997e-05,
+      "loss": 0.0114,
+      "step": 77550
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.96275382996815e-05,
+      "loss": 0.0124,
+      "step": 77560
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627490259943028e-05,
+      "loss": 0.0145,
+      "step": 77570
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627442220204554e-05,
+      "loss": 0.0117,
+      "step": 77580
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627394180466083e-05,
+      "loss": 0.0155,
+      "step": 77590
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627346140727613e-05,
+      "loss": 0.0127,
+      "step": 77600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627298100989142e-05,
+      "loss": 0.0092,
+      "step": 77610
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627250061250668e-05,
+      "loss": 0.0117,
+      "step": 77620
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627202021512197e-05,
+      "loss": 0.0082,
+      "step": 77630
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627153981773726e-05,
+      "loss": 0.0175,
+      "step": 77640
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9627105942035255e-05,
+      "loss": 0.0125,
+      "step": 77650
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962705790229678e-05,
+      "loss": 0.0147,
+      "step": 77660
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962700986255831e-05,
+      "loss": 0.0101,
+      "step": 77670
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962696182281984e-05,
+      "loss": 0.011,
+      "step": 77680
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626913783081366e-05,
+      "loss": 0.0103,
+      "step": 77690
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626865743342895e-05,
+      "loss": 0.0164,
+      "step": 77700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626817703604424e-05,
+      "loss": 0.0171,
+      "step": 77710
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626769663865954e-05,
+      "loss": 0.0083,
+      "step": 77720
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962672162412748e-05,
+      "loss": 0.0185,
+      "step": 77730
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962667358438901e-05,
+      "loss": 0.0125,
+      "step": 77740
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626625544650538e-05,
+      "loss": 0.0122,
+      "step": 77750
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626577504912064e-05,
+      "loss": 0.0128,
+      "step": 77760
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626529465173593e-05,
+      "loss": 0.0131,
+      "step": 77770
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626481425435122e-05,
+      "loss": 0.0127,
+      "step": 77780
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962643338569665e-05,
+      "loss": 0.0104,
+      "step": 77790
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626385345958178e-05,
+      "loss": 0.0161,
+      "step": 77800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626337306219707e-05,
+      "loss": 0.0146,
+      "step": 77810
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626289266481236e-05,
+      "loss": 0.0154,
+      "step": 77820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626241226742765e-05,
+      "loss": 0.0088,
+      "step": 77830
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962619318700429e-05,
+      "loss": 0.0108,
+      "step": 77840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962614514726582e-05,
+      "loss": 0.0132,
+      "step": 77850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962609710752735e-05,
+      "loss": 0.0111,
+      "step": 77860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626049067788876e-05,
+      "loss": 0.01,
+      "step": 77870
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9626001028050405e-05,
+      "loss": 0.0127,
+      "step": 77880
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9625952988311934e-05,
+      "loss": 0.0127,
+      "step": 77890
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9625904948573463e-05,
+      "loss": 0.0102,
+      "step": 77900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962585690883499e-05,
+      "loss": 0.0109,
+      "step": 77910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962580886909652e-05,
+      "loss": 0.0149,
+      "step": 77920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9625760829358048e-05,
+      "loss": 0.0124,
+      "step": 77930
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9625712789619574e-05,
+      "loss": 0.0118,
+      "step": 77940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9625664749881103e-05,
+      "loss": 0.0126,
+      "step": 77950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9625616710142632e-05,
+      "loss": 0.0106,
+      "step": 77960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962556867040416e-05,
+      "loss": 0.0119,
+      "step": 77970
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9625520630665687e-05,
+      "loss": 0.0153,
+      "step": 77980
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9625472590927217e-05,
+      "loss": 0.0126,
+      "step": 77990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9625424551188746e-05,
+      "loss": 0.0112,
+      "step": 78000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9625376511450275e-05,
+      "loss": 0.0116,
+      "step": 78010
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.96253284717118e-05,
+      "loss": 0.0094,
+      "step": 78020
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962528043197333e-05,
+      "loss": 0.0106,
+      "step": 78030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.962523239223486e-05,
+      "loss": 0.0163,
+      "step": 78040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9625184352496386e-05,
+      "loss": 0.0139,
+      "step": 78050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9625136312757915e-05,
+      "loss": 0.0112,
+      "step": 78060
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9625093076993293e-05,
+      "loss": 0.0103,
+      "step": 78070
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962504503725482e-05,
+      "loss": 0.0117,
+      "step": 78080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624996997516348e-05,
+      "loss": 0.0173,
+      "step": 78090
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624948957777877e-05,
+      "loss": 0.0092,
+      "step": 78100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624900918039403e-05,
+      "loss": 0.0133,
+      "step": 78110
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624852878300932e-05,
+      "loss": 0.0121,
+      "step": 78120
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624804838562462e-05,
+      "loss": 0.0115,
+      "step": 78130
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962475679882399e-05,
+      "loss": 0.0138,
+      "step": 78140
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624708759085517e-05,
+      "loss": 0.0091,
+      "step": 78150
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624660719347046e-05,
+      "loss": 0.0109,
+      "step": 78160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624612679608575e-05,
+      "loss": 0.0128,
+      "step": 78170
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.96245646398701e-05,
+      "loss": 0.0136,
+      "step": 78180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962451660013163e-05,
+      "loss": 0.0105,
+      "step": 78190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962446856039316e-05,
+      "loss": 0.0138,
+      "step": 78200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962442052065469e-05,
+      "loss": 0.0195,
+      "step": 78210
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624372480916215e-05,
+      "loss": 0.0127,
+      "step": 78220
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624324441177744e-05,
+      "loss": 0.0098,
+      "step": 78230
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624276401439274e-05,
+      "loss": 0.0174,
+      "step": 78240
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624228361700803e-05,
+      "loss": 0.0162,
+      "step": 78250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962418032196233e-05,
+      "loss": 0.0117,
+      "step": 78260
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624132282223858e-05,
+      "loss": 0.0156,
+      "step": 78270
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624084242485387e-05,
+      "loss": 0.021,
+      "step": 78280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9624036202746913e-05,
+      "loss": 0.0114,
+      "step": 78290
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9623988163008442e-05,
+      "loss": 0.0083,
+      "step": 78300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962394012326997e-05,
+      "loss": 0.0147,
+      "step": 78310
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.96238920835315e-05,
+      "loss": 0.0126,
+      "step": 78320
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9623844043793027e-05,
+      "loss": 0.0085,
+      "step": 78330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9623796004054556e-05,
+      "loss": 0.0073,
+      "step": 78340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9623747964316085e-05,
+      "loss": 0.0124,
+      "step": 78350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962369992457761e-05,
+      "loss": 0.0108,
+      "step": 78360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962365188483914e-05,
+      "loss": 0.0126,
+      "step": 78370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962360384510067e-05,
+      "loss": 0.0157,
+      "step": 78380
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.96235558053622e-05,
+      "loss": 0.0142,
+      "step": 78390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9623507765623725e-05,
+      "loss": 0.0101,
+      "step": 78400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9623459725885254e-05,
+      "loss": 0.0077,
+      "step": 78410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9623411686146783e-05,
+      "loss": 0.0135,
+      "step": 78420
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9623363646408313e-05,
+      "loss": 0.0155,
+      "step": 78430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962331560666984e-05,
+      "loss": 0.0114,
+      "step": 78440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9623267566931368e-05,
+      "loss": 0.0149,
+      "step": 78450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9623219527192897e-05,
+      "loss": 0.0119,
+      "step": 78460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9623171487454423e-05,
+      "loss": 0.0172,
+      "step": 78470
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9623123447715952e-05,
+      "loss": 0.0098,
+      "step": 78480
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962307540797748e-05,
+      "loss": 0.0154,
+      "step": 78490
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962302736823901e-05,
+      "loss": 0.0113,
+      "step": 78500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622979328500537e-05,
+      "loss": 0.0146,
+      "step": 78510
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622931288762066e-05,
+      "loss": 0.0139,
+      "step": 78520
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622883249023595e-05,
+      "loss": 0.0102,
+      "step": 78530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962283520928512e-05,
+      "loss": 0.0143,
+      "step": 78540
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.96227919735205e-05,
+      "loss": 0.0141,
+      "step": 78550
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962274393378203e-05,
+      "loss": 0.0177,
+      "step": 78560
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622695894043554e-05,
+      "loss": 0.0128,
+      "step": 78570
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622647854305084e-05,
+      "loss": 0.017,
+      "step": 78580
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622599814566613e-05,
+      "loss": 0.0133,
+      "step": 78590
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962255177482814e-05,
+      "loss": 0.0155,
+      "step": 78600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622503735089668e-05,
+      "loss": 0.012,
+      "step": 78610
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622455695351197e-05,
+      "loss": 0.0117,
+      "step": 78620
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622407655612727e-05,
+      "loss": 0.0138,
+      "step": 78630
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622359615874252e-05,
+      "loss": 0.0134,
+      "step": 78640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622311576135782e-05,
+      "loss": 0.0122,
+      "step": 78650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962226353639731e-05,
+      "loss": 0.0128,
+      "step": 78660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622215496658837e-05,
+      "loss": 0.0136,
+      "step": 78670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622167456920366e-05,
+      "loss": 0.0192,
+      "step": 78680
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622119417181895e-05,
+      "loss": 0.015,
+      "step": 78690
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9622071377443425e-05,
+      "loss": 0.0121,
+      "step": 78700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962202333770495e-05,
+      "loss": 0.012,
+      "step": 78710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962197529796648e-05,
+      "loss": 0.0158,
+      "step": 78720
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962192725822801e-05,
+      "loss": 0.0115,
+      "step": 78730
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.962187921848954e-05,
+      "loss": 0.0139,
+      "step": 78740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9621831178751064e-05,
+      "loss": 0.0174,
+      "step": 78750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9621783139012594e-05,
+      "loss": 0.0121,
+      "step": 78760
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9621735099274123e-05,
+      "loss": 0.0124,
+      "step": 78770
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.962168705953565e-05,
+      "loss": 0.0112,
+      "step": 78780
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9621639019797178e-05,
+      "loss": 0.0138,
+      "step": 78790
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9621590980058707e-05,
+      "loss": 0.0098,
+      "step": 78800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9621542940320236e-05,
+      "loss": 0.0108,
+      "step": 78810
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9621494900581762e-05,
+      "loss": 0.0113,
+      "step": 78820
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.962144686084329e-05,
+      "loss": 0.014,
+      "step": 78830
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.962139882110482e-05,
+      "loss": 0.0123,
+      "step": 78840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9621350781366347e-05,
+      "loss": 0.0139,
+      "step": 78850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9621302741627876e-05,
+      "loss": 0.017,
+      "step": 78860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9621254701889405e-05,
+      "loss": 0.0132,
+      "step": 78870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9621206662150935e-05,
+      "loss": 0.0112,
+      "step": 78880
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.962115862241246e-05,
+      "loss": 0.0112,
+      "step": 78890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.962111058267399e-05,
+      "loss": 0.0153,
+      "step": 78900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.962106254293552e-05,
+      "loss": 0.0124,
+      "step": 78910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9621014503197048e-05,
+      "loss": 0.0141,
+      "step": 78920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620966463458574e-05,
+      "loss": 0.0145,
+      "step": 78930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620918423720103e-05,
+      "loss": 0.0178,
+      "step": 78940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620870383981633e-05,
+      "loss": 0.0151,
+      "step": 78950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.962082234424316e-05,
+      "loss": 0.0106,
+      "step": 78960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620774304504688e-05,
+      "loss": 0.0126,
+      "step": 78970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620726264766217e-05,
+      "loss": 0.0068,
+      "step": 78980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620678225027746e-05,
+      "loss": 0.0153,
+      "step": 78990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620630185289272e-05,
+      "loss": 0.0138,
+      "step": 79000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.96205821455508e-05,
+      "loss": 0.0158,
+      "step": 79010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.962053410581233e-05,
+      "loss": 0.0099,
+      "step": 79020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620486066073857e-05,
+      "loss": 0.0104,
+      "step": 79030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620438026335386e-05,
+      "loss": 0.0154,
+      "step": 79040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620389986596915e-05,
+      "loss": 0.0134,
+      "step": 79050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620341946858445e-05,
+      "loss": 0.0139,
+      "step": 79060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.962029390711997e-05,
+      "loss": 0.0123,
+      "step": 79070
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.96202458673815e-05,
+      "loss": 0.0146,
+      "step": 79080
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.962019782764303e-05,
+      "loss": 0.0107,
+      "step": 79090
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620149787904558e-05,
+      "loss": 0.0139,
+      "step": 79100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620101748166084e-05,
+      "loss": 0.0127,
+      "step": 79110
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620053708427613e-05,
+      "loss": 0.0183,
+      "step": 79120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9620005668689143e-05,
+      "loss": 0.0128,
+      "step": 79130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.961995762895067e-05,
+      "loss": 0.0105,
+      "step": 79140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619909589212198e-05,
+      "loss": 0.0179,
+      "step": 79150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619861549473727e-05,
+      "loss": 0.0187,
+      "step": 79160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619813509735256e-05,
+      "loss": 0.0135,
+      "step": 79170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619765469996782e-05,
+      "loss": 0.0094,
+      "step": 79180
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.961971743025831e-05,
+      "loss": 0.016,
+      "step": 79190
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.961966939051984e-05,
+      "loss": 0.015,
+      "step": 79200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619621350781367e-05,
+      "loss": 0.0166,
+      "step": 79210
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619573311042896e-05,
+      "loss": 0.0123,
+      "step": 79220
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619525271304425e-05,
+      "loss": 0.0098,
+      "step": 79230
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619477231565954e-05,
+      "loss": 0.0151,
+      "step": 79240
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.961942919182748e-05,
+      "loss": 0.0128,
+      "step": 79250
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.961938115208901e-05,
+      "loss": 0.0104,
+      "step": 79260
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.961933311235054e-05,
+      "loss": 0.0099,
+      "step": 79270
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619285072612068e-05,
+      "loss": 0.0136,
+      "step": 79280
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619237032873594e-05,
+      "loss": 0.0072,
+      "step": 79290
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619188993135123e-05,
+      "loss": 0.0111,
+      "step": 79300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619140953396653e-05,
+      "loss": 0.0106,
+      "step": 79310
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.961909291365818e-05,
+      "loss": 0.0139,
+      "step": 79320
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9619044873919708e-05,
+      "loss": 0.0094,
+      "step": 79330
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9618996834181237e-05,
+      "loss": 0.0077,
+      "step": 79340
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9618948794442766e-05,
+      "loss": 0.0127,
+      "step": 79350
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9618900754704292e-05,
+      "loss": 0.0099,
+      "step": 79360
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.961885271496582e-05,
+      "loss": 0.0095,
+      "step": 79370
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.961880467522735e-05,
+      "loss": 0.0174,
+      "step": 79380
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9618756635488877e-05,
+      "loss": 0.01,
+      "step": 79390
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9618708595750406e-05,
+      "loss": 0.0115,
+      "step": 79400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9618660556011935e-05,
+      "loss": 0.0147,
+      "step": 79410
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9618612516273464e-05,
+      "loss": 0.0129,
+      "step": 79420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.961856447653499e-05,
+      "loss": 0.0151,
+      "step": 79430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.961851643679652e-05,
+      "loss": 0.0135,
+      "step": 79440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961846839705805e-05,
+      "loss": 0.0149,
+      "step": 79450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9618420357319578e-05,
+      "loss": 0.0118,
+      "step": 79460
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9618372317581104e-05,
+      "loss": 0.0159,
+      "step": 79470
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9618324277842633e-05,
+      "loss": 0.0115,
+      "step": 79480
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9618276238104162e-05,
+      "loss": 0.0104,
+      "step": 79490
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961822819836569e-05,
+      "loss": 0.0088,
+      "step": 79500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9618180158627218e-05,
+      "loss": 0.0143,
+      "step": 79510
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9618132118888747e-05,
+      "loss": 0.0133,
+      "step": 79520
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9618084079150276e-05,
+      "loss": 0.0116,
+      "step": 79530
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9618036039411802e-05,
+      "loss": 0.01,
+      "step": 79540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961798799967333e-05,
+      "loss": 0.0122,
+      "step": 79550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961793995993486e-05,
+      "loss": 0.0122,
+      "step": 79560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617891920196386e-05,
+      "loss": 0.0134,
+      "step": 79570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617843880457916e-05,
+      "loss": 0.0279,
+      "step": 79580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617795840719445e-05,
+      "loss": 0.0138,
+      "step": 79590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617747800980974e-05,
+      "loss": 0.0116,
+      "step": 79600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.96176997612425e-05,
+      "loss": 0.0122,
+      "step": 79610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961765172150403e-05,
+      "loss": 0.0123,
+      "step": 79620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961760368176556e-05,
+      "loss": 0.014,
+      "step": 79630
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617555642027088e-05,
+      "loss": 0.0113,
+      "step": 79640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617507602288614e-05,
+      "loss": 0.0162,
+      "step": 79650
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617459562550143e-05,
+      "loss": 0.0117,
+      "step": 79660
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617411522811672e-05,
+      "loss": 0.0108,
+      "step": 79670
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617363483073198e-05,
+      "loss": 0.0129,
+      "step": 79680
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617315443334728e-05,
+      "loss": 0.012,
+      "step": 79690
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617267403596257e-05,
+      "loss": 0.0118,
+      "step": 79700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617219363857786e-05,
+      "loss": 0.0101,
+      "step": 79710
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617171324119312e-05,
+      "loss": 0.0134,
+      "step": 79720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961712328438084e-05,
+      "loss": 0.0182,
+      "step": 79730
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961707524464237e-05,
+      "loss": 0.0131,
+      "step": 79740
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9617027204903896e-05,
+      "loss": 0.0117,
+      "step": 79750
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616979165165426e-05,
+      "loss": 0.0207,
+      "step": 79760
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616931125426955e-05,
+      "loss": 0.0153,
+      "step": 79770
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616883085688484e-05,
+      "loss": 0.0136,
+      "step": 79780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961683504595001e-05,
+      "loss": 0.0123,
+      "step": 79790
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961678700621154e-05,
+      "loss": 0.0127,
+      "step": 79800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961673896647307e-05,
+      "loss": 0.0102,
+      "step": 79810
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616690926734598e-05,
+      "loss": 0.0111,
+      "step": 79820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616642886996124e-05,
+      "loss": 0.0138,
+      "step": 79830
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616594847257653e-05,
+      "loss": 0.015,
+      "step": 79840
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616546807519182e-05,
+      "loss": 0.015,
+      "step": 79850
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616498767780708e-05,
+      "loss": 0.0155,
+      "step": 79860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616450728042237e-05,
+      "loss": 0.0118,
+      "step": 79870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616402688303767e-05,
+      "loss": 0.0098,
+      "step": 79880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616354648565296e-05,
+      "loss": 0.0117,
+      "step": 79890
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616306608826822e-05,
+      "loss": 0.011,
+      "step": 79900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961625856908835e-05,
+      "loss": 0.009,
+      "step": 79910
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961621052934988e-05,
+      "loss": 0.0134,
+      "step": 79920
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616162489611406e-05,
+      "loss": 0.0099,
+      "step": 79930
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616114449872936e-05,
+      "loss": 0.0107,
+      "step": 79940
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616066410134465e-05,
+      "loss": 0.0139,
+      "step": 79950
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9616018370395994e-05,
+      "loss": 0.0135,
+      "step": 79960
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961597033065752e-05,
+      "loss": 0.01,
+      "step": 79970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961592229091905e-05,
+      "loss": 0.0128,
+      "step": 79980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961587425118058e-05,
+      "loss": 0.0102,
+      "step": 79990
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615826211442108e-05,
+      "loss": 0.0104,
+      "step": 80000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615778171703634e-05,
+      "loss": 0.0104,
+      "step": 80010
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615730131965163e-05,
+      "loss": 0.0115,
+      "step": 80020
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615682092226692e-05,
+      "loss": 0.0121,
+      "step": 80030
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615634052488218e-05,
+      "loss": 0.0116,
+      "step": 80040
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615586012749747e-05,
+      "loss": 0.0111,
+      "step": 80050
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615537973011277e-05,
+      "loss": 0.0108,
+      "step": 80060
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615489933272806e-05,
+      "loss": 0.014,
+      "step": 80070
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615441893534332e-05,
+      "loss": 0.0142,
+      "step": 80080
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961539385379586e-05,
+      "loss": 0.0127,
+      "step": 80090
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.961534581405739e-05,
+      "loss": 0.017,
+      "step": 80100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615297774318916e-05,
+      "loss": 0.0122,
+      "step": 80110
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615249734580445e-05,
+      "loss": 0.0129,
+      "step": 80120
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615201694841975e-05,
+      "loss": 0.0133,
+      "step": 80130
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9615153655103504e-05,
+      "loss": 0.015,
+      "step": 80140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961510561536503e-05,
+      "loss": 0.0113,
+      "step": 80150
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961505757562656e-05,
+      "loss": 0.0196,
+      "step": 80160
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961500953588809e-05,
+      "loss": 0.0132,
+      "step": 80170
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614961496149614e-05,
+      "loss": 0.0142,
+      "step": 80180
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614913456411144e-05,
+      "loss": 0.0116,
+      "step": 80190
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614865416672673e-05,
+      "loss": 0.0107,
+      "step": 80200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614817376934202e-05,
+      "loss": 0.0147,
+      "step": 80210
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614769337195728e-05,
+      "loss": 0.0142,
+      "step": 80220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614721297457257e-05,
+      "loss": 0.0114,
+      "step": 80230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614673257718787e-05,
+      "loss": 0.0123,
+      "step": 80240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614625217980316e-05,
+      "loss": 0.0104,
+      "step": 80250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961457717824184e-05,
+      "loss": 0.0108,
+      "step": 80260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961452913850337e-05,
+      "loss": 0.015,
+      "step": 80270
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.96144810987649e-05,
+      "loss": 0.0119,
+      "step": 80280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614433059026426e-05,
+      "loss": 0.0141,
+      "step": 80290
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614385019287955e-05,
+      "loss": 0.0108,
+      "step": 80300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614336979549485e-05,
+      "loss": 0.0118,
+      "step": 80310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614288939811014e-05,
+      "loss": 0.0109,
+      "step": 80320
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961424090007254e-05,
+      "loss": 0.0125,
+      "step": 80330
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961419286033407e-05,
+      "loss": 0.0145,
+      "step": 80340
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.96141448205956e-05,
+      "loss": 0.0126,
+      "step": 80350
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614096780857124e-05,
+      "loss": 0.0111,
+      "step": 80360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614048741118653e-05,
+      "loss": 0.012,
+      "step": 80370
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9614000701380183e-05,
+      "loss": 0.0107,
+      "step": 80380
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613952661641712e-05,
+      "loss": 0.0105,
+      "step": 80390
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613904621903238e-05,
+      "loss": 0.0139,
+      "step": 80400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613856582164767e-05,
+      "loss": 0.0131,
+      "step": 80410
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613808542426296e-05,
+      "loss": 0.0144,
+      "step": 80420
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613760502687826e-05,
+      "loss": 0.0129,
+      "step": 80430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961371246294935e-05,
+      "loss": 0.0111,
+      "step": 80440
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961366442321088e-05,
+      "loss": 0.0174,
+      "step": 80450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961361638347241e-05,
+      "loss": 0.0131,
+      "step": 80460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613568343733936e-05,
+      "loss": 0.0131,
+      "step": 80470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613520303995465e-05,
+      "loss": 0.0147,
+      "step": 80480
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613472264256995e-05,
+      "loss": 0.0112,
+      "step": 80490
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613424224518524e-05,
+      "loss": 0.0134,
+      "step": 80500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961337618478005e-05,
+      "loss": 0.0128,
+      "step": 80510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961332814504158e-05,
+      "loss": 0.0164,
+      "step": 80520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613280105303108e-05,
+      "loss": 0.011,
+      "step": 80530
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613232065564634e-05,
+      "loss": 0.0089,
+      "step": 80540
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613184025826163e-05,
+      "loss": 0.0119,
+      "step": 80550
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613135986087693e-05,
+      "loss": 0.0115,
+      "step": 80560
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613087946349222e-05,
+      "loss": 0.0126,
+      "step": 80570
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9613039906610748e-05,
+      "loss": 0.0114,
+      "step": 80580
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612991866872277e-05,
+      "loss": 0.0099,
+      "step": 80590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612943827133806e-05,
+      "loss": 0.0116,
+      "step": 80600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612895787395336e-05,
+      "loss": 0.0169,
+      "step": 80610
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961284774765686e-05,
+      "loss": 0.0111,
+      "step": 80620
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961279970791839e-05,
+      "loss": 0.0102,
+      "step": 80630
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961275166817992e-05,
+      "loss": 0.0145,
+      "step": 80640
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612703628441446e-05,
+      "loss": 0.0113,
+      "step": 80650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612655588702975e-05,
+      "loss": 0.0147,
+      "step": 80660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612607548964504e-05,
+      "loss": 0.0156,
+      "step": 80670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612559509226034e-05,
+      "loss": 0.0136,
+      "step": 80680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961251146948756e-05,
+      "loss": 0.0145,
+      "step": 80690
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961246342974909e-05,
+      "loss": 0.0102,
+      "step": 80700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612415390010618e-05,
+      "loss": 0.0131,
+      "step": 80710
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612367350272144e-05,
+      "loss": 0.0115,
+      "step": 80720
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612319310533673e-05,
+      "loss": 0.0125,
+      "step": 80730
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612271270795203e-05,
+      "loss": 0.0188,
+      "step": 80740
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612223231056732e-05,
+      "loss": 0.0142,
+      "step": 80750
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612175191318258e-05,
+      "loss": 0.0147,
+      "step": 80760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612127151579787e-05,
+      "loss": 0.0141,
+      "step": 80770
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612079111841316e-05,
+      "loss": 0.016,
+      "step": 80780
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9612031072102846e-05,
+      "loss": 0.008,
+      "step": 80790
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9611983032364375e-05,
+      "loss": 0.0118,
+      "step": 80800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9611934992625904e-05,
+      "loss": 0.0135,
+      "step": 80810
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961188695288743e-05,
+      "loss": 0.0135,
+      "step": 80820
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.961183891314896e-05,
+      "loss": 0.0128,
+      "step": 80830
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.961179087341049e-05,
+      "loss": 0.009,
+      "step": 80840
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611742833672018e-05,
+      "loss": 0.0096,
+      "step": 80850
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611694793933544e-05,
+      "loss": 0.0146,
+      "step": 80860
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611646754195073e-05,
+      "loss": 0.0119,
+      "step": 80870
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611598714456602e-05,
+      "loss": 0.0115,
+      "step": 80880
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611550674718128e-05,
+      "loss": 0.0101,
+      "step": 80890
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611502634979657e-05,
+      "loss": 0.0156,
+      "step": 80900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611454595241187e-05,
+      "loss": 0.0142,
+      "step": 80910
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611406555502716e-05,
+      "loss": 0.0108,
+      "step": 80920
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611358515764242e-05,
+      "loss": 0.0133,
+      "step": 80930
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.961131047602577e-05,
+      "loss": 0.0149,
+      "step": 80940
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.96112624362873e-05,
+      "loss": 0.0169,
+      "step": 80950
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611214396548826e-05,
+      "loss": 0.0106,
+      "step": 80960
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611166356810355e-05,
+      "loss": 0.0154,
+      "step": 80970
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611118317071885e-05,
+      "loss": 0.0095,
+      "step": 80980
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9611070277333414e-05,
+      "loss": 0.0106,
+      "step": 80990
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.961102223759494e-05,
+      "loss": 0.0108,
+      "step": 81000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.961097419785647e-05,
+      "loss": 0.0098,
+      "step": 81010
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610926158118e-05,
+      "loss": 0.0157,
+      "step": 81020
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610878118379528e-05,
+      "loss": 0.0106,
+      "step": 81030
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610830078641054e-05,
+      "loss": 0.0168,
+      "step": 81040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610782038902583e-05,
+      "loss": 0.0109,
+      "step": 81050
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610733999164112e-05,
+      "loss": 0.02,
+      "step": 81060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610685959425638e-05,
+      "loss": 0.0187,
+      "step": 81070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610637919687167e-05,
+      "loss": 0.0121,
+      "step": 81080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610589879948697e-05,
+      "loss": 0.0125,
+      "step": 81090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610541840210226e-05,
+      "loss": 0.0153,
+      "step": 81100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.961049380047175e-05,
+      "loss": 0.0106,
+      "step": 81110
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.961044576073328e-05,
+      "loss": 0.0109,
+      "step": 81120
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.961039772099481e-05,
+      "loss": 0.0138,
+      "step": 81130
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610349681256336e-05,
+      "loss": 0.0163,
+      "step": 81140
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610301641517865e-05,
+      "loss": 0.0123,
+      "step": 81150
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610253601779395e-05,
+      "loss": 0.0153,
+      "step": 81160
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610205562040924e-05,
+      "loss": 0.0221,
+      "step": 81170
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.961015752230245e-05,
+      "loss": 0.0123,
+      "step": 81180
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.961010948256398e-05,
+      "loss": 0.0118,
+      "step": 81190
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.961006144282551e-05,
+      "loss": 0.0131,
+      "step": 81200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610013403087038e-05,
+      "loss": 0.012,
+      "step": 81210
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609965363348563e-05,
+      "loss": 0.0151,
+      "step": 81220
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609917323610093e-05,
+      "loss": 0.0156,
+      "step": 81230
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609869283871622e-05,
+      "loss": 0.0139,
+      "step": 81240
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609821244133148e-05,
+      "loss": 0.0103,
+      "step": 81250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609773204394677e-05,
+      "loss": 0.0097,
+      "step": 81260
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609725164656206e-05,
+      "loss": 0.0133,
+      "step": 81270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609677124917736e-05,
+      "loss": 0.0134,
+      "step": 81280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.960962908517926e-05,
+      "loss": 0.0129,
+      "step": 81290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.960958104544079e-05,
+      "loss": 0.0119,
+      "step": 81300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.960953300570232e-05,
+      "loss": 0.0151,
+      "step": 81310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609484965963846e-05,
+      "loss": 0.0121,
+      "step": 81320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609436926225375e-05,
+      "loss": 0.0168,
+      "step": 81330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609388886486905e-05,
+      "loss": 0.0102,
+      "step": 81340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609340846748434e-05,
+      "loss": 0.0115,
+      "step": 81350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.960929280700996e-05,
+      "loss": 0.0107,
+      "step": 81360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.960924476727149e-05,
+      "loss": 0.02,
+      "step": 81370
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609196727533018e-05,
+      "loss": 0.0111,
+      "step": 81380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609148687794548e-05,
+      "loss": 0.0193,
+      "step": 81390
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609100648056073e-05,
+      "loss": 0.0112,
+      "step": 81400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609052608317603e-05,
+      "loss": 0.0153,
+      "step": 81410
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9609004568579132e-05,
+      "loss": 0.0094,
+      "step": 81420
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9608956528840658e-05,
+      "loss": 0.016,
+      "step": 81430
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9608908489102187e-05,
+      "loss": 0.0113,
+      "step": 81440
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9608860449363716e-05,
+      "loss": 0.0145,
+      "step": 81450
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9608812409625246e-05,
+      "loss": 0.0158,
+      "step": 81460
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.960876436988677e-05,
+      "loss": 0.0139,
+      "step": 81470
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.96087163301483e-05,
+      "loss": 0.016,
+      "step": 81480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.960866829040983e-05,
+      "loss": 0.0137,
+      "step": 81490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9608620250671356e-05,
+      "loss": 0.0139,
+      "step": 81500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9608572210932885e-05,
+      "loss": 0.0136,
+      "step": 81510
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9608524171194414e-05,
+      "loss": 0.0143,
+      "step": 81520
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9608476131455944e-05,
+      "loss": 0.0152,
+      "step": 81530
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960842809171747e-05,
+      "loss": 0.0128,
+      "step": 81540
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9608380051979e-05,
+      "loss": 0.0132,
+      "step": 81550
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9608332012240528e-05,
+      "loss": 0.013,
+      "step": 81560
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9608283972502054e-05,
+      "loss": 0.0148,
+      "step": 81570
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9608235932763583e-05,
+      "loss": 0.015,
+      "step": 81580
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9608187893025113e-05,
+      "loss": 0.0113,
+      "step": 81590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9608139853286642e-05,
+      "loss": 0.0106,
+      "step": 81600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9608091813548168e-05,
+      "loss": 0.0153,
+      "step": 81610
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9608043773809697e-05,
+      "loss": 0.0099,
+      "step": 81620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607995734071226e-05,
+      "loss": 0.0123,
+      "step": 81630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607947694332756e-05,
+      "loss": 0.0112,
+      "step": 81640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960789965459428e-05,
+      "loss": 0.0111,
+      "step": 81650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960785161485581e-05,
+      "loss": 0.0079,
+      "step": 81660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960780357511734e-05,
+      "loss": 0.0098,
+      "step": 81670
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607755535378866e-05,
+      "loss": 0.0143,
+      "step": 81680
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607707495640395e-05,
+      "loss": 0.0142,
+      "step": 81690
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607659455901924e-05,
+      "loss": 0.0141,
+      "step": 81700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607611416163454e-05,
+      "loss": 0.0122,
+      "step": 81710
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960756337642498e-05,
+      "loss": 0.0114,
+      "step": 81720
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960751533668651e-05,
+      "loss": 0.0123,
+      "step": 81730
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607467296948038e-05,
+      "loss": 0.0123,
+      "step": 81740
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607419257209564e-05,
+      "loss": 0.016,
+      "step": 81750
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607371217471093e-05,
+      "loss": 0.0129,
+      "step": 81760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607323177732622e-05,
+      "loss": 0.0146,
+      "step": 81770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607275137994152e-05,
+      "loss": 0.0139,
+      "step": 81780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607227098255678e-05,
+      "loss": 0.0166,
+      "step": 81790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607179058517207e-05,
+      "loss": 0.0098,
+      "step": 81800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607131018778736e-05,
+      "loss": 0.0102,
+      "step": 81810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9607082979040265e-05,
+      "loss": 0.0148,
+      "step": 81820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960703493930179e-05,
+      "loss": 0.0152,
+      "step": 81830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960698689956332e-05,
+      "loss": 0.0132,
+      "step": 81840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960693885982485e-05,
+      "loss": 0.0136,
+      "step": 81850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606890820086376e-05,
+      "loss": 0.0148,
+      "step": 81860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606842780347905e-05,
+      "loss": 0.0157,
+      "step": 81870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606794740609434e-05,
+      "loss": 0.0126,
+      "step": 81880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606746700870964e-05,
+      "loss": 0.01,
+      "step": 81890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960669866113249e-05,
+      "loss": 0.012,
+      "step": 81900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960665062139402e-05,
+      "loss": 0.0157,
+      "step": 81910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606602581655548e-05,
+      "loss": 0.0107,
+      "step": 81920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606554541917074e-05,
+      "loss": 0.0132,
+      "step": 81930
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606506502178603e-05,
+      "loss": 0.0171,
+      "step": 81940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606458462440132e-05,
+      "loss": 0.0126,
+      "step": 81950
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960641042270166e-05,
+      "loss": 0.0119,
+      "step": 81960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606362382963188e-05,
+      "loss": 0.0159,
+      "step": 81970
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606314343224717e-05,
+      "loss": 0.0107,
+      "step": 81980
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606266303486246e-05,
+      "loss": 0.0158,
+      "step": 81990
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606218263747775e-05,
+      "loss": 0.016,
+      "step": 82000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.96061702240093e-05,
+      "loss": 0.0086,
+      "step": 82010
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960612218427083e-05,
+      "loss": 0.0116,
+      "step": 82020
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960607414453236e-05,
+      "loss": 0.0103,
+      "step": 82030
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9606026104793886e-05,
+      "loss": 0.0166,
+      "step": 82040
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9605978065055415e-05,
+      "loss": 0.0172,
+      "step": 82050
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9605930025316944e-05,
+      "loss": 0.0139,
+      "step": 82060
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9605881985578473e-05,
+      "loss": 0.0141,
+      "step": 82070
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960583394584e-05,
+      "loss": 0.0096,
+      "step": 82080
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960578590610153e-05,
+      "loss": 0.0152,
+      "step": 82090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9605737866363058e-05,
+      "loss": 0.0128,
+      "step": 82100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9605689826624584e-05,
+      "loss": 0.0186,
+      "step": 82110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9605641786886113e-05,
+      "loss": 0.0138,
+      "step": 82120
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9605593747147642e-05,
+      "loss": 0.0101,
+      "step": 82130
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960554570740917e-05,
+      "loss": 0.0114,
+      "step": 82140
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9605497667670697e-05,
+      "loss": 0.0124,
+      "step": 82150
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9605449627932227e-05,
+      "loss": 0.0158,
+      "step": 82160
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9605401588193756e-05,
+      "loss": 0.0122,
+      "step": 82170
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9605353548455285e-05,
+      "loss": 0.0114,
+      "step": 82180
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960530550871681e-05,
+      "loss": 0.0095,
+      "step": 82190
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960525746897834e-05,
+      "loss": 0.0114,
+      "step": 82200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.960520942923987e-05,
+      "loss": 0.0106,
+      "step": 82210
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9605161389501396e-05,
+      "loss": 0.0112,
+      "step": 82220
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9605113349762925e-05,
+      "loss": 0.0141,
+      "step": 82230
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9605065310024454e-05,
+      "loss": 0.0118,
+      "step": 82240
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9605017270285983e-05,
+      "loss": 0.011,
+      "step": 82250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960496923054751e-05,
+      "loss": 0.0119,
+      "step": 82260
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960492119080904e-05,
+      "loss": 0.0119,
+      "step": 82270
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604873151070568e-05,
+      "loss": 0.0131,
+      "step": 82280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604825111332094e-05,
+      "loss": 0.0102,
+      "step": 82290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604777071593623e-05,
+      "loss": 0.0098,
+      "step": 82300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604729031855152e-05,
+      "loss": 0.0115,
+      "step": 82310
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960468099211668e-05,
+      "loss": 0.0128,
+      "step": 82320
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604632952378207e-05,
+      "loss": 0.012,
+      "step": 82330
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604584912639737e-05,
+      "loss": 0.0131,
+      "step": 82340
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604536872901266e-05,
+      "loss": 0.0133,
+      "step": 82350
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604488833162795e-05,
+      "loss": 0.0138,
+      "step": 82360
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960444079342432e-05,
+      "loss": 0.0121,
+      "step": 82370
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960439275368585e-05,
+      "loss": 0.0126,
+      "step": 82380
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960434471394738e-05,
+      "loss": 0.0134,
+      "step": 82390
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604296674208906e-05,
+      "loss": 0.0122,
+      "step": 82400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604248634470435e-05,
+      "loss": 0.0101,
+      "step": 82410
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604200594731964e-05,
+      "loss": 0.0155,
+      "step": 82420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604152554993493e-05,
+      "loss": 0.0168,
+      "step": 82430
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960410451525502e-05,
+      "loss": 0.0104,
+      "step": 82440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960405647551655e-05,
+      "loss": 0.0177,
+      "step": 82450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9604008435778078e-05,
+      "loss": 0.0098,
+      "step": 82460
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603960396039604e-05,
+      "loss": 0.0148,
+      "step": 82470
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603912356301133e-05,
+      "loss": 0.0186,
+      "step": 82480
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603864316562662e-05,
+      "loss": 0.014,
+      "step": 82490
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960381627682419e-05,
+      "loss": 0.0126,
+      "step": 82500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603768237085717e-05,
+      "loss": 0.0188,
+      "step": 82510
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603720197347247e-05,
+      "loss": 0.0119,
+      "step": 82520
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603672157608776e-05,
+      "loss": 0.0126,
+      "step": 82530
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603624117870305e-05,
+      "loss": 0.0128,
+      "step": 82540
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960357607813183e-05,
+      "loss": 0.0143,
+      "step": 82550
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960352803839336e-05,
+      "loss": 0.0211,
+      "step": 82560
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960347999865489e-05,
+      "loss": 0.0093,
+      "step": 82570
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603431958916415e-05,
+      "loss": 0.0197,
+      "step": 82580
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603383919177945e-05,
+      "loss": 0.0115,
+      "step": 82590
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603335879439474e-05,
+      "loss": 0.0177,
+      "step": 82600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603287839701003e-05,
+      "loss": 0.0088,
+      "step": 82610
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960323979996253e-05,
+      "loss": 0.0147,
+      "step": 82620
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960319176022406e-05,
+      "loss": 0.0142,
+      "step": 82630
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603143720485588e-05,
+      "loss": 0.0092,
+      "step": 82640
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603095680747114e-05,
+      "loss": 0.0125,
+      "step": 82650
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9603047641008643e-05,
+      "loss": 0.0115,
+      "step": 82660
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602999601270172e-05,
+      "loss": 0.0103,
+      "step": 82670
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.96029515615317e-05,
+      "loss": 0.0123,
+      "step": 82680
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602903521793227e-05,
+      "loss": 0.0136,
+      "step": 82690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602855482054756e-05,
+      "loss": 0.0116,
+      "step": 82700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602807442316286e-05,
+      "loss": 0.0116,
+      "step": 82710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602759402577815e-05,
+      "loss": 0.0162,
+      "step": 82720
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960271136283934e-05,
+      "loss": 0.021,
+      "step": 82730
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960266332310087e-05,
+      "loss": 0.0157,
+      "step": 82740
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.96026152833624e-05,
+      "loss": 0.0134,
+      "step": 82750
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602567243623925e-05,
+      "loss": 0.017,
+      "step": 82760
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602519203885455e-05,
+      "loss": 0.0186,
+      "step": 82770
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602471164146984e-05,
+      "loss": 0.0146,
+      "step": 82780
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602423124408513e-05,
+      "loss": 0.0101,
+      "step": 82790
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960237508467004e-05,
+      "loss": 0.014,
+      "step": 82800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602327044931568e-05,
+      "loss": 0.0124,
+      "step": 82810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602279005193098e-05,
+      "loss": 0.0127,
+      "step": 82820
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602230965454623e-05,
+      "loss": 0.0067,
+      "step": 82830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602182925716153e-05,
+      "loss": 0.0158,
+      "step": 82840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602134885977682e-05,
+      "loss": 0.0125,
+      "step": 82850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960208684623921e-05,
+      "loss": 0.0101,
+      "step": 82860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9602038806500737e-05,
+      "loss": 0.0174,
+      "step": 82870
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9601990766762266e-05,
+      "loss": 0.0122,
+      "step": 82880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9601942727023796e-05,
+      "loss": 0.015,
+      "step": 82890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9601894687285325e-05,
+      "loss": 0.0122,
+      "step": 82900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.960184664754685e-05,
+      "loss": 0.0157,
+      "step": 82910
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.960179860780838e-05,
+      "loss": 0.0107,
+      "step": 82920
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.960175056806991e-05,
+      "loss": 0.0138,
+      "step": 82930
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601702528331435e-05,
+      "loss": 0.0135,
+      "step": 82940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601654488592965e-05,
+      "loss": 0.0116,
+      "step": 82950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601606448854494e-05,
+      "loss": 0.0147,
+      "step": 82960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601558409116023e-05,
+      "loss": 0.0128,
+      "step": 82970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.960151036937755e-05,
+      "loss": 0.022,
+      "step": 82980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601462329639078e-05,
+      "loss": 0.0123,
+      "step": 82990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601414289900607e-05,
+      "loss": 0.0164,
+      "step": 83000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601366250162133e-05,
+      "loss": 0.0159,
+      "step": 83010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601318210423663e-05,
+      "loss": 0.0157,
+      "step": 83020
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601270170685192e-05,
+      "loss": 0.0108,
+      "step": 83030
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.960122213094672e-05,
+      "loss": 0.0125,
+      "step": 83040
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601174091208247e-05,
+      "loss": 0.0142,
+      "step": 83050
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601126051469776e-05,
+      "loss": 0.0141,
+      "step": 83060
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601078011731306e-05,
+      "loss": 0.0121,
+      "step": 83070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9601029971992835e-05,
+      "loss": 0.0097,
+      "step": 83080
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.960098193225436e-05,
+      "loss": 0.015,
+      "step": 83090
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.960093389251589e-05,
+      "loss": 0.0119,
+      "step": 83100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.960088585277742e-05,
+      "loss": 0.0124,
+      "step": 83110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600837813038945e-05,
+      "loss": 0.0146,
+      "step": 83120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600789773300474e-05,
+      "loss": 0.012,
+      "step": 83130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600741733562004e-05,
+      "loss": 0.0158,
+      "step": 83140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600693693823533e-05,
+      "loss": 0.0121,
+      "step": 83150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.960064565408506e-05,
+      "loss": 0.0128,
+      "step": 83160
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600597614346588e-05,
+      "loss": 0.0122,
+      "step": 83170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600549574608117e-05,
+      "loss": 0.0129,
+      "step": 83180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600501534869643e-05,
+      "loss": 0.0135,
+      "step": 83190
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600453495131173e-05,
+      "loss": 0.0153,
+      "step": 83200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600405455392702e-05,
+      "loss": 0.0122,
+      "step": 83210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.960035741565423e-05,
+      "loss": 0.0191,
+      "step": 83220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600309375915757e-05,
+      "loss": 0.0077,
+      "step": 83230
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600261336177286e-05,
+      "loss": 0.0115,
+      "step": 83240
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600213296438816e-05,
+      "loss": 0.014,
+      "step": 83250
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9600165256700345e-05,
+      "loss": 0.0121,
+      "step": 83260
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.960011721696187e-05,
+      "loss": 0.0136,
+      "step": 83270
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.96000691772234e-05,
+      "loss": 0.0113,
+      "step": 83280
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.960002113748493e-05,
+      "loss": 0.01,
+      "step": 83290
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599973097746455e-05,
+      "loss": 0.0107,
+      "step": 83300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599925058007984e-05,
+      "loss": 0.0134,
+      "step": 83310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599877018269514e-05,
+      "loss": 0.0108,
+      "step": 83320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599828978531043e-05,
+      "loss": 0.0135,
+      "step": 83330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.959978093879257e-05,
+      "loss": 0.0134,
+      "step": 83340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.95997328990541e-05,
+      "loss": 0.018,
+      "step": 83350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599684859315627e-05,
+      "loss": 0.0114,
+      "step": 83360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599636819577157e-05,
+      "loss": 0.011,
+      "step": 83370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599588779838686e-05,
+      "loss": 0.0127,
+      "step": 83380
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599540740100215e-05,
+      "loss": 0.014,
+      "step": 83390
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.959949270036174e-05,
+      "loss": 0.0112,
+      "step": 83400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.959944466062327e-05,
+      "loss": 0.0127,
+      "step": 83410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.95993966208848e-05,
+      "loss": 0.0103,
+      "step": 83420
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599348581146325e-05,
+      "loss": 0.0117,
+      "step": 83430
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599300541407855e-05,
+      "loss": 0.0089,
+      "step": 83440
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599252501669384e-05,
+      "loss": 0.0116,
+      "step": 83450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599204461930913e-05,
+      "loss": 0.0136,
+      "step": 83460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.959915642219244e-05,
+      "loss": 0.0136,
+      "step": 83470
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.959910838245397e-05,
+      "loss": 0.0142,
+      "step": 83480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599060342715498e-05,
+      "loss": 0.0103,
+      "step": 83490
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9599012302977024e-05,
+      "loss": 0.0129,
+      "step": 83500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9598964263238553e-05,
+      "loss": 0.0139,
+      "step": 83510
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9598916223500082e-05,
+      "loss": 0.0118,
+      "step": 83520
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.959886818376161e-05,
+      "loss": 0.0095,
+      "step": 83530
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9598820144023137e-05,
+      "loss": 0.0135,
+      "step": 83540
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9598772104284666e-05,
+      "loss": 0.0125,
+      "step": 83550
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9598724064546196e-05,
+      "loss": 0.0109,
+      "step": 83560
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9598676024807725e-05,
+      "loss": 0.0157,
+      "step": 83570
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.959862798506925e-05,
+      "loss": 0.0125,
+      "step": 83580
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.959857994533078e-05,
+      "loss": 0.0109,
+      "step": 83590
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.959853190559231e-05,
+      "loss": 0.0108,
+      "step": 83600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9598483865853835e-05,
+      "loss": 0.0118,
+      "step": 83610
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9598435826115365e-05,
+      "loss": 0.0127,
+      "step": 83620
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959839259035074e-05,
+      "loss": 0.0136,
+      "step": 83630
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959834455061227e-05,
+      "loss": 0.0131,
+      "step": 83640
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9598296510873794e-05,
+      "loss": 0.011,
+      "step": 83650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9598248471135324e-05,
+      "loss": 0.0136,
+      "step": 83660
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9598200431396853e-05,
+      "loss": 0.015,
+      "step": 83670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959815239165838e-05,
+      "loss": 0.0162,
+      "step": 83680
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9598104351919908e-05,
+      "loss": 0.0112,
+      "step": 83690
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9598056312181437e-05,
+      "loss": 0.0139,
+      "step": 83700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9598008272442967e-05,
+      "loss": 0.0122,
+      "step": 83710
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9597960232704493e-05,
+      "loss": 0.0115,
+      "step": 83720
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9597912192966022e-05,
+      "loss": 0.0122,
+      "step": 83730
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959786415322755e-05,
+      "loss": 0.0107,
+      "step": 83740
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959781611348908e-05,
+      "loss": 0.0103,
+      "step": 83750
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9597768073750606e-05,
+      "loss": 0.0108,
+      "step": 83760
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9597720034012135e-05,
+      "loss": 0.008,
+      "step": 83770
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9597671994273665e-05,
+      "loss": 0.0143,
+      "step": 83780
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959762395453519e-05,
+      "loss": 0.0097,
+      "step": 83790
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959757591479672e-05,
+      "loss": 0.0151,
+      "step": 83800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959752787505825e-05,
+      "loss": 0.0151,
+      "step": 83810
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959747983531978e-05,
+      "loss": 0.0161,
+      "step": 83820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9597431795581304e-05,
+      "loss": 0.012,
+      "step": 83830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9597383755842834e-05,
+      "loss": 0.0127,
+      "step": 83840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9597335716104363e-05,
+      "loss": 0.0171,
+      "step": 83850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9597287676365892e-05,
+      "loss": 0.0121,
+      "step": 83860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959723963662742e-05,
+      "loss": 0.0116,
+      "step": 83870
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959719159688895e-05,
+      "loss": 0.0103,
+      "step": 83880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9597143557150477e-05,
+      "loss": 0.0129,
+      "step": 83890
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9597095517412006e-05,
+      "loss": 0.0112,
+      "step": 83900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9597047477673535e-05,
+      "loss": 0.0115,
+      "step": 83910
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959699943793506e-05,
+      "loss": 0.0139,
+      "step": 83920
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959695139819659e-05,
+      "loss": 0.0129,
+      "step": 83930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959690335845812e-05,
+      "loss": 0.0144,
+      "step": 83940
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959685531871965e-05,
+      "loss": 0.007,
+      "step": 83950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9596807278981175e-05,
+      "loss": 0.0155,
+      "step": 83960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9596759239242704e-05,
+      "loss": 0.0088,
+      "step": 83970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9596711199504233e-05,
+      "loss": 0.0124,
+      "step": 83980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9596663159765762e-05,
+      "loss": 0.0115,
+      "step": 83990
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959661512002729e-05,
+      "loss": 0.0136,
+      "step": 84000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9596567080288818e-05,
+      "loss": 0.0129,
+      "step": 84010
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9596519040550347e-05,
+      "loss": 0.0117,
+      "step": 84020
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9596471000811873e-05,
+      "loss": 0.0102,
+      "step": 84030
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9596422961073402e-05,
+      "loss": 0.0159,
+      "step": 84040
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959637492133493e-05,
+      "loss": 0.0096,
+      "step": 84050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959632688159646e-05,
+      "loss": 0.0115,
+      "step": 84060
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9596278841857986e-05,
+      "loss": 0.012,
+      "step": 84070
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9596230802119516e-05,
+      "loss": 0.0082,
+      "step": 84080
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9596182762381045e-05,
+      "loss": 0.0095,
+      "step": 84090
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959613472264257e-05,
+      "loss": 0.015,
+      "step": 84100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.95960866829041e-05,
+      "loss": 0.0146,
+      "step": 84110
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959603864316563e-05,
+      "loss": 0.0133,
+      "step": 84120
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959599060342716e-05,
+      "loss": 0.0136,
+      "step": 84130
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9595942563688685e-05,
+      "loss": 0.0141,
+      "step": 84140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9595894523950214e-05,
+      "loss": 0.0122,
+      "step": 84150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9595846484211743e-05,
+      "loss": 0.0145,
+      "step": 84160
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9595798444473272e-05,
+      "loss": 0.0131,
+      "step": 84170
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9595750404734798e-05,
+      "loss": 0.009,
+      "step": 84180
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9595702364996328e-05,
+      "loss": 0.0148,
+      "step": 84190
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9595654325257857e-05,
+      "loss": 0.0122,
+      "step": 84200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9595606285519383e-05,
+      "loss": 0.0134,
+      "step": 84210
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9595558245780912e-05,
+      "loss": 0.0114,
+      "step": 84220
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959551020604244e-05,
+      "loss": 0.0124,
+      "step": 84230
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959546216630397e-05,
+      "loss": 0.011,
+      "step": 84240
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9595414126565496e-05,
+      "loss": 0.0095,
+      "step": 84250
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9595366086827026e-05,
+      "loss": 0.0101,
+      "step": 84260
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9595318047088555e-05,
+      "loss": 0.0099,
+      "step": 84270
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959527000735008e-05,
+      "loss": 0.0104,
+      "step": 84280
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959522196761161e-05,
+      "loss": 0.0159,
+      "step": 84290
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.959517392787314e-05,
+      "loss": 0.0176,
+      "step": 84300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959512588813467e-05,
+      "loss": 0.0143,
+      "step": 84310
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9595077848396194e-05,
+      "loss": 0.0114,
+      "step": 84320
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9595029808657724e-05,
+      "loss": 0.0125,
+      "step": 84330
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594981768919253e-05,
+      "loss": 0.0114,
+      "step": 84340
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594933729180782e-05,
+      "loss": 0.0116,
+      "step": 84350
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594885689442308e-05,
+      "loss": 0.0149,
+      "step": 84360
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594837649703837e-05,
+      "loss": 0.0105,
+      "step": 84370
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594789609965367e-05,
+      "loss": 0.0105,
+      "step": 84380
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594741570226893e-05,
+      "loss": 0.0109,
+      "step": 84390
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594693530488422e-05,
+      "loss": 0.0146,
+      "step": 84400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959464549074995e-05,
+      "loss": 0.0149,
+      "step": 84410
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959459745101148e-05,
+      "loss": 0.0149,
+      "step": 84420
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594549411273006e-05,
+      "loss": 0.0113,
+      "step": 84430
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594501371534536e-05,
+      "loss": 0.0149,
+      "step": 84440
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594453331796065e-05,
+      "loss": 0.0122,
+      "step": 84450
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959440529205759e-05,
+      "loss": 0.0142,
+      "step": 84460
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959435725231912e-05,
+      "loss": 0.0111,
+      "step": 84470
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959430921258065e-05,
+      "loss": 0.0137,
+      "step": 84480
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959426117284218e-05,
+      "loss": 0.0105,
+      "step": 84490
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594213133103704e-05,
+      "loss": 0.0151,
+      "step": 84500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594165093365234e-05,
+      "loss": 0.0075,
+      "step": 84510
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594117053626763e-05,
+      "loss": 0.0149,
+      "step": 84520
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959406901388829e-05,
+      "loss": 0.0127,
+      "step": 84530
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9594020974149818e-05,
+      "loss": 0.0109,
+      "step": 84540
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593972934411347e-05,
+      "loss": 0.0101,
+      "step": 84550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593924894672877e-05,
+      "loss": 0.0176,
+      "step": 84560
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593876854934403e-05,
+      "loss": 0.0125,
+      "step": 84570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593828815195932e-05,
+      "loss": 0.0156,
+      "step": 84580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959378077545746e-05,
+      "loss": 0.0088,
+      "step": 84590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959373273571899e-05,
+      "loss": 0.0153,
+      "step": 84600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593684695980516e-05,
+      "loss": 0.0127,
+      "step": 84610
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593636656242045e-05,
+      "loss": 0.0119,
+      "step": 84620
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593588616503575e-05,
+      "loss": 0.016,
+      "step": 84630
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.95935405767651e-05,
+      "loss": 0.0107,
+      "step": 84640
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959349253702663e-05,
+      "loss": 0.0166,
+      "step": 84650
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959344449728816e-05,
+      "loss": 0.0158,
+      "step": 84660
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959339645754969e-05,
+      "loss": 0.014,
+      "step": 84670
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593348417811214e-05,
+      "loss": 0.0139,
+      "step": 84680
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593300378072744e-05,
+      "loss": 0.0114,
+      "step": 84690
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593252338334273e-05,
+      "loss": 0.0177,
+      "step": 84700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.95932042985958e-05,
+      "loss": 0.0167,
+      "step": 84710
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593156258857328e-05,
+      "loss": 0.0167,
+      "step": 84720
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593108219118857e-05,
+      "loss": 0.0141,
+      "step": 84730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593060179380387e-05,
+      "loss": 0.0111,
+      "step": 84740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9593012139641912e-05,
+      "loss": 0.0134,
+      "step": 84750
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9592964099903442e-05,
+      "loss": 0.0092,
+      "step": 84760
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959291606016497e-05,
+      "loss": 0.0118,
+      "step": 84770
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.95928680204265e-05,
+      "loss": 0.0145,
+      "step": 84780
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9592819980688026e-05,
+      "loss": 0.0107,
+      "step": 84790
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9592771940949555e-05,
+      "loss": 0.0122,
+      "step": 84800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9592723901211085e-05,
+      "loss": 0.0108,
+      "step": 84810
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959267586147261e-05,
+      "loss": 0.0115,
+      "step": 84820
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959262782173414e-05,
+      "loss": 0.0144,
+      "step": 84830
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959257978199567e-05,
+      "loss": 0.0102,
+      "step": 84840
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.95925317422572e-05,
+      "loss": 0.0118,
+      "step": 84850
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9592483702518724e-05,
+      "loss": 0.011,
+      "step": 84860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9592435662780254e-05,
+      "loss": 0.0101,
+      "step": 84870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9592387623041783e-05,
+      "loss": 0.0128,
+      "step": 84880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959233958330331e-05,
+      "loss": 0.0108,
+      "step": 84890
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9592291543564838e-05,
+      "loss": 0.0146,
+      "step": 84900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9592243503826367e-05,
+      "loss": 0.0093,
+      "step": 84910
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9592195464087896e-05,
+      "loss": 0.0099,
+      "step": 84920
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9592147424349422e-05,
+      "loss": 0.0145,
+      "step": 84930
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959209938461095e-05,
+      "loss": 0.012,
+      "step": 84940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959205134487248e-05,
+      "loss": 0.0144,
+      "step": 84950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.959200330513401e-05,
+      "loss": 0.0126,
+      "step": 84960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9591955265395536e-05,
+      "loss": 0.0118,
+      "step": 84970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9591907225657065e-05,
+      "loss": 0.0165,
+      "step": 84980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9591859185918595e-05,
+      "loss": 0.0123,
+      "step": 84990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959181114618012e-05,
+      "loss": 0.0109,
+      "step": 85000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959176310644165e-05,
+      "loss": 0.009,
+      "step": 85010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959171506670318e-05,
+      "loss": 0.0195,
+      "step": 85020
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9591667026964708e-05,
+      "loss": 0.0131,
+      "step": 85030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9591618987226234e-05,
+      "loss": 0.0117,
+      "step": 85040
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9591570947487763e-05,
+      "loss": 0.0153,
+      "step": 85050
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9591522907749293e-05,
+      "loss": 0.0096,
+      "step": 85060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959147486801082e-05,
+      "loss": 0.0128,
+      "step": 85070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9591426828272348e-05,
+      "loss": 0.0143,
+      "step": 85080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9591378788533877e-05,
+      "loss": 0.0106,
+      "step": 85090
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9591330748795406e-05,
+      "loss": 0.0189,
+      "step": 85100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9591282709056932e-05,
+      "loss": 0.0134,
+      "step": 85110
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959123466931846e-05,
+      "loss": 0.0144,
+      "step": 85120
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959118662957999e-05,
+      "loss": 0.0126,
+      "step": 85130
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959113858984152e-05,
+      "loss": 0.0131,
+      "step": 85140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9591090550103046e-05,
+      "loss": 0.0151,
+      "step": 85150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9591042510364575e-05,
+      "loss": 0.0104,
+      "step": 85160
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590994470626104e-05,
+      "loss": 0.0133,
+      "step": 85170
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959094643088763e-05,
+      "loss": 0.0121,
+      "step": 85180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959089839114916e-05,
+      "loss": 0.0088,
+      "step": 85190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959085035141069e-05,
+      "loss": 0.0152,
+      "step": 85200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590802311672218e-05,
+      "loss": 0.016,
+      "step": 85210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590754271933744e-05,
+      "loss": 0.0169,
+      "step": 85220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590706232195273e-05,
+      "loss": 0.0098,
+      "step": 85230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590658192456803e-05,
+      "loss": 0.0165,
+      "step": 85240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959061015271833e-05,
+      "loss": 0.0158,
+      "step": 85250
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590562112979858e-05,
+      "loss": 0.0099,
+      "step": 85260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590514073241387e-05,
+      "loss": 0.0151,
+      "step": 85270
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590466033502916e-05,
+      "loss": 0.0101,
+      "step": 85280
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590417993764442e-05,
+      "loss": 0.0136,
+      "step": 85290
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959036995402597e-05,
+      "loss": 0.0143,
+      "step": 85300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.95903219142875e-05,
+      "loss": 0.0114,
+      "step": 85310
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959027387454903e-05,
+      "loss": 0.0156,
+      "step": 85320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590225834810556e-05,
+      "loss": 0.0107,
+      "step": 85330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590177795072085e-05,
+      "loss": 0.0115,
+      "step": 85340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9590129755333614e-05,
+      "loss": 0.0177,
+      "step": 85350
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959008171559514e-05,
+      "loss": 0.0127,
+      "step": 85360
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.959003367585667e-05,
+      "loss": 0.0127,
+      "step": 85370
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.95899856361182e-05,
+      "loss": 0.0121,
+      "step": 85380
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589937596379728e-05,
+      "loss": 0.0097,
+      "step": 85390
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589889556641254e-05,
+      "loss": 0.0136,
+      "step": 85400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589841516902783e-05,
+      "loss": 0.0109,
+      "step": 85410
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589793477164313e-05,
+      "loss": 0.0109,
+      "step": 85420
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.958974543742584e-05,
+      "loss": 0.0106,
+      "step": 85430
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589697397687368e-05,
+      "loss": 0.0098,
+      "step": 85440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589649357948897e-05,
+      "loss": 0.0131,
+      "step": 85450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589601318210426e-05,
+      "loss": 0.0148,
+      "step": 85460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589553278471952e-05,
+      "loss": 0.0133,
+      "step": 85470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.958950523873348e-05,
+      "loss": 0.0112,
+      "step": 85480
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.958945719899501e-05,
+      "loss": 0.0126,
+      "step": 85490
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.958940915925654e-05,
+      "loss": 0.0114,
+      "step": 85500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589361119518066e-05,
+      "loss": 0.0133,
+      "step": 85510
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589313079779595e-05,
+      "loss": 0.0107,
+      "step": 85520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589265040041124e-05,
+      "loss": 0.009,
+      "step": 85530
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.958921700030265e-05,
+      "loss": 0.0115,
+      "step": 85540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.958916896056418e-05,
+      "loss": 0.0118,
+      "step": 85550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.958912092082571e-05,
+      "loss": 0.0148,
+      "step": 85560
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589072881087238e-05,
+      "loss": 0.0103,
+      "step": 85570
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9589024841348764e-05,
+      "loss": 0.0139,
+      "step": 85580
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9588976801610293e-05,
+      "loss": 0.0123,
+      "step": 85590
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9588928761871822e-05,
+      "loss": 0.0163,
+      "step": 85600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.958888072213335e-05,
+      "loss": 0.0144,
+      "step": 85610
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9588832682394878e-05,
+      "loss": 0.0121,
+      "step": 85620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9588784642656407e-05,
+      "loss": 0.0088,
+      "step": 85630
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9588736602917936e-05,
+      "loss": 0.0096,
+      "step": 85640
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.958869336715331e-05,
+      "loss": 0.0148,
+      "step": 85650
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.958864532741484e-05,
+      "loss": 0.01,
+      "step": 85660
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9588597287676366e-05,
+      "loss": 0.0093,
+      "step": 85670
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9588549247937895e-05,
+      "loss": 0.0109,
+      "step": 85680
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9588501208199424e-05,
+      "loss": 0.0125,
+      "step": 85690
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9588453168460954e-05,
+      "loss": 0.0138,
+      "step": 85700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958840512872248e-05,
+      "loss": 0.0121,
+      "step": 85710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958835708898401e-05,
+      "loss": 0.0105,
+      "step": 85720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9588309049245538e-05,
+      "loss": 0.0173,
+      "step": 85730
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9588261009507067e-05,
+      "loss": 0.0119,
+      "step": 85740
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9588212969768593e-05,
+      "loss": 0.0119,
+      "step": 85750
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9588164930030123e-05,
+      "loss": 0.0106,
+      "step": 85760
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9588116890291652e-05,
+      "loss": 0.0148,
+      "step": 85770
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9588068850553178e-05,
+      "loss": 0.0133,
+      "step": 85780
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9588020810814707e-05,
+      "loss": 0.0146,
+      "step": 85790
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587972771076236e-05,
+      "loss": 0.0111,
+      "step": 85800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587924731337766e-05,
+      "loss": 0.0257,
+      "step": 85810
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958787669159929e-05,
+      "loss": 0.0137,
+      "step": 85820
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958782865186082e-05,
+      "loss": 0.0119,
+      "step": 85830
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958778061212235e-05,
+      "loss": 0.0103,
+      "step": 85840
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587732572383876e-05,
+      "loss": 0.0114,
+      "step": 85850
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587684532645405e-05,
+      "loss": 0.0165,
+      "step": 85860
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587636492906934e-05,
+      "loss": 0.0125,
+      "step": 85870
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587588453168464e-05,
+      "loss": 0.0125,
+      "step": 85880
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958754041342999e-05,
+      "loss": 0.0137,
+      "step": 85890
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958749237369152e-05,
+      "loss": 0.0135,
+      "step": 85900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587444333953048e-05,
+      "loss": 0.0144,
+      "step": 85910
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587396294214577e-05,
+      "loss": 0.0087,
+      "step": 85920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587348254476103e-05,
+      "loss": 0.013,
+      "step": 85930
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587300214737632e-05,
+      "loss": 0.0104,
+      "step": 85940
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587252174999162e-05,
+      "loss": 0.0102,
+      "step": 85950
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587204135260688e-05,
+      "loss": 0.0159,
+      "step": 85960
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587156095522217e-05,
+      "loss": 0.0128,
+      "step": 85970
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587108055783746e-05,
+      "loss": 0.0098,
+      "step": 85980
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9587060016045275e-05,
+      "loss": 0.014,
+      "step": 85990
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.95870119763068e-05,
+      "loss": 0.012,
+      "step": 86000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958696393656833e-05,
+      "loss": 0.0145,
+      "step": 86010
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958691589682986e-05,
+      "loss": 0.0089,
+      "step": 86020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586867857091386e-05,
+      "loss": 0.0101,
+      "step": 86030
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586819817352915e-05,
+      "loss": 0.0132,
+      "step": 86040
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586771777614444e-05,
+      "loss": 0.0181,
+      "step": 86050
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586723737875974e-05,
+      "loss": 0.0128,
+      "step": 86060
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.95866756981375e-05,
+      "loss": 0.0141,
+      "step": 86070
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958662765839903e-05,
+      "loss": 0.0136,
+      "step": 86080
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586579618660558e-05,
+      "loss": 0.0118,
+      "step": 86090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586531578922087e-05,
+      "loss": 0.0113,
+      "step": 86100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586483539183613e-05,
+      "loss": 0.0136,
+      "step": 86110
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586435499445142e-05,
+      "loss": 0.0108,
+      "step": 86120
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958638745970667e-05,
+      "loss": 0.0131,
+      "step": 86130
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586339419968198e-05,
+      "loss": 0.01,
+      "step": 86140
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586291380229727e-05,
+      "loss": 0.0122,
+      "step": 86150
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586243340491256e-05,
+      "loss": 0.0113,
+      "step": 86160
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586195300752785e-05,
+      "loss": 0.0131,
+      "step": 86170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958614726101431e-05,
+      "loss": 0.0146,
+      "step": 86180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958609922127584e-05,
+      "loss": 0.0145,
+      "step": 86190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958605118153737e-05,
+      "loss": 0.0142,
+      "step": 86200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9586003141798896e-05,
+      "loss": 0.0146,
+      "step": 86210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9585955102060425e-05,
+      "loss": 0.0111,
+      "step": 86220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9585907062321954e-05,
+      "loss": 0.0151,
+      "step": 86230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9585859022583483e-05,
+      "loss": 0.0104,
+      "step": 86240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958581098284501e-05,
+      "loss": 0.0131,
+      "step": 86250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958576294310654e-05,
+      "loss": 0.0129,
+      "step": 86260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9585714903368068e-05,
+      "loss": 0.0081,
+      "step": 86270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9585666863629594e-05,
+      "loss": 0.0131,
+      "step": 86280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9585618823891123e-05,
+      "loss": 0.0155,
+      "step": 86290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9585570784152652e-05,
+      "loss": 0.0115,
+      "step": 86300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958552274441418e-05,
+      "loss": 0.0127,
+      "step": 86310
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9585474704675707e-05,
+      "loss": 0.0088,
+      "step": 86320
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9585426664937237e-05,
+      "loss": 0.0214,
+      "step": 86330
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9585378625198766e-05,
+      "loss": 0.0121,
+      "step": 86340
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9585330585460295e-05,
+      "loss": 0.0125,
+      "step": 86350
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958528254572182e-05,
+      "loss": 0.0093,
+      "step": 86360
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958523450598335e-05,
+      "loss": 0.0144,
+      "step": 86370
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.958518646624488e-05,
+      "loss": 0.014,
+      "step": 86380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9585138426506406e-05,
+      "loss": 0.0131,
+      "step": 86390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9585090386767935e-05,
+      "loss": 0.0084,
+      "step": 86400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9585042347029464e-05,
+      "loss": 0.0105,
+      "step": 86410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584994307290993e-05,
+      "loss": 0.0134,
+      "step": 86420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958494626755252e-05,
+      "loss": 0.0132,
+      "step": 86430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958489822781405e-05,
+      "loss": 0.0106,
+      "step": 86440
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584850188075578e-05,
+      "loss": 0.0139,
+      "step": 86450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584802148337104e-05,
+      "loss": 0.0128,
+      "step": 86460
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584754108598633e-05,
+      "loss": 0.0116,
+      "step": 86470
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584706068860162e-05,
+      "loss": 0.0115,
+      "step": 86480
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958465802912169e-05,
+      "loss": 0.0098,
+      "step": 86490
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584609989383217e-05,
+      "loss": 0.0107,
+      "step": 86500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584561949644747e-05,
+      "loss": 0.0118,
+      "step": 86510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584513909906276e-05,
+      "loss": 0.0111,
+      "step": 86520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584465870167805e-05,
+      "loss": 0.0123,
+      "step": 86530
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958441783042933e-05,
+      "loss": 0.0105,
+      "step": 86540
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958436979069086e-05,
+      "loss": 0.0104,
+      "step": 86550
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958432175095239e-05,
+      "loss": 0.0144,
+      "step": 86560
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584273711213916e-05,
+      "loss": 0.0133,
+      "step": 86570
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584225671475445e-05,
+      "loss": 0.0135,
+      "step": 86580
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584177631736974e-05,
+      "loss": 0.0102,
+      "step": 86590
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9584129591998503e-05,
+      "loss": 0.0136,
+      "step": 86600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958408155226003e-05,
+      "loss": 0.0101,
+      "step": 86610
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958403351252156e-05,
+      "loss": 0.0082,
+      "step": 86620
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583985472783088e-05,
+      "loss": 0.0109,
+      "step": 86630
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583937433044614e-05,
+      "loss": 0.0187,
+      "step": 86640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583889393306143e-05,
+      "loss": 0.0112,
+      "step": 86650
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583841353567672e-05,
+      "loss": 0.0135,
+      "step": 86660
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.95837933138292e-05,
+      "loss": 0.0109,
+      "step": 86670
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583745274090727e-05,
+      "loss": 0.0084,
+      "step": 86680
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583697234352257e-05,
+      "loss": 0.0132,
+      "step": 86690
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583649194613786e-05,
+      "loss": 0.0149,
+      "step": 86700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583601154875315e-05,
+      "loss": 0.0141,
+      "step": 86710
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958355311513684e-05,
+      "loss": 0.0085,
+      "step": 86720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958350507539837e-05,
+      "loss": 0.0134,
+      "step": 86730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.95834570356599e-05,
+      "loss": 0.0109,
+      "step": 86740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583408995921425e-05,
+      "loss": 0.012,
+      "step": 86750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583360956182955e-05,
+      "loss": 0.0154,
+      "step": 86760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583312916444484e-05,
+      "loss": 0.0096,
+      "step": 86770
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583264876706013e-05,
+      "loss": 0.0134,
+      "step": 86780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958321683696754e-05,
+      "loss": 0.0113,
+      "step": 86790
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958316879722907e-05,
+      "loss": 0.0098,
+      "step": 86800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583120757490598e-05,
+      "loss": 0.0109,
+      "step": 86810
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583072717752124e-05,
+      "loss": 0.0185,
+      "step": 86820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9583024678013653e-05,
+      "loss": 0.0107,
+      "step": 86830
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582976638275182e-05,
+      "loss": 0.0127,
+      "step": 86840
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958292859853671e-05,
+      "loss": 0.0137,
+      "step": 86850
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582880558798237e-05,
+      "loss": 0.0139,
+      "step": 86860
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582832519059766e-05,
+      "loss": 0.0088,
+      "step": 86870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582784479321296e-05,
+      "loss": 0.0127,
+      "step": 86880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582736439582825e-05,
+      "loss": 0.0152,
+      "step": 86890
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958268839984435e-05,
+      "loss": 0.011,
+      "step": 86900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958264036010588e-05,
+      "loss": 0.012,
+      "step": 86910
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958259232036741e-05,
+      "loss": 0.0128,
+      "step": 86920
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582544280628935e-05,
+      "loss": 0.0122,
+      "step": 86930
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582496240890468e-05,
+      "loss": 0.0123,
+      "step": 86940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582448201151997e-05,
+      "loss": 0.0109,
+      "step": 86950
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582400161413523e-05,
+      "loss": 0.0141,
+      "step": 86960
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582352121675052e-05,
+      "loss": 0.0151,
+      "step": 86970
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958230408193658e-05,
+      "loss": 0.0115,
+      "step": 86980
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582256042198108e-05,
+      "loss": 0.0122,
+      "step": 86990
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582208002459637e-05,
+      "loss": 0.0099,
+      "step": 87000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582159962721166e-05,
+      "loss": 0.01,
+      "step": 87010
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582111922982695e-05,
+      "loss": 0.0149,
+      "step": 87020
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958206388324422e-05,
+      "loss": 0.0102,
+      "step": 87030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958201584350575e-05,
+      "loss": 0.013,
+      "step": 87040
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.958196780376728e-05,
+      "loss": 0.0111,
+      "step": 87050
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9581919764028806e-05,
+      "loss": 0.0101,
+      "step": 87060
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9581871724290335e-05,
+      "loss": 0.009,
+      "step": 87070
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9581823684551864e-05,
+      "loss": 0.0162,
+      "step": 87080
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9581775644813393e-05,
+      "loss": 0.0141,
+      "step": 87090
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958172760507492e-05,
+      "loss": 0.0108,
+      "step": 87100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958167956533645e-05,
+      "loss": 0.0131,
+      "step": 87110
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9581631525597978e-05,
+      "loss": 0.0124,
+      "step": 87120
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9581583485859507e-05,
+      "loss": 0.0121,
+      "step": 87130
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9581535446121033e-05,
+      "loss": 0.0121,
+      "step": 87140
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9581487406382562e-05,
+      "loss": 0.0157,
+      "step": 87150
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958143936664409e-05,
+      "loss": 0.0087,
+      "step": 87160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9581391326905617e-05,
+      "loss": 0.0113,
+      "step": 87170
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9581343287167147e-05,
+      "loss": 0.0153,
+      "step": 87180
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9581295247428676e-05,
+      "loss": 0.0156,
+      "step": 87190
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9581247207690205e-05,
+      "loss": 0.0114,
+      "step": 87200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958119916795173e-05,
+      "loss": 0.0124,
+      "step": 87210
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958115112821326e-05,
+      "loss": 0.0129,
+      "step": 87220
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9581107892448635e-05,
+      "loss": 0.0151,
+      "step": 87230
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958105985271016e-05,
+      "loss": 0.0112,
+      "step": 87240
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958101181297169e-05,
+      "loss": 0.0106,
+      "step": 87250
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958096377323322e-05,
+      "loss": 0.0113,
+      "step": 87260
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958091573349475e-05,
+      "loss": 0.0082,
+      "step": 87270
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9580867693756275e-05,
+      "loss": 0.0159,
+      "step": 87280
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9580819654017804e-05,
+      "loss": 0.0147,
+      "step": 87290
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9580771614279333e-05,
+      "loss": 0.0094,
+      "step": 87300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9580723574540862e-05,
+      "loss": 0.0112,
+      "step": 87310
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958067553480239e-05,
+      "loss": 0.016,
+      "step": 87320
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9580627495063918e-05,
+      "loss": 0.0092,
+      "step": 87330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9580579455325447e-05,
+      "loss": 0.015,
+      "step": 87340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9580531415586973e-05,
+      "loss": 0.0175,
+      "step": 87350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9580483375848502e-05,
+      "loss": 0.0126,
+      "step": 87360
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958043533611003e-05,
+      "loss": 0.0152,
+      "step": 87370
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958038729637156e-05,
+      "loss": 0.0158,
+      "step": 87380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9580339256633086e-05,
+      "loss": 0.0128,
+      "step": 87390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9580291216894616e-05,
+      "loss": 0.0099,
+      "step": 87400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9580243177156145e-05,
+      "loss": 0.0086,
+      "step": 87410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958019513741767e-05,
+      "loss": 0.0139,
+      "step": 87420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.95801470976792e-05,
+      "loss": 0.0091,
+      "step": 87430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958009905794073e-05,
+      "loss": 0.0135,
+      "step": 87440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.958005101820226e-05,
+      "loss": 0.0155,
+      "step": 87450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9580002978463788e-05,
+      "loss": 0.0141,
+      "step": 87460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579954938725317e-05,
+      "loss": 0.0096,
+      "step": 87470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579906898986843e-05,
+      "loss": 0.0108,
+      "step": 87480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579858859248372e-05,
+      "loss": 0.0129,
+      "step": 87490
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.95798108195099e-05,
+      "loss": 0.0088,
+      "step": 87500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.957976277977143e-05,
+      "loss": 0.0096,
+      "step": 87510
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579714740032957e-05,
+      "loss": 0.0135,
+      "step": 87520
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579666700294486e-05,
+      "loss": 0.019,
+      "step": 87530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579618660556015e-05,
+      "loss": 0.0094,
+      "step": 87540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.957957062081754e-05,
+      "loss": 0.0131,
+      "step": 87550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.957952258107907e-05,
+      "loss": 0.0108,
+      "step": 87560
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.95794745413406e-05,
+      "loss": 0.0115,
+      "step": 87570
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.957942650160213e-05,
+      "loss": 0.0127,
+      "step": 87580
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579378461863655e-05,
+      "loss": 0.0148,
+      "step": 87590
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579330422125184e-05,
+      "loss": 0.0088,
+      "step": 87600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579282382386713e-05,
+      "loss": 0.0142,
+      "step": 87610
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579234342648243e-05,
+      "loss": 0.0144,
+      "step": 87620
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.957918630290977e-05,
+      "loss": 0.0125,
+      "step": 87630
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579138263171298e-05,
+      "loss": 0.0152,
+      "step": 87640
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579090223432827e-05,
+      "loss": 0.0111,
+      "step": 87650
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9579042183694353e-05,
+      "loss": 0.0111,
+      "step": 87660
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9578994143955882e-05,
+      "loss": 0.0118,
+      "step": 87670
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.957894610421741e-05,
+      "loss": 0.0113,
+      "step": 87680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.957889806447894e-05,
+      "loss": 0.0118,
+      "step": 87690
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9578850024740467e-05,
+      "loss": 0.0159,
+      "step": 87700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9578801985001996e-05,
+      "loss": 0.012,
+      "step": 87710
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9578753945263525e-05,
+      "loss": 0.0124,
+      "step": 87720
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.957870590552505e-05,
+      "loss": 0.007,
+      "step": 87730
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.957865786578658e-05,
+      "loss": 0.031,
+      "step": 87740
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.957860982604811e-05,
+      "loss": 0.0115,
+      "step": 87750
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.957856178630964e-05,
+      "loss": 0.0121,
+      "step": 87760
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9578513746571165e-05,
+      "loss": 0.01,
+      "step": 87770
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9578465706832694e-05,
+      "loss": 0.0082,
+      "step": 87780
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9578417667094223e-05,
+      "loss": 0.0143,
+      "step": 87790
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9578369627355753e-05,
+      "loss": 0.0148,
+      "step": 87800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957832158761728e-05,
+      "loss": 0.0085,
+      "step": 87810
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9578273547878808e-05,
+      "loss": 0.0103,
+      "step": 87820
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9578225508140337e-05,
+      "loss": 0.014,
+      "step": 87830
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9578177468401863e-05,
+      "loss": 0.0124,
+      "step": 87840
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9578129428663392e-05,
+      "loss": 0.0113,
+      "step": 87850
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957808138892492e-05,
+      "loss": 0.0107,
+      "step": 87860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957803334918645e-05,
+      "loss": 0.0111,
+      "step": 87870
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577985309447977e-05,
+      "loss": 0.0123,
+      "step": 87880
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577937269709506e-05,
+      "loss": 0.0099,
+      "step": 87890
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577889229971035e-05,
+      "loss": 0.0148,
+      "step": 87900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957784119023256e-05,
+      "loss": 0.0101,
+      "step": 87910
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957779315049409e-05,
+      "loss": 0.0127,
+      "step": 87920
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957774511075562e-05,
+      "loss": 0.0121,
+      "step": 87930
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957769707101715e-05,
+      "loss": 0.0126,
+      "step": 87940
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577649031278675e-05,
+      "loss": 0.0118,
+      "step": 87950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577600991540204e-05,
+      "loss": 0.0208,
+      "step": 87960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577552951801733e-05,
+      "loss": 0.0093,
+      "step": 87970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577504912063263e-05,
+      "loss": 0.0141,
+      "step": 87980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957745687232479e-05,
+      "loss": 0.0129,
+      "step": 87990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577408832586318e-05,
+      "loss": 0.014,
+      "step": 88000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577360792847847e-05,
+      "loss": 0.0127,
+      "step": 88010
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577312753109373e-05,
+      "loss": 0.0097,
+      "step": 88020
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577264713370902e-05,
+      "loss": 0.0153,
+      "step": 88030
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957721667363243e-05,
+      "loss": 0.0123,
+      "step": 88040
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957716863389396e-05,
+      "loss": 0.0218,
+      "step": 88050
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577120594155487e-05,
+      "loss": 0.0151,
+      "step": 88060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577072554417016e-05,
+      "loss": 0.0161,
+      "step": 88070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9577024514678545e-05,
+      "loss": 0.0094,
+      "step": 88080
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957697647494007e-05,
+      "loss": 0.0127,
+      "step": 88090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.95769284352016e-05,
+      "loss": 0.0152,
+      "step": 88100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957688039546313e-05,
+      "loss": 0.0079,
+      "step": 88110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957683235572466e-05,
+      "loss": 0.0136,
+      "step": 88120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9576784315986185e-05,
+      "loss": 0.0082,
+      "step": 88130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9576736276247714e-05,
+      "loss": 0.0126,
+      "step": 88140
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9576688236509243e-05,
+      "loss": 0.0148,
+      "step": 88150
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9576640196770772e-05,
+      "loss": 0.0115,
+      "step": 88160
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.95765921570323e-05,
+      "loss": 0.0122,
+      "step": 88170
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9576544117293828e-05,
+      "loss": 0.0118,
+      "step": 88180
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9576496077555357e-05,
+      "loss": 0.0152,
+      "step": 88190
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9576448037816883e-05,
+      "loss": 0.0129,
+      "step": 88200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9576399998078412e-05,
+      "loss": 0.0119,
+      "step": 88210
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957635195833994e-05,
+      "loss": 0.0178,
+      "step": 88220
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957630391860147e-05,
+      "loss": 0.0119,
+      "step": 88230
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9576255878862996e-05,
+      "loss": 0.0131,
+      "step": 88240
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9576207839124526e-05,
+      "loss": 0.0112,
+      "step": 88250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.95761646033599e-05,
+      "loss": 0.0223,
+      "step": 88260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957611656362143e-05,
+      "loss": 0.0115,
+      "step": 88270
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957606852388296e-05,
+      "loss": 0.0145,
+      "step": 88280
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9576020484144488e-05,
+      "loss": 0.0103,
+      "step": 88290
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9575972444406014e-05,
+      "loss": 0.013,
+      "step": 88300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9575924404667543e-05,
+      "loss": 0.0101,
+      "step": 88310
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9575876364929073e-05,
+      "loss": 0.0126,
+      "step": 88320
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.95758283251906e-05,
+      "loss": 0.0119,
+      "step": 88330
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9575780285452128e-05,
+      "loss": 0.0123,
+      "step": 88340
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9575732245713657e-05,
+      "loss": 0.0074,
+      "step": 88350
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9575684205975186e-05,
+      "loss": 0.0146,
+      "step": 88360
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9575636166236712e-05,
+      "loss": 0.0084,
+      "step": 88370
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957558812649824e-05,
+      "loss": 0.0095,
+      "step": 88380
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957554008675977e-05,
+      "loss": 0.0158,
+      "step": 88390
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.95754920470213e-05,
+      "loss": 0.0141,
+      "step": 88400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9575444007282826e-05,
+      "loss": 0.0107,
+      "step": 88410
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9575395967544355e-05,
+      "loss": 0.0086,
+      "step": 88420
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9575347927805884e-05,
+      "loss": 0.0121,
+      "step": 88430
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957529988806741e-05,
+      "loss": 0.0123,
+      "step": 88440
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957525184832894e-05,
+      "loss": 0.011,
+      "step": 88450
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.957520380859047e-05,
+      "loss": 0.0148,
+      "step": 88460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9575155768851998e-05,
+      "loss": 0.0103,
+      "step": 88470
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9575107729113524e-05,
+      "loss": 0.016,
+      "step": 88480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9575059689375053e-05,
+      "loss": 0.0131,
+      "step": 88490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9575011649636583e-05,
+      "loss": 0.0107,
+      "step": 88500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957496360989811e-05,
+      "loss": 0.0116,
+      "step": 88510
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574915570159638e-05,
+      "loss": 0.013,
+      "step": 88520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574867530421167e-05,
+      "loss": 0.011,
+      "step": 88530
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574819490682696e-05,
+      "loss": 0.0123,
+      "step": 88540
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574771450944222e-05,
+      "loss": 0.0132,
+      "step": 88550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957472341120575e-05,
+      "loss": 0.0108,
+      "step": 88560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957467537146728e-05,
+      "loss": 0.0113,
+      "step": 88570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957462733172881e-05,
+      "loss": 0.0118,
+      "step": 88580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574579291990336e-05,
+      "loss": 0.0111,
+      "step": 88590
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574531252251865e-05,
+      "loss": 0.0131,
+      "step": 88600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574483212513394e-05,
+      "loss": 0.0115,
+      "step": 88610
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957443517277492e-05,
+      "loss": 0.0119,
+      "step": 88620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957438713303645e-05,
+      "loss": 0.0158,
+      "step": 88630
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957433909329798e-05,
+      "loss": 0.0114,
+      "step": 88640
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574291053559508e-05,
+      "loss": 0.0158,
+      "step": 88650
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574243013821034e-05,
+      "loss": 0.0118,
+      "step": 88660
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574194974082563e-05,
+      "loss": 0.0113,
+      "step": 88670
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574146934344092e-05,
+      "loss": 0.0103,
+      "step": 88680
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957409889460562e-05,
+      "loss": 0.0142,
+      "step": 88690
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574050854867148e-05,
+      "loss": 0.0077,
+      "step": 88700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9574002815128677e-05,
+      "loss": 0.0096,
+      "step": 88710
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573954775390206e-05,
+      "loss": 0.0156,
+      "step": 88720
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573906735651732e-05,
+      "loss": 0.0139,
+      "step": 88730
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957385869591326e-05,
+      "loss": 0.0138,
+      "step": 88740
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957381065617479e-05,
+      "loss": 0.0094,
+      "step": 88750
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957376261643632e-05,
+      "loss": 0.0075,
+      "step": 88760
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573714576697846e-05,
+      "loss": 0.0135,
+      "step": 88770
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573666536959375e-05,
+      "loss": 0.0122,
+      "step": 88780
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573618497220904e-05,
+      "loss": 0.0123,
+      "step": 88790
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957357045748243e-05,
+      "loss": 0.0107,
+      "step": 88800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957352241774396e-05,
+      "loss": 0.0109,
+      "step": 88810
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957347437800549e-05,
+      "loss": 0.0147,
+      "step": 88820
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573426338267018e-05,
+      "loss": 0.0135,
+      "step": 88830
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573378298528544e-05,
+      "loss": 0.0101,
+      "step": 88840
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573330258790073e-05,
+      "loss": 0.016,
+      "step": 88850
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573282219051602e-05,
+      "loss": 0.0115,
+      "step": 88860
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573234179313128e-05,
+      "loss": 0.0131,
+      "step": 88870
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573186139574658e-05,
+      "loss": 0.0121,
+      "step": 88880
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573138099836187e-05,
+      "loss": 0.0105,
+      "step": 88890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573090060097716e-05,
+      "loss": 0.0126,
+      "step": 88900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9573042020359242e-05,
+      "loss": 0.0118,
+      "step": 88910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957299398062077e-05,
+      "loss": 0.009,
+      "step": 88920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.95729459408823e-05,
+      "loss": 0.0107,
+      "step": 88930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572897901143826e-05,
+      "loss": 0.0128,
+      "step": 88940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572849861405356e-05,
+      "loss": 0.0133,
+      "step": 88950
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572801821666885e-05,
+      "loss": 0.0105,
+      "step": 88960
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572753781928414e-05,
+      "loss": 0.0123,
+      "step": 88970
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957270574218994e-05,
+      "loss": 0.0124,
+      "step": 88980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957265770245147e-05,
+      "loss": 0.0097,
+      "step": 88990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572609662713e-05,
+      "loss": 0.0101,
+      "step": 89000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572561622974528e-05,
+      "loss": 0.011,
+      "step": 89010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572513583236054e-05,
+      "loss": 0.0122,
+      "step": 89020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572465543497583e-05,
+      "loss": 0.0126,
+      "step": 89030
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572417503759112e-05,
+      "loss": 0.0135,
+      "step": 89040
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572369464020638e-05,
+      "loss": 0.0128,
+      "step": 89050
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572321424282167e-05,
+      "loss": 0.0128,
+      "step": 89060
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572273384543697e-05,
+      "loss": 0.013,
+      "step": 89070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572225344805226e-05,
+      "loss": 0.0142,
+      "step": 89080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572177305066752e-05,
+      "loss": 0.0117,
+      "step": 89090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957212926532828e-05,
+      "loss": 0.0128,
+      "step": 89100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957208122558981e-05,
+      "loss": 0.0143,
+      "step": 89110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9572033185851336e-05,
+      "loss": 0.0123,
+      "step": 89120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9571985146112866e-05,
+      "loss": 0.0122,
+      "step": 89130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9571937106374395e-05,
+      "loss": 0.0098,
+      "step": 89140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9571889066635924e-05,
+      "loss": 0.0146,
+      "step": 89150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.957184102689745e-05,
+      "loss": 0.0126,
+      "step": 89160
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.957179298715898e-05,
+      "loss": 0.0146,
+      "step": 89170
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.957174494742051e-05,
+      "loss": 0.0116,
+      "step": 89180
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571696907682038e-05,
+      "loss": 0.0106,
+      "step": 89190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571648867943564e-05,
+      "loss": 0.0105,
+      "step": 89200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571600828205093e-05,
+      "loss": 0.0129,
+      "step": 89210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571552788466622e-05,
+      "loss": 0.0184,
+      "step": 89220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571504748728148e-05,
+      "loss": 0.0114,
+      "step": 89230
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571456708989677e-05,
+      "loss": 0.0092,
+      "step": 89240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571408669251207e-05,
+      "loss": 0.0215,
+      "step": 89250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571360629512736e-05,
+      "loss": 0.0104,
+      "step": 89260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571312589774262e-05,
+      "loss": 0.013,
+      "step": 89270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.957126455003579e-05,
+      "loss": 0.0171,
+      "step": 89280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.957121651029732e-05,
+      "loss": 0.0115,
+      "step": 89290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571168470558846e-05,
+      "loss": 0.0105,
+      "step": 89300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571120430820375e-05,
+      "loss": 0.0109,
+      "step": 89310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571072391081905e-05,
+      "loss": 0.012,
+      "step": 89320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9571024351343434e-05,
+      "loss": 0.0108,
+      "step": 89330
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.957097631160496e-05,
+      "loss": 0.0152,
+      "step": 89340
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.957092827186649e-05,
+      "loss": 0.0134,
+      "step": 89350
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.957088023212802e-05,
+      "loss": 0.0096,
+      "step": 89360
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570832192389548e-05,
+      "loss": 0.0116,
+      "step": 89370
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570784152651074e-05,
+      "loss": 0.0147,
+      "step": 89380
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570736112912603e-05,
+      "loss": 0.0108,
+      "step": 89390
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570688073174132e-05,
+      "loss": 0.0108,
+      "step": 89400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570640033435658e-05,
+      "loss": 0.0115,
+      "step": 89410
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570591993697187e-05,
+      "loss": 0.0104,
+      "step": 89420
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570543953958717e-05,
+      "loss": 0.0105,
+      "step": 89430
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570495914220246e-05,
+      "loss": 0.0147,
+      "step": 89440
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.957044787448177e-05,
+      "loss": 0.0123,
+      "step": 89450
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.95703998347433e-05,
+      "loss": 0.011,
+      "step": 89460
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.957035179500483e-05,
+      "loss": 0.0174,
+      "step": 89470
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570303755266356e-05,
+      "loss": 0.011,
+      "step": 89480
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570255715527885e-05,
+      "loss": 0.0107,
+      "step": 89490
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570207675789415e-05,
+      "loss": 0.0095,
+      "step": 89500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570159636050944e-05,
+      "loss": 0.0104,
+      "step": 89510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.957011159631247e-05,
+      "loss": 0.0096,
+      "step": 89520
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9570063556574e-05,
+      "loss": 0.0122,
+      "step": 89530
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.957001551683553e-05,
+      "loss": 0.0104,
+      "step": 89540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569967477097058e-05,
+      "loss": 0.0115,
+      "step": 89550
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569919437358583e-05,
+      "loss": 0.0134,
+      "step": 89560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569871397620113e-05,
+      "loss": 0.0117,
+      "step": 89570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569823357881642e-05,
+      "loss": 0.0148,
+      "step": 89580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569775318143168e-05,
+      "loss": 0.0114,
+      "step": 89590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569727278404697e-05,
+      "loss": 0.0089,
+      "step": 89600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569679238666226e-05,
+      "loss": 0.0101,
+      "step": 89610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569631198927756e-05,
+      "loss": 0.0117,
+      "step": 89620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.956958315918928e-05,
+      "loss": 0.0138,
+      "step": 89630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.956953511945081e-05,
+      "loss": 0.0133,
+      "step": 89640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.956948707971234e-05,
+      "loss": 0.0161,
+      "step": 89650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569439039973866e-05,
+      "loss": 0.011,
+      "step": 89660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569391000235395e-05,
+      "loss": 0.012,
+      "step": 89670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569342960496925e-05,
+      "loss": 0.0132,
+      "step": 89680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569294920758454e-05,
+      "loss": 0.0121,
+      "step": 89690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.956924688101998e-05,
+      "loss": 0.0124,
+      "step": 89700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.956919884128151e-05,
+      "loss": 0.01,
+      "step": 89710
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569150801543038e-05,
+      "loss": 0.0138,
+      "step": 89720
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569102761804568e-05,
+      "loss": 0.0161,
+      "step": 89730
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569054722066093e-05,
+      "loss": 0.0112,
+      "step": 89740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9569006682327623e-05,
+      "loss": 0.0111,
+      "step": 89750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9568958642589152e-05,
+      "loss": 0.0142,
+      "step": 89760
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9568910602850678e-05,
+      "loss": 0.0139,
+      "step": 89770
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9568862563112207e-05,
+      "loss": 0.0124,
+      "step": 89780
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9568814523373736e-05,
+      "loss": 0.0133,
+      "step": 89790
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9568766483635266e-05,
+      "loss": 0.0104,
+      "step": 89800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.956871844389679e-05,
+      "loss": 0.011,
+      "step": 89810
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.956867040415832e-05,
+      "loss": 0.01,
+      "step": 89820
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.956862236441985e-05,
+      "loss": 0.013,
+      "step": 89830
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9568574324681376e-05,
+      "loss": 0.0121,
+      "step": 89840
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9568526284942905e-05,
+      "loss": 0.0114,
+      "step": 89850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9568478245204434e-05,
+      "loss": 0.013,
+      "step": 89860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9568430205465964e-05,
+      "loss": 0.011,
+      "step": 89870
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956838216572749e-05,
+      "loss": 0.0166,
+      "step": 89880
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956833412598902e-05,
+      "loss": 0.0124,
+      "step": 89890
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9568286086250548e-05,
+      "loss": 0.0125,
+      "step": 89900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9568238046512077e-05,
+      "loss": 0.0102,
+      "step": 89910
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9568190006773603e-05,
+      "loss": 0.0129,
+      "step": 89920
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9568141967035133e-05,
+      "loss": 0.0147,
+      "step": 89930
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9568093927296662e-05,
+      "loss": 0.0117,
+      "step": 89940
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9568045887558188e-05,
+      "loss": 0.0114,
+      "step": 89950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567997847819717e-05,
+      "loss": 0.0195,
+      "step": 89960
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567949808081246e-05,
+      "loss": 0.0115,
+      "step": 89970
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567901768342776e-05,
+      "loss": 0.0149,
+      "step": 89980
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.95678537286043e-05,
+      "loss": 0.0111,
+      "step": 89990
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956780568886583e-05,
+      "loss": 0.0105,
+      "step": 90000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956775764912736e-05,
+      "loss": 0.0077,
+      "step": 90010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567709609388886e-05,
+      "loss": 0.0176,
+      "step": 90020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567661569650415e-05,
+      "loss": 0.0134,
+      "step": 90030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567613529911944e-05,
+      "loss": 0.0108,
+      "step": 90040
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567565490173474e-05,
+      "loss": 0.017,
+      "step": 90050
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567517450435e-05,
+      "loss": 0.0182,
+      "step": 90060
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956746941069653e-05,
+      "loss": 0.0115,
+      "step": 90070
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567421370958058e-05,
+      "loss": 0.0122,
+      "step": 90080
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567373331219587e-05,
+      "loss": 0.0136,
+      "step": 90090
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567325291481113e-05,
+      "loss": 0.0108,
+      "step": 90100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567277251742642e-05,
+      "loss": 0.0118,
+      "step": 90110
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567229212004172e-05,
+      "loss": 0.0107,
+      "step": 90120
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567181172265698e-05,
+      "loss": 0.01,
+      "step": 90130
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567133132527227e-05,
+      "loss": 0.0092,
+      "step": 90140
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567085092788756e-05,
+      "loss": 0.0126,
+      "step": 90150
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9567037053050285e-05,
+      "loss": 0.011,
+      "step": 90160
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956698901331181e-05,
+      "loss": 0.0157,
+      "step": 90170
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956694097357334e-05,
+      "loss": 0.0143,
+      "step": 90180
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956689293383487e-05,
+      "loss": 0.0132,
+      "step": 90190
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9566844894096396e-05,
+      "loss": 0.0104,
+      "step": 90200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9566796854357925e-05,
+      "loss": 0.0131,
+      "step": 90210
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9566748814619454e-05,
+      "loss": 0.013,
+      "step": 90220
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9566700774880984e-05,
+      "loss": 0.0125,
+      "step": 90230
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956665273514251e-05,
+      "loss": 0.0158,
+      "step": 90240
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956660469540404e-05,
+      "loss": 0.0109,
+      "step": 90250
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9566556655665568e-05,
+      "loss": 0.0224,
+      "step": 90260
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9566508615927097e-05,
+      "loss": 0.0155,
+      "step": 90270
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9566460576188623e-05,
+      "loss": 0.0141,
+      "step": 90280
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9566412536450152e-05,
+      "loss": 0.0106,
+      "step": 90290
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956636449671168e-05,
+      "loss": 0.0126,
+      "step": 90300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9566316456973208e-05,
+      "loss": 0.0082,
+      "step": 90310
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9566268417234737e-05,
+      "loss": 0.0154,
+      "step": 90320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9566220377496266e-05,
+      "loss": 0.0108,
+      "step": 90330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9566172337757795e-05,
+      "loss": 0.0126,
+      "step": 90340
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956612429801932e-05,
+      "loss": 0.0103,
+      "step": 90350
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956607625828085e-05,
+      "loss": 0.013,
+      "step": 90360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956602821854238e-05,
+      "loss": 0.0106,
+      "step": 90370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565980178803906e-05,
+      "loss": 0.0172,
+      "step": 90380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565932139065435e-05,
+      "loss": 0.0141,
+      "step": 90390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565884099326964e-05,
+      "loss": 0.0106,
+      "step": 90400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565836059588493e-05,
+      "loss": 0.0086,
+      "step": 90410
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956578801985002e-05,
+      "loss": 0.0136,
+      "step": 90420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956573998011155e-05,
+      "loss": 0.0117,
+      "step": 90430
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565691940373078e-05,
+      "loss": 0.0086,
+      "step": 90440
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565643900634607e-05,
+      "loss": 0.0129,
+      "step": 90450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565595860896133e-05,
+      "loss": 0.0156,
+      "step": 90460
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565547821157662e-05,
+      "loss": 0.0145,
+      "step": 90470
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956549978141919e-05,
+      "loss": 0.0092,
+      "step": 90480
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565451741680717e-05,
+      "loss": 0.0127,
+      "step": 90490
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565403701942247e-05,
+      "loss": 0.0133,
+      "step": 90500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565355662203776e-05,
+      "loss": 0.0096,
+      "step": 90510
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565307622465305e-05,
+      "loss": 0.0153,
+      "step": 90520
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565259582726835e-05,
+      "loss": 0.0132,
+      "step": 90530
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9565211542988364e-05,
+      "loss": 0.0152,
+      "step": 90540
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.956516350324989e-05,
+      "loss": 0.0111,
+      "step": 90550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956511546351142e-05,
+      "loss": 0.0143,
+      "step": 90560
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9565067423772948e-05,
+      "loss": 0.0106,
+      "step": 90570
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9565019384034478e-05,
+      "loss": 0.014,
+      "step": 90580
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564971344296003e-05,
+      "loss": 0.0109,
+      "step": 90590
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564923304557533e-05,
+      "loss": 0.0127,
+      "step": 90600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564875264819062e-05,
+      "loss": 0.0147,
+      "step": 90610
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564827225080588e-05,
+      "loss": 0.0088,
+      "step": 90620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564779185342117e-05,
+      "loss": 0.0122,
+      "step": 90630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564731145603646e-05,
+      "loss": 0.0097,
+      "step": 90640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564683105865176e-05,
+      "loss": 0.0129,
+      "step": 90650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.95646350661267e-05,
+      "loss": 0.0108,
+      "step": 90660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956458702638823e-05,
+      "loss": 0.011,
+      "step": 90670
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956453898664976e-05,
+      "loss": 0.0106,
+      "step": 90680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564490946911286e-05,
+      "loss": 0.009,
+      "step": 90690
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564442907172815e-05,
+      "loss": 0.017,
+      "step": 90700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564394867434344e-05,
+      "loss": 0.0091,
+      "step": 90710
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564346827695874e-05,
+      "loss": 0.0141,
+      "step": 90720
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.95642987879574e-05,
+      "loss": 0.0163,
+      "step": 90730
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956425074821893e-05,
+      "loss": 0.0115,
+      "step": 90740
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564202708480458e-05,
+      "loss": 0.0145,
+      "step": 90750
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564154668741987e-05,
+      "loss": 0.009,
+      "step": 90760
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564106629003513e-05,
+      "loss": 0.0149,
+      "step": 90770
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564058589265043e-05,
+      "loss": 0.012,
+      "step": 90780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9564010549526572e-05,
+      "loss": 0.012,
+      "step": 90790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563962509788098e-05,
+      "loss": 0.0113,
+      "step": 90800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563914470049627e-05,
+      "loss": 0.0151,
+      "step": 90810
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563866430311156e-05,
+      "loss": 0.0184,
+      "step": 90820
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563818390572686e-05,
+      "loss": 0.011,
+      "step": 90830
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956377035083421e-05,
+      "loss": 0.0104,
+      "step": 90840
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956372231109574e-05,
+      "loss": 0.0095,
+      "step": 90850
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956367427135727e-05,
+      "loss": 0.0098,
+      "step": 90860
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563626231618796e-05,
+      "loss": 0.0113,
+      "step": 90870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563578191880325e-05,
+      "loss": 0.0142,
+      "step": 90880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563530152141854e-05,
+      "loss": 0.0181,
+      "step": 90890
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563482112403384e-05,
+      "loss": 0.0118,
+      "step": 90900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956343407266491e-05,
+      "loss": 0.0137,
+      "step": 90910
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956338603292644e-05,
+      "loss": 0.0085,
+      "step": 90920
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563337993187968e-05,
+      "loss": 0.0106,
+      "step": 90930
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563289953449497e-05,
+      "loss": 0.0126,
+      "step": 90940
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563241913711023e-05,
+      "loss": 0.0165,
+      "step": 90950
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563193873972552e-05,
+      "loss": 0.0152,
+      "step": 90960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563145834234082e-05,
+      "loss": 0.012,
+      "step": 90970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563097794495608e-05,
+      "loss": 0.0129,
+      "step": 90980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563049754757137e-05,
+      "loss": 0.0163,
+      "step": 90990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9563001715018666e-05,
+      "loss": 0.0146,
+      "step": 91000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562953675280195e-05,
+      "loss": 0.0163,
+      "step": 91010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956290563554172e-05,
+      "loss": 0.0128,
+      "step": 91020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956285759580325e-05,
+      "loss": 0.0157,
+      "step": 91030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956280955606478e-05,
+      "loss": 0.0129,
+      "step": 91040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562761516326306e-05,
+      "loss": 0.0117,
+      "step": 91050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562713476587835e-05,
+      "loss": 0.0108,
+      "step": 91060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562665436849364e-05,
+      "loss": 0.012,
+      "step": 91070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562617397110894e-05,
+      "loss": 0.0104,
+      "step": 91080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956256935737242e-05,
+      "loss": 0.0118,
+      "step": 91090
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956252131763395e-05,
+      "loss": 0.0116,
+      "step": 91100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562473277895478e-05,
+      "loss": 0.0079,
+      "step": 91110
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562425238157007e-05,
+      "loss": 0.0109,
+      "step": 91120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562377198418533e-05,
+      "loss": 0.0122,
+      "step": 91130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562329158680062e-05,
+      "loss": 0.0159,
+      "step": 91140
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956228111894159e-05,
+      "loss": 0.0167,
+      "step": 91150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562233079203118e-05,
+      "loss": 0.0141,
+      "step": 91160
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562185039464647e-05,
+      "loss": 0.0118,
+      "step": 91170
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562136999726176e-05,
+      "loss": 0.0168,
+      "step": 91180
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9562088959987705e-05,
+      "loss": 0.0133,
+      "step": 91190
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956204092024923e-05,
+      "loss": 0.012,
+      "step": 91200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956199288051076e-05,
+      "loss": 0.0084,
+      "step": 91210
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.956194484077229e-05,
+      "loss": 0.0095,
+      "step": 91220
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9561896801033816e-05,
+      "loss": 0.0105,
+      "step": 91230
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9561848761295345e-05,
+      "loss": 0.0116,
+      "step": 91240
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9561800721556874e-05,
+      "loss": 0.0117,
+      "step": 91250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9561752681818403e-05,
+      "loss": 0.0116,
+      "step": 91260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.956170464207993e-05,
+      "loss": 0.0131,
+      "step": 91270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.956165660234146e-05,
+      "loss": 0.0136,
+      "step": 91280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9561608562602988e-05,
+      "loss": 0.0086,
+      "step": 91290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9561560522864517e-05,
+      "loss": 0.0146,
+      "step": 91300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9561512483126043e-05,
+      "loss": 0.0162,
+      "step": 91310
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9561464443387572e-05,
+      "loss": 0.0104,
+      "step": 91320
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.95614164036491e-05,
+      "loss": 0.0146,
+      "step": 91330
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9561368363910627e-05,
+      "loss": 0.0097,
+      "step": 91340
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9561320324172157e-05,
+      "loss": 0.0121,
+      "step": 91350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9561272284433686e-05,
+      "loss": 0.0075,
+      "step": 91360
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9561224244695215e-05,
+      "loss": 0.0188,
+      "step": 91370
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.956117620495674e-05,
+      "loss": 0.0112,
+      "step": 91380
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.956112816521827e-05,
+      "loss": 0.012,
+      "step": 91390
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.95610801254798e-05,
+      "loss": 0.0103,
+      "step": 91400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9561032085741326e-05,
+      "loss": 0.013,
+      "step": 91410
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560984046002855e-05,
+      "loss": 0.0137,
+      "step": 91420
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560936006264384e-05,
+      "loss": 0.0086,
+      "step": 91430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560887966525913e-05,
+      "loss": 0.01,
+      "step": 91440
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.956083992678744e-05,
+      "loss": 0.0081,
+      "step": 91450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.956079188704897e-05,
+      "loss": 0.0113,
+      "step": 91460
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560743847310498e-05,
+      "loss": 0.0098,
+      "step": 91470
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560695807572027e-05,
+      "loss": 0.0149,
+      "step": 91480
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560647767833553e-05,
+      "loss": 0.0107,
+      "step": 91490
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560599728095082e-05,
+      "loss": 0.0116,
+      "step": 91500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.956055168835661e-05,
+      "loss": 0.0144,
+      "step": 91510
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560503648618137e-05,
+      "loss": 0.0125,
+      "step": 91520
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560455608879667e-05,
+      "loss": 0.0123,
+      "step": 91530
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560407569141196e-05,
+      "loss": 0.0116,
+      "step": 91540
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560359529402725e-05,
+      "loss": 0.0101,
+      "step": 91550
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.956031148966425e-05,
+      "loss": 0.0132,
+      "step": 91560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.956026344992578e-05,
+      "loss": 0.0122,
+      "step": 91570
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.956021541018731e-05,
+      "loss": 0.0102,
+      "step": 91580
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560167370448836e-05,
+      "loss": 0.0124,
+      "step": 91590
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560119330710365e-05,
+      "loss": 0.0155,
+      "step": 91600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560071290971894e-05,
+      "loss": 0.0152,
+      "step": 91610
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9560023251233423e-05,
+      "loss": 0.013,
+      "step": 91620
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.955997521149495e-05,
+      "loss": 0.0132,
+      "step": 91630
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.955992717175648e-05,
+      "loss": 0.0103,
+      "step": 91640
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559879132018008e-05,
+      "loss": 0.0128,
+      "step": 91650
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559831092279537e-05,
+      "loss": 0.0098,
+      "step": 91660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559783052541063e-05,
+      "loss": 0.0086,
+      "step": 91670
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559735012802592e-05,
+      "loss": 0.0153,
+      "step": 91680
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.955968697306412e-05,
+      "loss": 0.0113,
+      "step": 91690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559638933325647e-05,
+      "loss": 0.0107,
+      "step": 91700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559590893587177e-05,
+      "loss": 0.0112,
+      "step": 91710
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559542853848706e-05,
+      "loss": 0.0129,
+      "step": 91720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559494814110235e-05,
+      "loss": 0.0083,
+      "step": 91730
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.955944677437176e-05,
+      "loss": 0.0119,
+      "step": 91740
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.955939873463329e-05,
+      "loss": 0.0133,
+      "step": 91750
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.955935069489482e-05,
+      "loss": 0.01,
+      "step": 91760
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559302655156345e-05,
+      "loss": 0.0137,
+      "step": 91770
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559254615417875e-05,
+      "loss": 0.0143,
+      "step": 91780
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559206575679404e-05,
+      "loss": 0.0154,
+      "step": 91790
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559158535940933e-05,
+      "loss": 0.0088,
+      "step": 91800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.955911049620246e-05,
+      "loss": 0.0141,
+      "step": 91810
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.955906245646399e-05,
+      "loss": 0.0101,
+      "step": 91820
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9559014416725518e-05,
+      "loss": 0.0117,
+      "step": 91830
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9558966376987047e-05,
+      "loss": 0.0121,
+      "step": 91840
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9558918337248573e-05,
+      "loss": 0.0115,
+      "step": 91850
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9558870297510102e-05,
+      "loss": 0.0131,
+      "step": 91860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.955882225777163e-05,
+      "loss": 0.0096,
+      "step": 91870
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9558774218033157e-05,
+      "loss": 0.0124,
+      "step": 91880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9558726178294686e-05,
+      "loss": 0.012,
+      "step": 91890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9558678138556216e-05,
+      "loss": 0.0118,
+      "step": 91900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9558630098817745e-05,
+      "loss": 0.016,
+      "step": 91910
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.955858205907927e-05,
+      "loss": 0.0128,
+      "step": 91920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.95585340193408e-05,
+      "loss": 0.0098,
+      "step": 91930
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955848597960233e-05,
+      "loss": 0.013,
+      "step": 91940
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9558437939863855e-05,
+      "loss": 0.0152,
+      "step": 91950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9558389900125385e-05,
+      "loss": 0.011,
+      "step": 91960
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9558341860386914e-05,
+      "loss": 0.0112,
+      "step": 91970
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9558293820648443e-05,
+      "loss": 0.0149,
+      "step": 91980
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955824578090997e-05,
+      "loss": 0.0109,
+      "step": 91990
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9558197741171498e-05,
+      "loss": 0.0115,
+      "step": 92000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9558149701433028e-05,
+      "loss": 0.0134,
+      "step": 92010
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9558101661694557e-05,
+      "loss": 0.011,
+      "step": 92020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9558053621956083e-05,
+      "loss": 0.0124,
+      "step": 92030
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9558005582217612e-05,
+      "loss": 0.0146,
+      "step": 92040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955795754247914e-05,
+      "loss": 0.0098,
+      "step": 92050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557909502740667e-05,
+      "loss": 0.0141,
+      "step": 92060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557861463002196e-05,
+      "loss": 0.0118,
+      "step": 92070
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557813423263726e-05,
+      "loss": 0.01,
+      "step": 92080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557765383525255e-05,
+      "loss": 0.0186,
+      "step": 92090
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955771734378678e-05,
+      "loss": 0.0117,
+      "step": 92100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955766930404831e-05,
+      "loss": 0.0098,
+      "step": 92110
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955762126430984e-05,
+      "loss": 0.0098,
+      "step": 92120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557573224571365e-05,
+      "loss": 0.0123,
+      "step": 92130
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557525184832895e-05,
+      "loss": 0.0096,
+      "step": 92140
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557477145094424e-05,
+      "loss": 0.0135,
+      "step": 92150
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557429105355953e-05,
+      "loss": 0.0123,
+      "step": 92160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955738106561748e-05,
+      "loss": 0.0096,
+      "step": 92170
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557333025879008e-05,
+      "loss": 0.0104,
+      "step": 92180
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557284986140537e-05,
+      "loss": 0.0119,
+      "step": 92190
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557236946402063e-05,
+      "loss": 0.0137,
+      "step": 92200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557188906663593e-05,
+      "loss": 0.0117,
+      "step": 92210
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557140866925122e-05,
+      "loss": 0.0141,
+      "step": 92220
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955709282718665e-05,
+      "loss": 0.0097,
+      "step": 92230
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9557044787448177e-05,
+      "loss": 0.0131,
+      "step": 92240
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556996747709706e-05,
+      "loss": 0.0084,
+      "step": 92250
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556948707971236e-05,
+      "loss": 0.0096,
+      "step": 92260
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556900668232765e-05,
+      "loss": 0.01,
+      "step": 92270
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955685262849429e-05,
+      "loss": 0.0136,
+      "step": 92280
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955680458875582e-05,
+      "loss": 0.0116,
+      "step": 92290
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955675654901735e-05,
+      "loss": 0.0161,
+      "step": 92300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556708509278875e-05,
+      "loss": 0.0197,
+      "step": 92310
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556660469540404e-05,
+      "loss": 0.0114,
+      "step": 92320
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556612429801934e-05,
+      "loss": 0.013,
+      "step": 92330
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556564390063463e-05,
+      "loss": 0.0108,
+      "step": 92340
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955651635032499e-05,
+      "loss": 0.011,
+      "step": 92350
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556468310586518e-05,
+      "loss": 0.0106,
+      "step": 92360
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556420270848047e-05,
+      "loss": 0.0095,
+      "step": 92370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556372231109573e-05,
+      "loss": 0.0149,
+      "step": 92380
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556324191371103e-05,
+      "loss": 0.0107,
+      "step": 92390
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556276151632632e-05,
+      "loss": 0.0132,
+      "step": 92400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955622811189416e-05,
+      "loss": 0.0123,
+      "step": 92410
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556180072155687e-05,
+      "loss": 0.0079,
+      "step": 92420
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556132032417216e-05,
+      "loss": 0.0068,
+      "step": 92430
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556083992678746e-05,
+      "loss": 0.0109,
+      "step": 92440
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9556035952940275e-05,
+      "loss": 0.0161,
+      "step": 92450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.95559879132018e-05,
+      "loss": 0.0124,
+      "step": 92460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955593987346333e-05,
+      "loss": 0.0103,
+      "step": 92470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955589183372486e-05,
+      "loss": 0.0098,
+      "step": 92480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9555843793986385e-05,
+      "loss": 0.0127,
+      "step": 92490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9555795754247914e-05,
+      "loss": 0.0135,
+      "step": 92500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9555747714509444e-05,
+      "loss": 0.0137,
+      "step": 92510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9555699674770973e-05,
+      "loss": 0.0103,
+      "step": 92520
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.95556516350325e-05,
+      "loss": 0.0116,
+      "step": 92530
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9555603595294028e-05,
+      "loss": 0.0124,
+      "step": 92540
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 0.011,
+      "step": 92550
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9555507515817083e-05,
+      "loss": 0.0139,
+      "step": 92560
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9555459476078612e-05,
+      "loss": 0.0098,
+      "step": 92570
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9555411436340142e-05,
+      "loss": 0.0141,
+      "step": 92580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955536339660167e-05,
+      "loss": 0.0115,
+      "step": 92590
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9555320160837046e-05,
+      "loss": 0.0118,
+      "step": 92600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9555272121098575e-05,
+      "loss": 0.01,
+      "step": 92610
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.95552240813601e-05,
+      "loss": 0.013,
+      "step": 92620
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.955517604162163e-05,
+      "loss": 0.0115,
+      "step": 92630
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955512800188316e-05,
+      "loss": 0.0119,
+      "step": 92640
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955507996214469e-05,
+      "loss": 0.0126,
+      "step": 92650
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9555031922406214e-05,
+      "loss": 0.009,
+      "step": 92660
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554983882667744e-05,
+      "loss": 0.0126,
+      "step": 92670
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554935842929273e-05,
+      "loss": 0.0119,
+      "step": 92680
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554887803190802e-05,
+      "loss": 0.0142,
+      "step": 92690
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554839763452328e-05,
+      "loss": 0.0119,
+      "step": 92700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554791723713857e-05,
+      "loss": 0.0149,
+      "step": 92710
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554743683975387e-05,
+      "loss": 0.0115,
+      "step": 92720
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554695644236913e-05,
+      "loss": 0.0112,
+      "step": 92730
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554647604498442e-05,
+      "loss": 0.014,
+      "step": 92740
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955459956475997e-05,
+      "loss": 0.014,
+      "step": 92750
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.95545515250215e-05,
+      "loss": 0.0115,
+      "step": 92760
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554503485283026e-05,
+      "loss": 0.0129,
+      "step": 92770
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554455445544556e-05,
+      "loss": 0.0135,
+      "step": 92780
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554407405806085e-05,
+      "loss": 0.0116,
+      "step": 92790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955435936606761e-05,
+      "loss": 0.0168,
+      "step": 92800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955431132632914e-05,
+      "loss": 0.0131,
+      "step": 92810
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955426328659067e-05,
+      "loss": 0.0095,
+      "step": 92820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.95542152468522e-05,
+      "loss": 0.0127,
+      "step": 92830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554167207113724e-05,
+      "loss": 0.012,
+      "step": 92840
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554119167375254e-05,
+      "loss": 0.0192,
+      "step": 92850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554071127636783e-05,
+      "loss": 0.0123,
+      "step": 92860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9554023087898312e-05,
+      "loss": 0.0114,
+      "step": 92870
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553975048159838e-05,
+      "loss": 0.0093,
+      "step": 92880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553927008421367e-05,
+      "loss": 0.0109,
+      "step": 92890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553878968682897e-05,
+      "loss": 0.0106,
+      "step": 92900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553830928944423e-05,
+      "loss": 0.0126,
+      "step": 92910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553782889205952e-05,
+      "loss": 0.0129,
+      "step": 92920
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955373484946748e-05,
+      "loss": 0.0108,
+      "step": 92930
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955368680972901e-05,
+      "loss": 0.0144,
+      "step": 92940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553638769990536e-05,
+      "loss": 0.0111,
+      "step": 92950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553590730252065e-05,
+      "loss": 0.0121,
+      "step": 92960
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553542690513595e-05,
+      "loss": 0.0098,
+      "step": 92970
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955349465077512e-05,
+      "loss": 0.0119,
+      "step": 92980
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955344661103665e-05,
+      "loss": 0.0124,
+      "step": 92990
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955339857129818e-05,
+      "loss": 0.0173,
+      "step": 93000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955335053155971e-05,
+      "loss": 0.0149,
+      "step": 93010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553302491821234e-05,
+      "loss": 0.0094,
+      "step": 93020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553254452082764e-05,
+      "loss": 0.0076,
+      "step": 93030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553206412344293e-05,
+      "loss": 0.0099,
+      "step": 93040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553158372605822e-05,
+      "loss": 0.0123,
+      "step": 93050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553110332867348e-05,
+      "loss": 0.0085,
+      "step": 93060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553062293128877e-05,
+      "loss": 0.0082,
+      "step": 93070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9553014253390407e-05,
+      "loss": 0.0109,
+      "step": 93080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552966213651932e-05,
+      "loss": 0.0091,
+      "step": 93090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552918173913462e-05,
+      "loss": 0.0133,
+      "step": 93100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955287013417499e-05,
+      "loss": 0.0093,
+      "step": 93110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955282209443652e-05,
+      "loss": 0.0125,
+      "step": 93120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552774054698046e-05,
+      "loss": 0.0129,
+      "step": 93130
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552726014959575e-05,
+      "loss": 0.0128,
+      "step": 93140
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552677975221105e-05,
+      "loss": 0.0112,
+      "step": 93150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955262993548263e-05,
+      "loss": 0.0123,
+      "step": 93160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955258189574416e-05,
+      "loss": 0.0097,
+      "step": 93170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955253385600569e-05,
+      "loss": 0.0096,
+      "step": 93180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955248581626722e-05,
+      "loss": 0.0107,
+      "step": 93190
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552437776528744e-05,
+      "loss": 0.0103,
+      "step": 93200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552389736790274e-05,
+      "loss": 0.0118,
+      "step": 93210
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552341697051803e-05,
+      "loss": 0.0122,
+      "step": 93220
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552293657313332e-05,
+      "loss": 0.0148,
+      "step": 93230
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552245617574858e-05,
+      "loss": 0.0109,
+      "step": 93240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552197577836387e-05,
+      "loss": 0.0094,
+      "step": 93250
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552149538097916e-05,
+      "loss": 0.0149,
+      "step": 93260
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9552101498359442e-05,
+      "loss": 0.0133,
+      "step": 93270
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955205345862097e-05,
+      "loss": 0.012,
+      "step": 93280
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.95520054188825e-05,
+      "loss": 0.0133,
+      "step": 93290
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.955195737914403e-05,
+      "loss": 0.0116,
+      "step": 93300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9551909339405556e-05,
+      "loss": 0.0108,
+      "step": 93310
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9551861299667085e-05,
+      "loss": 0.0153,
+      "step": 93320
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9551813259928615e-05,
+      "loss": 0.014,
+      "step": 93330
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955176522019014e-05,
+      "loss": 0.0118,
+      "step": 93340
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955171718045167e-05,
+      "loss": 0.0111,
+      "step": 93350
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.95516691407132e-05,
+      "loss": 0.0136,
+      "step": 93360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9551621100974728e-05,
+      "loss": 0.0091,
+      "step": 93370
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9551573061236254e-05,
+      "loss": 0.0143,
+      "step": 93380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9551525021497783e-05,
+      "loss": 0.0121,
+      "step": 93390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9551476981759313e-05,
+      "loss": 0.0136,
+      "step": 93400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9551428942020842e-05,
+      "loss": 0.0097,
+      "step": 93410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9551380902282368e-05,
+      "loss": 0.0096,
+      "step": 93420
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9551332862543897e-05,
+      "loss": 0.0087,
+      "step": 93430
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9551284822805426e-05,
+      "loss": 0.014,
+      "step": 93440
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9551236783066952e-05,
+      "loss": 0.014,
+      "step": 93450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955118874332848e-05,
+      "loss": 0.0124,
+      "step": 93460
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955114070359001e-05,
+      "loss": 0.012,
+      "step": 93470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955109266385154e-05,
+      "loss": 0.0118,
+      "step": 93480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9551044624113066e-05,
+      "loss": 0.0152,
+      "step": 93490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550996584374595e-05,
+      "loss": 0.0096,
+      "step": 93500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550948544636124e-05,
+      "loss": 0.0117,
+      "step": 93510
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955090050489765e-05,
+      "loss": 0.0106,
+      "step": 93520
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955085246515918e-05,
+      "loss": 0.01,
+      "step": 93530
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955080442542071e-05,
+      "loss": 0.0208,
+      "step": 93540
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550756385682238e-05,
+      "loss": 0.0077,
+      "step": 93550
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550708345943764e-05,
+      "loss": 0.0117,
+      "step": 93560
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550660306205293e-05,
+      "loss": 0.0155,
+      "step": 93570
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550612266466823e-05,
+      "loss": 0.0075,
+      "step": 93580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550564226728352e-05,
+      "loss": 0.0174,
+      "step": 93590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550516186989878e-05,
+      "loss": 0.0079,
+      "step": 93600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955046814725141e-05,
+      "loss": 0.018,
+      "step": 93610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550420107512936e-05,
+      "loss": 0.0128,
+      "step": 93620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550372067774466e-05,
+      "loss": 0.012,
+      "step": 93630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550324028035995e-05,
+      "loss": 0.012,
+      "step": 93640
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955027598829752e-05,
+      "loss": 0.0117,
+      "step": 93650
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955022794855905e-05,
+      "loss": 0.0098,
+      "step": 93660
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955017990882058e-05,
+      "loss": 0.0118,
+      "step": 93670
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.955013186908211e-05,
+      "loss": 0.0101,
+      "step": 93680
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550083829343634e-05,
+      "loss": 0.0128,
+      "step": 93690
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9550035789605164e-05,
+      "loss": 0.0079,
+      "step": 93700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549987749866693e-05,
+      "loss": 0.0145,
+      "step": 93710
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549939710128222e-05,
+      "loss": 0.0085,
+      "step": 93720
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549891670389748e-05,
+      "loss": 0.0126,
+      "step": 93730
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549843630651277e-05,
+      "loss": 0.0139,
+      "step": 93740
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549795590912807e-05,
+      "loss": 0.0112,
+      "step": 93750
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549747551174333e-05,
+      "loss": 0.0107,
+      "step": 93760
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549699511435862e-05,
+      "loss": 0.0084,
+      "step": 93770
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.954965147169739e-05,
+      "loss": 0.0111,
+      "step": 93780
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.954960343195892e-05,
+      "loss": 0.015,
+      "step": 93790
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549555392220446e-05,
+      "loss": 0.0091,
+      "step": 93800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549507352481975e-05,
+      "loss": 0.0122,
+      "step": 93810
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549459312743505e-05,
+      "loss": 0.0119,
+      "step": 93820
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.954941127300503e-05,
+      "loss": 0.012,
+      "step": 93830
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.954936323326656e-05,
+      "loss": 0.0134,
+      "step": 93840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.954931519352809e-05,
+      "loss": 0.0102,
+      "step": 93850
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.954926715378962e-05,
+      "loss": 0.0137,
+      "step": 93860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549219114051144e-05,
+      "loss": 0.0093,
+      "step": 93870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549171074312674e-05,
+      "loss": 0.0104,
+      "step": 93880
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549123034574203e-05,
+      "loss": 0.009,
+      "step": 93890
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549074994835732e-05,
+      "loss": 0.01,
+      "step": 93900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9549026955097258e-05,
+      "loss": 0.0129,
+      "step": 93910
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9548978915358787e-05,
+      "loss": 0.0091,
+      "step": 93920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9548930875620317e-05,
+      "loss": 0.0113,
+      "step": 93930
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9548882835881842e-05,
+      "loss": 0.0106,
+      "step": 93940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9548834796143372e-05,
+      "loss": 0.0132,
+      "step": 93950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.95487867564049e-05,
+      "loss": 0.0103,
+      "step": 93960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.954873871666643e-05,
+      "loss": 0.0123,
+      "step": 93970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9548690676927956e-05,
+      "loss": 0.009,
+      "step": 93980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9548642637189485e-05,
+      "loss": 0.0143,
+      "step": 93990
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9548594597451015e-05,
+      "loss": 0.011,
+      "step": 94000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.954854655771254e-05,
+      "loss": 0.0096,
+      "step": 94010
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954849851797407e-05,
+      "loss": 0.0084,
+      "step": 94020
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.95484504782356e-05,
+      "loss": 0.0134,
+      "step": 94030
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954840243849713e-05,
+      "loss": 0.0123,
+      "step": 94040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9548354398758654e-05,
+      "loss": 0.0106,
+      "step": 94050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9548306359020184e-05,
+      "loss": 0.0089,
+      "step": 94060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9548258319281713e-05,
+      "loss": 0.0114,
+      "step": 94070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9548210279543242e-05,
+      "loss": 0.0117,
+      "step": 94080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9548162239804768e-05,
+      "loss": 0.0122,
+      "step": 94090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9548114200066297e-05,
+      "loss": 0.0095,
+      "step": 94100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9548066160327826e-05,
+      "loss": 0.0114,
+      "step": 94110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9548018120589352e-05,
+      "loss": 0.01,
+      "step": 94120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954797008085088e-05,
+      "loss": 0.0121,
+      "step": 94130
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954792204111241e-05,
+      "loss": 0.0097,
+      "step": 94140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954787400137394e-05,
+      "loss": 0.0154,
+      "step": 94150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9547825961635466e-05,
+      "loss": 0.0123,
+      "step": 94160
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9547777921896995e-05,
+      "loss": 0.0124,
+      "step": 94170
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9547729882158525e-05,
+      "loss": 0.0088,
+      "step": 94180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954768184242005e-05,
+      "loss": 0.0106,
+      "step": 94190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954763380268158e-05,
+      "loss": 0.0109,
+      "step": 94200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954758576294311e-05,
+      "loss": 0.0107,
+      "step": 94210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9547537723204638e-05,
+      "loss": 0.0135,
+      "step": 94220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9547489683466164e-05,
+      "loss": 0.016,
+      "step": 94230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9547441643727693e-05,
+      "loss": 0.016,
+      "step": 94240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9547393603989223e-05,
+      "loss": 0.0143,
+      "step": 94250
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9547345564250752e-05,
+      "loss": 0.0086,
+      "step": 94260
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9547297524512278e-05,
+      "loss": 0.0113,
+      "step": 94270
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9547249484773807e-05,
+      "loss": 0.0111,
+      "step": 94280
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9547201445035336e-05,
+      "loss": 0.0127,
+      "step": 94290
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9547153405296862e-05,
+      "loss": 0.0104,
+      "step": 94300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954710536555839e-05,
+      "loss": 0.0111,
+      "step": 94310
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954705732581992e-05,
+      "loss": 0.0117,
+      "step": 94320
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954700928608145e-05,
+      "loss": 0.0139,
+      "step": 94330
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546961246342976e-05,
+      "loss": 0.0123,
+      "step": 94340
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546913206604505e-05,
+      "loss": 0.0127,
+      "step": 94350
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546865166866034e-05,
+      "loss": 0.0102,
+      "step": 94360
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954681712712756e-05,
+      "loss": 0.0113,
+      "step": 94370
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954676908738909e-05,
+      "loss": 0.0103,
+      "step": 94380
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954672104765062e-05,
+      "loss": 0.0101,
+      "step": 94390
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546673007912148e-05,
+      "loss": 0.0116,
+      "step": 94400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546624968173674e-05,
+      "loss": 0.0116,
+      "step": 94410
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546576928435203e-05,
+      "loss": 0.0099,
+      "step": 94420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546528888696733e-05,
+      "loss": 0.0247,
+      "step": 94430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546480848958262e-05,
+      "loss": 0.0097,
+      "step": 94440
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546437613193637e-05,
+      "loss": 0.011,
+      "step": 94450
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546389573455166e-05,
+      "loss": 0.0131,
+      "step": 94460
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954634153371669e-05,
+      "loss": 0.012,
+      "step": 94470
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954629349397822e-05,
+      "loss": 0.0164,
+      "step": 94480
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954624545423975e-05,
+      "loss": 0.0186,
+      "step": 94490
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954619741450128e-05,
+      "loss": 0.009,
+      "step": 94500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546149374762805e-05,
+      "loss": 0.0147,
+      "step": 94510
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546101335024335e-05,
+      "loss": 0.0163,
+      "step": 94520
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9546053295285864e-05,
+      "loss": 0.0104,
+      "step": 94530
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954600525554739e-05,
+      "loss": 0.0102,
+      "step": 94540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954595721580892e-05,
+      "loss": 0.0141,
+      "step": 94550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954590917607045e-05,
+      "loss": 0.0091,
+      "step": 94560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9545861136331978e-05,
+      "loss": 0.0104,
+      "step": 94570
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9545813096593503e-05,
+      "loss": 0.0123,
+      "step": 94580
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9545765056855033e-05,
+      "loss": 0.0146,
+      "step": 94590
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9545717017116562e-05,
+      "loss": 0.01,
+      "step": 94600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9545668977378088e-05,
+      "loss": 0.0106,
+      "step": 94610
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9545620937639617e-05,
+      "loss": 0.0112,
+      "step": 94620
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9545572897901146e-05,
+      "loss": 0.0147,
+      "step": 94630
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9545524858162676e-05,
+      "loss": 0.0078,
+      "step": 94640
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.95454768184242e-05,
+      "loss": 0.0173,
+      "step": 94650
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954542877868573e-05,
+      "loss": 0.0128,
+      "step": 94660
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954538073894726e-05,
+      "loss": 0.0105,
+      "step": 94670
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.954533269920879e-05,
+      "loss": 0.0114,
+      "step": 94680
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9545284659470315e-05,
+      "loss": 0.01,
+      "step": 94690
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9545236619731845e-05,
+      "loss": 0.0101,
+      "step": 94700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9545188579993374e-05,
+      "loss": 0.0123,
+      "step": 94710
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.95451405402549e-05,
+      "loss": 0.0102,
+      "step": 94720
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954509250051643e-05,
+      "loss": 0.0116,
+      "step": 94730
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9545044460777958e-05,
+      "loss": 0.0104,
+      "step": 94740
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544996421039488e-05,
+      "loss": 0.0097,
+      "step": 94750
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544948381301013e-05,
+      "loss": 0.0121,
+      "step": 94760
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544900341562543e-05,
+      "loss": 0.0126,
+      "step": 94770
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544852301824072e-05,
+      "loss": 0.0102,
+      "step": 94780
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544804262085598e-05,
+      "loss": 0.0094,
+      "step": 94790
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544756222347127e-05,
+      "loss": 0.0103,
+      "step": 94800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544708182608656e-05,
+      "loss": 0.0086,
+      "step": 94810
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544660142870186e-05,
+      "loss": 0.0105,
+      "step": 94820
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954461210313171e-05,
+      "loss": 0.0101,
+      "step": 94830
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954456406339324e-05,
+      "loss": 0.0098,
+      "step": 94840
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954451602365477e-05,
+      "loss": 0.0129,
+      "step": 94850
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.95444679839163e-05,
+      "loss": 0.0137,
+      "step": 94860
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544419944177825e-05,
+      "loss": 0.0087,
+      "step": 94870
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544371904439354e-05,
+      "loss": 0.012,
+      "step": 94880
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544323864700884e-05,
+      "loss": 0.0113,
+      "step": 94890
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954427582496241e-05,
+      "loss": 0.0133,
+      "step": 94900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954422778522394e-05,
+      "loss": 0.0111,
+      "step": 94910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544179745485468e-05,
+      "loss": 0.01,
+      "step": 94920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544131705746997e-05,
+      "loss": 0.01,
+      "step": 94930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544083666008523e-05,
+      "loss": 0.0092,
+      "step": 94940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9544035626270053e-05,
+      "loss": 0.011,
+      "step": 94950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543987586531582e-05,
+      "loss": 0.0146,
+      "step": 94960
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543939546793108e-05,
+      "loss": 0.0144,
+      "step": 94970
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543891507054637e-05,
+      "loss": 0.0095,
+      "step": 94980
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543843467316166e-05,
+      "loss": 0.0163,
+      "step": 94990
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543795427577696e-05,
+      "loss": 0.0152,
+      "step": 95000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954374738783922e-05,
+      "loss": 0.0074,
+      "step": 95010
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954369934810075e-05,
+      "loss": 0.0119,
+      "step": 95020
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954365130836228e-05,
+      "loss": 0.0103,
+      "step": 95030
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543603268623806e-05,
+      "loss": 0.0107,
+      "step": 95040
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543555228885335e-05,
+      "loss": 0.0139,
+      "step": 95050
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543507189146864e-05,
+      "loss": 0.0119,
+      "step": 95060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543459149408394e-05,
+      "loss": 0.0118,
+      "step": 95070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954341110966992e-05,
+      "loss": 0.0204,
+      "step": 95080
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954336306993145e-05,
+      "loss": 0.008,
+      "step": 95090
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543315030192978e-05,
+      "loss": 0.0123,
+      "step": 95100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543266990454507e-05,
+      "loss": 0.0144,
+      "step": 95110
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543218950716033e-05,
+      "loss": 0.0164,
+      "step": 95120
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543170910977562e-05,
+      "loss": 0.0148,
+      "step": 95130
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543122871239092e-05,
+      "loss": 0.0103,
+      "step": 95140
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543074831500618e-05,
+      "loss": 0.0101,
+      "step": 95150
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9543026791762147e-05,
+      "loss": 0.0107,
+      "step": 95160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542978752023676e-05,
+      "loss": 0.0107,
+      "step": 95170
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542930712285205e-05,
+      "loss": 0.0075,
+      "step": 95180
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954288267254673e-05,
+      "loss": 0.0173,
+      "step": 95190
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954283463280826e-05,
+      "loss": 0.0172,
+      "step": 95200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954278659306979e-05,
+      "loss": 0.0127,
+      "step": 95210
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542738553331316e-05,
+      "loss": 0.0081,
+      "step": 95220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542690513592845e-05,
+      "loss": 0.0147,
+      "step": 95230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542642473854374e-05,
+      "loss": 0.0116,
+      "step": 95240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542594434115904e-05,
+      "loss": 0.0144,
+      "step": 95250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954254639437743e-05,
+      "loss": 0.0112,
+      "step": 95260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954249835463896e-05,
+      "loss": 0.0108,
+      "step": 95270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542450314900488e-05,
+      "loss": 0.0102,
+      "step": 95280
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542402275162017e-05,
+      "loss": 0.0096,
+      "step": 95290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542354235423543e-05,
+      "loss": 0.0118,
+      "step": 95300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542306195685072e-05,
+      "loss": 0.0078,
+      "step": 95310
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.95422581559466e-05,
+      "loss": 0.015,
+      "step": 95320
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542210116208128e-05,
+      "loss": 0.0129,
+      "step": 95330
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542162076469657e-05,
+      "loss": 0.0113,
+      "step": 95340
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542114036731186e-05,
+      "loss": 0.0092,
+      "step": 95350
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9542065996992715e-05,
+      "loss": 0.0094,
+      "step": 95360
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954201795725424e-05,
+      "loss": 0.0141,
+      "step": 95370
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.954196991751577e-05,
+      "loss": 0.0104,
+      "step": 95380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.95419218777773e-05,
+      "loss": 0.0176,
+      "step": 95390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9541873838038826e-05,
+      "loss": 0.0114,
+      "step": 95400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9541825798300355e-05,
+      "loss": 0.0094,
+      "step": 95410
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9541777758561884e-05,
+      "loss": 0.0122,
+      "step": 95420
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9541729718823413e-05,
+      "loss": 0.0103,
+      "step": 95430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.954168167908494e-05,
+      "loss": 0.0116,
+      "step": 95440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.954163363934647e-05,
+      "loss": 0.0163,
+      "step": 95450
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9541585599607998e-05,
+      "loss": 0.0102,
+      "step": 95460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9541537559869527e-05,
+      "loss": 0.0104,
+      "step": 95470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9541489520131053e-05,
+      "loss": 0.0151,
+      "step": 95480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9541441480392582e-05,
+      "loss": 0.0106,
+      "step": 95490
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.954139344065411e-05,
+      "loss": 0.0128,
+      "step": 95500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9541345400915637e-05,
+      "loss": 0.0138,
+      "step": 95510
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9541297361177167e-05,
+      "loss": 0.0122,
+      "step": 95520
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9541249321438696e-05,
+      "loss": 0.0112,
+      "step": 95530
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9541201281700225e-05,
+      "loss": 0.0104,
+      "step": 95540
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.954115324196175e-05,
+      "loss": 0.0087,
+      "step": 95550
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.954110520222328e-05,
+      "loss": 0.0114,
+      "step": 95560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.954105716248481e-05,
+      "loss": 0.01,
+      "step": 95570
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9541009122746336e-05,
+      "loss": 0.0128,
+      "step": 95580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540961083007865e-05,
+      "loss": 0.0101,
+      "step": 95590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540913043269394e-05,
+      "loss": 0.0164,
+      "step": 95600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540865003530923e-05,
+      "loss": 0.0092,
+      "step": 95610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.954081696379245e-05,
+      "loss": 0.0136,
+      "step": 95620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.954076892405398e-05,
+      "loss": 0.0119,
+      "step": 95630
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540720884315508e-05,
+      "loss": 0.0117,
+      "step": 95640
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540672844577037e-05,
+      "loss": 0.0097,
+      "step": 95650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540624804838563e-05,
+      "loss": 0.0099,
+      "step": 95660
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540576765100092e-05,
+      "loss": 0.0125,
+      "step": 95670
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.954052872536162e-05,
+      "loss": 0.0112,
+      "step": 95680
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540480685623147e-05,
+      "loss": 0.0121,
+      "step": 95690
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540432645884677e-05,
+      "loss": 0.0094,
+      "step": 95700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540384606146206e-05,
+      "loss": 0.0126,
+      "step": 95710
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540336566407735e-05,
+      "loss": 0.0117,
+      "step": 95720
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.954028852666926e-05,
+      "loss": 0.0125,
+      "step": 95730
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.954024048693079e-05,
+      "loss": 0.0112,
+      "step": 95740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.954019244719232e-05,
+      "loss": 0.0127,
+      "step": 95750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540144407453845e-05,
+      "loss": 0.0097,
+      "step": 95760
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540096367715375e-05,
+      "loss": 0.0103,
+      "step": 95770
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540048327976904e-05,
+      "loss": 0.0075,
+      "step": 95780
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9540000288238433e-05,
+      "loss": 0.0138,
+      "step": 95790
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.953995224849996e-05,
+      "loss": 0.0078,
+      "step": 95800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.953990420876149e-05,
+      "loss": 0.0106,
+      "step": 95810
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539856169023018e-05,
+      "loss": 0.0097,
+      "step": 95820
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539808129284547e-05,
+      "loss": 0.0086,
+      "step": 95830
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539760089546073e-05,
+      "loss": 0.0114,
+      "step": 95840
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539712049807602e-05,
+      "loss": 0.0152,
+      "step": 95850
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.953966401006913e-05,
+      "loss": 0.0167,
+      "step": 95860
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539615970330657e-05,
+      "loss": 0.0114,
+      "step": 95870
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539567930592187e-05,
+      "loss": 0.0116,
+      "step": 95880
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539519890853716e-05,
+      "loss": 0.0113,
+      "step": 95890
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539471851115245e-05,
+      "loss": 0.0144,
+      "step": 95900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.953942381137677e-05,
+      "loss": 0.0088,
+      "step": 95910
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.95393757716383e-05,
+      "loss": 0.01,
+      "step": 95920
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.953932773189983e-05,
+      "loss": 0.0154,
+      "step": 95930
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539279692161355e-05,
+      "loss": 0.0103,
+      "step": 95940
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539231652422885e-05,
+      "loss": 0.011,
+      "step": 95950
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539183612684414e-05,
+      "loss": 0.0084,
+      "step": 95960
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539135572945943e-05,
+      "loss": 0.012,
+      "step": 95970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.953908753320747e-05,
+      "loss": 0.0117,
+      "step": 95980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9539039493469e-05,
+      "loss": 0.0132,
+      "step": 95990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9538991453730528e-05,
+      "loss": 0.0099,
+      "step": 96000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9538943413992057e-05,
+      "loss": 0.0077,
+      "step": 96010
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9538895374253583e-05,
+      "loss": 0.0185,
+      "step": 96020
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9538847334515112e-05,
+      "loss": 0.0123,
+      "step": 96030
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.953879929477664e-05,
+      "loss": 0.0138,
+      "step": 96040
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9538751255038167e-05,
+      "loss": 0.0095,
+      "step": 96050
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9538703215299696e-05,
+      "loss": 0.0124,
+      "step": 96060
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9538655175561226e-05,
+      "loss": 0.013,
+      "step": 96070
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9538607135822755e-05,
+      "loss": 0.0074,
+      "step": 96080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.953855909608428e-05,
+      "loss": 0.0075,
+      "step": 96090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.953851105634581e-05,
+      "loss": 0.0117,
+      "step": 96100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953846301660734e-05,
+      "loss": 0.0099,
+      "step": 96110
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9538414976868865e-05,
+      "loss": 0.0114,
+      "step": 96120
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9538366937130395e-05,
+      "loss": 0.0099,
+      "step": 96130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9538318897391924e-05,
+      "loss": 0.0103,
+      "step": 96140
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9538270857653453e-05,
+      "loss": 0.0079,
+      "step": 96150
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953822281791498e-05,
+      "loss": 0.0119,
+      "step": 96160
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9538174778176508e-05,
+      "loss": 0.0133,
+      "step": 96170
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9538126738438038e-05,
+      "loss": 0.0102,
+      "step": 96180
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9538078698699567e-05,
+      "loss": 0.0098,
+      "step": 96190
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9538030658961093e-05,
+      "loss": 0.0127,
+      "step": 96200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537982619222622e-05,
+      "loss": 0.013,
+      "step": 96210
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953793457948415e-05,
+      "loss": 0.0142,
+      "step": 96220
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537886539745677e-05,
+      "loss": 0.0072,
+      "step": 96230
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537838500007206e-05,
+      "loss": 0.0105,
+      "step": 96240
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537790460268736e-05,
+      "loss": 0.0126,
+      "step": 96250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537742420530265e-05,
+      "loss": 0.0083,
+      "step": 96260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953769438079179e-05,
+      "loss": 0.0111,
+      "step": 96270
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953764634105332e-05,
+      "loss": 0.0126,
+      "step": 96280
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953759830131485e-05,
+      "loss": 0.0111,
+      "step": 96290
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537550261576375e-05,
+      "loss": 0.0108,
+      "step": 96300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537502221837905e-05,
+      "loss": 0.0124,
+      "step": 96310
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537454182099434e-05,
+      "loss": 0.0122,
+      "step": 96320
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537406142360963e-05,
+      "loss": 0.0122,
+      "step": 96330
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953735810262249e-05,
+      "loss": 0.0095,
+      "step": 96340
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537310062884018e-05,
+      "loss": 0.011,
+      "step": 96350
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537262023145547e-05,
+      "loss": 0.0134,
+      "step": 96360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537213983407077e-05,
+      "loss": 0.011,
+      "step": 96370
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537165943668603e-05,
+      "loss": 0.0118,
+      "step": 96380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537117903930132e-05,
+      "loss": 0.0125,
+      "step": 96390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953706986419166e-05,
+      "loss": 0.0114,
+      "step": 96400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9537021824453187e-05,
+      "loss": 0.0114,
+      "step": 96410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536973784714716e-05,
+      "loss": 0.0122,
+      "step": 96420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536925744976246e-05,
+      "loss": 0.011,
+      "step": 96430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536877705237775e-05,
+      "loss": 0.0141,
+      "step": 96440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.95368296654993e-05,
+      "loss": 0.0097,
+      "step": 96450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953678162576083e-05,
+      "loss": 0.0139,
+      "step": 96460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953673358602236e-05,
+      "loss": 0.0118,
+      "step": 96470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536685546283885e-05,
+      "loss": 0.0137,
+      "step": 96480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536637506545414e-05,
+      "loss": 0.0131,
+      "step": 96490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536589466806944e-05,
+      "loss": 0.0106,
+      "step": 96500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536541427068473e-05,
+      "loss": 0.0125,
+      "step": 96510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953649338733e-05,
+      "loss": 0.0136,
+      "step": 96520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536445347591528e-05,
+      "loss": 0.0125,
+      "step": 96530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536397307853057e-05,
+      "loss": 0.0104,
+      "step": 96540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536349268114587e-05,
+      "loss": 0.0175,
+      "step": 96550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536301228376113e-05,
+      "loss": 0.0141,
+      "step": 96560
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536253188637642e-05,
+      "loss": 0.0104,
+      "step": 96570
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953620514889917e-05,
+      "loss": 0.0141,
+      "step": 96580
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536157109160697e-05,
+      "loss": 0.0149,
+      "step": 96590
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536109069422226e-05,
+      "loss": 0.0112,
+      "step": 96600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536061029683755e-05,
+      "loss": 0.0123,
+      "step": 96610
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9536012989945285e-05,
+      "loss": 0.0116,
+      "step": 96620
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953596495020681e-05,
+      "loss": 0.0124,
+      "step": 96630
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953591691046834e-05,
+      "loss": 0.0098,
+      "step": 96640
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953586887072987e-05,
+      "loss": 0.0112,
+      "step": 96650
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9535820830991395e-05,
+      "loss": 0.0124,
+      "step": 96660
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9535772791252924e-05,
+      "loss": 0.0129,
+      "step": 96670
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9535724751514454e-05,
+      "loss": 0.0136,
+      "step": 96680
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9535676711775983e-05,
+      "loss": 0.0148,
+      "step": 96690
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9535628672037512e-05,
+      "loss": 0.0112,
+      "step": 96700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953558063229904e-05,
+      "loss": 0.0108,
+      "step": 96710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9535532592560567e-05,
+      "loss": 0.0118,
+      "step": 96720
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9535484552822097e-05,
+      "loss": 0.0098,
+      "step": 96730
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9535436513083626e-05,
+      "loss": 0.0147,
+      "step": 96740
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9535388473345155e-05,
+      "loss": 0.0155,
+      "step": 96750
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953534043360668e-05,
+      "loss": 0.0114,
+      "step": 96760
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953529239386821e-05,
+      "loss": 0.0089,
+      "step": 96770
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.953524435412974e-05,
+      "loss": 0.0137,
+      "step": 96780
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9535196314391265e-05,
+      "loss": 0.0085,
+      "step": 96790
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9535148274652795e-05,
+      "loss": 0.0112,
+      "step": 96800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9535100234914324e-05,
+      "loss": 0.0135,
+      "step": 96810
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9535052195175853e-05,
+      "loss": 0.0121,
+      "step": 96820
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953500415543738e-05,
+      "loss": 0.0102,
+      "step": 96830
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953495611569891e-05,
+      "loss": 0.0114,
+      "step": 96840
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534908075960438e-05,
+      "loss": 0.0086,
+      "step": 96850
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534860036221967e-05,
+      "loss": 0.0225,
+      "step": 96860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534811996483493e-05,
+      "loss": 0.0133,
+      "step": 96870
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534763956745022e-05,
+      "loss": 0.0096,
+      "step": 96880
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953471591700655e-05,
+      "loss": 0.0115,
+      "step": 96890
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534667877268077e-05,
+      "loss": 0.0109,
+      "step": 96900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534619837529606e-05,
+      "loss": 0.0139,
+      "step": 96910
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534571797791136e-05,
+      "loss": 0.0121,
+      "step": 96920
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534523758052665e-05,
+      "loss": 0.0109,
+      "step": 96930
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953447571831419e-05,
+      "loss": 0.0092,
+      "step": 96940
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953442767857572e-05,
+      "loss": 0.0123,
+      "step": 96950
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953437963883725e-05,
+      "loss": 0.0089,
+      "step": 96960
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534331599098775e-05,
+      "loss": 0.0128,
+      "step": 96970
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534283559360305e-05,
+      "loss": 0.0128,
+      "step": 96980
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534235519621834e-05,
+      "loss": 0.0103,
+      "step": 96990
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534187479883363e-05,
+      "loss": 0.0126,
+      "step": 97000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953413944014489e-05,
+      "loss": 0.0104,
+      "step": 97010
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534091400406418e-05,
+      "loss": 0.0096,
+      "step": 97020
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9534043360667948e-05,
+      "loss": 0.0098,
+      "step": 97030
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533995320929477e-05,
+      "loss": 0.0109,
+      "step": 97040
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533947281191003e-05,
+      "loss": 0.0078,
+      "step": 97050
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533899241452532e-05,
+      "loss": 0.0115,
+      "step": 97060
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953385120171406e-05,
+      "loss": 0.0122,
+      "step": 97070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533803161975587e-05,
+      "loss": 0.0096,
+      "step": 97080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533755122237116e-05,
+      "loss": 0.0118,
+      "step": 97090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533707082498646e-05,
+      "loss": 0.0115,
+      "step": 97100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533659042760175e-05,
+      "loss": 0.0137,
+      "step": 97110
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.95336110030217e-05,
+      "loss": 0.0118,
+      "step": 97120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953356296328323e-05,
+      "loss": 0.0103,
+      "step": 97130
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953351492354476e-05,
+      "loss": 0.0132,
+      "step": 97140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533466883806285e-05,
+      "loss": 0.0111,
+      "step": 97150
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533418844067815e-05,
+      "loss": 0.0163,
+      "step": 97160
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533370804329344e-05,
+      "loss": 0.0123,
+      "step": 97170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533322764590873e-05,
+      "loss": 0.0152,
+      "step": 97180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.95332747248524e-05,
+      "loss": 0.0093,
+      "step": 97190
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533226685113928e-05,
+      "loss": 0.0126,
+      "step": 97200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533178645375457e-05,
+      "loss": 0.0104,
+      "step": 97210
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533130605636987e-05,
+      "loss": 0.0113,
+      "step": 97220
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533082565898513e-05,
+      "loss": 0.0161,
+      "step": 97230
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9533034526160042e-05,
+      "loss": 0.0121,
+      "step": 97240
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953298648642157e-05,
+      "loss": 0.0166,
+      "step": 97250
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532938446683097e-05,
+      "loss": 0.0098,
+      "step": 97260
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532890406944626e-05,
+      "loss": 0.0093,
+      "step": 97270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532842367206156e-05,
+      "loss": 0.0093,
+      "step": 97280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532794327467685e-05,
+      "loss": 0.0097,
+      "step": 97290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953274628772921e-05,
+      "loss": 0.0156,
+      "step": 97300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953269824799074e-05,
+      "loss": 0.0157,
+      "step": 97310
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953265020825227e-05,
+      "loss": 0.0121,
+      "step": 97320
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532602168513795e-05,
+      "loss": 0.0106,
+      "step": 97330
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532554128775324e-05,
+      "loss": 0.0102,
+      "step": 97340
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532506089036854e-05,
+      "loss": 0.0141,
+      "step": 97350
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532458049298383e-05,
+      "loss": 0.0117,
+      "step": 97360
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953241000955991e-05,
+      "loss": 0.014,
+      "step": 97370
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532361969821438e-05,
+      "loss": 0.009,
+      "step": 97380
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532313930082967e-05,
+      "loss": 0.0128,
+      "step": 97390
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532265890344497e-05,
+      "loss": 0.0101,
+      "step": 97400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532217850606023e-05,
+      "loss": 0.0194,
+      "step": 97410
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532169810867552e-05,
+      "loss": 0.0102,
+      "step": 97420
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953212177112908e-05,
+      "loss": 0.0127,
+      "step": 97430
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532073731390607e-05,
+      "loss": 0.0103,
+      "step": 97440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9532025691652136e-05,
+      "loss": 0.0068,
+      "step": 97450
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9531977651913666e-05,
+      "loss": 0.0104,
+      "step": 97460
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9531929612175195e-05,
+      "loss": 0.0132,
+      "step": 97470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.953188157243672e-05,
+      "loss": 0.0119,
+      "step": 97480
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.953183353269825e-05,
+      "loss": 0.0104,
+      "step": 97490
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.953178549295978e-05,
+      "loss": 0.0081,
+      "step": 97500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531737453221305e-05,
+      "loss": 0.0116,
+      "step": 97510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531689413482834e-05,
+      "loss": 0.0122,
+      "step": 97520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531641373744364e-05,
+      "loss": 0.0131,
+      "step": 97530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531593334005893e-05,
+      "loss": 0.0162,
+      "step": 97540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.953154529426742e-05,
+      "loss": 0.0109,
+      "step": 97550
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531497254528948e-05,
+      "loss": 0.0088,
+      "step": 97560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531449214790477e-05,
+      "loss": 0.0154,
+      "step": 97570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531401175052007e-05,
+      "loss": 0.0087,
+      "step": 97580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531353135313532e-05,
+      "loss": 0.0157,
+      "step": 97590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531305095575062e-05,
+      "loss": 0.0096,
+      "step": 97600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.953125705583659e-05,
+      "loss": 0.007,
+      "step": 97610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531209016098117e-05,
+      "loss": 0.0078,
+      "step": 97620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531160976359646e-05,
+      "loss": 0.0113,
+      "step": 97630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531112936621175e-05,
+      "loss": 0.0121,
+      "step": 97640
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9531064896882705e-05,
+      "loss": 0.0091,
+      "step": 97650
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.953101685714423e-05,
+      "loss": 0.0091,
+      "step": 97660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.953096881740576e-05,
+      "loss": 0.0096,
+      "step": 97670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.953092077766729e-05,
+      "loss": 0.0159,
+      "step": 97680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530872737928815e-05,
+      "loss": 0.0098,
+      "step": 97690
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530824698190344e-05,
+      "loss": 0.0133,
+      "step": 97700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530776658451874e-05,
+      "loss": 0.0096,
+      "step": 97710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530728618713403e-05,
+      "loss": 0.0124,
+      "step": 97720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.953068057897493e-05,
+      "loss": 0.0118,
+      "step": 97730
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530632539236458e-05,
+      "loss": 0.0126,
+      "step": 97740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530584499497987e-05,
+      "loss": 0.0128,
+      "step": 97750
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530536459759516e-05,
+      "loss": 0.0127,
+      "step": 97760
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530488420021042e-05,
+      "loss": 0.0171,
+      "step": 97770
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.953044038028257e-05,
+      "loss": 0.0131,
+      "step": 97780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.95303923405441e-05,
+      "loss": 0.0091,
+      "step": 97790
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530344300805627e-05,
+      "loss": 0.0132,
+      "step": 97800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530296261067156e-05,
+      "loss": 0.0101,
+      "step": 97810
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530248221328685e-05,
+      "loss": 0.0099,
+      "step": 97820
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530200181590215e-05,
+      "loss": 0.0105,
+      "step": 97830
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.953015214185174e-05,
+      "loss": 0.0121,
+      "step": 97840
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.953010410211327e-05,
+      "loss": 0.0109,
+      "step": 97850
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.95300560623748e-05,
+      "loss": 0.0157,
+      "step": 97860
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9530008022636325e-05,
+      "loss": 0.0132,
+      "step": 97870
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529959982897854e-05,
+      "loss": 0.0131,
+      "step": 97880
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529911943159383e-05,
+      "loss": 0.0115,
+      "step": 97890
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529863903420913e-05,
+      "loss": 0.0084,
+      "step": 97900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.952981586368244e-05,
+      "loss": 0.0097,
+      "step": 97910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529767823943968e-05,
+      "loss": 0.0089,
+      "step": 97920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529719784205497e-05,
+      "loss": 0.0132,
+      "step": 97930
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529671744467026e-05,
+      "loss": 0.0158,
+      "step": 97940
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529623704728552e-05,
+      "loss": 0.0127,
+      "step": 97950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.952957566499008e-05,
+      "loss": 0.0106,
+      "step": 97960
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.952952762525161e-05,
+      "loss": 0.0115,
+      "step": 97970
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529479585513137e-05,
+      "loss": 0.0138,
+      "step": 97980
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529431545774666e-05,
+      "loss": 0.012,
+      "step": 97990
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529383506036195e-05,
+      "loss": 0.0118,
+      "step": 98000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529335466297725e-05,
+      "loss": 0.0119,
+      "step": 98010
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.952928742655925e-05,
+      "loss": 0.0101,
+      "step": 98020
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.952923938682078e-05,
+      "loss": 0.0148,
+      "step": 98030
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.952919134708231e-05,
+      "loss": 0.0112,
+      "step": 98040
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529143307343835e-05,
+      "loss": 0.0135,
+      "step": 98050
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529095267605364e-05,
+      "loss": 0.0143,
+      "step": 98060
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9529047227866893e-05,
+      "loss": 0.0107,
+      "step": 98070
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9528999188128423e-05,
+      "loss": 0.0123,
+      "step": 98080
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.952895114838995e-05,
+      "loss": 0.0125,
+      "step": 98090
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9528903108651478e-05,
+      "loss": 0.0086,
+      "step": 98100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9528855068913007e-05,
+      "loss": 0.0083,
+      "step": 98110
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9528807029174536e-05,
+      "loss": 0.0116,
+      "step": 98120
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9528758989436062e-05,
+      "loss": 0.011,
+      "step": 98130
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.952871094969759e-05,
+      "loss": 0.014,
+      "step": 98140
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.952866290995912e-05,
+      "loss": 0.0108,
+      "step": 98150
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9528614870220647e-05,
+      "loss": 0.0132,
+      "step": 98160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9528566830482176e-05,
+      "loss": 0.0084,
+      "step": 98170
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9528518790743705e-05,
+      "loss": 0.0105,
+      "step": 98180
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9528470751005234e-05,
+      "loss": 0.0121,
+      "step": 98190
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952842271126676e-05,
+      "loss": 0.0101,
+      "step": 98200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952837467152829e-05,
+      "loss": 0.0088,
+      "step": 98210
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952832663178982e-05,
+      "loss": 0.0112,
+      "step": 98220
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9528278592051345e-05,
+      "loss": 0.0128,
+      "step": 98230
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9528230552312874e-05,
+      "loss": 0.0144,
+      "step": 98240
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9528182512574403e-05,
+      "loss": 0.0118,
+      "step": 98250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9528134472835933e-05,
+      "loss": 0.0127,
+      "step": 98260
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952808643309746e-05,
+      "loss": 0.01,
+      "step": 98270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9528038393358988e-05,
+      "loss": 0.009,
+      "step": 98280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527990353620517e-05,
+      "loss": 0.0145,
+      "step": 98290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527942313882043e-05,
+      "loss": 0.0137,
+      "step": 98300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527894274143572e-05,
+      "loss": 0.0176,
+      "step": 98310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.95278462344051e-05,
+      "loss": 0.0098,
+      "step": 98320
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952779819466663e-05,
+      "loss": 0.0103,
+      "step": 98330
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527750154928157e-05,
+      "loss": 0.0126,
+      "step": 98340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527702115189686e-05,
+      "loss": 0.016,
+      "step": 98350
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527654075451215e-05,
+      "loss": 0.0115,
+      "step": 98360
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527606035712744e-05,
+      "loss": 0.0105,
+      "step": 98370
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952755799597427e-05,
+      "loss": 0.011,
+      "step": 98380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.95275099562358e-05,
+      "loss": 0.0155,
+      "step": 98390
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952746191649733e-05,
+      "loss": 0.0123,
+      "step": 98400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527413876758855e-05,
+      "loss": 0.0135,
+      "step": 98410
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527365837020384e-05,
+      "loss": 0.0132,
+      "step": 98420
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527317797281913e-05,
+      "loss": 0.0159,
+      "step": 98430
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527269757543442e-05,
+      "loss": 0.0129,
+      "step": 98440
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952722171780497e-05,
+      "loss": 0.0143,
+      "step": 98450
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527173678066498e-05,
+      "loss": 0.0123,
+      "step": 98460
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527125638328027e-05,
+      "loss": 0.012,
+      "step": 98470
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527077598589553e-05,
+      "loss": 0.0143,
+      "step": 98480
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9527029558851082e-05,
+      "loss": 0.0138,
+      "step": 98490
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952698151911261e-05,
+      "loss": 0.0108,
+      "step": 98500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952693347937414e-05,
+      "loss": 0.0129,
+      "step": 98510
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526885439635666e-05,
+      "loss": 0.0122,
+      "step": 98520
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526837399897196e-05,
+      "loss": 0.0122,
+      "step": 98530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526789360158725e-05,
+      "loss": 0.015,
+      "step": 98540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526741320420254e-05,
+      "loss": 0.009,
+      "step": 98550
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952669328068178e-05,
+      "loss": 0.012,
+      "step": 98560
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952664524094331e-05,
+      "loss": 0.0114,
+      "step": 98570
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952659720120484e-05,
+      "loss": 0.0112,
+      "step": 98580
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526549161466365e-05,
+      "loss": 0.014,
+      "step": 98590
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526501121727894e-05,
+      "loss": 0.0092,
+      "step": 98600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526453081989423e-05,
+      "loss": 0.0134,
+      "step": 98610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526405042250952e-05,
+      "loss": 0.0118,
+      "step": 98620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526357002512478e-05,
+      "loss": 0.0125,
+      "step": 98630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526308962774008e-05,
+      "loss": 0.0111,
+      "step": 98640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526260923035537e-05,
+      "loss": 0.0106,
+      "step": 98650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526212883297063e-05,
+      "loss": 0.01,
+      "step": 98660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526164843558592e-05,
+      "loss": 0.0091,
+      "step": 98670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952611680382012e-05,
+      "loss": 0.0124,
+      "step": 98680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952606876408165e-05,
+      "loss": 0.0079,
+      "step": 98690
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9526020724343176e-05,
+      "loss": 0.015,
+      "step": 98700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9525972684604706e-05,
+      "loss": 0.0088,
+      "step": 98710
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9525924644866235e-05,
+      "loss": 0.0136,
+      "step": 98720
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9525876605127764e-05,
+      "loss": 0.0089,
+      "step": 98730
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952582856538929e-05,
+      "loss": 0.0085,
+      "step": 98740
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952578052565082e-05,
+      "loss": 0.0107,
+      "step": 98750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952573248591235e-05,
+      "loss": 0.0125,
+      "step": 98760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9525684446173874e-05,
+      "loss": 0.0114,
+      "step": 98770
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9525636406435404e-05,
+      "loss": 0.0125,
+      "step": 98780
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9525588366696933e-05,
+      "loss": 0.0119,
+      "step": 98790
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9525540326958462e-05,
+      "loss": 0.0136,
+      "step": 98800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9525492287219988e-05,
+      "loss": 0.0142,
+      "step": 98810
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9525444247481517e-05,
+      "loss": 0.014,
+      "step": 98820
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9525396207743047e-05,
+      "loss": 0.0086,
+      "step": 98830
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9525348168004573e-05,
+      "loss": 0.0103,
+      "step": 98840
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9525300128266102e-05,
+      "loss": 0.0143,
+      "step": 98850
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952525208852763e-05,
+      "loss": 0.0125,
+      "step": 98860
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.952520404878916e-05,
+      "loss": 0.0129,
+      "step": 98870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9525156009050686e-05,
+      "loss": 0.01,
+      "step": 98880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9525107969312216e-05,
+      "loss": 0.0141,
+      "step": 98890
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952506473354759e-05,
+      "loss": 0.0139,
+      "step": 98900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952501669380912e-05,
+      "loss": 0.0088,
+      "step": 98910
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952496865407065e-05,
+      "loss": 0.0139,
+      "step": 98920
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524920614332178e-05,
+      "loss": 0.0084,
+      "step": 98930
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524872574593704e-05,
+      "loss": 0.0107,
+      "step": 98940
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524824534855233e-05,
+      "loss": 0.0107,
+      "step": 98950
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524776495116762e-05,
+      "loss": 0.0104,
+      "step": 98960
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524728455378292e-05,
+      "loss": 0.0115,
+      "step": 98970
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524680415639818e-05,
+      "loss": 0.0083,
+      "step": 98980
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524632375901347e-05,
+      "loss": 0.0118,
+      "step": 98990
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524584336162876e-05,
+      "loss": 0.011,
+      "step": 99000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524536296424402e-05,
+      "loss": 0.0149,
+      "step": 99010
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952448825668593e-05,
+      "loss": 0.0157,
+      "step": 99020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952444021694746e-05,
+      "loss": 0.0145,
+      "step": 99030
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952439217720899e-05,
+      "loss": 0.0097,
+      "step": 99040
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524344137470516e-05,
+      "loss": 0.0141,
+      "step": 99050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524296097732045e-05,
+      "loss": 0.0133,
+      "step": 99060
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524248057993574e-05,
+      "loss": 0.0144,
+      "step": 99070
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.95242000182551e-05,
+      "loss": 0.0089,
+      "step": 99080
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952415197851663e-05,
+      "loss": 0.0086,
+      "step": 99090
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952410393877816e-05,
+      "loss": 0.0108,
+      "step": 99100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524055899039688e-05,
+      "loss": 0.0104,
+      "step": 99110
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9524007859301214e-05,
+      "loss": 0.0113,
+      "step": 99120
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523959819562743e-05,
+      "loss": 0.0095,
+      "step": 99130
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523911779824272e-05,
+      "loss": 0.0163,
+      "step": 99140
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.95238637400858e-05,
+      "loss": 0.0087,
+      "step": 99150
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523815700347327e-05,
+      "loss": 0.0145,
+      "step": 99160
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523767660608857e-05,
+      "loss": 0.0128,
+      "step": 99170
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523719620870386e-05,
+      "loss": 0.012,
+      "step": 99180
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523671581131912e-05,
+      "loss": 0.0122,
+      "step": 99190
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952362354139344e-05,
+      "loss": 0.0104,
+      "step": 99200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952357550165497e-05,
+      "loss": 0.013,
+      "step": 99210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.95235274619165e-05,
+      "loss": 0.011,
+      "step": 99220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523479422178026e-05,
+      "loss": 0.0101,
+      "step": 99230
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523431382439555e-05,
+      "loss": 0.0097,
+      "step": 99240
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523383342701084e-05,
+      "loss": 0.0117,
+      "step": 99250
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952333530296261e-05,
+      "loss": 0.0132,
+      "step": 99260
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952328726322414e-05,
+      "loss": 0.0082,
+      "step": 99270
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952323922348567e-05,
+      "loss": 0.0095,
+      "step": 99280
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523191183747198e-05,
+      "loss": 0.0112,
+      "step": 99290
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523143144008724e-05,
+      "loss": 0.0088,
+      "step": 99300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523095104270253e-05,
+      "loss": 0.013,
+      "step": 99310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9523047064531782e-05,
+      "loss": 0.011,
+      "step": 99320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952299902479331e-05,
+      "loss": 0.009,
+      "step": 99330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522950985054837e-05,
+      "loss": 0.0087,
+      "step": 99340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522902945316367e-05,
+      "loss": 0.0149,
+      "step": 99350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522854905577896e-05,
+      "loss": 0.0125,
+      "step": 99360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522806865839422e-05,
+      "loss": 0.011,
+      "step": 99370
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952275882610095e-05,
+      "loss": 0.0088,
+      "step": 99380
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952271078636248e-05,
+      "loss": 0.0137,
+      "step": 99390
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952266274662401e-05,
+      "loss": 0.0118,
+      "step": 99400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522614706885536e-05,
+      "loss": 0.0138,
+      "step": 99410
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522566667147065e-05,
+      "loss": 0.0151,
+      "step": 99420
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522518627408594e-05,
+      "loss": 0.0122,
+      "step": 99430
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952247058767012e-05,
+      "loss": 0.0108,
+      "step": 99440
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952242254793165e-05,
+      "loss": 0.0144,
+      "step": 99450
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952237450819318e-05,
+      "loss": 0.0187,
+      "step": 99460
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522326468454708e-05,
+      "loss": 0.0131,
+      "step": 99470
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522278428716234e-05,
+      "loss": 0.0095,
+      "step": 99480
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522230388977763e-05,
+      "loss": 0.0107,
+      "step": 99490
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522182349239292e-05,
+      "loss": 0.0143,
+      "step": 99500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952213430950082e-05,
+      "loss": 0.0097,
+      "step": 99510
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522086269762347e-05,
+      "loss": 0.013,
+      "step": 99520
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9522038230023877e-05,
+      "loss": 0.01,
+      "step": 99530
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9521990190285406e-05,
+      "loss": 0.0122,
+      "step": 99540
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9521942150546932e-05,
+      "loss": 0.0121,
+      "step": 99550
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952189411080846e-05,
+      "loss": 0.0096,
+      "step": 99560
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.952184607106999e-05,
+      "loss": 0.0116,
+      "step": 99570
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952179803133152e-05,
+      "loss": 0.01,
+      "step": 99580
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9521749991593045e-05,
+      "loss": 0.0146,
+      "step": 99590
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9521701951854575e-05,
+      "loss": 0.0137,
+      "step": 99600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9521653912116104e-05,
+      "loss": 0.0098,
+      "step": 99610
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952160587237763e-05,
+      "loss": 0.0138,
+      "step": 99620
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952155783263916e-05,
+      "loss": 0.0088,
+      "step": 99630
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952150979290069e-05,
+      "loss": 0.0101,
+      "step": 99640
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9521461753162218e-05,
+      "loss": 0.0094,
+      "step": 99650
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9521413713423744e-05,
+      "loss": 0.0098,
+      "step": 99660
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9521365673685273e-05,
+      "loss": 0.0086,
+      "step": 99670
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9521317633946802e-05,
+      "loss": 0.0099,
+      "step": 99680
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952126959420833e-05,
+      "loss": 0.0133,
+      "step": 99690
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9521221554469857e-05,
+      "loss": 0.0151,
+      "step": 99700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9521173514731387e-05,
+      "loss": 0.0106,
+      "step": 99710
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9521125474992916e-05,
+      "loss": 0.0122,
+      "step": 99720
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952107743525444e-05,
+      "loss": 0.0102,
+      "step": 99730
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952102939551597e-05,
+      "loss": 0.01,
+      "step": 99740
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.95209813557775e-05,
+      "loss": 0.0104,
+      "step": 99750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952093331603903e-05,
+      "loss": 0.0126,
+      "step": 99760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952088527630056e-05,
+      "loss": 0.0092,
+      "step": 99770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9520837236562088e-05,
+      "loss": 0.0113,
+      "step": 99780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9520789196823614e-05,
+      "loss": 0.0099,
+      "step": 99790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9520741157085143e-05,
+      "loss": 0.0129,
+      "step": 99800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9520693117346672e-05,
+      "loss": 0.011,
+      "step": 99810
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9520645077608202e-05,
+      "loss": 0.0131,
+      "step": 99820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9520597037869728e-05,
+      "loss": 0.0114,
+      "step": 99830
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9520548998131257e-05,
+      "loss": 0.0101,
+      "step": 99840
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9520500958392786e-05,
+      "loss": 0.0094,
+      "step": 99850
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9520452918654312e-05,
+      "loss": 0.0111,
+      "step": 99860
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952040487891584e-05,
+      "loss": 0.009,
+      "step": 99870
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952035683917737e-05,
+      "loss": 0.0087,
+      "step": 99880
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.95203087994389e-05,
+      "loss": 0.0106,
+      "step": 99890
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9520260759700426e-05,
+      "loss": 0.0117,
+      "step": 99900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9520212719961955e-05,
+      "loss": 0.0105,
+      "step": 99910
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9520164680223484e-05,
+      "loss": 0.0141,
+      "step": 99920
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952011664048501e-05,
+      "loss": 0.0138,
+      "step": 99930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952006860074654e-05,
+      "loss": 0.011,
+      "step": 99940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.952002056100807e-05,
+      "loss": 0.0094,
+      "step": 99950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519972521269598e-05,
+      "loss": 0.0068,
+      "step": 99960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519924481531124e-05,
+      "loss": 0.0102,
+      "step": 99970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519876441792653e-05,
+      "loss": 0.0086,
+      "step": 99980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519828402054182e-05,
+      "loss": 0.0088,
+      "step": 99990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.951978036231571e-05,
+      "loss": 0.0107,
+      "step": 100000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519732322577237e-05,
+      "loss": 0.0127,
+      "step": 100010
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519684282838767e-05,
+      "loss": 0.0096,
+      "step": 100020
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519636243100296e-05,
+      "loss": 0.012,
+      "step": 100030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519588203361822e-05,
+      "loss": 0.014,
+      "step": 100040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.951954016362335e-05,
+      "loss": 0.0081,
+      "step": 100050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.951949212388488e-05,
+      "loss": 0.0155,
+      "step": 100060
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.951944408414641e-05,
+      "loss": 0.0154,
+      "step": 100070
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519396044407936e-05,
+      "loss": 0.0155,
+      "step": 100080
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519348004669465e-05,
+      "loss": 0.0104,
+      "step": 100090
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519299964930994e-05,
+      "loss": 0.0137,
+      "step": 100100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.951925192519252e-05,
+      "loss": 0.0098,
+      "step": 100110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.951920388545405e-05,
+      "loss": 0.0127,
+      "step": 100120
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.951915584571558e-05,
+      "loss": 0.0105,
+      "step": 100130
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519107805977108e-05,
+      "loss": 0.0163,
+      "step": 100140
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519059766238634e-05,
+      "loss": 0.0102,
+      "step": 100150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9519011726500163e-05,
+      "loss": 0.0116,
+      "step": 100160
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9518963686761692e-05,
+      "loss": 0.0079,
+      "step": 100170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.951891564702322e-05,
+      "loss": 0.0119,
+      "step": 100180
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9518867607284747e-05,
+      "loss": 0.0084,
+      "step": 100190
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9518819567546277e-05,
+      "loss": 0.0106,
+      "step": 100200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9518771527807806e-05,
+      "loss": 0.0091,
+      "step": 100210
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9518723488069332e-05,
+      "loss": 0.0083,
+      "step": 100220
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.951867544833086e-05,
+      "loss": 0.0111,
+      "step": 100230
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.951862740859239e-05,
+      "loss": 0.0084,
+      "step": 100240
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.951857936885392e-05,
+      "loss": 0.0092,
+      "step": 100250
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9518531329115446e-05,
+      "loss": 0.0114,
+      "step": 100260
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9518483289376975e-05,
+      "loss": 0.0141,
+      "step": 100270
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9518435249638504e-05,
+      "loss": 0.0167,
+      "step": 100280
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951838720990003e-05,
+      "loss": 0.0139,
+      "step": 100290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951833917016156e-05,
+      "loss": 0.0094,
+      "step": 100300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951829113042309e-05,
+      "loss": 0.0076,
+      "step": 100310
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9518243090684618e-05,
+      "loss": 0.0096,
+      "step": 100320
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9518195050946144e-05,
+      "loss": 0.0097,
+      "step": 100330
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9518147011207673e-05,
+      "loss": 0.0105,
+      "step": 100340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9518098971469202e-05,
+      "loss": 0.013,
+      "step": 100350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951805093173073e-05,
+      "loss": 0.0114,
+      "step": 100360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9518002891992257e-05,
+      "loss": 0.0075,
+      "step": 100370
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517954852253787e-05,
+      "loss": 0.0127,
+      "step": 100380
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517906812515316e-05,
+      "loss": 0.0094,
+      "step": 100390
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517858772776842e-05,
+      "loss": 0.0128,
+      "step": 100400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951781073303837e-05,
+      "loss": 0.0131,
+      "step": 100410
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.95177626932999e-05,
+      "loss": 0.0128,
+      "step": 100420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951771465356143e-05,
+      "loss": 0.0122,
+      "step": 100430
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517666613822955e-05,
+      "loss": 0.0071,
+      "step": 100440
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517618574084485e-05,
+      "loss": 0.0144,
+      "step": 100450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517570534346014e-05,
+      "loss": 0.0123,
+      "step": 100460
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951752249460754e-05,
+      "loss": 0.0167,
+      "step": 100470
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951747445486907e-05,
+      "loss": 0.0143,
+      "step": 100480
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.95174264151306e-05,
+      "loss": 0.0131,
+      "step": 100490
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517378375392128e-05,
+      "loss": 0.012,
+      "step": 100500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517330335653654e-05,
+      "loss": 0.0081,
+      "step": 100510
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517282295915183e-05,
+      "loss": 0.0153,
+      "step": 100520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517234256176712e-05,
+      "loss": 0.0106,
+      "step": 100530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951718621643824e-05,
+      "loss": 0.0143,
+      "step": 100540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517138176699767e-05,
+      "loss": 0.0118,
+      "step": 100550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517090136961297e-05,
+      "loss": 0.0141,
+      "step": 100560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9517042097222826e-05,
+      "loss": 0.0126,
+      "step": 100570
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951699405748435e-05,
+      "loss": 0.012,
+      "step": 100580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951694601774588e-05,
+      "loss": 0.0098,
+      "step": 100590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951689797800741e-05,
+      "loss": 0.0131,
+      "step": 100600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951684993826894e-05,
+      "loss": 0.0127,
+      "step": 100610
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9516801898530465e-05,
+      "loss": 0.0137,
+      "step": 100620
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9516753858791995e-05,
+      "loss": 0.0121,
+      "step": 100630
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9516705819053524e-05,
+      "loss": 0.0117,
+      "step": 100640
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951665777931505e-05,
+      "loss": 0.0106,
+      "step": 100650
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951660973957658e-05,
+      "loss": 0.0095,
+      "step": 100660
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951656169983811e-05,
+      "loss": 0.0091,
+      "step": 100670
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9516513660099638e-05,
+      "loss": 0.0116,
+      "step": 100680
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9516465620361163e-05,
+      "loss": 0.01,
+      "step": 100690
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9516417580622693e-05,
+      "loss": 0.0111,
+      "step": 100700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9516369540884222e-05,
+      "loss": 0.0118,
+      "step": 100710
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951632150114575e-05,
+      "loss": 0.015,
+      "step": 100720
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9516273461407277e-05,
+      "loss": 0.012,
+      "step": 100730
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9516225421668806e-05,
+      "loss": 0.0131,
+      "step": 100740
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9516177381930336e-05,
+      "loss": 0.0101,
+      "step": 100750
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951612934219186e-05,
+      "loss": 0.0144,
+      "step": 100760
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951608130245339e-05,
+      "loss": 0.0157,
+      "step": 100770
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951603326271492e-05,
+      "loss": 0.0142,
+      "step": 100780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951598522297645e-05,
+      "loss": 0.0105,
+      "step": 100790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9515937183237975e-05,
+      "loss": 0.0109,
+      "step": 100800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9515889143499505e-05,
+      "loss": 0.0125,
+      "step": 100810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9515841103761034e-05,
+      "loss": 0.0134,
+      "step": 100820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951579306402256e-05,
+      "loss": 0.0092,
+      "step": 100830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951574502428409e-05,
+      "loss": 0.0165,
+      "step": 100840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9515696984545618e-05,
+      "loss": 0.013,
+      "step": 100850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9515648944807147e-05,
+      "loss": 0.0112,
+      "step": 100860
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9515600905068673e-05,
+      "loss": 0.0105,
+      "step": 100870
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9515552865330203e-05,
+      "loss": 0.0131,
+      "step": 100880
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9515504825591732e-05,
+      "loss": 0.0115,
+      "step": 100890
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951545678585326e-05,
+      "loss": 0.0099,
+      "step": 100900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9515408746114787e-05,
+      "loss": 0.0137,
+      "step": 100910
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9515360706376316e-05,
+      "loss": 0.0125,
+      "step": 100920
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9515312666637846e-05,
+      "loss": 0.0107,
+      "step": 100930
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.951526462689937e-05,
+      "loss": 0.0137,
+      "step": 100940
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.95152165871609e-05,
+      "loss": 0.0098,
+      "step": 100950
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951516854742243e-05,
+      "loss": 0.0143,
+      "step": 100960
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951512050768396e-05,
+      "loss": 0.0118,
+      "step": 100970
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9515072467945485e-05,
+      "loss": 0.0134,
+      "step": 100980
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9515024428207014e-05,
+      "loss": 0.0117,
+      "step": 100990
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9514976388468544e-05,
+      "loss": 0.0119,
+      "step": 101000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951492834873007e-05,
+      "loss": 0.0113,
+      "step": 101010
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.95148803089916e-05,
+      "loss": 0.0142,
+      "step": 101020
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9514832269253128e-05,
+      "loss": 0.0091,
+      "step": 101030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9514784229514657e-05,
+      "loss": 0.0101,
+      "step": 101040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9514736189776183e-05,
+      "loss": 0.0106,
+      "step": 101050
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9514688150037713e-05,
+      "loss": 0.0119,
+      "step": 101060
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9514640110299242e-05,
+      "loss": 0.0109,
+      "step": 101070
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951459207056077e-05,
+      "loss": 0.0104,
+      "step": 101080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9514544030822297e-05,
+      "loss": 0.0117,
+      "step": 101090
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9514495991083826e-05,
+      "loss": 0.013,
+      "step": 101100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9514447951345356e-05,
+      "loss": 0.0087,
+      "step": 101110
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951439991160688e-05,
+      "loss": 0.0127,
+      "step": 101120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951435187186841e-05,
+      "loss": 0.0073,
+      "step": 101130
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951430383212994e-05,
+      "loss": 0.0104,
+      "step": 101140
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951425579239147e-05,
+      "loss": 0.0112,
+      "step": 101150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9514207752652995e-05,
+      "loss": 0.0121,
+      "step": 101160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9514159712914524e-05,
+      "loss": 0.0118,
+      "step": 101170
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9514111673176054e-05,
+      "loss": 0.011,
+      "step": 101180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951406363343758e-05,
+      "loss": 0.0096,
+      "step": 101190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951401559369911e-05,
+      "loss": 0.0088,
+      "step": 101200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513967553960638e-05,
+      "loss": 0.0141,
+      "step": 101210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513919514222167e-05,
+      "loss": 0.0142,
+      "step": 101220
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513871474483693e-05,
+      "loss": 0.0096,
+      "step": 101230
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513823434745222e-05,
+      "loss": 0.0119,
+      "step": 101240
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513775395006752e-05,
+      "loss": 0.0146,
+      "step": 101250
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951372735526828e-05,
+      "loss": 0.0162,
+      "step": 101260
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513679315529807e-05,
+      "loss": 0.011,
+      "step": 101270
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513631275791336e-05,
+      "loss": 0.0096,
+      "step": 101280
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513583236052865e-05,
+      "loss": 0.0145,
+      "step": 101290
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951353519631439e-05,
+      "loss": 0.0087,
+      "step": 101300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951348715657592e-05,
+      "loss": 0.0122,
+      "step": 101310
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951343911683745e-05,
+      "loss": 0.0078,
+      "step": 101320
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951339107709898e-05,
+      "loss": 0.0123,
+      "step": 101330
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513343037360505e-05,
+      "loss": 0.0103,
+      "step": 101340
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513294997622034e-05,
+      "loss": 0.01,
+      "step": 101350
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513246957883564e-05,
+      "loss": 0.0095,
+      "step": 101360
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951319891814509e-05,
+      "loss": 0.0063,
+      "step": 101370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951315087840662e-05,
+      "loss": 0.0126,
+      "step": 101380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513102838668148e-05,
+      "loss": 0.0075,
+      "step": 101390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513054798929677e-05,
+      "loss": 0.0129,
+      "step": 101400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9513006759191203e-05,
+      "loss": 0.0113,
+      "step": 101410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9512958719452732e-05,
+      "loss": 0.0082,
+      "step": 101420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951291067971426e-05,
+      "loss": 0.0109,
+      "step": 101430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9512862639975788e-05,
+      "loss": 0.015,
+      "step": 101440
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9512814600237317e-05,
+      "loss": 0.0121,
+      "step": 101450
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9512766560498846e-05,
+      "loss": 0.0111,
+      "step": 101460
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9512718520760375e-05,
+      "loss": 0.0069,
+      "step": 101470
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.95126704810219e-05,
+      "loss": 0.0079,
+      "step": 101480
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951262244128343e-05,
+      "loss": 0.0111,
+      "step": 101490
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951257440154496e-05,
+      "loss": 0.0095,
+      "step": 101500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951252636180649e-05,
+      "loss": 0.0154,
+      "step": 101510
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9512478322068015e-05,
+      "loss": 0.0124,
+      "step": 101520
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9512430282329544e-05,
+      "loss": 0.0081,
+      "step": 101530
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9512382242591073e-05,
+      "loss": 0.0085,
+      "step": 101540
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.95123342028526e-05,
+      "loss": 0.0161,
+      "step": 101550
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951228616311413e-05,
+      "loss": 0.0121,
+      "step": 101560
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9512238123375658e-05,
+      "loss": 0.0088,
+      "step": 101570
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9512190083637187e-05,
+      "loss": 0.0095,
+      "step": 101580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9512142043898713e-05,
+      "loss": 0.0092,
+      "step": 101590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9512094004160242e-05,
+      "loss": 0.0092,
+      "step": 101600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.951204596442177e-05,
+      "loss": 0.0121,
+      "step": 101610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9511997924683297e-05,
+      "loss": 0.0083,
+      "step": 101620
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9511949884944827e-05,
+      "loss": 0.0122,
+      "step": 101630
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9511901845206356e-05,
+      "loss": 0.0144,
+      "step": 101640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9511853805467885e-05,
+      "loss": 0.0116,
+      "step": 101650
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951180576572941e-05,
+      "loss": 0.01,
+      "step": 101660
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951175772599094e-05,
+      "loss": 0.0088,
+      "step": 101670
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951170968625247e-05,
+      "loss": 0.0093,
+      "step": 101680
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9511661646514e-05,
+      "loss": 0.0108,
+      "step": 101690
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9511613606775525e-05,
+      "loss": 0.0142,
+      "step": 101700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9511565567037054e-05,
+      "loss": 0.0131,
+      "step": 101710
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9511517527298583e-05,
+      "loss": 0.0106,
+      "step": 101720
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951146948756011e-05,
+      "loss": 0.0085,
+      "step": 101730
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951142144782164e-05,
+      "loss": 0.0114,
+      "step": 101740
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9511373408083168e-05,
+      "loss": 0.0122,
+      "step": 101750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9511325368344697e-05,
+      "loss": 0.0161,
+      "step": 101760
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9511277328606223e-05,
+      "loss": 0.0089,
+      "step": 101770
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9511229288867752e-05,
+      "loss": 0.0123,
+      "step": 101780
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951118124912928e-05,
+      "loss": 0.015,
+      "step": 101790
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9511133209390807e-05,
+      "loss": 0.0101,
+      "step": 101800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9511085169652337e-05,
+      "loss": 0.0157,
+      "step": 101810
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9511037129913866e-05,
+      "loss": 0.0132,
+      "step": 101820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9510989090175395e-05,
+      "loss": 0.0141,
+      "step": 101830
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951094105043692e-05,
+      "loss": 0.0101,
+      "step": 101840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951089301069845e-05,
+      "loss": 0.0105,
+      "step": 101850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951084497095998e-05,
+      "loss": 0.0126,
+      "step": 101860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951079693122151e-05,
+      "loss": 0.0114,
+      "step": 101870
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9510748891483035e-05,
+      "loss": 0.0102,
+      "step": 101880
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9510700851744564e-05,
+      "loss": 0.0163,
+      "step": 101890
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9510652812006093e-05,
+      "loss": 0.0121,
+      "step": 101900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951060477226762e-05,
+      "loss": 0.0095,
+      "step": 101910
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951055673252915e-05,
+      "loss": 0.0119,
+      "step": 101920
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9510508692790678e-05,
+      "loss": 0.0119,
+      "step": 101930
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9510460653052207e-05,
+      "loss": 0.0079,
+      "step": 101940
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9510412613313733e-05,
+      "loss": 0.0122,
+      "step": 101950
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9510364573575262e-05,
+      "loss": 0.0106,
+      "step": 101960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951031653383679e-05,
+      "loss": 0.0135,
+      "step": 101970
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9510268494098317e-05,
+      "loss": 0.0153,
+      "step": 101980
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9510220454359847e-05,
+      "loss": 0.0139,
+      "step": 101990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9510172414621376e-05,
+      "loss": 0.0093,
+      "step": 102000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9510124374882905e-05,
+      "loss": 0.0075,
+      "step": 102010
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951007633514443e-05,
+      "loss": 0.009,
+      "step": 102020
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.951003309937981e-05,
+      "loss": 0.0097,
+      "step": 102030
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509985059641335e-05,
+      "loss": 0.0111,
+      "step": 102040
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509937019902864e-05,
+      "loss": 0.0098,
+      "step": 102050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509888980164393e-05,
+      "loss": 0.0091,
+      "step": 102060
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509840940425923e-05,
+      "loss": 0.0114,
+      "step": 102070
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.950979290068745e-05,
+      "loss": 0.014,
+      "step": 102080
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509744860948978e-05,
+      "loss": 0.0131,
+      "step": 102090
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509696821210507e-05,
+      "loss": 0.01,
+      "step": 102100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509648781472036e-05,
+      "loss": 0.0125,
+      "step": 102110
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509600741733562e-05,
+      "loss": 0.0147,
+      "step": 102120
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.950955270199509e-05,
+      "loss": 0.0146,
+      "step": 102130
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.950950466225662e-05,
+      "loss": 0.0141,
+      "step": 102140
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509456622518147e-05,
+      "loss": 0.0072,
+      "step": 102150
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509408582779676e-05,
+      "loss": 0.0149,
+      "step": 102160
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509360543041205e-05,
+      "loss": 0.0118,
+      "step": 102170
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509312503302735e-05,
+      "loss": 0.0107,
+      "step": 102180
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.950926446356426e-05,
+      "loss": 0.0126,
+      "step": 102190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.950921642382579e-05,
+      "loss": 0.0089,
+      "step": 102200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.950916838408732e-05,
+      "loss": 0.0117,
+      "step": 102210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509120344348845e-05,
+      "loss": 0.0101,
+      "step": 102220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509072304610374e-05,
+      "loss": 0.0127,
+      "step": 102230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9509024264871903e-05,
+      "loss": 0.0124,
+      "step": 102240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9508976225133433e-05,
+      "loss": 0.01,
+      "step": 102250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.950892818539496e-05,
+      "loss": 0.0106,
+      "step": 102260
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9508880145656488e-05,
+      "loss": 0.0133,
+      "step": 102270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9508832105918017e-05,
+      "loss": 0.0124,
+      "step": 102280
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9508784066179546e-05,
+      "loss": 0.0098,
+      "step": 102290
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9508736026441072e-05,
+      "loss": 0.0138,
+      "step": 102300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.95086879867026e-05,
+      "loss": 0.0144,
+      "step": 102310
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.950863994696413e-05,
+      "loss": 0.0131,
+      "step": 102320
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9508591907225657e-05,
+      "loss": 0.0169,
+      "step": 102330
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9508543867487186e-05,
+      "loss": 0.0111,
+      "step": 102340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9508495827748715e-05,
+      "loss": 0.0101,
+      "step": 102350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9508447788010244e-05,
+      "loss": 0.0113,
+      "step": 102360
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950839974827177e-05,
+      "loss": 0.0123,
+      "step": 102370
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.95083517085333e-05,
+      "loss": 0.0157,
+      "step": 102380
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950830366879483e-05,
+      "loss": 0.0116,
+      "step": 102390
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9508255629056355e-05,
+      "loss": 0.0109,
+      "step": 102400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9508207589317884e-05,
+      "loss": 0.0093,
+      "step": 102410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9508159549579413e-05,
+      "loss": 0.0188,
+      "step": 102420
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9508111509840943e-05,
+      "loss": 0.0137,
+      "step": 102430
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950806347010247e-05,
+      "loss": 0.013,
+      "step": 102440
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9508015430363998e-05,
+      "loss": 0.0114,
+      "step": 102450
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507967390625527e-05,
+      "loss": 0.0095,
+      "step": 102460
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507919350887056e-05,
+      "loss": 0.0139,
+      "step": 102470
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507871311148582e-05,
+      "loss": 0.0116,
+      "step": 102480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950782327141011e-05,
+      "loss": 0.0097,
+      "step": 102490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950777523167164e-05,
+      "loss": 0.0139,
+      "step": 102500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507727191933167e-05,
+      "loss": 0.008,
+      "step": 102510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507679152194696e-05,
+      "loss": 0.0103,
+      "step": 102520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507631112456225e-05,
+      "loss": 0.0073,
+      "step": 102530
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507583072717754e-05,
+      "loss": 0.0092,
+      "step": 102540
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950753503297928e-05,
+      "loss": 0.0103,
+      "step": 102550
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950748699324081e-05,
+      "loss": 0.0153,
+      "step": 102560
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950743895350234e-05,
+      "loss": 0.0103,
+      "step": 102570
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507390913763865e-05,
+      "loss": 0.0118,
+      "step": 102580
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507342874025394e-05,
+      "loss": 0.0092,
+      "step": 102590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507294834286923e-05,
+      "loss": 0.0075,
+      "step": 102600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507246794548452e-05,
+      "loss": 0.0107,
+      "step": 102610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950719875480998e-05,
+      "loss": 0.0133,
+      "step": 102620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507150715071508e-05,
+      "loss": 0.0132,
+      "step": 102630
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507102675333037e-05,
+      "loss": 0.013,
+      "step": 102640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507054635594566e-05,
+      "loss": 0.014,
+      "step": 102650
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9507006595856092e-05,
+      "loss": 0.0094,
+      "step": 102660
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950695855611762e-05,
+      "loss": 0.0117,
+      "step": 102670
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950691051637915e-05,
+      "loss": 0.0109,
+      "step": 102680
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506862476640676e-05,
+      "loss": 0.0129,
+      "step": 102690
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506814436902206e-05,
+      "loss": 0.0125,
+      "step": 102700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506766397163735e-05,
+      "loss": 0.0111,
+      "step": 102710
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506718357425264e-05,
+      "loss": 0.01,
+      "step": 102720
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950667031768679e-05,
+      "loss": 0.0107,
+      "step": 102730
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950662227794832e-05,
+      "loss": 0.0132,
+      "step": 102740
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950657423820985e-05,
+      "loss": 0.0165,
+      "step": 102750
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506526198471375e-05,
+      "loss": 0.009,
+      "step": 102760
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506478158732904e-05,
+      "loss": 0.012,
+      "step": 102770
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506430118994433e-05,
+      "loss": 0.0109,
+      "step": 102780
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506382079255962e-05,
+      "loss": 0.0078,
+      "step": 102790
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506334039517488e-05,
+      "loss": 0.0145,
+      "step": 102800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506285999779018e-05,
+      "loss": 0.0154,
+      "step": 102810
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506237960040547e-05,
+      "loss": 0.0093,
+      "step": 102820
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506189920302076e-05,
+      "loss": 0.0134,
+      "step": 102830
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506141880563605e-05,
+      "loss": 0.013,
+      "step": 102840
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9506093840825135e-05,
+      "loss": 0.01,
+      "step": 102850
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950604580108666e-05,
+      "loss": 0.018,
+      "step": 102860
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950599776134819e-05,
+      "loss": 0.0086,
+      "step": 102870
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950594972160972e-05,
+      "loss": 0.0092,
+      "step": 102880
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505901681871245e-05,
+      "loss": 0.0132,
+      "step": 102890
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505853642132774e-05,
+      "loss": 0.0146,
+      "step": 102900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505805602394303e-05,
+      "loss": 0.0108,
+      "step": 102910
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505757562655833e-05,
+      "loss": 0.0165,
+      "step": 102920
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950570952291736e-05,
+      "loss": 0.0085,
+      "step": 102930
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505661483178888e-05,
+      "loss": 0.0088,
+      "step": 102940
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505613443440417e-05,
+      "loss": 0.0075,
+      "step": 102950
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505565403701946e-05,
+      "loss": 0.0116,
+      "step": 102960
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505517363963472e-05,
+      "loss": 0.0168,
+      "step": 102970
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505469324225e-05,
+      "loss": 0.0132,
+      "step": 102980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.950542128448653e-05,
+      "loss": 0.0125,
+      "step": 102990
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505373244748057e-05,
+      "loss": 0.0087,
+      "step": 103000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505325205009586e-05,
+      "loss": 0.0114,
+      "step": 103010
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505277165271115e-05,
+      "loss": 0.0118,
+      "step": 103020
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9505229125532645e-05,
+      "loss": 0.0123,
+      "step": 103030
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950518108579417e-05,
+      "loss": 0.0093,
+      "step": 103040
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.95051330460557e-05,
+      "loss": 0.0096,
+      "step": 103050
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950508500631723e-05,
+      "loss": 0.0079,
+      "step": 103060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9505036966578755e-05,
+      "loss": 0.0107,
+      "step": 103070
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504988926840284e-05,
+      "loss": 0.0115,
+      "step": 103080
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504940887101813e-05,
+      "loss": 0.0099,
+      "step": 103090
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504892847363343e-05,
+      "loss": 0.0088,
+      "step": 103100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950484480762487e-05,
+      "loss": 0.0108,
+      "step": 103110
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504796767886398e-05,
+      "loss": 0.0123,
+      "step": 103120
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504748728147927e-05,
+      "loss": 0.0113,
+      "step": 103130
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504700688409456e-05,
+      "loss": 0.0108,
+      "step": 103140
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504652648670982e-05,
+      "loss": 0.0068,
+      "step": 103150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950460460893251e-05,
+      "loss": 0.0161,
+      "step": 103160
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950455656919404e-05,
+      "loss": 0.012,
+      "step": 103170
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504508529455567e-05,
+      "loss": 0.0123,
+      "step": 103180
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504460489717096e-05,
+      "loss": 0.0152,
+      "step": 103190
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504412449978625e-05,
+      "loss": 0.0116,
+      "step": 103200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504364410240154e-05,
+      "loss": 0.0107,
+      "step": 103210
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950431637050168e-05,
+      "loss": 0.0216,
+      "step": 103220
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950426833076321e-05,
+      "loss": 0.0112,
+      "step": 103230
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950422029102474e-05,
+      "loss": 0.0102,
+      "step": 103240
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504172251286265e-05,
+      "loss": 0.0155,
+      "step": 103250
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504124211547794e-05,
+      "loss": 0.0107,
+      "step": 103260
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504076171809323e-05,
+      "loss": 0.0121,
+      "step": 103270
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9504028132070853e-05,
+      "loss": 0.0079,
+      "step": 103280
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950398009233238e-05,
+      "loss": 0.0132,
+      "step": 103290
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503932052593908e-05,
+      "loss": 0.0083,
+      "step": 103300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503884012855437e-05,
+      "loss": 0.0103,
+      "step": 103310
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503835973116966e-05,
+      "loss": 0.0125,
+      "step": 103320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503787933378492e-05,
+      "loss": 0.0173,
+      "step": 103330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950373989364002e-05,
+      "loss": 0.0095,
+      "step": 103340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950369185390155e-05,
+      "loss": 0.0093,
+      "step": 103350
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503643814163077e-05,
+      "loss": 0.0139,
+      "step": 103360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503595774424606e-05,
+      "loss": 0.0078,
+      "step": 103370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503547734686135e-05,
+      "loss": 0.0162,
+      "step": 103380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503499694947664e-05,
+      "loss": 0.0094,
+      "step": 103390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950345165520919e-05,
+      "loss": 0.0098,
+      "step": 103400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950340361547072e-05,
+      "loss": 0.0089,
+      "step": 103410
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950335557573225e-05,
+      "loss": 0.0155,
+      "step": 103420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503307535993775e-05,
+      "loss": 0.0113,
+      "step": 103430
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503259496255304e-05,
+      "loss": 0.0102,
+      "step": 103440
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503211456516833e-05,
+      "loss": 0.0111,
+      "step": 103450
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503163416778362e-05,
+      "loss": 0.0098,
+      "step": 103460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950311537703989e-05,
+      "loss": 0.011,
+      "step": 103470
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503067337301418e-05,
+      "loss": 0.0086,
+      "step": 103480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9503019297562947e-05,
+      "loss": 0.0126,
+      "step": 103490
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502971257824476e-05,
+      "loss": 0.0075,
+      "step": 103500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502923218086002e-05,
+      "loss": 0.029,
+      "step": 103510
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950287517834753e-05,
+      "loss": 0.0107,
+      "step": 103520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950282713860906e-05,
+      "loss": 0.0166,
+      "step": 103530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502779098870586e-05,
+      "loss": 0.0082,
+      "step": 103540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502731059132116e-05,
+      "loss": 0.0102,
+      "step": 103550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502683019393645e-05,
+      "loss": 0.0147,
+      "step": 103560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502634979655174e-05,
+      "loss": 0.0094,
+      "step": 103570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.95025869399167e-05,
+      "loss": 0.0096,
+      "step": 103580
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950253890017823e-05,
+      "loss": 0.009,
+      "step": 103590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950249086043976e-05,
+      "loss": 0.0101,
+      "step": 103600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502442820701285e-05,
+      "loss": 0.0117,
+      "step": 103610
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502394780962814e-05,
+      "loss": 0.0113,
+      "step": 103620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502346741224343e-05,
+      "loss": 0.0102,
+      "step": 103630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502298701485872e-05,
+      "loss": 0.0081,
+      "step": 103640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502250661747398e-05,
+      "loss": 0.0117,
+      "step": 103650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502202622008928e-05,
+      "loss": 0.0122,
+      "step": 103660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502154582270457e-05,
+      "loss": 0.0128,
+      "step": 103670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502106542531986e-05,
+      "loss": 0.0114,
+      "step": 103680
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9502058502793512e-05,
+      "loss": 0.0109,
+      "step": 103690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950201046305504e-05,
+      "loss": 0.0077,
+      "step": 103700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.950196242331657e-05,
+      "loss": 0.011,
+      "step": 103710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9501914383578096e-05,
+      "loss": 0.0116,
+      "step": 103720
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9501866343839626e-05,
+      "loss": 0.0124,
+      "step": 103730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501818304101155e-05,
+      "loss": 0.0095,
+      "step": 103740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501770264362684e-05,
+      "loss": 0.0187,
+      "step": 103750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.950172222462421e-05,
+      "loss": 0.011,
+      "step": 103760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.950167418488574e-05,
+      "loss": 0.0108,
+      "step": 103770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.950162614514727e-05,
+      "loss": 0.0121,
+      "step": 103780
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501578105408794e-05,
+      "loss": 0.0078,
+      "step": 103790
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501530065670324e-05,
+      "loss": 0.012,
+      "step": 103800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501482025931853e-05,
+      "loss": 0.014,
+      "step": 103810
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501433986193382e-05,
+      "loss": 0.014,
+      "step": 103820
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501385946454908e-05,
+      "loss": 0.0154,
+      "step": 103830
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501337906716437e-05,
+      "loss": 0.0132,
+      "step": 103840
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501289866977967e-05,
+      "loss": 0.0096,
+      "step": 103850
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501241827239496e-05,
+      "loss": 0.0132,
+      "step": 103860
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501193787501022e-05,
+      "loss": 0.0122,
+      "step": 103870
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.950114574776255e-05,
+      "loss": 0.0082,
+      "step": 103880
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.950109770802408e-05,
+      "loss": 0.0088,
+      "step": 103890
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501049668285606e-05,
+      "loss": 0.0096,
+      "step": 103900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9501001628547136e-05,
+      "loss": 0.0121,
+      "step": 103910
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500953588808665e-05,
+      "loss": 0.0135,
+      "step": 103920
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500905549070194e-05,
+      "loss": 0.0104,
+      "step": 103930
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.950085750933172e-05,
+      "loss": 0.0119,
+      "step": 103940
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.950080946959325e-05,
+      "loss": 0.0118,
+      "step": 103950
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.950076142985478e-05,
+      "loss": 0.0076,
+      "step": 103960
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500713390116304e-05,
+      "loss": 0.0111,
+      "step": 103970
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500665350377834e-05,
+      "loss": 0.0107,
+      "step": 103980
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500617310639363e-05,
+      "loss": 0.0122,
+      "step": 103990
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500569270900892e-05,
+      "loss": 0.0081,
+      "step": 104000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500521231162418e-05,
+      "loss": 0.0101,
+      "step": 104010
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500473191423947e-05,
+      "loss": 0.0135,
+      "step": 104020
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500425151685477e-05,
+      "loss": 0.0102,
+      "step": 104030
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500377111947006e-05,
+      "loss": 0.0122,
+      "step": 104040
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500329072208532e-05,
+      "loss": 0.0134,
+      "step": 104050
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.950028103247006e-05,
+      "loss": 0.0155,
+      "step": 104060
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.950023299273159e-05,
+      "loss": 0.0089,
+      "step": 104070
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500184952993116e-05,
+      "loss": 0.0117,
+      "step": 104080
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500136913254645e-05,
+      "loss": 0.0105,
+      "step": 104090
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500088873516175e-05,
+      "loss": 0.0094,
+      "step": 104100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9500040833777704e-05,
+      "loss": 0.0058,
+      "step": 104110
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.949999279403923e-05,
+      "loss": 0.0096,
+      "step": 104120
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.949994475430076e-05,
+      "loss": 0.0124,
+      "step": 104130
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.949989671456229e-05,
+      "loss": 0.0087,
+      "step": 104140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9499848674823814e-05,
+      "loss": 0.0107,
+      "step": 104150
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9499800635085344e-05,
+      "loss": 0.0073,
+      "step": 104160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9499752595346873e-05,
+      "loss": 0.014,
+      "step": 104170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9499704555608402e-05,
+      "loss": 0.0102,
+      "step": 104180
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9499656515869928e-05,
+      "loss": 0.0099,
+      "step": 104190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9499608476131457e-05,
+      "loss": 0.0109,
+      "step": 104200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9499560436392987e-05,
+      "loss": 0.0103,
+      "step": 104210
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9499512396654516e-05,
+      "loss": 0.01,
+      "step": 104220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.949946435691604e-05,
+      "loss": 0.0112,
+      "step": 104230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.949941631717757e-05,
+      "loss": 0.0102,
+      "step": 104240
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.94993682774391e-05,
+      "loss": 0.0072,
+      "step": 104250
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9499320237700626e-05,
+      "loss": 0.0076,
+      "step": 104260
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9499272197962155e-05,
+      "loss": 0.012,
+      "step": 104270
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9499224158223685e-05,
+      "loss": 0.013,
+      "step": 104280
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9499176118485214e-05,
+      "loss": 0.0125,
+      "step": 104290
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.949912807874674e-05,
+      "loss": 0.0088,
+      "step": 104300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.949908003900827e-05,
+      "loss": 0.011,
+      "step": 104310
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.94990319992698e-05,
+      "loss": 0.0113,
+      "step": 104320
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9498983959531324e-05,
+      "loss": 0.0075,
+      "step": 104330
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9498935919792853e-05,
+      "loss": 0.0112,
+      "step": 104340
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9498887880054383e-05,
+      "loss": 0.0106,
+      "step": 104350
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9498839840315912e-05,
+      "loss": 0.0159,
+      "step": 104360
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9498791800577438e-05,
+      "loss": 0.0113,
+      "step": 104370
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9498743760838967e-05,
+      "loss": 0.0095,
+      "step": 104380
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9498695721100496e-05,
+      "loss": 0.0066,
+      "step": 104390
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9498647681362022e-05,
+      "loss": 0.0114,
+      "step": 104400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.949859964162355e-05,
+      "loss": 0.0098,
+      "step": 104410
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.949855160188508e-05,
+      "loss": 0.0066,
+      "step": 104420
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949850356214661e-05,
+      "loss": 0.0087,
+      "step": 104430
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9498455522408136e-05,
+      "loss": 0.0135,
+      "step": 104440
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9498407482669665e-05,
+      "loss": 0.0124,
+      "step": 104450
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9498359442931195e-05,
+      "loss": 0.0138,
+      "step": 104460
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9498311403192724e-05,
+      "loss": 0.0096,
+      "step": 104470
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949826336345425e-05,
+      "loss": 0.0138,
+      "step": 104480
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949821532371578e-05,
+      "loss": 0.0116,
+      "step": 104490
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9498167283977308e-05,
+      "loss": 0.0111,
+      "step": 104500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9498119244238834e-05,
+      "loss": 0.0118,
+      "step": 104510
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9498071204500363e-05,
+      "loss": 0.0129,
+      "step": 104520
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9498023164761893e-05,
+      "loss": 0.012,
+      "step": 104530
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497975125023422e-05,
+      "loss": 0.0103,
+      "step": 104540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497927085284948e-05,
+      "loss": 0.0118,
+      "step": 104550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497879045546477e-05,
+      "loss": 0.0137,
+      "step": 104560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497831005808006e-05,
+      "loss": 0.0134,
+      "step": 104570
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497782966069532e-05,
+      "loss": 0.0099,
+      "step": 104580
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949773492633106e-05,
+      "loss": 0.0111,
+      "step": 104590
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949768688659259e-05,
+      "loss": 0.0106,
+      "step": 104600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949763884685412e-05,
+      "loss": 0.0131,
+      "step": 104610
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497590807115646e-05,
+      "loss": 0.0131,
+      "step": 104620
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497542767377175e-05,
+      "loss": 0.0179,
+      "step": 104630
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497494727638704e-05,
+      "loss": 0.0119,
+      "step": 104640
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497446687900234e-05,
+      "loss": 0.0159,
+      "step": 104650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949739864816176e-05,
+      "loss": 0.0118,
+      "step": 104660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949735060842329e-05,
+      "loss": 0.0118,
+      "step": 104670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497302568684818e-05,
+      "loss": 0.0117,
+      "step": 104680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497254528946344e-05,
+      "loss": 0.0122,
+      "step": 104690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497206489207873e-05,
+      "loss": 0.0138,
+      "step": 104700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497158449469403e-05,
+      "loss": 0.0096,
+      "step": 104710
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497110409730932e-05,
+      "loss": 0.0108,
+      "step": 104720
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497062369992458e-05,
+      "loss": 0.0081,
+      "step": 104730
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9497014330253987e-05,
+      "loss": 0.0134,
+      "step": 104740
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496966290515516e-05,
+      "loss": 0.0142,
+      "step": 104750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496918250777042e-05,
+      "loss": 0.0112,
+      "step": 104760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949687021103857e-05,
+      "loss": 0.0116,
+      "step": 104770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.94968221713001e-05,
+      "loss": 0.0115,
+      "step": 104780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949677413156163e-05,
+      "loss": 0.0182,
+      "step": 104790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496726091823156e-05,
+      "loss": 0.0134,
+      "step": 104800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496678052084685e-05,
+      "loss": 0.0142,
+      "step": 104810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496630012346214e-05,
+      "loss": 0.0102,
+      "step": 104820
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496581972607744e-05,
+      "loss": 0.01,
+      "step": 104830
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949653393286927e-05,
+      "loss": 0.0106,
+      "step": 104840
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.94964858931308e-05,
+      "loss": 0.0099,
+      "step": 104850
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496437853392328e-05,
+      "loss": 0.0084,
+      "step": 104860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496389813653854e-05,
+      "loss": 0.0142,
+      "step": 104870
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496341773915383e-05,
+      "loss": 0.0093,
+      "step": 104880
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496293734176913e-05,
+      "loss": 0.0146,
+      "step": 104890
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496245694438442e-05,
+      "loss": 0.0103,
+      "step": 104900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496197654699968e-05,
+      "loss": 0.0079,
+      "step": 104910
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496149614961497e-05,
+      "loss": 0.0102,
+      "step": 104920
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496101575223026e-05,
+      "loss": 0.0068,
+      "step": 104930
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9496053535484552e-05,
+      "loss": 0.0155,
+      "step": 104940
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949600549574608e-05,
+      "loss": 0.0112,
+      "step": 104950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949595745600761e-05,
+      "loss": 0.0096,
+      "step": 104960
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949590941626914e-05,
+      "loss": 0.0101,
+      "step": 104970
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9495861376530666e-05,
+      "loss": 0.0119,
+      "step": 104980
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9495813336792195e-05,
+      "loss": 0.0155,
+      "step": 104990
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9495765297053724e-05,
+      "loss": 0.0127,
+      "step": 105000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9495717257315254e-05,
+      "loss": 0.0126,
+      "step": 105010
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949566921757678e-05,
+      "loss": 0.0126,
+      "step": 105020
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949562117783831e-05,
+      "loss": 0.0108,
+      "step": 105030
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9495573138099838e-05,
+      "loss": 0.0112,
+      "step": 105040
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9495525098361364e-05,
+      "loss": 0.0127,
+      "step": 105050
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9495477058622893e-05,
+      "loss": 0.0119,
+      "step": 105060
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9495429018884422e-05,
+      "loss": 0.0127,
+      "step": 105070
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.949538097914595e-05,
+      "loss": 0.0091,
+      "step": 105080
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9495332939407478e-05,
+      "loss": 0.0106,
+      "step": 105090
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9495284899669007e-05,
+      "loss": 0.0094,
+      "step": 105100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9495236859930536e-05,
+      "loss": 0.0095,
+      "step": 105110
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9495188820192062e-05,
+      "loss": 0.0121,
+      "step": 105120
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949514078045359e-05,
+      "loss": 0.0136,
+      "step": 105130
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949509274071512e-05,
+      "loss": 0.0081,
+      "step": 105140
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949504470097665e-05,
+      "loss": 0.0126,
+      "step": 105150
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494996661238176e-05,
+      "loss": 0.0225,
+      "step": 105160
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494948621499705e-05,
+      "loss": 0.0099,
+      "step": 105170
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494900581761234e-05,
+      "loss": 0.0122,
+      "step": 105180
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494852542022763e-05,
+      "loss": 0.0123,
+      "step": 105190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949480450228429e-05,
+      "loss": 0.0132,
+      "step": 105200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949475646254582e-05,
+      "loss": 0.0074,
+      "step": 105210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494708422807348e-05,
+      "loss": 0.0147,
+      "step": 105220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494665187042723e-05,
+      "loss": 0.008,
+      "step": 105230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494617147304252e-05,
+      "loss": 0.01,
+      "step": 105240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949456910756578e-05,
+      "loss": 0.0106,
+      "step": 105250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494521067827307e-05,
+      "loss": 0.013,
+      "step": 105260
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494473028088836e-05,
+      "loss": 0.0151,
+      "step": 105270
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494424988350366e-05,
+      "loss": 0.0111,
+      "step": 105280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949437694861189e-05,
+      "loss": 0.0101,
+      "step": 105290
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949432890887342e-05,
+      "loss": 0.0103,
+      "step": 105300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949428086913495e-05,
+      "loss": 0.0134,
+      "step": 105310
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949423282939648e-05,
+      "loss": 0.0093,
+      "step": 105320
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494184789658005e-05,
+      "loss": 0.015,
+      "step": 105330
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494136749919534e-05,
+      "loss": 0.0135,
+      "step": 105340
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9494088710181064e-05,
+      "loss": 0.0101,
+      "step": 105350
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949404067044259e-05,
+      "loss": 0.008,
+      "step": 105360
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949399263070412e-05,
+      "loss": 0.013,
+      "step": 105370
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493944590965648e-05,
+      "loss": 0.0094,
+      "step": 105380
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493896551227177e-05,
+      "loss": 0.008,
+      "step": 105390
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493848511488703e-05,
+      "loss": 0.011,
+      "step": 105400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493800471750232e-05,
+      "loss": 0.0114,
+      "step": 105410
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493752432011762e-05,
+      "loss": 0.0089,
+      "step": 105420
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949370439227329e-05,
+      "loss": 0.0087,
+      "step": 105430
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493656352534817e-05,
+      "loss": 0.0142,
+      "step": 105440
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493608312796346e-05,
+      "loss": 0.0129,
+      "step": 105450
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493560273057875e-05,
+      "loss": 0.0124,
+      "step": 105460
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.94935122333194e-05,
+      "loss": 0.0087,
+      "step": 105470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949346419358093e-05,
+      "loss": 0.0084,
+      "step": 105480
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949341615384246e-05,
+      "loss": 0.0104,
+      "step": 105490
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949336811410399e-05,
+      "loss": 0.0114,
+      "step": 105500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493320074365515e-05,
+      "loss": 0.0107,
+      "step": 105510
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493272034627044e-05,
+      "loss": 0.0129,
+      "step": 105520
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493223994888574e-05,
+      "loss": 0.0085,
+      "step": 105530
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.94931759551501e-05,
+      "loss": 0.0139,
+      "step": 105540
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949312791541163e-05,
+      "loss": 0.0105,
+      "step": 105550
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493079875673158e-05,
+      "loss": 0.0129,
+      "step": 105560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9493031835934687e-05,
+      "loss": 0.0109,
+      "step": 105570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9492983796196213e-05,
+      "loss": 0.0148,
+      "step": 105580
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9492935756457742e-05,
+      "loss": 0.011,
+      "step": 105590
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949288771671927e-05,
+      "loss": 0.0111,
+      "step": 105600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.94928396769808e-05,
+      "loss": 0.0119,
+      "step": 105610
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9492791637242327e-05,
+      "loss": 0.0103,
+      "step": 105620
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9492743597503856e-05,
+      "loss": 0.0131,
+      "step": 105630
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9492695557765385e-05,
+      "loss": 0.0122,
+      "step": 105640
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949264751802691e-05,
+      "loss": 0.0144,
+      "step": 105650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949259947828844e-05,
+      "loss": 0.0109,
+      "step": 105660
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949255143854997e-05,
+      "loss": 0.0093,
+      "step": 105670
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.94925033988115e-05,
+      "loss": 0.013,
+      "step": 105680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9492455359073025e-05,
+      "loss": 0.0093,
+      "step": 105690
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9492407319334554e-05,
+      "loss": 0.0118,
+      "step": 105700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9492359279596083e-05,
+      "loss": 0.0169,
+      "step": 105710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949231123985761e-05,
+      "loss": 0.0109,
+      "step": 105720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949226320011914e-05,
+      "loss": 0.0123,
+      "step": 105730
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9492215160380668e-05,
+      "loss": 0.0091,
+      "step": 105740
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9492167120642197e-05,
+      "loss": 0.0084,
+      "step": 105750
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9492119080903723e-05,
+      "loss": 0.0131,
+      "step": 105760
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9492071041165252e-05,
+      "loss": 0.0125,
+      "step": 105770
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949202300142678e-05,
+      "loss": 0.0134,
+      "step": 105780
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.949197496168831e-05,
+      "loss": 0.0124,
+      "step": 105790
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9491926921949837e-05,
+      "loss": 0.0149,
+      "step": 105800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9491878882211366e-05,
+      "loss": 0.0162,
+      "step": 105810
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9491830842472895e-05,
+      "loss": 0.0104,
+      "step": 105820
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949178280273442e-05,
+      "loss": 0.0109,
+      "step": 105830
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949173476299595e-05,
+      "loss": 0.0131,
+      "step": 105840
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949168672325748e-05,
+      "loss": 0.0106,
+      "step": 105850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949163868351901e-05,
+      "loss": 0.0133,
+      "step": 105860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9491590643780535e-05,
+      "loss": 0.0096,
+      "step": 105870
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9491542604042064e-05,
+      "loss": 0.0108,
+      "step": 105880
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9491494564303593e-05,
+      "loss": 0.0113,
+      "step": 105890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949144652456512e-05,
+      "loss": 0.0113,
+      "step": 105900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949139848482665e-05,
+      "loss": 0.0083,
+      "step": 105910
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949135044508818e-05,
+      "loss": 0.0119,
+      "step": 105920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9491302405349707e-05,
+      "loss": 0.0133,
+      "step": 105930
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9491254365611236e-05,
+      "loss": 0.0116,
+      "step": 105940
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9491206325872766e-05,
+      "loss": 0.011,
+      "step": 105950
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949115828613429e-05,
+      "loss": 0.0125,
+      "step": 105960
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949111024639582e-05,
+      "loss": 0.0119,
+      "step": 105970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949106220665735e-05,
+      "loss": 0.013,
+      "step": 105980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949101416691888e-05,
+      "loss": 0.0113,
+      "step": 105990
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490966127180405e-05,
+      "loss": 0.014,
+      "step": 106000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490918087441934e-05,
+      "loss": 0.0138,
+      "step": 106010
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490870047703464e-05,
+      "loss": 0.012,
+      "step": 106020
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949082200796499e-05,
+      "loss": 0.0131,
+      "step": 106030
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949077396822652e-05,
+      "loss": 0.0098,
+      "step": 106040
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490725928488048e-05,
+      "loss": 0.0105,
+      "step": 106050
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490677888749577e-05,
+      "loss": 0.0154,
+      "step": 106060
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490629849011103e-05,
+      "loss": 0.0099,
+      "step": 106070
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490581809272633e-05,
+      "loss": 0.0073,
+      "step": 106080
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490533769534162e-05,
+      "loss": 0.0119,
+      "step": 106090
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949048572979569e-05,
+      "loss": 0.0089,
+      "step": 106100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490437690057217e-05,
+      "loss": 0.0107,
+      "step": 106110
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490389650318746e-05,
+      "loss": 0.0097,
+      "step": 106120
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490341610580276e-05,
+      "loss": 0.0153,
+      "step": 106130
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.94902935708418e-05,
+      "loss": 0.0149,
+      "step": 106140
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949024553110333e-05,
+      "loss": 0.0116,
+      "step": 106150
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949019749136486e-05,
+      "loss": 0.0107,
+      "step": 106160
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.949014945162639e-05,
+      "loss": 0.0122,
+      "step": 106170
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490101411887915e-05,
+      "loss": 0.0144,
+      "step": 106180
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490053372149444e-05,
+      "loss": 0.015,
+      "step": 106190
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9490005332410974e-05,
+      "loss": 0.0071,
+      "step": 106200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.94899572926725e-05,
+      "loss": 0.0122,
+      "step": 106210
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.948990925293403e-05,
+      "loss": 0.0117,
+      "step": 106220
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9489861213195558e-05,
+      "loss": 0.0105,
+      "step": 106230
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9489813173457087e-05,
+      "loss": 0.0127,
+      "step": 106240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9489765133718613e-05,
+      "loss": 0.0105,
+      "step": 106250
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9489717093980142e-05,
+      "loss": 0.0097,
+      "step": 106260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9489669054241672e-05,
+      "loss": 0.012,
+      "step": 106270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.94896210145032e-05,
+      "loss": 0.0119,
+      "step": 106280
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9489572974764727e-05,
+      "loss": 0.0112,
+      "step": 106290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9489524935026256e-05,
+      "loss": 0.0109,
+      "step": 106300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9489476895287785e-05,
+      "loss": 0.0083,
+      "step": 106310
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.948942885554931e-05,
+      "loss": 0.0126,
+      "step": 106320
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.948938081581084e-05,
+      "loss": 0.0152,
+      "step": 106330
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.948933277607237e-05,
+      "loss": 0.0087,
+      "step": 106340
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.94892847363339e-05,
+      "loss": 0.0115,
+      "step": 106350
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9489236696595425e-05,
+      "loss": 0.0129,
+      "step": 106360
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9489188656856954e-05,
+      "loss": 0.0142,
+      "step": 106370
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9489140617118484e-05,
+      "loss": 0.0093,
+      "step": 106380
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.948909257738001e-05,
+      "loss": 0.0087,
+      "step": 106390
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.948904453764154e-05,
+      "loss": 0.0104,
+      "step": 106400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9488996497903068e-05,
+      "loss": 0.0102,
+      "step": 106410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9488948458164597e-05,
+      "loss": 0.0137,
+      "step": 106420
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9488900418426123e-05,
+      "loss": 0.0102,
+      "step": 106430
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9488852378687652e-05,
+      "loss": 0.0091,
+      "step": 106440
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.948880433894918e-05,
+      "loss": 0.0095,
+      "step": 106450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.948875629921071e-05,
+      "loss": 0.0083,
+      "step": 106460
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9488708259472237e-05,
+      "loss": 0.0115,
+      "step": 106470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9488660219733766e-05,
+      "loss": 0.0083,
+      "step": 106480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9488612179995295e-05,
+      "loss": 0.0081,
+      "step": 106490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.948856414025682e-05,
+      "loss": 0.0087,
+      "step": 106500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948851610051835e-05,
+      "loss": 0.0112,
+      "step": 106510
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948846806077988e-05,
+      "loss": 0.0082,
+      "step": 106520
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948842002104141e-05,
+      "loss": 0.0085,
+      "step": 106530
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9488371981302935e-05,
+      "loss": 0.0089,
+      "step": 106540
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9488323941564464e-05,
+      "loss": 0.0136,
+      "step": 106550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9488275901825993e-05,
+      "loss": 0.0112,
+      "step": 106560
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948822786208752e-05,
+      "loss": 0.0103,
+      "step": 106570
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948817982234905e-05,
+      "loss": 0.0127,
+      "step": 106580
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9488131782610578e-05,
+      "loss": 0.0094,
+      "step": 106590
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9488083742872107e-05,
+      "loss": 0.0097,
+      "step": 106600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9488035703133633e-05,
+      "loss": 0.0134,
+      "step": 106610
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9487987663395162e-05,
+      "loss": 0.0089,
+      "step": 106620
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948793962365669e-05,
+      "loss": 0.0121,
+      "step": 106630
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948789158391822e-05,
+      "loss": 0.0145,
+      "step": 106640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9487843544179747e-05,
+      "loss": 0.0092,
+      "step": 106650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9487795504441276e-05,
+      "loss": 0.014,
+      "step": 106660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9487747464702805e-05,
+      "loss": 0.012,
+      "step": 106670
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948769942496433e-05,
+      "loss": 0.009,
+      "step": 106680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948765138522586e-05,
+      "loss": 0.015,
+      "step": 106690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948760334548739e-05,
+      "loss": 0.0158,
+      "step": 106700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948755530574892e-05,
+      "loss": 0.0127,
+      "step": 106710
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9487507266010445e-05,
+      "loss": 0.0115,
+      "step": 106720
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9487459226271974e-05,
+      "loss": 0.0067,
+      "step": 106730
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9487411186533503e-05,
+      "loss": 0.0101,
+      "step": 106740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948736314679503e-05,
+      "loss": 0.0122,
+      "step": 106750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948731510705656e-05,
+      "loss": 0.0104,
+      "step": 106760
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9487267067318088e-05,
+      "loss": 0.0117,
+      "step": 106770
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9487219027579617e-05,
+      "loss": 0.0115,
+      "step": 106780
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9487170987841143e-05,
+      "loss": 0.0112,
+      "step": 106790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9487122948102672e-05,
+      "loss": 0.0096,
+      "step": 106800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.94870749083642e-05,
+      "loss": 0.0106,
+      "step": 106810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948702686862573e-05,
+      "loss": 0.0095,
+      "step": 106820
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486978828887257e-05,
+      "loss": 0.0092,
+      "step": 106830
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486930789148786e-05,
+      "loss": 0.0097,
+      "step": 106840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486882749410315e-05,
+      "loss": 0.0147,
+      "step": 106850
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948683470967184e-05,
+      "loss": 0.0101,
+      "step": 106860
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948678666993337e-05,
+      "loss": 0.018,
+      "step": 106870
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.94867386301949e-05,
+      "loss": 0.012,
+      "step": 106880
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948669059045643e-05,
+      "loss": 0.0116,
+      "step": 106890
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486642550717955e-05,
+      "loss": 0.0139,
+      "step": 106900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486594510979484e-05,
+      "loss": 0.0104,
+      "step": 106910
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486546471241013e-05,
+      "loss": 0.0112,
+      "step": 106920
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948649843150254e-05,
+      "loss": 0.0114,
+      "step": 106930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948645039176407e-05,
+      "loss": 0.0124,
+      "step": 106940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486402352025598e-05,
+      "loss": 0.0115,
+      "step": 106950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486354312287127e-05,
+      "loss": 0.0106,
+      "step": 106960
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486306272548653e-05,
+      "loss": 0.0112,
+      "step": 106970
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486258232810182e-05,
+      "loss": 0.0106,
+      "step": 106980
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948621019307171e-05,
+      "loss": 0.0115,
+      "step": 106990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948616215333324e-05,
+      "loss": 0.0098,
+      "step": 107000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486114113594767e-05,
+      "loss": 0.0076,
+      "step": 107010
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486066073856296e-05,
+      "loss": 0.0076,
+      "step": 107020
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9486018034117825e-05,
+      "loss": 0.0096,
+      "step": 107030
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948596999437935e-05,
+      "loss": 0.0148,
+      "step": 107040
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948592195464088e-05,
+      "loss": 0.0147,
+      "step": 107050
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948587391490241e-05,
+      "loss": 0.0127,
+      "step": 107060
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948582587516394e-05,
+      "loss": 0.0111,
+      "step": 107070
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9485777835425465e-05,
+      "loss": 0.0113,
+      "step": 107080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9485729795686994e-05,
+      "loss": 0.0088,
+      "step": 107090
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9485681755948523e-05,
+      "loss": 0.0156,
+      "step": 107100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948563371621005e-05,
+      "loss": 0.0141,
+      "step": 107110
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948558567647158e-05,
+      "loss": 0.0124,
+      "step": 107120
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9485537636733108e-05,
+      "loss": 0.0094,
+      "step": 107130
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9485489596994637e-05,
+      "loss": 0.0141,
+      "step": 107140
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9485441557256163e-05,
+      "loss": 0.0104,
+      "step": 107150
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9485393517517692e-05,
+      "loss": 0.0113,
+      "step": 107160
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948534547777922e-05,
+      "loss": 0.011,
+      "step": 107170
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.948529743804075e-05,
+      "loss": 0.0102,
+      "step": 107180
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9485249398302276e-05,
+      "loss": 0.01,
+      "step": 107190
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9485201358563806e-05,
+      "loss": 0.0092,
+      "step": 107200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9485153318825335e-05,
+      "loss": 0.0056,
+      "step": 107210
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948510527908686e-05,
+      "loss": 0.0094,
+      "step": 107220
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948505723934839e-05,
+      "loss": 0.008,
+      "step": 107230
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948500919960992e-05,
+      "loss": 0.0082,
+      "step": 107240
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948496115987145e-05,
+      "loss": 0.0107,
+      "step": 107250
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484913120132975e-05,
+      "loss": 0.0107,
+      "step": 107260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484865080394504e-05,
+      "loss": 0.0087,
+      "step": 107270
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484817040656033e-05,
+      "loss": 0.0095,
+      "step": 107280
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948476900091756e-05,
+      "loss": 0.0079,
+      "step": 107290
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484720961179088e-05,
+      "loss": 0.0162,
+      "step": 107300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484672921440618e-05,
+      "loss": 0.0126,
+      "step": 107310
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484624881702147e-05,
+      "loss": 0.0094,
+      "step": 107320
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484576841963673e-05,
+      "loss": 0.0085,
+      "step": 107330
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484528802225202e-05,
+      "loss": 0.0128,
+      "step": 107340
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948448076248673e-05,
+      "loss": 0.0092,
+      "step": 107350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484432722748257e-05,
+      "loss": 0.0089,
+      "step": 107360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484384683009786e-05,
+      "loss": 0.0129,
+      "step": 107370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484336643271316e-05,
+      "loss": 0.0092,
+      "step": 107380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484288603532845e-05,
+      "loss": 0.0089,
+      "step": 107390
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948424056379437e-05,
+      "loss": 0.0094,
+      "step": 107400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.94841925240559e-05,
+      "loss": 0.0167,
+      "step": 107410
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948414448431743e-05,
+      "loss": 0.013,
+      "step": 107420
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948409644457896e-05,
+      "loss": 0.007,
+      "step": 107430
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484048404840485e-05,
+      "loss": 0.0108,
+      "step": 107440
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9484000365102014e-05,
+      "loss": 0.0132,
+      "step": 107450
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483952325363543e-05,
+      "loss": 0.0108,
+      "step": 107460
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948390428562507e-05,
+      "loss": 0.0091,
+      "step": 107470
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483861049860447e-05,
+      "loss": 0.0092,
+      "step": 107480
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483813010121976e-05,
+      "loss": 0.0165,
+      "step": 107490
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483764970383502e-05,
+      "loss": 0.0131,
+      "step": 107500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948371693064503e-05,
+      "loss": 0.0138,
+      "step": 107510
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948366889090656e-05,
+      "loss": 0.0119,
+      "step": 107520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483620851168087e-05,
+      "loss": 0.011,
+      "step": 107530
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483572811429616e-05,
+      "loss": 0.0115,
+      "step": 107540
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483524771691145e-05,
+      "loss": 0.0113,
+      "step": 107550
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483476731952674e-05,
+      "loss": 0.011,
+      "step": 107560
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.94834286922142e-05,
+      "loss": 0.0077,
+      "step": 107570
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948338065247573e-05,
+      "loss": 0.0128,
+      "step": 107580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948333261273726e-05,
+      "loss": 0.0079,
+      "step": 107590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483284572998785e-05,
+      "loss": 0.0112,
+      "step": 107600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483236533260314e-05,
+      "loss": 0.0102,
+      "step": 107610
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483188493521843e-05,
+      "loss": 0.0113,
+      "step": 107620
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483140453783372e-05,
+      "loss": 0.0105,
+      "step": 107630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.94830924140449e-05,
+      "loss": 0.0089,
+      "step": 107640
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9483044374306428e-05,
+      "loss": 0.0086,
+      "step": 107650
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482996334567957e-05,
+      "loss": 0.0106,
+      "step": 107660
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482948294829486e-05,
+      "loss": 0.0111,
+      "step": 107670
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482900255091012e-05,
+      "loss": 0.0122,
+      "step": 107680
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948285221535254e-05,
+      "loss": 0.0075,
+      "step": 107690
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948280417561407e-05,
+      "loss": 0.0098,
+      "step": 107700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482756135875596e-05,
+      "loss": 0.0155,
+      "step": 107710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482708096137126e-05,
+      "loss": 0.0097,
+      "step": 107720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482660056398655e-05,
+      "loss": 0.0099,
+      "step": 107730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482612016660184e-05,
+      "loss": 0.0093,
+      "step": 107740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948256397692171e-05,
+      "loss": 0.0138,
+      "step": 107750
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948251593718324e-05,
+      "loss": 0.0107,
+      "step": 107760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948246789744477e-05,
+      "loss": 0.0088,
+      "step": 107770
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482419857706295e-05,
+      "loss": 0.0116,
+      "step": 107780
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482371817967824e-05,
+      "loss": 0.011,
+      "step": 107790
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482323778229353e-05,
+      "loss": 0.0094,
+      "step": 107800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482275738490882e-05,
+      "loss": 0.01,
+      "step": 107810
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482227698752408e-05,
+      "loss": 0.0101,
+      "step": 107820
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482179659013938e-05,
+      "loss": 0.0131,
+      "step": 107830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482131619275467e-05,
+      "loss": 0.0116,
+      "step": 107840
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482083579536996e-05,
+      "loss": 0.0098,
+      "step": 107850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9482035539798522e-05,
+      "loss": 0.0131,
+      "step": 107860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948198750006005e-05,
+      "loss": 0.013,
+      "step": 107870
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.948193946032158e-05,
+      "loss": 0.0094,
+      "step": 107880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9481891420583106e-05,
+      "loss": 0.0105,
+      "step": 107890
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481843380844636e-05,
+      "loss": 0.0107,
+      "step": 107900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481795341106165e-05,
+      "loss": 0.0091,
+      "step": 107910
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481747301367694e-05,
+      "loss": 0.0099,
+      "step": 107920
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.948169926162922e-05,
+      "loss": 0.0092,
+      "step": 107930
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.948165122189075e-05,
+      "loss": 0.0127,
+      "step": 107940
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.948160318215228e-05,
+      "loss": 0.0123,
+      "step": 107950
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481555142413804e-05,
+      "loss": 0.012,
+      "step": 107960
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481507102675334e-05,
+      "loss": 0.0133,
+      "step": 107970
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481459062936863e-05,
+      "loss": 0.011,
+      "step": 107980
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481411023198392e-05,
+      "loss": 0.01,
+      "step": 107990
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481362983459918e-05,
+      "loss": 0.009,
+      "step": 108000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481314943721447e-05,
+      "loss": 0.0129,
+      "step": 108010
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481266903982977e-05,
+      "loss": 0.0079,
+      "step": 108020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481218864244506e-05,
+      "loss": 0.007,
+      "step": 108030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481170824506032e-05,
+      "loss": 0.0103,
+      "step": 108040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.948112278476756e-05,
+      "loss": 0.0124,
+      "step": 108050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.948107474502909e-05,
+      "loss": 0.0133,
+      "step": 108060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9481026705290616e-05,
+      "loss": 0.0098,
+      "step": 108070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480978665552146e-05,
+      "loss": 0.0139,
+      "step": 108080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480930625813675e-05,
+      "loss": 0.0095,
+      "step": 108090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480882586075204e-05,
+      "loss": 0.0147,
+      "step": 108100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.948083454633673e-05,
+      "loss": 0.0126,
+      "step": 108110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.948078650659826e-05,
+      "loss": 0.0143,
+      "step": 108120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.948073846685979e-05,
+      "loss": 0.013,
+      "step": 108130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480690427121314e-05,
+      "loss": 0.0145,
+      "step": 108140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480642387382844e-05,
+      "loss": 0.0124,
+      "step": 108150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480594347644373e-05,
+      "loss": 0.0125,
+      "step": 108160
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480546307905902e-05,
+      "loss": 0.0094,
+      "step": 108170
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480498268167428e-05,
+      "loss": 0.0105,
+      "step": 108180
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480450228428957e-05,
+      "loss": 0.0151,
+      "step": 108190
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480402188690487e-05,
+      "loss": 0.0097,
+      "step": 108200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480354148952016e-05,
+      "loss": 0.0092,
+      "step": 108210
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480306109213542e-05,
+      "loss": 0.0092,
+      "step": 108220
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.948025806947507e-05,
+      "loss": 0.0162,
+      "step": 108230
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.94802100297366e-05,
+      "loss": 0.0089,
+      "step": 108240
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480161989998126e-05,
+      "loss": 0.0139,
+      "step": 108250
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480113950259655e-05,
+      "loss": 0.0125,
+      "step": 108260
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480065910521185e-05,
+      "loss": 0.0097,
+      "step": 108270
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9480017870782714e-05,
+      "loss": 0.0128,
+      "step": 108280
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.947996983104424e-05,
+      "loss": 0.0087,
+      "step": 108290
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.947992179130577e-05,
+      "loss": 0.0085,
+      "step": 108300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.94798737515673e-05,
+      "loss": 0.0138,
+      "step": 108310
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9479825711828824e-05,
+      "loss": 0.0111,
+      "step": 108320
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9479777672090354e-05,
+      "loss": 0.0125,
+      "step": 108330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9479729632351883e-05,
+      "loss": 0.0107,
+      "step": 108340
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9479681592613412e-05,
+      "loss": 0.0114,
+      "step": 108350
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9479633552874938e-05,
+      "loss": 0.0082,
+      "step": 108360
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9479585513136467e-05,
+      "loss": 0.0119,
+      "step": 108370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9479537473397997e-05,
+      "loss": 0.0127,
+      "step": 108380
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9479489433659526e-05,
+      "loss": 0.0078,
+      "step": 108390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.947944139392105e-05,
+      "loss": 0.0088,
+      "step": 108400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.947939335418258e-05,
+      "loss": 0.0092,
+      "step": 108410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.947934531444411e-05,
+      "loss": 0.0118,
+      "step": 108420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9479297274705636e-05,
+      "loss": 0.0082,
+      "step": 108430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9479249234967165e-05,
+      "loss": 0.0093,
+      "step": 108440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9479201195228695e-05,
+      "loss": 0.0085,
+      "step": 108450
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9479153155490224e-05,
+      "loss": 0.0075,
+      "step": 108460
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.947910511575175e-05,
+      "loss": 0.0104,
+      "step": 108470
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.947905707601328e-05,
+      "loss": 0.0098,
+      "step": 108480
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.947900903627481e-05,
+      "loss": 0.0109,
+      "step": 108490
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9478960996536334e-05,
+      "loss": 0.013,
+      "step": 108500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9478912956797863e-05,
+      "loss": 0.012,
+      "step": 108510
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9478864917059393e-05,
+      "loss": 0.0134,
+      "step": 108520
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9478816877320922e-05,
+      "loss": 0.0131,
+      "step": 108530
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9478768837582448e-05,
+      "loss": 0.0112,
+      "step": 108540
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9478720797843977e-05,
+      "loss": 0.0087,
+      "step": 108550
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9478672758105506e-05,
+      "loss": 0.0112,
+      "step": 108560
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9478624718367036e-05,
+      "loss": 0.012,
+      "step": 108570
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.947857667862856e-05,
+      "loss": 0.0107,
+      "step": 108580
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.947852863889009e-05,
+      "loss": 0.0124,
+      "step": 108590
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947848059915162e-05,
+      "loss": 0.0136,
+      "step": 108600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9478432559413146e-05,
+      "loss": 0.0102,
+      "step": 108610
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9478384519674675e-05,
+      "loss": 0.0122,
+      "step": 108620
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9478336479936205e-05,
+      "loss": 0.0121,
+      "step": 108630
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9478288440197734e-05,
+      "loss": 0.0135,
+      "step": 108640
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947824040045926e-05,
+      "loss": 0.0118,
+      "step": 108650
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947819236072079e-05,
+      "loss": 0.0116,
+      "step": 108660
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9478144320982318e-05,
+      "loss": 0.0113,
+      "step": 108670
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9478096281243844e-05,
+      "loss": 0.0107,
+      "step": 108680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9478048241505373e-05,
+      "loss": 0.0105,
+      "step": 108690
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9478000201766903e-05,
+      "loss": 0.01,
+      "step": 108700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477952162028432e-05,
+      "loss": 0.0097,
+      "step": 108710
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477904122289958e-05,
+      "loss": 0.0167,
+      "step": 108720
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477856082551487e-05,
+      "loss": 0.0108,
+      "step": 108730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477808042813016e-05,
+      "loss": 0.0107,
+      "step": 108740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477760003074546e-05,
+      "loss": 0.0133,
+      "step": 108750
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947771196333607e-05,
+      "loss": 0.0104,
+      "step": 108760
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.94776639235976e-05,
+      "loss": 0.0132,
+      "step": 108770
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947761588385913e-05,
+      "loss": 0.0098,
+      "step": 108780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477567844120656e-05,
+      "loss": 0.0129,
+      "step": 108790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477519804382185e-05,
+      "loss": 0.013,
+      "step": 108800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477471764643714e-05,
+      "loss": 0.0101,
+      "step": 108810
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477423724905244e-05,
+      "loss": 0.0117,
+      "step": 108820
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947737568516677e-05,
+      "loss": 0.0097,
+      "step": 108830
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.94773276454283e-05,
+      "loss": 0.0117,
+      "step": 108840
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477279605689828e-05,
+      "loss": 0.0097,
+      "step": 108850
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477231565951354e-05,
+      "loss": 0.0151,
+      "step": 108860
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477183526212883e-05,
+      "loss": 0.013,
+      "step": 108870
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477135486474413e-05,
+      "loss": 0.0111,
+      "step": 108880
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477087446735942e-05,
+      "loss": 0.0099,
+      "step": 108890
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9477039406997468e-05,
+      "loss": 0.0139,
+      "step": 108900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476991367258997e-05,
+      "loss": 0.0101,
+      "step": 108910
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476943327520526e-05,
+      "loss": 0.0101,
+      "step": 108920
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476895287782052e-05,
+      "loss": 0.0113,
+      "step": 108930
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947684724804358e-05,
+      "loss": 0.0098,
+      "step": 108940
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947679920830511e-05,
+      "loss": 0.0116,
+      "step": 108950
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947675116856664e-05,
+      "loss": 0.0091,
+      "step": 108960
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476703128828166e-05,
+      "loss": 0.0127,
+      "step": 108970
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476655089089695e-05,
+      "loss": 0.0115,
+      "step": 108980
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476607049351224e-05,
+      "loss": 0.0103,
+      "step": 108990
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476559009612754e-05,
+      "loss": 0.0108,
+      "step": 109000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476510969874283e-05,
+      "loss": 0.0098,
+      "step": 109010
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476462930135812e-05,
+      "loss": 0.0092,
+      "step": 109020
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476414890397338e-05,
+      "loss": 0.0107,
+      "step": 109030
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476366850658867e-05,
+      "loss": 0.0124,
+      "step": 109040
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476318810920397e-05,
+      "loss": 0.0105,
+      "step": 109050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476270771181926e-05,
+      "loss": 0.0121,
+      "step": 109060
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476222731443452e-05,
+      "loss": 0.0145,
+      "step": 109070
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947617469170498e-05,
+      "loss": 0.0101,
+      "step": 109080
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947612665196651e-05,
+      "loss": 0.0106,
+      "step": 109090
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476078612228036e-05,
+      "loss": 0.0131,
+      "step": 109100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9476030572489565e-05,
+      "loss": 0.0152,
+      "step": 109110
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9475982532751095e-05,
+      "loss": 0.0115,
+      "step": 109120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9475934493012624e-05,
+      "loss": 0.0124,
+      "step": 109130
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947588645327415e-05,
+      "loss": 0.0098,
+      "step": 109140
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947583841353568e-05,
+      "loss": 0.0092,
+      "step": 109150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947579037379721e-05,
+      "loss": 0.0119,
+      "step": 109160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9475742334058734e-05,
+      "loss": 0.0113,
+      "step": 109170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9475694294320264e-05,
+      "loss": 0.0109,
+      "step": 109180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9475646254581793e-05,
+      "loss": 0.0101,
+      "step": 109190
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9475598214843322e-05,
+      "loss": 0.0079,
+      "step": 109200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9475550175104848e-05,
+      "loss": 0.0108,
+      "step": 109210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9475502135366377e-05,
+      "loss": 0.0085,
+      "step": 109220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9475454095627907e-05,
+      "loss": 0.0104,
+      "step": 109230
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9475406055889436e-05,
+      "loss": 0.0128,
+      "step": 109240
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947535801615096e-05,
+      "loss": 0.0124,
+      "step": 109250
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947530997641249e-05,
+      "loss": 0.0106,
+      "step": 109260
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.947526193667402e-05,
+      "loss": 0.0122,
+      "step": 109270
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9475213896935546e-05,
+      "loss": 0.0116,
+      "step": 109280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9475165857197075e-05,
+      "loss": 0.0107,
+      "step": 109290
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9475117817458605e-05,
+      "loss": 0.0119,
+      "step": 109300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9475069777720134e-05,
+      "loss": 0.0109,
+      "step": 109310
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947502173798166e-05,
+      "loss": 0.0067,
+      "step": 109320
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947497369824319e-05,
+      "loss": 0.0094,
+      "step": 109330
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947492565850472e-05,
+      "loss": 0.013,
+      "step": 109340
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474877618766244e-05,
+      "loss": 0.008,
+      "step": 109350
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474829579027773e-05,
+      "loss": 0.0097,
+      "step": 109360
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474781539289303e-05,
+      "loss": 0.0097,
+      "step": 109370
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474733499550832e-05,
+      "loss": 0.009,
+      "step": 109380
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474685459812358e-05,
+      "loss": 0.0129,
+      "step": 109390
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474637420073887e-05,
+      "loss": 0.0118,
+      "step": 109400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474589380335416e-05,
+      "loss": 0.0083,
+      "step": 109410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474541340596946e-05,
+      "loss": 0.0117,
+      "step": 109420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947449330085847e-05,
+      "loss": 0.0115,
+      "step": 109430
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947444526112e-05,
+      "loss": 0.0102,
+      "step": 109440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947439722138153e-05,
+      "loss": 0.0098,
+      "step": 109450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474349181643056e-05,
+      "loss": 0.0092,
+      "step": 109460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474301141904585e-05,
+      "loss": 0.0113,
+      "step": 109470
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474253102166115e-05,
+      "loss": 0.0082,
+      "step": 109480
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474205062427644e-05,
+      "loss": 0.0109,
+      "step": 109490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947415702268917e-05,
+      "loss": 0.0073,
+      "step": 109500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.94741089829507e-05,
+      "loss": 0.0083,
+      "step": 109510
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474060943212228e-05,
+      "loss": 0.0101,
+      "step": 109520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9474012903473754e-05,
+      "loss": 0.0108,
+      "step": 109530
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473964863735283e-05,
+      "loss": 0.0107,
+      "step": 109540
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473916823996813e-05,
+      "loss": 0.0126,
+      "step": 109550
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473868784258342e-05,
+      "loss": 0.0125,
+      "step": 109560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473820744519868e-05,
+      "loss": 0.0123,
+      "step": 109570
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473772704781397e-05,
+      "loss": 0.0135,
+      "step": 109580
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473724665042926e-05,
+      "loss": 0.0133,
+      "step": 109590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473676625304456e-05,
+      "loss": 0.0111,
+      "step": 109600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947362858556598e-05,
+      "loss": 0.009,
+      "step": 109610
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947358054582751e-05,
+      "loss": 0.0108,
+      "step": 109620
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947353250608904e-05,
+      "loss": 0.0128,
+      "step": 109630
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473484466350566e-05,
+      "loss": 0.0094,
+      "step": 109640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473436426612095e-05,
+      "loss": 0.0129,
+      "step": 109650
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473388386873624e-05,
+      "loss": 0.0128,
+      "step": 109660
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473340347135154e-05,
+      "loss": 0.0072,
+      "step": 109670
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947329230739668e-05,
+      "loss": 0.0093,
+      "step": 109680
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947324426765821e-05,
+      "loss": 0.0096,
+      "step": 109690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473196227919738e-05,
+      "loss": 0.0073,
+      "step": 109700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473148188181264e-05,
+      "loss": 0.0086,
+      "step": 109710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473100148442793e-05,
+      "loss": 0.0088,
+      "step": 109720
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473052108704323e-05,
+      "loss": 0.0103,
+      "step": 109730
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9473004068965852e-05,
+      "loss": 0.0109,
+      "step": 109740
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472956029227378e-05,
+      "loss": 0.0105,
+      "step": 109750
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472907989488907e-05,
+      "loss": 0.0176,
+      "step": 109760
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472859949750436e-05,
+      "loss": 0.0117,
+      "step": 109770
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472811910011966e-05,
+      "loss": 0.0131,
+      "step": 109780
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947276387027349e-05,
+      "loss": 0.0099,
+      "step": 109790
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947271583053502e-05,
+      "loss": 0.0147,
+      "step": 109800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947266779079655e-05,
+      "loss": 0.0153,
+      "step": 109810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472619751058076e-05,
+      "loss": 0.0098,
+      "step": 109820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472571711319605e-05,
+      "loss": 0.0091,
+      "step": 109830
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472523671581134e-05,
+      "loss": 0.0146,
+      "step": 109840
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472475631842664e-05,
+      "loss": 0.0081,
+      "step": 109850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947242759210419e-05,
+      "loss": 0.0124,
+      "step": 109860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.947237955236572e-05,
+      "loss": 0.015,
+      "step": 109870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472331512627248e-05,
+      "loss": 0.0113,
+      "step": 109880
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472283472888774e-05,
+      "loss": 0.0084,
+      "step": 109890
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472235433150303e-05,
+      "loss": 0.0106,
+      "step": 109900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472187393411833e-05,
+      "loss": 0.0101,
+      "step": 109910
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472139353673362e-05,
+      "loss": 0.01,
+      "step": 109920
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472091313934888e-05,
+      "loss": 0.0118,
+      "step": 109930
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9472043274196417e-05,
+      "loss": 0.0087,
+      "step": 109940
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9471995234457946e-05,
+      "loss": 0.0087,
+      "step": 109950
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9471947194719475e-05,
+      "loss": 0.0125,
+      "step": 109960
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9471899154981e-05,
+      "loss": 0.0095,
+      "step": 109970
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.947185111524253e-05,
+      "loss": 0.0103,
+      "step": 109980
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.947180307550406e-05,
+      "loss": 0.0087,
+      "step": 109990
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9471755035765586e-05,
+      "loss": 0.0109,
+      "step": 110000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9471706996027115e-05,
+      "loss": 0.0115,
+      "step": 110010
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9471658956288644e-05,
+      "loss": 0.0133,
+      "step": 110020
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9471610916550174e-05,
+      "loss": 0.0107,
+      "step": 110030
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.94715628768117e-05,
+      "loss": 0.0098,
+      "step": 110040
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.947151483707323e-05,
+      "loss": 0.012,
+      "step": 110050
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9471466797334758e-05,
+      "loss": 0.0087,
+      "step": 110060
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9471418757596284e-05,
+      "loss": 0.0098,
+      "step": 110070
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9471370717857813e-05,
+      "loss": 0.0094,
+      "step": 110080
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9471322678119342e-05,
+      "loss": 0.0102,
+      "step": 110090
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.947127463838087e-05,
+      "loss": 0.013,
+      "step": 110100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9471226598642398e-05,
+      "loss": 0.0123,
+      "step": 110110
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9471178558903927e-05,
+      "loss": 0.0117,
+      "step": 110120
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9471130519165456e-05,
+      "loss": 0.01,
+      "step": 110130
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9471082479426985e-05,
+      "loss": 0.0114,
+      "step": 110140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.947103443968851e-05,
+      "loss": 0.0114,
+      "step": 110150
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.947098639995004e-05,
+      "loss": 0.0142,
+      "step": 110160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.947093836021157e-05,
+      "loss": 0.0108,
+      "step": 110170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9470890320473096e-05,
+      "loss": 0.0106,
+      "step": 110180
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9470842280734625e-05,
+      "loss": 0.0105,
+      "step": 110190
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9470794240996154e-05,
+      "loss": 0.0099,
+      "step": 110200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9470746201257683e-05,
+      "loss": 0.0117,
+      "step": 110210
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.947069816151921e-05,
+      "loss": 0.0108,
+      "step": 110220
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.947065012178074e-05,
+      "loss": 0.0103,
+      "step": 110230
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9470606886016113e-05,
+      "loss": 0.0108,
+      "step": 110240
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9470558846277643e-05,
+      "loss": 0.0116,
+      "step": 110250
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9470510806539172e-05,
+      "loss": 0.0115,
+      "step": 110260
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.94704627668007e-05,
+      "loss": 0.0121,
+      "step": 110270
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9470414727062227e-05,
+      "loss": 0.0118,
+      "step": 110280
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9470366687323756e-05,
+      "loss": 0.011,
+      "step": 110290
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9470318647585286e-05,
+      "loss": 0.0121,
+      "step": 110300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.947027060784681e-05,
+      "loss": 0.0102,
+      "step": 110310
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.947022256810834e-05,
+      "loss": 0.0163,
+      "step": 110320
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.947017452836987e-05,
+      "loss": 0.0131,
+      "step": 110330
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.94701264886314e-05,
+      "loss": 0.0099,
+      "step": 110340
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9470078448892925e-05,
+      "loss": 0.0134,
+      "step": 110350
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9470030409154454e-05,
+      "loss": 0.0121,
+      "step": 110360
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9469982369415984e-05,
+      "loss": 0.0134,
+      "step": 110370
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.946993432967751e-05,
+      "loss": 0.0087,
+      "step": 110380
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.946988628993904e-05,
+      "loss": 0.0081,
+      "step": 110390
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9469838250200568e-05,
+      "loss": 0.0066,
+      "step": 110400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9469790210462097e-05,
+      "loss": 0.0112,
+      "step": 110410
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9469742170723623e-05,
+      "loss": 0.0162,
+      "step": 110420
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9469694130985152e-05,
+      "loss": 0.0112,
+      "step": 110430
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9469646091246682e-05,
+      "loss": 0.0113,
+      "step": 110440
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.946959805150821e-05,
+      "loss": 0.0126,
+      "step": 110450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9469550011769737e-05,
+      "loss": 0.0104,
+      "step": 110460
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9469501972031266e-05,
+      "loss": 0.0099,
+      "step": 110470
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9469453932292795e-05,
+      "loss": 0.0093,
+      "step": 110480
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.946940589255432e-05,
+      "loss": 0.0117,
+      "step": 110490
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.946935785281585e-05,
+      "loss": 0.0107,
+      "step": 110500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.946930981307738e-05,
+      "loss": 0.0084,
+      "step": 110510
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.946926177333891e-05,
+      "loss": 0.0131,
+      "step": 110520
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9469213733600435e-05,
+      "loss": 0.0126,
+      "step": 110530
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9469165693861964e-05,
+      "loss": 0.0128,
+      "step": 110540
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9469117654123494e-05,
+      "loss": 0.0111,
+      "step": 110550
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.946906961438502e-05,
+      "loss": 0.012,
+      "step": 110560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.946902157464655e-05,
+      "loss": 0.013,
+      "step": 110570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9468973534908078e-05,
+      "loss": 0.0086,
+      "step": 110580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9468925495169607e-05,
+      "loss": 0.0095,
+      "step": 110590
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9468877455431133e-05,
+      "loss": 0.0101,
+      "step": 110600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9468829415692662e-05,
+      "loss": 0.0093,
+      "step": 110610
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.946878137595419e-05,
+      "loss": 0.0105,
+      "step": 110620
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.946873333621572e-05,
+      "loss": 0.0108,
+      "step": 110630
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9468685296477247e-05,
+      "loss": 0.014,
+      "step": 110640
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9468637256738776e-05,
+      "loss": 0.0102,
+      "step": 110650
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9468589217000305e-05,
+      "loss": 0.0089,
+      "step": 110660
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.946854117726183e-05,
+      "loss": 0.0097,
+      "step": 110670
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946849313752336e-05,
+      "loss": 0.0086,
+      "step": 110680
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946844509778489e-05,
+      "loss": 0.0119,
+      "step": 110690
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946839705804642e-05,
+      "loss": 0.009,
+      "step": 110700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9468349018307945e-05,
+      "loss": 0.0098,
+      "step": 110710
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9468300978569474e-05,
+      "loss": 0.0082,
+      "step": 110720
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9468252938831003e-05,
+      "loss": 0.014,
+      "step": 110730
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946820489909253e-05,
+      "loss": 0.0108,
+      "step": 110740
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946815685935406e-05,
+      "loss": 0.0116,
+      "step": 110750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9468108819615588e-05,
+      "loss": 0.0122,
+      "step": 110760
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9468060779877117e-05,
+      "loss": 0.0128,
+      "step": 110770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9468012740138643e-05,
+      "loss": 0.0098,
+      "step": 110780
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9467964700400172e-05,
+      "loss": 0.0111,
+      "step": 110790
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.94679166606617e-05,
+      "loss": 0.0135,
+      "step": 110800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946786862092323e-05,
+      "loss": 0.0122,
+      "step": 110810
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9467820581184757e-05,
+      "loss": 0.008,
+      "step": 110820
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9467772541446286e-05,
+      "loss": 0.0109,
+      "step": 110830
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9467724501707815e-05,
+      "loss": 0.0093,
+      "step": 110840
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946767646196934e-05,
+      "loss": 0.011,
+      "step": 110850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946762842223087e-05,
+      "loss": 0.0129,
+      "step": 110860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.94675803824924e-05,
+      "loss": 0.0082,
+      "step": 110870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946753234275393e-05,
+      "loss": 0.0134,
+      "step": 110880
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9467484303015455e-05,
+      "loss": 0.0141,
+      "step": 110890
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9467436263276984e-05,
+      "loss": 0.0087,
+      "step": 110900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9467388223538513e-05,
+      "loss": 0.0151,
+      "step": 110910
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946734018380004e-05,
+      "loss": 0.0095,
+      "step": 110920
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946729214406157e-05,
+      "loss": 0.0146,
+      "step": 110930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9467244104323098e-05,
+      "loss": 0.0122,
+      "step": 110940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9467196064584627e-05,
+      "loss": 0.0128,
+      "step": 110950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9467148024846153e-05,
+      "loss": 0.0094,
+      "step": 110960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9467099985107682e-05,
+      "loss": 0.008,
+      "step": 110970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946705194536921e-05,
+      "loss": 0.0111,
+      "step": 110980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946700390563074e-05,
+      "loss": 0.01,
+      "step": 110990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9466955865892267e-05,
+      "loss": 0.009,
+      "step": 111000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9466907826153796e-05,
+      "loss": 0.0094,
+      "step": 111010
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9466859786415325e-05,
+      "loss": 0.0153,
+      "step": 111020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946681174667685e-05,
+      "loss": 0.0123,
+      "step": 111030
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946676370693838e-05,
+      "loss": 0.0103,
+      "step": 111040
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946671566719991e-05,
+      "loss": 0.014,
+      "step": 111050
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946666762746144e-05,
+      "loss": 0.009,
+      "step": 111060
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9466619587722965e-05,
+      "loss": 0.0106,
+      "step": 111070
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9466571547984494e-05,
+      "loss": 0.0082,
+      "step": 111080
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9466523508246023e-05,
+      "loss": 0.0157,
+      "step": 111090
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946647546850755e-05,
+      "loss": 0.0096,
+      "step": 111100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946642742876908e-05,
+      "loss": 0.011,
+      "step": 111110
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9466379389030608e-05,
+      "loss": 0.0096,
+      "step": 111120
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9466331349292137e-05,
+      "loss": 0.0119,
+      "step": 111130
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9466283309553663e-05,
+      "loss": 0.0099,
+      "step": 111140
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9466235269815192e-05,
+      "loss": 0.0104,
+      "step": 111150
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946618723007672e-05,
+      "loss": 0.0125,
+      "step": 111160
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946613919033825e-05,
+      "loss": 0.016,
+      "step": 111170
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9466091150599777e-05,
+      "loss": 0.0109,
+      "step": 111180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9466043110861306e-05,
+      "loss": 0.015,
+      "step": 111190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9465995071122835e-05,
+      "loss": 0.0056,
+      "step": 111200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946594703138436e-05,
+      "loss": 0.0093,
+      "step": 111210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946589899164589e-05,
+      "loss": 0.011,
+      "step": 111220
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946585095190742e-05,
+      "loss": 0.0157,
+      "step": 111230
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946580291216895e-05,
+      "loss": 0.0108,
+      "step": 111240
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9465754872430475e-05,
+      "loss": 0.0123,
+      "step": 111250
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9465706832692004e-05,
+      "loss": 0.0115,
+      "step": 111260
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9465658792953533e-05,
+      "loss": 0.012,
+      "step": 111270
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946561075321506e-05,
+      "loss": 0.009,
+      "step": 111280
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946556271347659e-05,
+      "loss": 0.0125,
+      "step": 111290
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9465514673738118e-05,
+      "loss": 0.013,
+      "step": 111300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9465466633999647e-05,
+      "loss": 0.0136,
+      "step": 111310
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9465418594261173e-05,
+      "loss": 0.0094,
+      "step": 111320
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9465370554522702e-05,
+      "loss": 0.0081,
+      "step": 111330
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946532251478423e-05,
+      "loss": 0.0111,
+      "step": 111340
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.946527447504576e-05,
+      "loss": 0.0086,
+      "step": 111350
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9465226435307286e-05,
+      "loss": 0.0086,
+      "step": 111360
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9465178395568816e-05,
+      "loss": 0.0091,
+      "step": 111370
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9465130355830345e-05,
+      "loss": 0.0122,
+      "step": 111380
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946508231609187e-05,
+      "loss": 0.0078,
+      "step": 111390
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.94650342763534e-05,
+      "loss": 0.0105,
+      "step": 111400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946498623661493e-05,
+      "loss": 0.0094,
+      "step": 111410
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946493819687646e-05,
+      "loss": 0.0106,
+      "step": 111420
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9464890157137985e-05,
+      "loss": 0.0085,
+      "step": 111430
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9464842117399514e-05,
+      "loss": 0.0138,
+      "step": 111440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9464794077661043e-05,
+      "loss": 0.0086,
+      "step": 111450
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946474603792257e-05,
+      "loss": 0.0105,
+      "step": 111460
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9464697998184098e-05,
+      "loss": 0.0113,
+      "step": 111470
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9464649958445628e-05,
+      "loss": 0.0122,
+      "step": 111480
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9464601918707157e-05,
+      "loss": 0.0104,
+      "step": 111490
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9464553878968683e-05,
+      "loss": 0.0108,
+      "step": 111500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9464505839230212e-05,
+      "loss": 0.0075,
+      "step": 111510
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946445779949174e-05,
+      "loss": 0.014,
+      "step": 111520
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946440975975327e-05,
+      "loss": 0.0123,
+      "step": 111530
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9464361720014796e-05,
+      "loss": 0.0147,
+      "step": 111540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9464313680276326e-05,
+      "loss": 0.0088,
+      "step": 111550
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9464265640537855e-05,
+      "loss": 0.0114,
+      "step": 111560
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946421760079938e-05,
+      "loss": 0.0134,
+      "step": 111570
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946416956106091e-05,
+      "loss": 0.0091,
+      "step": 111580
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946412152132244e-05,
+      "loss": 0.0126,
+      "step": 111590
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946407348158397e-05,
+      "loss": 0.0098,
+      "step": 111600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9464025441845494e-05,
+      "loss": 0.0186,
+      "step": 111610
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463977402107024e-05,
+      "loss": 0.0103,
+      "step": 111620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463929362368553e-05,
+      "loss": 0.0101,
+      "step": 111630
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946388132263008e-05,
+      "loss": 0.0092,
+      "step": 111640
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463833282891608e-05,
+      "loss": 0.0164,
+      "step": 111650
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463785243153137e-05,
+      "loss": 0.012,
+      "step": 111660
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463737203414667e-05,
+      "loss": 0.0098,
+      "step": 111670
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463689163676193e-05,
+      "loss": 0.0082,
+      "step": 111680
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463641123937722e-05,
+      "loss": 0.0102,
+      "step": 111690
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946359308419925e-05,
+      "loss": 0.0105,
+      "step": 111700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946354504446078e-05,
+      "loss": 0.0105,
+      "step": 111710
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463497004722306e-05,
+      "loss": 0.0134,
+      "step": 111720
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463448964983836e-05,
+      "loss": 0.009,
+      "step": 111730
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463400925245365e-05,
+      "loss": 0.0102,
+      "step": 111740
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946335288550689e-05,
+      "loss": 0.0121,
+      "step": 111750
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946330484576842e-05,
+      "loss": 0.0137,
+      "step": 111760
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946325680602995e-05,
+      "loss": 0.0086,
+      "step": 111770
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946320876629148e-05,
+      "loss": 0.0102,
+      "step": 111780
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463160726553004e-05,
+      "loss": 0.0105,
+      "step": 111790
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463112686814534e-05,
+      "loss": 0.0095,
+      "step": 111800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9463064647076063e-05,
+      "loss": 0.0082,
+      "step": 111810
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946301660733759e-05,
+      "loss": 0.0124,
+      "step": 111820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462968567599118e-05,
+      "loss": 0.0101,
+      "step": 111830
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462920527860647e-05,
+      "loss": 0.0124,
+      "step": 111840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462872488122177e-05,
+      "loss": 0.0138,
+      "step": 111850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462824448383703e-05,
+      "loss": 0.0074,
+      "step": 111860
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462776408645232e-05,
+      "loss": 0.0114,
+      "step": 111870
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946272836890676e-05,
+      "loss": 0.0078,
+      "step": 111880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946268032916829e-05,
+      "loss": 0.0088,
+      "step": 111890
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462632289429816e-05,
+      "loss": 0.0097,
+      "step": 111900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462584249691345e-05,
+      "loss": 0.0243,
+      "step": 111910
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462536209952875e-05,
+      "loss": 0.0079,
+      "step": 111920
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.94624881702144e-05,
+      "loss": 0.0125,
+      "step": 111930
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946244013047593e-05,
+      "loss": 0.0073,
+      "step": 111940
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946239209073746e-05,
+      "loss": 0.0112,
+      "step": 111950
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946234405099899e-05,
+      "loss": 0.0129,
+      "step": 111960
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462296011260514e-05,
+      "loss": 0.0105,
+      "step": 111970
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462247971522044e-05,
+      "loss": 0.0077,
+      "step": 111980
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462199931783573e-05,
+      "loss": 0.009,
+      "step": 111990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.94621518920451e-05,
+      "loss": 0.01,
+      "step": 112000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462103852306628e-05,
+      "loss": 0.01,
+      "step": 112010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462055812568157e-05,
+      "loss": 0.0099,
+      "step": 112020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462007772829687e-05,
+      "loss": 0.0115,
+      "step": 112030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9461959733091212e-05,
+      "loss": 0.0095,
+      "step": 112040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9461911693352742e-05,
+      "loss": 0.0113,
+      "step": 112050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946186365361427e-05,
+      "loss": 0.0128,
+      "step": 112060
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.94618156138758e-05,
+      "loss": 0.0109,
+      "step": 112070
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946176757413733e-05,
+      "loss": 0.0088,
+      "step": 112080
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946171953439886e-05,
+      "loss": 0.0104,
+      "step": 112090
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9461671494660385e-05,
+      "loss": 0.0077,
+      "step": 112100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9461623454921914e-05,
+      "loss": 0.0099,
+      "step": 112110
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9461575415183443e-05,
+      "loss": 0.0109,
+      "step": 112120
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946152737544497e-05,
+      "loss": 0.0099,
+      "step": 112130
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.94614793357065e-05,
+      "loss": 0.011,
+      "step": 112140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9461431295968028e-05,
+      "loss": 0.0144,
+      "step": 112150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9461383256229557e-05,
+      "loss": 0.0101,
+      "step": 112160
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9461335216491083e-05,
+      "loss": 0.0127,
+      "step": 112170
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9461287176752612e-05,
+      "loss": 0.0116,
+      "step": 112180
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946123913701414e-05,
+      "loss": 0.0078,
+      "step": 112190
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946119109727567e-05,
+      "loss": 0.0105,
+      "step": 112200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9461143057537196e-05,
+      "loss": 0.0116,
+      "step": 112210
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9461095017798726e-05,
+      "loss": 0.011,
+      "step": 112220
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9461046978060255e-05,
+      "loss": 0.0082,
+      "step": 112230
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946099893832178e-05,
+      "loss": 0.0088,
+      "step": 112240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946095089858331e-05,
+      "loss": 0.0123,
+      "step": 112250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946090285884484e-05,
+      "loss": 0.0092,
+      "step": 112260
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946085481910637e-05,
+      "loss": 0.0118,
+      "step": 112270
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9460806779367895e-05,
+      "loss": 0.0101,
+      "step": 112280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9460758739629424e-05,
+      "loss": 0.0116,
+      "step": 112290
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9460710699890953e-05,
+      "loss": 0.0095,
+      "step": 112300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946066266015248e-05,
+      "loss": 0.0093,
+      "step": 112310
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9460614620414008e-05,
+      "loss": 0.0118,
+      "step": 112320
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9460566580675538e-05,
+      "loss": 0.0117,
+      "step": 112330
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9460518540937067e-05,
+      "loss": 0.0076,
+      "step": 112340
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9460470501198593e-05,
+      "loss": 0.0136,
+      "step": 112350
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9460422461460122e-05,
+      "loss": 0.0121,
+      "step": 112360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946037442172165e-05,
+      "loss": 0.0105,
+      "step": 112370
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946032638198318e-05,
+      "loss": 0.01,
+      "step": 112380
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9460278342244706e-05,
+      "loss": 0.0134,
+      "step": 112390
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9460230302506236e-05,
+      "loss": 0.009,
+      "step": 112400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9460182262767765e-05,
+      "loss": 0.0147,
+      "step": 112410
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946013422302929e-05,
+      "loss": 0.0075,
+      "step": 112420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946008618329082e-05,
+      "loss": 0.0096,
+      "step": 112430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.946003814355235e-05,
+      "loss": 0.0117,
+      "step": 112440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.945999010381388e-05,
+      "loss": 0.012,
+      "step": 112450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459942064075404e-05,
+      "loss": 0.0108,
+      "step": 112460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459894024336934e-05,
+      "loss": 0.0084,
+      "step": 112470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459845984598463e-05,
+      "loss": 0.0101,
+      "step": 112480
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.945979794485999e-05,
+      "loss": 0.0105,
+      "step": 112490
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459749905121518e-05,
+      "loss": 0.0113,
+      "step": 112500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459701865383047e-05,
+      "loss": 0.0085,
+      "step": 112510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459653825644577e-05,
+      "loss": 0.0115,
+      "step": 112520
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459605785906103e-05,
+      "loss": 0.0124,
+      "step": 112530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459557746167632e-05,
+      "loss": 0.0109,
+      "step": 112540
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.945950970642916e-05,
+      "loss": 0.0096,
+      "step": 112550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.945946166669069e-05,
+      "loss": 0.0134,
+      "step": 112560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459413626952216e-05,
+      "loss": 0.0132,
+      "step": 112570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459365587213746e-05,
+      "loss": 0.0079,
+      "step": 112580
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459317547475275e-05,
+      "loss": 0.0094,
+      "step": 112590
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.94592695077368e-05,
+      "loss": 0.0101,
+      "step": 112600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.945922146799833e-05,
+      "loss": 0.0135,
+      "step": 112610
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.945917342825986e-05,
+      "loss": 0.0102,
+      "step": 112620
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.945912538852139e-05,
+      "loss": 0.0117,
+      "step": 112630
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459077348782914e-05,
+      "loss": 0.0094,
+      "step": 112640
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9459029309044444e-05,
+      "loss": 0.0097,
+      "step": 112650
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9458981269305973e-05,
+      "loss": 0.0105,
+      "step": 112660
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.94589332295675e-05,
+      "loss": 0.0112,
+      "step": 112670
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9458885189829028e-05,
+      "loss": 0.0126,
+      "step": 112680
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9458837150090557e-05,
+      "loss": 0.0131,
+      "step": 112690
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9458789110352087e-05,
+      "loss": 0.0116,
+      "step": 112700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9458741070613613e-05,
+      "loss": 0.0137,
+      "step": 112710
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9458693030875142e-05,
+      "loss": 0.0133,
+      "step": 112720
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.945864499113667e-05,
+      "loss": 0.0148,
+      "step": 112730
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.94585969513982e-05,
+      "loss": 0.0115,
+      "step": 112740
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9458548911659726e-05,
+      "loss": 0.0127,
+      "step": 112750
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9458500871921255e-05,
+      "loss": 0.0113,
+      "step": 112760
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9458452832182785e-05,
+      "loss": 0.0073,
+      "step": 112770
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945840479244431e-05,
+      "loss": 0.0121,
+      "step": 112780
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945835675270584e-05,
+      "loss": 0.008,
+      "step": 112790
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945830871296737e-05,
+      "loss": 0.0074,
+      "step": 112800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.94582606732289e-05,
+      "loss": 0.0087,
+      "step": 112810
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9458212633490424e-05,
+      "loss": 0.0107,
+      "step": 112820
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9458164593751954e-05,
+      "loss": 0.0086,
+      "step": 112830
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9458116554013483e-05,
+      "loss": 0.0124,
+      "step": 112840
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945806851427501e-05,
+      "loss": 0.0112,
+      "step": 112850
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9458020474536538e-05,
+      "loss": 0.0107,
+      "step": 112860
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457972434798067e-05,
+      "loss": 0.0086,
+      "step": 112870
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457924395059597e-05,
+      "loss": 0.0099,
+      "step": 112880
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457876355321122e-05,
+      "loss": 0.019,
+      "step": 112890
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457828315582652e-05,
+      "loss": 0.0098,
+      "step": 112900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945778027584418e-05,
+      "loss": 0.0125,
+      "step": 112910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945773223610571e-05,
+      "loss": 0.0083,
+      "step": 112920
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457684196367236e-05,
+      "loss": 0.0096,
+      "step": 112930
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457636156628765e-05,
+      "loss": 0.0104,
+      "step": 112940
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457588116890295e-05,
+      "loss": 0.0097,
+      "step": 112950
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945754007715182e-05,
+      "loss": 0.0137,
+      "step": 112960
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945749203741335e-05,
+      "loss": 0.0143,
+      "step": 112970
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945744399767488e-05,
+      "loss": 0.0102,
+      "step": 112980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945739595793641e-05,
+      "loss": 0.0091,
+      "step": 112990
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457347918197934e-05,
+      "loss": 0.0099,
+      "step": 113000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457299878459464e-05,
+      "loss": 0.0143,
+      "step": 113010
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457251838720993e-05,
+      "loss": 0.0124,
+      "step": 113020
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945720379898252e-05,
+      "loss": 0.01,
+      "step": 113030
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457155759244048e-05,
+      "loss": 0.0106,
+      "step": 113040
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457107719505577e-05,
+      "loss": 0.0077,
+      "step": 113050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457059679767106e-05,
+      "loss": 0.0108,
+      "step": 113060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9457011640028632e-05,
+      "loss": 0.0123,
+      "step": 113070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945696360029016e-05,
+      "loss": 0.0108,
+      "step": 113080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945691556055169e-05,
+      "loss": 0.0146,
+      "step": 113090
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945686752081322e-05,
+      "loss": 0.0101,
+      "step": 113100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9456819481074746e-05,
+      "loss": 0.0117,
+      "step": 113110
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9456771441336275e-05,
+      "loss": 0.0081,
+      "step": 113120
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9456723401597805e-05,
+      "loss": 0.0121,
+      "step": 113130
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945667536185933e-05,
+      "loss": 0.0135,
+      "step": 113140
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945662732212086e-05,
+      "loss": 0.0145,
+      "step": 113150
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945657928238239e-05,
+      "loss": 0.0116,
+      "step": 113160
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9456531242643918e-05,
+      "loss": 0.0089,
+      "step": 113170
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9456483202905444e-05,
+      "loss": 0.0124,
+      "step": 113180
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9456435163166973e-05,
+      "loss": 0.0117,
+      "step": 113190
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9456387123428503e-05,
+      "loss": 0.0147,
+      "step": 113200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945633908369003e-05,
+      "loss": 0.0083,
+      "step": 113210
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9456291043951558e-05,
+      "loss": 0.0127,
+      "step": 113220
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9456243004213087e-05,
+      "loss": 0.013,
+      "step": 113230
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9456194964474616e-05,
+      "loss": 0.0128,
+      "step": 113240
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9456146924736142e-05,
+      "loss": 0.0125,
+      "step": 113250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945609888499767e-05,
+      "loss": 0.0099,
+      "step": 113260
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.94560508452592e-05,
+      "loss": 0.0113,
+      "step": 113270
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945600280552073e-05,
+      "loss": 0.0098,
+      "step": 113280
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9455954765782256e-05,
+      "loss": 0.0091,
+      "step": 113290
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9455906726043785e-05,
+      "loss": 0.0144,
+      "step": 113300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9455858686305315e-05,
+      "loss": 0.0093,
+      "step": 113310
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945581064656684e-05,
+      "loss": 0.0093,
+      "step": 113320
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945576260682837e-05,
+      "loss": 0.0106,
+      "step": 113330
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.94557145670899e-05,
+      "loss": 0.0115,
+      "step": 113340
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9455666527351428e-05,
+      "loss": 0.0077,
+      "step": 113350
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9455618487612954e-05,
+      "loss": 0.0092,
+      "step": 113360
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9455570447874483e-05,
+      "loss": 0.0109,
+      "step": 113370
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9455522408136013e-05,
+      "loss": 0.01,
+      "step": 113380
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945547436839754e-05,
+      "loss": 0.0142,
+      "step": 113390
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9455426328659068e-05,
+      "loss": 0.0108,
+      "step": 113400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9455378288920597e-05,
+      "loss": 0.0106,
+      "step": 113410
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9455330249182126e-05,
+      "loss": 0.0103,
+      "step": 113420
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9455282209443652e-05,
+      "loss": 0.0093,
+      "step": 113430
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.945523416970518e-05,
+      "loss": 0.0099,
+      "step": 113440
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945518612996671e-05,
+      "loss": 0.009,
+      "step": 113450
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9455138090228237e-05,
+      "loss": 0.0167,
+      "step": 113460
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9455090050489766e-05,
+      "loss": 0.0089,
+      "step": 113470
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9455042010751295e-05,
+      "loss": 0.0119,
+      "step": 113480
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454993971012824e-05,
+      "loss": 0.0098,
+      "step": 113490
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945494593127435e-05,
+      "loss": 0.0158,
+      "step": 113500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945489789153588e-05,
+      "loss": 0.0089,
+      "step": 113510
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945484985179741e-05,
+      "loss": 0.0135,
+      "step": 113520
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454801812058938e-05,
+      "loss": 0.0145,
+      "step": 113530
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454753772320464e-05,
+      "loss": 0.0075,
+      "step": 113540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454705732581993e-05,
+      "loss": 0.0093,
+      "step": 113550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454657692843523e-05,
+      "loss": 0.0095,
+      "step": 113560
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945460965310505e-05,
+      "loss": 0.0149,
+      "step": 113570
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454561613366578e-05,
+      "loss": 0.0109,
+      "step": 113580
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454513573628107e-05,
+      "loss": 0.0167,
+      "step": 113590
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454465533889636e-05,
+      "loss": 0.013,
+      "step": 113600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454417494151162e-05,
+      "loss": 0.011,
+      "step": 113610
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945436945441269e-05,
+      "loss": 0.0112,
+      "step": 113620
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945432141467422e-05,
+      "loss": 0.0113,
+      "step": 113630
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454273374935747e-05,
+      "loss": 0.0112,
+      "step": 113640
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454225335197276e-05,
+      "loss": 0.0107,
+      "step": 113650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454177295458805e-05,
+      "loss": 0.009,
+      "step": 113660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9454129255720334e-05,
+      "loss": 0.0128,
+      "step": 113670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945408121598186e-05,
+      "loss": 0.0103,
+      "step": 113680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945403317624339e-05,
+      "loss": 0.0082,
+      "step": 113690
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945398513650492e-05,
+      "loss": 0.0118,
+      "step": 113700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453937096766448e-05,
+      "loss": 0.0176,
+      "step": 113710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453889057027974e-05,
+      "loss": 0.0109,
+      "step": 113720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453841017289503e-05,
+      "loss": 0.0122,
+      "step": 113730
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453792977551032e-05,
+      "loss": 0.014,
+      "step": 113740
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945374493781256e-05,
+      "loss": 0.0127,
+      "step": 113750
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453696898074088e-05,
+      "loss": 0.0115,
+      "step": 113760
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453648858335617e-05,
+      "loss": 0.0136,
+      "step": 113770
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453600818597146e-05,
+      "loss": 0.0118,
+      "step": 113780
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453552778858672e-05,
+      "loss": 0.0095,
+      "step": 113790
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.94535047391202e-05,
+      "loss": 0.0074,
+      "step": 113800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945345669938173e-05,
+      "loss": 0.0097,
+      "step": 113810
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453408659643256e-05,
+      "loss": 0.0161,
+      "step": 113820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453360619904786e-05,
+      "loss": 0.0077,
+      "step": 113830
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453312580166315e-05,
+      "loss": 0.0102,
+      "step": 113840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453264540427844e-05,
+      "loss": 0.0071,
+      "step": 113850
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945321650068937e-05,
+      "loss": 0.011,
+      "step": 113860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.94531684609509e-05,
+      "loss": 0.0107,
+      "step": 113870
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945312042121243e-05,
+      "loss": 0.0096,
+      "step": 113880
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453072381473958e-05,
+      "loss": 0.0116,
+      "step": 113890
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9453024341735484e-05,
+      "loss": 0.0121,
+      "step": 113900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452976301997013e-05,
+      "loss": 0.012,
+      "step": 113910
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452928262258542e-05,
+      "loss": 0.0117,
+      "step": 113920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452880222520068e-05,
+      "loss": 0.0134,
+      "step": 113930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452832182781598e-05,
+      "loss": 0.0132,
+      "step": 113940
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452784143043127e-05,
+      "loss": 0.0099,
+      "step": 113950
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452736103304656e-05,
+      "loss": 0.0182,
+      "step": 113960
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452688063566182e-05,
+      "loss": 0.0117,
+      "step": 113970
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945264002382771e-05,
+      "loss": 0.0098,
+      "step": 113980
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945259198408924e-05,
+      "loss": 0.0104,
+      "step": 113990
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452543944350766e-05,
+      "loss": 0.0103,
+      "step": 114000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452495904612296e-05,
+      "loss": 0.0087,
+      "step": 114010
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452447864873825e-05,
+      "loss": 0.0115,
+      "step": 114020
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452399825135354e-05,
+      "loss": 0.0084,
+      "step": 114030
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945235178539688e-05,
+      "loss": 0.0114,
+      "step": 114040
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945230374565841e-05,
+      "loss": 0.0149,
+      "step": 114050
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.945225570591994e-05,
+      "loss": 0.0099,
+      "step": 114060
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452207666181468e-05,
+      "loss": 0.0088,
+      "step": 114070
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452159626442994e-05,
+      "loss": 0.0109,
+      "step": 114080
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452111586704523e-05,
+      "loss": 0.0098,
+      "step": 114090
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452063546966052e-05,
+      "loss": 0.0102,
+      "step": 114100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9452015507227578e-05,
+      "loss": 0.0115,
+      "step": 114110
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9451967467489107e-05,
+      "loss": 0.0096,
+      "step": 114120
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9451919427750637e-05,
+      "loss": 0.0104,
+      "step": 114130
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9451871388012166e-05,
+      "loss": 0.0115,
+      "step": 114140
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9451823348273692e-05,
+      "loss": 0.0077,
+      "step": 114150
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945177530853522e-05,
+      "loss": 0.0165,
+      "step": 114160
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945172726879675e-05,
+      "loss": 0.014,
+      "step": 114170
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9451679229058276e-05,
+      "loss": 0.0109,
+      "step": 114180
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9451631189319806e-05,
+      "loss": 0.0108,
+      "step": 114190
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9451583149581335e-05,
+      "loss": 0.0112,
+      "step": 114200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9451535109842864e-05,
+      "loss": 0.0096,
+      "step": 114210
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945148707010439e-05,
+      "loss": 0.0065,
+      "step": 114220
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945143903036592e-05,
+      "loss": 0.0115,
+      "step": 114230
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945139099062745e-05,
+      "loss": 0.0101,
+      "step": 114240
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9451342950888978e-05,
+      "loss": 0.01,
+      "step": 114250
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9451294911150504e-05,
+      "loss": 0.0099,
+      "step": 114260
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9451246871412033e-05,
+      "loss": 0.0108,
+      "step": 114270
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9451198831673562e-05,
+      "loss": 0.016,
+      "step": 114280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9451150791935088e-05,
+      "loss": 0.0074,
+      "step": 114290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9451102752196617e-05,
+      "loss": 0.0105,
+      "step": 114300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9451054712458147e-05,
+      "loss": 0.01,
+      "step": 114310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945101147669352e-05,
+      "loss": 0.0115,
+      "step": 114320
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945096343695505e-05,
+      "loss": 0.0094,
+      "step": 114330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945091539721658e-05,
+      "loss": 0.0077,
+      "step": 114340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450867357478106e-05,
+      "loss": 0.0108,
+      "step": 114350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450819317739635e-05,
+      "loss": 0.0102,
+      "step": 114360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450771278001164e-05,
+      "loss": 0.0096,
+      "step": 114370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450723238262693e-05,
+      "loss": 0.0113,
+      "step": 114380
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945067519852422e-05,
+      "loss": 0.0121,
+      "step": 114390
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945062715878575e-05,
+      "loss": 0.01,
+      "step": 114400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450579119047278e-05,
+      "loss": 0.0148,
+      "step": 114410
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450531079308804e-05,
+      "loss": 0.0094,
+      "step": 114420
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450483039570333e-05,
+      "loss": 0.0106,
+      "step": 114430
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450434999831862e-05,
+      "loss": 0.0146,
+      "step": 114440
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945038696009339e-05,
+      "loss": 0.0105,
+      "step": 114450
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450338920354917e-05,
+      "loss": 0.0105,
+      "step": 114460
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450290880616447e-05,
+      "loss": 0.0128,
+      "step": 114470
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450242840877976e-05,
+      "loss": 0.0096,
+      "step": 114480
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450194801139505e-05,
+      "loss": 0.0113,
+      "step": 114490
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945014676140103e-05,
+      "loss": 0.0117,
+      "step": 114500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945009872166256e-05,
+      "loss": 0.0089,
+      "step": 114510
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.945005068192409e-05,
+      "loss": 0.0123,
+      "step": 114520
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9450002642185616e-05,
+      "loss": 0.0096,
+      "step": 114530
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449954602447145e-05,
+      "loss": 0.0088,
+      "step": 114540
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449906562708674e-05,
+      "loss": 0.0094,
+      "step": 114550
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449858522970203e-05,
+      "loss": 0.0132,
+      "step": 114560
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.944981048323173e-05,
+      "loss": 0.0122,
+      "step": 114570
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.944976244349326e-05,
+      "loss": 0.0148,
+      "step": 114580
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449714403754788e-05,
+      "loss": 0.0099,
+      "step": 114590
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449666364016314e-05,
+      "loss": 0.0111,
+      "step": 114600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449618324277843e-05,
+      "loss": 0.0146,
+      "step": 114610
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449570284539372e-05,
+      "loss": 0.0106,
+      "step": 114620
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.94495222448009e-05,
+      "loss": 0.0127,
+      "step": 114630
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449474205062427e-05,
+      "loss": 0.0092,
+      "step": 114640
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449426165323957e-05,
+      "loss": 0.0098,
+      "step": 114650
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449378125585486e-05,
+      "loss": 0.008,
+      "step": 114660
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449330085847015e-05,
+      "loss": 0.0151,
+      "step": 114670
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.944928204610854e-05,
+      "loss": 0.0127,
+      "step": 114680
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.944923400637007e-05,
+      "loss": 0.007,
+      "step": 114690
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.94491859666316e-05,
+      "loss": 0.0091,
+      "step": 114700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449137926893126e-05,
+      "loss": 0.0071,
+      "step": 114710
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449089887154655e-05,
+      "loss": 0.011,
+      "step": 114720
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9449041847416184e-05,
+      "loss": 0.012,
+      "step": 114730
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9448993807677713e-05,
+      "loss": 0.015,
+      "step": 114740
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.944894576793924e-05,
+      "loss": 0.0094,
+      "step": 114750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.944889772820077e-05,
+      "loss": 0.0149,
+      "step": 114760
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9448849688462298e-05,
+      "loss": 0.0101,
+      "step": 114770
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9448801648723824e-05,
+      "loss": 0.0132,
+      "step": 114780
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9448753608985353e-05,
+      "loss": 0.0096,
+      "step": 114790
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9448705569246882e-05,
+      "loss": 0.0104,
+      "step": 114800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.944865752950841e-05,
+      "loss": 0.0082,
+      "step": 114810
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9448609489769937e-05,
+      "loss": 0.0142,
+      "step": 114820
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9448561450031467e-05,
+      "loss": 0.0093,
+      "step": 114830
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9448513410292996e-05,
+      "loss": 0.0074,
+      "step": 114840
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9448465370554525e-05,
+      "loss": 0.0121,
+      "step": 114850
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944841733081605e-05,
+      "loss": 0.0108,
+      "step": 114860
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944836929107758e-05,
+      "loss": 0.0082,
+      "step": 114870
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944832125133911e-05,
+      "loss": 0.0079,
+      "step": 114880
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9448273211600635e-05,
+      "loss": 0.0066,
+      "step": 114890
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9448225171862165e-05,
+      "loss": 0.0117,
+      "step": 114900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9448177132123694e-05,
+      "loss": 0.0101,
+      "step": 114910
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9448129092385223e-05,
+      "loss": 0.0073,
+      "step": 114920
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944808105264675e-05,
+      "loss": 0.0081,
+      "step": 114930
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944803301290828e-05,
+      "loss": 0.0126,
+      "step": 114940
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447984973169808e-05,
+      "loss": 0.0122,
+      "step": 114950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447936933431334e-05,
+      "loss": 0.0127,
+      "step": 114960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447888893692863e-05,
+      "loss": 0.0093,
+      "step": 114970
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447840853954392e-05,
+      "loss": 0.0115,
+      "step": 114980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944779281421592e-05,
+      "loss": 0.0128,
+      "step": 114990
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447744774477447e-05,
+      "loss": 0.0106,
+      "step": 115000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447696734738976e-05,
+      "loss": 0.0094,
+      "step": 115010
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447648695000506e-05,
+      "loss": 0.0081,
+      "step": 115020
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944760065526203e-05,
+      "loss": 0.0121,
+      "step": 115030
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944755261552356e-05,
+      "loss": 0.0121,
+      "step": 115040
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944750457578509e-05,
+      "loss": 0.0113,
+      "step": 115050
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944745653604662e-05,
+      "loss": 0.0101,
+      "step": 115060
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447408496308145e-05,
+      "loss": 0.0085,
+      "step": 115070
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447360456569675e-05,
+      "loss": 0.0099,
+      "step": 115080
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447312416831204e-05,
+      "loss": 0.0108,
+      "step": 115090
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447264377092733e-05,
+      "loss": 0.0128,
+      "step": 115100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944721633735426e-05,
+      "loss": 0.0084,
+      "step": 115110
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447168297615788e-05,
+      "loss": 0.0112,
+      "step": 115120
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447120257877318e-05,
+      "loss": 0.0114,
+      "step": 115130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447072218138843e-05,
+      "loss": 0.0091,
+      "step": 115140
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9447024178400376e-05,
+      "loss": 0.0183,
+      "step": 115150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446976138661905e-05,
+      "loss": 0.0124,
+      "step": 115160
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944692809892343e-05,
+      "loss": 0.0121,
+      "step": 115170
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944688005918496e-05,
+      "loss": 0.0096,
+      "step": 115180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944683201944649e-05,
+      "loss": 0.0107,
+      "step": 115190
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446783979708016e-05,
+      "loss": 0.0108,
+      "step": 115200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446735939969545e-05,
+      "loss": 0.0093,
+      "step": 115210
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446687900231074e-05,
+      "loss": 0.0074,
+      "step": 115220
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446639860492603e-05,
+      "loss": 0.0097,
+      "step": 115230
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944659182075413e-05,
+      "loss": 0.0107,
+      "step": 115240
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944654378101566e-05,
+      "loss": 0.0148,
+      "step": 115250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446495741277188e-05,
+      "loss": 0.0124,
+      "step": 115260
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446447701538714e-05,
+      "loss": 0.0082,
+      "step": 115270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446399661800243e-05,
+      "loss": 0.0089,
+      "step": 115280
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446351622061772e-05,
+      "loss": 0.0117,
+      "step": 115290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.94463035823233e-05,
+      "loss": 0.0149,
+      "step": 115300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446255542584827e-05,
+      "loss": 0.0096,
+      "step": 115310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446207502846357e-05,
+      "loss": 0.0092,
+      "step": 115320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446159463107886e-05,
+      "loss": 0.0097,
+      "step": 115330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9446111423369415e-05,
+      "loss": 0.0162,
+      "step": 115340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944606338363094e-05,
+      "loss": 0.0172,
+      "step": 115350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944601534389247e-05,
+      "loss": 0.0115,
+      "step": 115360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445967304154e-05,
+      "loss": 0.0077,
+      "step": 115370
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445919264415526e-05,
+      "loss": 0.01,
+      "step": 115380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445871224677055e-05,
+      "loss": 0.0119,
+      "step": 115390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445823184938584e-05,
+      "loss": 0.0088,
+      "step": 115400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445775145200113e-05,
+      "loss": 0.0093,
+      "step": 115410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944572710546164e-05,
+      "loss": 0.0083,
+      "step": 115420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944567906572317e-05,
+      "loss": 0.0113,
+      "step": 115430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445631025984698e-05,
+      "loss": 0.0089,
+      "step": 115440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445582986246224e-05,
+      "loss": 0.0098,
+      "step": 115450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445534946507753e-05,
+      "loss": 0.0079,
+      "step": 115460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445486906769282e-05,
+      "loss": 0.012,
+      "step": 115470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.944543886703081e-05,
+      "loss": 0.009,
+      "step": 115480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445390827292337e-05,
+      "loss": 0.0114,
+      "step": 115490
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445342787553867e-05,
+      "loss": 0.0079,
+      "step": 115500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445294747815396e-05,
+      "loss": 0.0122,
+      "step": 115510
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9445246708076925e-05,
+      "loss": 0.0072,
+      "step": 115520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944519866833845e-05,
+      "loss": 0.0141,
+      "step": 115530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944515062859998e-05,
+      "loss": 0.0108,
+      "step": 115540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944510258886151e-05,
+      "loss": 0.0125,
+      "step": 115550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9445054549123036e-05,
+      "loss": 0.0117,
+      "step": 115560
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9445006509384565e-05,
+      "loss": 0.0088,
+      "step": 115570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444958469646094e-05,
+      "loss": 0.0141,
+      "step": 115580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444910429907623e-05,
+      "loss": 0.009,
+      "step": 115590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944486239016915e-05,
+      "loss": 0.0107,
+      "step": 115600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944481435043068e-05,
+      "loss": 0.0098,
+      "step": 115610
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444766310692208e-05,
+      "loss": 0.0089,
+      "step": 115620
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444718270953734e-05,
+      "loss": 0.0105,
+      "step": 115630
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444670231215263e-05,
+      "loss": 0.0085,
+      "step": 115640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444622191476792e-05,
+      "loss": 0.0125,
+      "step": 115650
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944457415173832e-05,
+      "loss": 0.0117,
+      "step": 115660
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444526111999847e-05,
+      "loss": 0.0124,
+      "step": 115670
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444478072261377e-05,
+      "loss": 0.0125,
+      "step": 115680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444430032522906e-05,
+      "loss": 0.0093,
+      "step": 115690
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444381992784435e-05,
+      "loss": 0.0065,
+      "step": 115700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944433395304596e-05,
+      "loss": 0.0147,
+      "step": 115710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944428591330749e-05,
+      "loss": 0.0153,
+      "step": 115720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944423787356902e-05,
+      "loss": 0.0096,
+      "step": 115730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444189833830545e-05,
+      "loss": 0.0099,
+      "step": 115740
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444141794092075e-05,
+      "loss": 0.0115,
+      "step": 115750
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444093754353604e-05,
+      "loss": 0.0089,
+      "step": 115760
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444045714615133e-05,
+      "loss": 0.0079,
+      "step": 115770
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944399767487666e-05,
+      "loss": 0.0099,
+      "step": 115780
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944394963513819e-05,
+      "loss": 0.0106,
+      "step": 115790
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443901595399718e-05,
+      "loss": 0.0082,
+      "step": 115800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443853555661244e-05,
+      "loss": 0.016,
+      "step": 115810
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443805515922773e-05,
+      "loss": 0.0084,
+      "step": 115820
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443757476184302e-05,
+      "loss": 0.0112,
+      "step": 115830
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944370943644583e-05,
+      "loss": 0.0113,
+      "step": 115840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443661396707357e-05,
+      "loss": 0.0084,
+      "step": 115850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443613356968886e-05,
+      "loss": 0.0124,
+      "step": 115860
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443565317230416e-05,
+      "loss": 0.0111,
+      "step": 115870
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443517277491945e-05,
+      "loss": 0.0104,
+      "step": 115880
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944346923775347e-05,
+      "loss": 0.0123,
+      "step": 115890
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443421198015e-05,
+      "loss": 0.007,
+      "step": 115900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944337315827653e-05,
+      "loss": 0.0112,
+      "step": 115910
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443325118538055e-05,
+      "loss": 0.0056,
+      "step": 115920
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443277078799585e-05,
+      "loss": 0.0151,
+      "step": 115930
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443229039061114e-05,
+      "loss": 0.0093,
+      "step": 115940
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443180999322643e-05,
+      "loss": 0.0097,
+      "step": 115950
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944313295958417e-05,
+      "loss": 0.0078,
+      "step": 115960
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443084919845698e-05,
+      "loss": 0.0121,
+      "step": 115970
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9443036880107228e-05,
+      "loss": 0.0101,
+      "step": 115980
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442988840368753e-05,
+      "loss": 0.0107,
+      "step": 115990
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442940800630283e-05,
+      "loss": 0.017,
+      "step": 116000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442892760891812e-05,
+      "loss": 0.0149,
+      "step": 116010
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944284472115334e-05,
+      "loss": 0.0115,
+      "step": 116020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442796681414867e-05,
+      "loss": 0.0131,
+      "step": 116030
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442748641676396e-05,
+      "loss": 0.0119,
+      "step": 116040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442700601937926e-05,
+      "loss": 0.0084,
+      "step": 116050
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442652562199455e-05,
+      "loss": 0.0097,
+      "step": 116060
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944260452246098e-05,
+      "loss": 0.0101,
+      "step": 116070
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944255648272251e-05,
+      "loss": 0.0114,
+      "step": 116080
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944250844298404e-05,
+      "loss": 0.0089,
+      "step": 116090
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442460403245565e-05,
+      "loss": 0.0088,
+      "step": 116100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442412363507095e-05,
+      "loss": 0.0116,
+      "step": 116110
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442364323768624e-05,
+      "loss": 0.0096,
+      "step": 116120
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442316284030153e-05,
+      "loss": 0.0111,
+      "step": 116130
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944226824429168e-05,
+      "loss": 0.0088,
+      "step": 116140
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442220204553208e-05,
+      "loss": 0.0117,
+      "step": 116150
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442172164814737e-05,
+      "loss": 0.0078,
+      "step": 116160
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442124125076263e-05,
+      "loss": 0.0091,
+      "step": 116170
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442076085337793e-05,
+      "loss": 0.0105,
+      "step": 116180
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442028045599322e-05,
+      "loss": 0.0143,
+      "step": 116190
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944198000586085e-05,
+      "loss": 0.0132,
+      "step": 116200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9441931966122377e-05,
+      "loss": 0.0122,
+      "step": 116210
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9441883926383906e-05,
+      "loss": 0.0143,
+      "step": 116220
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441835886645436e-05,
+      "loss": 0.0099,
+      "step": 116230
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441787846906965e-05,
+      "loss": 0.0115,
+      "step": 116240
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.944173980716849e-05,
+      "loss": 0.0117,
+      "step": 116250
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.944169176743002e-05,
+      "loss": 0.0086,
+      "step": 116260
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.944164372769155e-05,
+      "loss": 0.0113,
+      "step": 116270
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441595687953075e-05,
+      "loss": 0.0097,
+      "step": 116280
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441547648214604e-05,
+      "loss": 0.0097,
+      "step": 116290
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441499608476134e-05,
+      "loss": 0.0117,
+      "step": 116300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441451568737663e-05,
+      "loss": 0.0086,
+      "step": 116310
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.944140352899919e-05,
+      "loss": 0.0093,
+      "step": 116320
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441355489260718e-05,
+      "loss": 0.0112,
+      "step": 116330
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441307449522247e-05,
+      "loss": 0.0079,
+      "step": 116340
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441259409783773e-05,
+      "loss": 0.009,
+      "step": 116350
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441211370045303e-05,
+      "loss": 0.0095,
+      "step": 116360
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441163330306832e-05,
+      "loss": 0.0082,
+      "step": 116370
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.944111529056836e-05,
+      "loss": 0.0083,
+      "step": 116380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441067250829887e-05,
+      "loss": 0.0131,
+      "step": 116390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9441019211091416e-05,
+      "loss": 0.0116,
+      "step": 116400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440971171352946e-05,
+      "loss": 0.011,
+      "step": 116410
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440923131614475e-05,
+      "loss": 0.012,
+      "step": 116420
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440875091876e-05,
+      "loss": 0.0113,
+      "step": 116430
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.944082705213753e-05,
+      "loss": 0.0104,
+      "step": 116440
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.944077901239906e-05,
+      "loss": 0.0087,
+      "step": 116450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440730972660585e-05,
+      "loss": 0.0079,
+      "step": 116460
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440682932922114e-05,
+      "loss": 0.0165,
+      "step": 116470
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440634893183644e-05,
+      "loss": 0.0098,
+      "step": 116480
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440586853445173e-05,
+      "loss": 0.0091,
+      "step": 116490
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.94405388137067e-05,
+      "loss": 0.0073,
+      "step": 116500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440490773968228e-05,
+      "loss": 0.0085,
+      "step": 116510
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440442734229757e-05,
+      "loss": 0.0115,
+      "step": 116520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440394694491283e-05,
+      "loss": 0.0125,
+      "step": 116530
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440346654752812e-05,
+      "loss": 0.0128,
+      "step": 116540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440298615014342e-05,
+      "loss": 0.0087,
+      "step": 116550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.944025057527587e-05,
+      "loss": 0.0097,
+      "step": 116560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440202535537397e-05,
+      "loss": 0.011,
+      "step": 116570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440154495798926e-05,
+      "loss": 0.0096,
+      "step": 116580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9440106456060455e-05,
+      "loss": 0.0131,
+      "step": 116590
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.944005841632198e-05,
+      "loss": 0.0118,
+      "step": 116600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.944001037658351e-05,
+      "loss": 0.0114,
+      "step": 116610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.943996233684504e-05,
+      "loss": 0.0094,
+      "step": 116620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.943991429710657e-05,
+      "loss": 0.0102,
+      "step": 116630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9439866257368095e-05,
+      "loss": 0.0097,
+      "step": 116640
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9439818217629624e-05,
+      "loss": 0.0092,
+      "step": 116650
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9439770177891154e-05,
+      "loss": 0.0104,
+      "step": 116660
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9439722138152683e-05,
+      "loss": 0.0111,
+      "step": 116670
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.943967409841421e-05,
+      "loss": 0.0091,
+      "step": 116680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9439626058675738e-05,
+      "loss": 0.0157,
+      "step": 116690
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9439578018937267e-05,
+      "loss": 0.016,
+      "step": 116700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9439529979198793e-05,
+      "loss": 0.0117,
+      "step": 116710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9439481939460322e-05,
+      "loss": 0.0146,
+      "step": 116720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.943943389972185e-05,
+      "loss": 0.0155,
+      "step": 116730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.943938585998338e-05,
+      "loss": 0.0117,
+      "step": 116740
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9439337820244907e-05,
+      "loss": 0.0099,
+      "step": 116750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9439289780506436e-05,
+      "loss": 0.012,
+      "step": 116760
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9439241740767965e-05,
+      "loss": 0.0105,
+      "step": 116770
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.943919370102949e-05,
+      "loss": 0.0104,
+      "step": 116780
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.943914566129102e-05,
+      "loss": 0.008,
+      "step": 116790
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.943909762155255e-05,
+      "loss": 0.0106,
+      "step": 116800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.943904958181408e-05,
+      "loss": 0.0113,
+      "step": 116810
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9439001542075605e-05,
+      "loss": 0.0111,
+      "step": 116820
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9438953502337134e-05,
+      "loss": 0.0133,
+      "step": 116830
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9438905462598663e-05,
+      "loss": 0.0105,
+      "step": 116840
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9438857422860193e-05,
+      "loss": 0.0101,
+      "step": 116850
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.943880938312172e-05,
+      "loss": 0.0121,
+      "step": 116860
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9438761343383248e-05,
+      "loss": 0.0088,
+      "step": 116870
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9438713303644777e-05,
+      "loss": 0.0111,
+      "step": 116880
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9438665263906303e-05,
+      "loss": 0.0104,
+      "step": 116890
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9438617224167832e-05,
+      "loss": 0.0117,
+      "step": 116900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.943856918442936e-05,
+      "loss": 0.0117,
+      "step": 116910
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943852114469089e-05,
+      "loss": 0.0084,
+      "step": 116920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9438473104952417e-05,
+      "loss": 0.0113,
+      "step": 116930
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9438425065213946e-05,
+      "loss": 0.0107,
+      "step": 116940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9438377025475475e-05,
+      "loss": 0.0136,
+      "step": 116950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9438328985737e-05,
+      "loss": 0.0094,
+      "step": 116960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943828094599853e-05,
+      "loss": 0.0106,
+      "step": 116970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943823290626006e-05,
+      "loss": 0.0104,
+      "step": 116980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943818486652159e-05,
+      "loss": 0.0087,
+      "step": 116990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9438136826783115e-05,
+      "loss": 0.0134,
+      "step": 117000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9438088787044644e-05,
+      "loss": 0.0111,
+      "step": 117010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9438040747306173e-05,
+      "loss": 0.0096,
+      "step": 117020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437992707567703e-05,
+      "loss": 0.0123,
+      "step": 117030
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943794466782923e-05,
+      "loss": 0.0102,
+      "step": 117040
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437896628090758e-05,
+      "loss": 0.0088,
+      "step": 117050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437848588352287e-05,
+      "loss": 0.0138,
+      "step": 117060
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437800548613813e-05,
+      "loss": 0.0119,
+      "step": 117070
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437752508875342e-05,
+      "loss": 0.0078,
+      "step": 117080
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943770446913687e-05,
+      "loss": 0.0148,
+      "step": 117090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.94376564293984e-05,
+      "loss": 0.0104,
+      "step": 117100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437608389659927e-05,
+      "loss": 0.0096,
+      "step": 117110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437560349921456e-05,
+      "loss": 0.007,
+      "step": 117120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437512310182985e-05,
+      "loss": 0.0115,
+      "step": 117130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943746427044451e-05,
+      "loss": 0.0098,
+      "step": 117140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943741623070604e-05,
+      "loss": 0.0094,
+      "step": 117150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943736819096757e-05,
+      "loss": 0.0076,
+      "step": 117160
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.94373201512291e-05,
+      "loss": 0.0111,
+      "step": 117170
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437272111490625e-05,
+      "loss": 0.0075,
+      "step": 117180
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437224071752154e-05,
+      "loss": 0.0108,
+      "step": 117190
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437176032013683e-05,
+      "loss": 0.0121,
+      "step": 117200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437127992275213e-05,
+      "loss": 0.0077,
+      "step": 117210
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943707995253674e-05,
+      "loss": 0.0154,
+      "step": 117220
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9437031912798268e-05,
+      "loss": 0.0097,
+      "step": 117230
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9436983873059797e-05,
+      "loss": 0.0096,
+      "step": 117240
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943694063729517e-05,
+      "loss": 0.0135,
+      "step": 117250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.94368925975567e-05,
+      "loss": 0.0092,
+      "step": 117260
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943684455781823e-05,
+      "loss": 0.0122,
+      "step": 117270
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9436796518079756e-05,
+      "loss": 0.0103,
+      "step": 117280
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9436748478341285e-05,
+      "loss": 0.0081,
+      "step": 117290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9436700438602815e-05,
+      "loss": 0.0123,
+      "step": 117300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943665239886434e-05,
+      "loss": 0.0114,
+      "step": 117310
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943660435912587e-05,
+      "loss": 0.013,
+      "step": 117320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.94365563193874e-05,
+      "loss": 0.0123,
+      "step": 117330
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9436508279648928e-05,
+      "loss": 0.0135,
+      "step": 117340
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9436460239910454e-05,
+      "loss": 0.012,
+      "step": 117350
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9436412200171983e-05,
+      "loss": 0.0083,
+      "step": 117360
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9436364160433513e-05,
+      "loss": 0.0103,
+      "step": 117370
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943631612069504e-05,
+      "loss": 0.0135,
+      "step": 117380
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9436268080956568e-05,
+      "loss": 0.0101,
+      "step": 117390
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9436220041218097e-05,
+      "loss": 0.0117,
+      "step": 117400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9436172001479626e-05,
+      "loss": 0.0092,
+      "step": 117410
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9436123961741152e-05,
+      "loss": 0.0083,
+      "step": 117420
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943607592200268e-05,
+      "loss": 0.0077,
+      "step": 117430
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943602788226421e-05,
+      "loss": 0.013,
+      "step": 117440
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943597984252574e-05,
+      "loss": 0.0144,
+      "step": 117450
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9435931802787266e-05,
+      "loss": 0.0079,
+      "step": 117460
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9435883763048795e-05,
+      "loss": 0.0107,
+      "step": 117470
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9435835723310324e-05,
+      "loss": 0.0091,
+      "step": 117480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943578768357185e-05,
+      "loss": 0.0115,
+      "step": 117490
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943573964383338e-05,
+      "loss": 0.0077,
+      "step": 117500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943569160409491e-05,
+      "loss": 0.009,
+      "step": 117510
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9435643564356438e-05,
+      "loss": 0.007,
+      "step": 117520
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9435595524617964e-05,
+      "loss": 0.0103,
+      "step": 117530
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9435547484879493e-05,
+      "loss": 0.0091,
+      "step": 117540
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9435499445141023e-05,
+      "loss": 0.0103,
+      "step": 117550
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943545140540255e-05,
+      "loss": 0.0077,
+      "step": 117560
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9435403365664078e-05,
+      "loss": 0.0133,
+      "step": 117570
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9435355325925607e-05,
+      "loss": 0.0132,
+      "step": 117580
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9435307286187136e-05,
+      "loss": 0.0103,
+      "step": 117590
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9435259246448662e-05,
+      "loss": 0.0076,
+      "step": 117600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.943521120671019e-05,
+      "loss": 0.0095,
+      "step": 117610
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943516316697172e-05,
+      "loss": 0.0109,
+      "step": 117620
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943511512723325e-05,
+      "loss": 0.0074,
+      "step": 117630
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9435067087494776e-05,
+      "loss": 0.0126,
+      "step": 117640
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9435019047756305e-05,
+      "loss": 0.0078,
+      "step": 117650
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9434971008017834e-05,
+      "loss": 0.0107,
+      "step": 117660
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943492296827936e-05,
+      "loss": 0.0123,
+      "step": 117670
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943487492854089e-05,
+      "loss": 0.0065,
+      "step": 117680
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943482688880242e-05,
+      "loss": 0.01,
+      "step": 117690
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9434778849063948e-05,
+      "loss": 0.0131,
+      "step": 117700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9434730809325474e-05,
+      "loss": 0.0117,
+      "step": 117710
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9434682769587003e-05,
+      "loss": 0.0132,
+      "step": 117720
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9434634729848533e-05,
+      "loss": 0.0096,
+      "step": 117730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943458669011006e-05,
+      "loss": 0.0097,
+      "step": 117740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9434538650371588e-05,
+      "loss": 0.0114,
+      "step": 117750
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9434490610633117e-05,
+      "loss": 0.0109,
+      "step": 117760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9434442570894646e-05,
+      "loss": 0.0107,
+      "step": 117770
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9434394531156172e-05,
+      "loss": 0.011,
+      "step": 117780
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.94343464914177e-05,
+      "loss": 0.0104,
+      "step": 117790
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943429845167923e-05,
+      "loss": 0.0095,
+      "step": 117800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943425041194076e-05,
+      "loss": 0.0111,
+      "step": 117810
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9434202372202286e-05,
+      "loss": 0.0126,
+      "step": 117820
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9434154332463815e-05,
+      "loss": 0.0092,
+      "step": 117830
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9434106292725344e-05,
+      "loss": 0.0135,
+      "step": 117840
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943405825298687e-05,
+      "loss": 0.0104,
+      "step": 117850
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.94340102132484e-05,
+      "loss": 0.0106,
+      "step": 117860
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943396217350993e-05,
+      "loss": 0.0118,
+      "step": 117870
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433914133771458e-05,
+      "loss": 0.0133,
+      "step": 117880
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433866094032984e-05,
+      "loss": 0.0144,
+      "step": 117890
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433818054294513e-05,
+      "loss": 0.0151,
+      "step": 117900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433770014556042e-05,
+      "loss": 0.0098,
+      "step": 117910
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943372197481757e-05,
+      "loss": 0.0142,
+      "step": 117920
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433673935079098e-05,
+      "loss": 0.0137,
+      "step": 117930
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433625895340627e-05,
+      "loss": 0.0124,
+      "step": 117940
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433577855602156e-05,
+      "loss": 0.0097,
+      "step": 117950
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433529815863682e-05,
+      "loss": 0.0094,
+      "step": 117960
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943348177612521e-05,
+      "loss": 0.0187,
+      "step": 117970
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943343373638674e-05,
+      "loss": 0.0105,
+      "step": 117980
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943338569664827e-05,
+      "loss": 0.0126,
+      "step": 117990
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433337656909796e-05,
+      "loss": 0.0094,
+      "step": 118000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433289617171325e-05,
+      "loss": 0.0108,
+      "step": 118010
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433241577432854e-05,
+      "loss": 0.0077,
+      "step": 118020
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943319353769438e-05,
+      "loss": 0.0083,
+      "step": 118030
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943314549795591e-05,
+      "loss": 0.0085,
+      "step": 118040
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943309745821744e-05,
+      "loss": 0.0099,
+      "step": 118050
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433049418478968e-05,
+      "loss": 0.011,
+      "step": 118060
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9433001378740494e-05,
+      "loss": 0.0103,
+      "step": 118070
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432953339002023e-05,
+      "loss": 0.0086,
+      "step": 118080
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432905299263552e-05,
+      "loss": 0.0074,
+      "step": 118090
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432857259525078e-05,
+      "loss": 0.0099,
+      "step": 118100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432809219786608e-05,
+      "loss": 0.0104,
+      "step": 118110
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432761180048137e-05,
+      "loss": 0.0073,
+      "step": 118120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432713140309666e-05,
+      "loss": 0.0137,
+      "step": 118130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432665100571192e-05,
+      "loss": 0.0122,
+      "step": 118140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943261706083272e-05,
+      "loss": 0.0098,
+      "step": 118150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943256902109425e-05,
+      "loss": 0.0102,
+      "step": 118160
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432520981355776e-05,
+      "loss": 0.0085,
+      "step": 118170
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432472941617306e-05,
+      "loss": 0.0104,
+      "step": 118180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432424901878835e-05,
+      "loss": 0.0086,
+      "step": 118190
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432376862140364e-05,
+      "loss": 0.0087,
+      "step": 118200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943232882240189e-05,
+      "loss": 0.0088,
+      "step": 118210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943228078266342e-05,
+      "loss": 0.0106,
+      "step": 118220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943223274292495e-05,
+      "loss": 0.0071,
+      "step": 118230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432184703186478e-05,
+      "loss": 0.0125,
+      "step": 118240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432136663448007e-05,
+      "loss": 0.0085,
+      "step": 118250
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432088623709536e-05,
+      "loss": 0.0097,
+      "step": 118260
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9432040583971062e-05,
+      "loss": 0.0117,
+      "step": 118270
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943199254423259e-05,
+      "loss": 0.009,
+      "step": 118280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943194450449412e-05,
+      "loss": 0.0119,
+      "step": 118290
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.943189646475565e-05,
+      "loss": 0.0083,
+      "step": 118300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9431848425017176e-05,
+      "loss": 0.0145,
+      "step": 118310
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9431800385278705e-05,
+      "loss": 0.0085,
+      "step": 118320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9431752345540234e-05,
+      "loss": 0.008,
+      "step": 118330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.943170430580176e-05,
+      "loss": 0.0099,
+      "step": 118340
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.943165626606329e-05,
+      "loss": 0.0111,
+      "step": 118350
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.943160822632482e-05,
+      "loss": 0.0084,
+      "step": 118360
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9431560186586348e-05,
+      "loss": 0.0142,
+      "step": 118370
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9431512146847874e-05,
+      "loss": 0.0106,
+      "step": 118380
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9431464107109403e-05,
+      "loss": 0.013,
+      "step": 118390
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9431416067370933e-05,
+      "loss": 0.0117,
+      "step": 118400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.943136802763246e-05,
+      "loss": 0.0095,
+      "step": 118410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9431319987893988e-05,
+      "loss": 0.015,
+      "step": 118420
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9431271948155517e-05,
+      "loss": 0.0087,
+      "step": 118430
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9431223908417046e-05,
+      "loss": 0.0123,
+      "step": 118440
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9431175868678572e-05,
+      "loss": 0.0118,
+      "step": 118450
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.94311278289401e-05,
+      "loss": 0.0091,
+      "step": 118460
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.943107978920163e-05,
+      "loss": 0.0093,
+      "step": 118470
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.943103174946316e-05,
+      "loss": 0.0079,
+      "step": 118480
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430983709724686e-05,
+      "loss": 0.0128,
+      "step": 118490
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430935669986215e-05,
+      "loss": 0.0132,
+      "step": 118500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430887630247744e-05,
+      "loss": 0.0108,
+      "step": 118510
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.943083959050927e-05,
+      "loss": 0.0088,
+      "step": 118520
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.94307915507708e-05,
+      "loss": 0.0093,
+      "step": 118530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.943074351103233e-05,
+      "loss": 0.0105,
+      "step": 118540
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430695471293858e-05,
+      "loss": 0.0085,
+      "step": 118550
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430647431555384e-05,
+      "loss": 0.0092,
+      "step": 118560
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430599391816913e-05,
+      "loss": 0.0111,
+      "step": 118570
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430551352078443e-05,
+      "loss": 0.0055,
+      "step": 118580
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.943050331233997e-05,
+      "loss": 0.0106,
+      "step": 118590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430460076575343e-05,
+      "loss": 0.0092,
+      "step": 118600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430412036836872e-05,
+      "loss": 0.01,
+      "step": 118610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.94303639970984e-05,
+      "loss": 0.0086,
+      "step": 118620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430315957359927e-05,
+      "loss": 0.0119,
+      "step": 118630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430267917621457e-05,
+      "loss": 0.0107,
+      "step": 118640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430219877882986e-05,
+      "loss": 0.0115,
+      "step": 118650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9430171838144515e-05,
+      "loss": 0.0102,
+      "step": 118660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.943012379840604e-05,
+      "loss": 0.012,
+      "step": 118670
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.943007575866757e-05,
+      "loss": 0.0107,
+      "step": 118680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.94300277189291e-05,
+      "loss": 0.0076,
+      "step": 118690
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429979679190626e-05,
+      "loss": 0.0078,
+      "step": 118700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429931639452155e-05,
+      "loss": 0.0128,
+      "step": 118710
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429888403687533e-05,
+      "loss": 0.01,
+      "step": 118720
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.942984036394906e-05,
+      "loss": 0.0098,
+      "step": 118730
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429792324210588e-05,
+      "loss": 0.0096,
+      "step": 118740
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429744284472117e-05,
+      "loss": 0.01,
+      "step": 118750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429696244733643e-05,
+      "loss": 0.0113,
+      "step": 118760
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429648204995172e-05,
+      "loss": 0.0072,
+      "step": 118770
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429600165256702e-05,
+      "loss": 0.0109,
+      "step": 118780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.942955212551823e-05,
+      "loss": 0.0081,
+      "step": 118790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429504085779757e-05,
+      "loss": 0.0088,
+      "step": 118800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429456046041286e-05,
+      "loss": 0.0095,
+      "step": 118810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429408006302815e-05,
+      "loss": 0.0093,
+      "step": 118820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.942935996656434e-05,
+      "loss": 0.0074,
+      "step": 118830
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.942931192682587e-05,
+      "loss": 0.0101,
+      "step": 118840
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.94292638870874e-05,
+      "loss": 0.0092,
+      "step": 118850
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.942921584734893e-05,
+      "loss": 0.0091,
+      "step": 118860
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429167807610455e-05,
+      "loss": 0.0109,
+      "step": 118870
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429119767871984e-05,
+      "loss": 0.0096,
+      "step": 118880
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429071728133514e-05,
+      "loss": 0.0135,
+      "step": 118890
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9429023688395043e-05,
+      "loss": 0.0088,
+      "step": 118900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.942897564865657e-05,
+      "loss": 0.0104,
+      "step": 118910
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9428927608918098e-05,
+      "loss": 0.0124,
+      "step": 118920
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9428879569179627e-05,
+      "loss": 0.0125,
+      "step": 118930
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9428831529441153e-05,
+      "loss": 0.0125,
+      "step": 118940
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9428783489702682e-05,
+      "loss": 0.0095,
+      "step": 118950
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.942873544996421e-05,
+      "loss": 0.0113,
+      "step": 118960
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.942868741022574e-05,
+      "loss": 0.0115,
+      "step": 118970
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9428639370487267e-05,
+      "loss": 0.0135,
+      "step": 118980
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9428591330748796e-05,
+      "loss": 0.0091,
+      "step": 118990
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9428543291010325e-05,
+      "loss": 0.009,
+      "step": 119000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942849525127185e-05,
+      "loss": 0.0134,
+      "step": 119010
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942844721153338e-05,
+      "loss": 0.0092,
+      "step": 119020
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942839917179491e-05,
+      "loss": 0.0108,
+      "step": 119030
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942835113205644e-05,
+      "loss": 0.012,
+      "step": 119040
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9428303092317965e-05,
+      "loss": 0.0068,
+      "step": 119050
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9428255052579494e-05,
+      "loss": 0.0109,
+      "step": 119060
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9428207012841023e-05,
+      "loss": 0.0103,
+      "step": 119070
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9428158973102553e-05,
+      "loss": 0.0093,
+      "step": 119080
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942811093336408e-05,
+      "loss": 0.0125,
+      "step": 119090
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9428062893625608e-05,
+      "loss": 0.008,
+      "step": 119100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9428014853887137e-05,
+      "loss": 0.009,
+      "step": 119110
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427966814148663e-05,
+      "loss": 0.0089,
+      "step": 119120
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427918774410192e-05,
+      "loss": 0.0076,
+      "step": 119130
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942787073467172e-05,
+      "loss": 0.0127,
+      "step": 119140
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942782269493325e-05,
+      "loss": 0.0075,
+      "step": 119150
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427774655194777e-05,
+      "loss": 0.0086,
+      "step": 119160
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427726615456306e-05,
+      "loss": 0.0072,
+      "step": 119170
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427678575717835e-05,
+      "loss": 0.0118,
+      "step": 119180
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942763053597936e-05,
+      "loss": 0.0102,
+      "step": 119190
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942758249624089e-05,
+      "loss": 0.011,
+      "step": 119200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942753445650242e-05,
+      "loss": 0.0089,
+      "step": 119210
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942748641676395e-05,
+      "loss": 0.0095,
+      "step": 119220
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427438377025475e-05,
+      "loss": 0.0072,
+      "step": 119230
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427390337287004e-05,
+      "loss": 0.0086,
+      "step": 119240
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427342297548533e-05,
+      "loss": 0.0082,
+      "step": 119250
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427294257810063e-05,
+      "loss": 0.0112,
+      "step": 119260
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427246218071592e-05,
+      "loss": 0.0082,
+      "step": 119270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942719817833312e-05,
+      "loss": 0.0134,
+      "step": 119280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427150138594647e-05,
+      "loss": 0.0071,
+      "step": 119290
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427102098856176e-05,
+      "loss": 0.0129,
+      "step": 119300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9427054059117706e-05,
+      "loss": 0.0068,
+      "step": 119310
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942700601937923e-05,
+      "loss": 0.0091,
+      "step": 119320
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942695797964076e-05,
+      "loss": 0.0139,
+      "step": 119330
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942690993990229e-05,
+      "loss": 0.0222,
+      "step": 119340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942686190016382e-05,
+      "loss": 0.0098,
+      "step": 119350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9426813860425345e-05,
+      "loss": 0.01,
+      "step": 119360
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9426765820686874e-05,
+      "loss": 0.0131,
+      "step": 119370
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9426717780948404e-05,
+      "loss": 0.0141,
+      "step": 119380
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9426669741209933e-05,
+      "loss": 0.0084,
+      "step": 119390
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942662170147146e-05,
+      "loss": 0.0106,
+      "step": 119400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9426573661732988e-05,
+      "loss": 0.0137,
+      "step": 119410
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9426525621994517e-05,
+      "loss": 0.011,
+      "step": 119420
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9426477582256043e-05,
+      "loss": 0.0107,
+      "step": 119430
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9426429542517573e-05,
+      "loss": 0.0088,
+      "step": 119440
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9426381502779102e-05,
+      "loss": 0.0087,
+      "step": 119450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942633346304063e-05,
+      "loss": 0.0096,
+      "step": 119460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9426285423302157e-05,
+      "loss": 0.0099,
+      "step": 119470
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9426237383563686e-05,
+      "loss": 0.0105,
+      "step": 119480
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9426189343825216e-05,
+      "loss": 0.009,
+      "step": 119490
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942614130408674e-05,
+      "loss": 0.0114,
+      "step": 119500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942609326434827e-05,
+      "loss": 0.0121,
+      "step": 119510
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.94260452246098e-05,
+      "loss": 0.0092,
+      "step": 119520
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942599718487133e-05,
+      "loss": 0.0117,
+      "step": 119530
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9425949145132855e-05,
+      "loss": 0.0091,
+      "step": 119540
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9425901105394384e-05,
+      "loss": 0.0114,
+      "step": 119550
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9425853065655914e-05,
+      "loss": 0.0125,
+      "step": 119560
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9425805025917443e-05,
+      "loss": 0.0085,
+      "step": 119570
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942575698617897e-05,
+      "loss": 0.0111,
+      "step": 119580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9425708946440498e-05,
+      "loss": 0.011,
+      "step": 119590
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9425660906702027e-05,
+      "loss": 0.0113,
+      "step": 119600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9425612866963553e-05,
+      "loss": 0.0096,
+      "step": 119610
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9425564827225082e-05,
+      "loss": 0.009,
+      "step": 119620
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9425516787486612e-05,
+      "loss": 0.0138,
+      "step": 119630
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942546874774814e-05,
+      "loss": 0.0108,
+      "step": 119640
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9425420708009667e-05,
+      "loss": 0.0111,
+      "step": 119650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9425372668271196e-05,
+      "loss": 0.0111,
+      "step": 119660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9425324628532725e-05,
+      "loss": 0.0091,
+      "step": 119670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942527658879425e-05,
+      "loss": 0.0077,
+      "step": 119680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.942522854905578e-05,
+      "loss": 0.0087,
+      "step": 119690
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942518050931731e-05,
+      "loss": 0.01,
+      "step": 119700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942513246957884e-05,
+      "loss": 0.0121,
+      "step": 119710
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9425084429840365e-05,
+      "loss": 0.0135,
+      "step": 119720
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9425036390101894e-05,
+      "loss": 0.0177,
+      "step": 119730
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424988350363424e-05,
+      "loss": 0.0128,
+      "step": 119740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424940310624953e-05,
+      "loss": 0.0114,
+      "step": 119750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942489227088648e-05,
+      "loss": 0.0105,
+      "step": 119760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424844231148008e-05,
+      "loss": 0.0105,
+      "step": 119770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424796191409537e-05,
+      "loss": 0.0087,
+      "step": 119780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424748151671063e-05,
+      "loss": 0.0062,
+      "step": 119790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424700111932592e-05,
+      "loss": 0.0118,
+      "step": 119800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942465207219412e-05,
+      "loss": 0.0084,
+      "step": 119810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942460403245565e-05,
+      "loss": 0.0111,
+      "step": 119820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424555992717177e-05,
+      "loss": 0.0084,
+      "step": 119830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424507952978706e-05,
+      "loss": 0.0081,
+      "step": 119840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424459913240235e-05,
+      "loss": 0.0131,
+      "step": 119850
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942441187350176e-05,
+      "loss": 0.0127,
+      "step": 119860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942436383376329e-05,
+      "loss": 0.0084,
+      "step": 119870
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942431579402482e-05,
+      "loss": 0.0106,
+      "step": 119880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942426775428635e-05,
+      "loss": 0.0112,
+      "step": 119890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424219714547875e-05,
+      "loss": 0.0129,
+      "step": 119900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424171674809404e-05,
+      "loss": 0.0092,
+      "step": 119910
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424123635070933e-05,
+      "loss": 0.0115,
+      "step": 119920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9424075595332463e-05,
+      "loss": 0.0127,
+      "step": 119930
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942402755559399e-05,
+      "loss": 0.0092,
+      "step": 119940
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423979515855518e-05,
+      "loss": 0.0098,
+      "step": 119950
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423931476117047e-05,
+      "loss": 0.0106,
+      "step": 119960
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423883436378573e-05,
+      "loss": 0.01,
+      "step": 119970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423835396640102e-05,
+      "loss": 0.0104,
+      "step": 119980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942378735690163e-05,
+      "loss": 0.0126,
+      "step": 119990
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942373931716316e-05,
+      "loss": 0.0081,
+      "step": 120000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423691277424687e-05,
+      "loss": 0.0127,
+      "step": 120010
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423643237686216e-05,
+      "loss": 0.0102,
+      "step": 120020
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423595197947745e-05,
+      "loss": 0.0151,
+      "step": 120030
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942354715820927e-05,
+      "loss": 0.016,
+      "step": 120040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.94234991184708e-05,
+      "loss": 0.0084,
+      "step": 120050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942345107873233e-05,
+      "loss": 0.0162,
+      "step": 120060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942340303899386e-05,
+      "loss": 0.0093,
+      "step": 120070
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423354999255385e-05,
+      "loss": 0.0075,
+      "step": 120080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423306959516914e-05,
+      "loss": 0.0106,
+      "step": 120090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423258919778443e-05,
+      "loss": 0.0111,
+      "step": 120100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423210880039973e-05,
+      "loss": 0.0136,
+      "step": 120110
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.94231628403015e-05,
+      "loss": 0.0126,
+      "step": 120120
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423114800563028e-05,
+      "loss": 0.0114,
+      "step": 120130
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423066760824557e-05,
+      "loss": 0.0101,
+      "step": 120140
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9423018721086083e-05,
+      "loss": 0.0089,
+      "step": 120150
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9422970681347612e-05,
+      "loss": 0.0095,
+      "step": 120160
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942292264160914e-05,
+      "loss": 0.0105,
+      "step": 120170
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942287460187067e-05,
+      "loss": 0.0107,
+      "step": 120180
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9422826562132197e-05,
+      "loss": 0.0114,
+      "step": 120190
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9422778522393726e-05,
+      "loss": 0.0113,
+      "step": 120200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9422730482655255e-05,
+      "loss": 0.0113,
+      "step": 120210
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942268244291678e-05,
+      "loss": 0.0086,
+      "step": 120220
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942263440317831e-05,
+      "loss": 0.011,
+      "step": 120230
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942258636343984e-05,
+      "loss": 0.0094,
+      "step": 120240
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942253832370137e-05,
+      "loss": 0.012,
+      "step": 120250
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9422490283962895e-05,
+      "loss": 0.0095,
+      "step": 120260
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9422442244224424e-05,
+      "loss": 0.0093,
+      "step": 120270
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9422394204485953e-05,
+      "loss": 0.0116,
+      "step": 120280
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9422346164747483e-05,
+      "loss": 0.0098,
+      "step": 120290
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942229812500901e-05,
+      "loss": 0.0092,
+      "step": 120300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9422250085270538e-05,
+      "loss": 0.0072,
+      "step": 120310
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9422202045532067e-05,
+      "loss": 0.0146,
+      "step": 120320
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9422154005793593e-05,
+      "loss": 0.0105,
+      "step": 120330
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9422105966055122e-05,
+      "loss": 0.01,
+      "step": 120340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942205792631665e-05,
+      "loss": 0.012,
+      "step": 120350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.942200988657818e-05,
+      "loss": 0.0084,
+      "step": 120360
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9421961846839707e-05,
+      "loss": 0.01,
+      "step": 120370
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9421913807101236e-05,
+      "loss": 0.0134,
+      "step": 120380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9421865767362765e-05,
+      "loss": 0.008,
+      "step": 120390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942181772762429e-05,
+      "loss": 0.0111,
+      "step": 120400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942176968788582e-05,
+      "loss": 0.0087,
+      "step": 120410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942172164814735e-05,
+      "loss": 0.0144,
+      "step": 120420
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942167360840888e-05,
+      "loss": 0.0111,
+      "step": 120430
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9421625568670405e-05,
+      "loss": 0.0109,
+      "step": 120440
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9421577528931934e-05,
+      "loss": 0.0133,
+      "step": 120450
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9421529489193463e-05,
+      "loss": 0.0117,
+      "step": 120460
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9421481449454992e-05,
+      "loss": 0.011,
+      "step": 120470
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942143340971652e-05,
+      "loss": 0.0078,
+      "step": 120480
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9421385369978048e-05,
+      "loss": 0.0099,
+      "step": 120490
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9421337330239577e-05,
+      "loss": 0.0101,
+      "step": 120500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9421289290501103e-05,
+      "loss": 0.0094,
+      "step": 120510
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9421241250762632e-05,
+      "loss": 0.0119,
+      "step": 120520
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942119321102416e-05,
+      "loss": 0.0092,
+      "step": 120530
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942114517128569e-05,
+      "loss": 0.0112,
+      "step": 120540
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9421097131547216e-05,
+      "loss": 0.009,
+      "step": 120550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9421049091808746e-05,
+      "loss": 0.0061,
+      "step": 120560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9421001052070275e-05,
+      "loss": 0.0066,
+      "step": 120570
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.94209530123318e-05,
+      "loss": 0.0131,
+      "step": 120580
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942090497259333e-05,
+      "loss": 0.0098,
+      "step": 120590
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942085693285486e-05,
+      "loss": 0.0107,
+      "step": 120600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942080889311639e-05,
+      "loss": 0.0097,
+      "step": 120610
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9420760853377915e-05,
+      "loss": 0.007,
+      "step": 120620
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9420712813639444e-05,
+      "loss": 0.0125,
+      "step": 120630
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9420664773900973e-05,
+      "loss": 0.0084,
+      "step": 120640
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9420616734162502e-05,
+      "loss": 0.0105,
+      "step": 120650
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9420568694424028e-05,
+      "loss": 0.0113,
+      "step": 120660
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9420520654685558e-05,
+      "loss": 0.0081,
+      "step": 120670
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9420472614947087e-05,
+      "loss": 0.012,
+      "step": 120680
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9420424575208613e-05,
+      "loss": 0.0079,
+      "step": 120690
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9420376535470142e-05,
+      "loss": 0.0072,
+      "step": 120700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942032849573167e-05,
+      "loss": 0.0077,
+      "step": 120710
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.94202804559932e-05,
+      "loss": 0.01,
+      "step": 120720
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9420232416254726e-05,
+      "loss": 0.0165,
+      "step": 120730
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9420184376516256e-05,
+      "loss": 0.0115,
+      "step": 120740
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9420136336777785e-05,
+      "loss": 0.0134,
+      "step": 120750
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942008829703931e-05,
+      "loss": 0.0095,
+      "step": 120760
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.942004025730084e-05,
+      "loss": 0.0111,
+      "step": 120770
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.941999221756237e-05,
+      "loss": 0.0119,
+      "step": 120780
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.94199441778239e-05,
+      "loss": 0.0092,
+      "step": 120790
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9419896138085424e-05,
+      "loss": 0.0102,
+      "step": 120800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9419848098346954e-05,
+      "loss": 0.0099,
+      "step": 120810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9419800058608483e-05,
+      "loss": 0.0114,
+      "step": 120820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.941975201887001e-05,
+      "loss": 0.0101,
+      "step": 120830
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9419703979131538e-05,
+      "loss": 0.0078,
+      "step": 120840
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9419655939393067e-05,
+      "loss": 0.0084,
+      "step": 120850
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9419607899654597e-05,
+      "loss": 0.0089,
+      "step": 120860
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9419559859916123e-05,
+      "loss": 0.0135,
+      "step": 120870
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9419511820177652e-05,
+      "loss": 0.0079,
+      "step": 120880
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.941946378043918e-05,
+      "loss": 0.0065,
+      "step": 120890
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.941941574070071e-05,
+      "loss": 0.0103,
+      "step": 120900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9419367700962236e-05,
+      "loss": 0.0118,
+      "step": 120910
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9419319661223766e-05,
+      "loss": 0.012,
+      "step": 120920
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9419271621485295e-05,
+      "loss": 0.0075,
+      "step": 120930
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.941922358174682e-05,
+      "loss": 0.0099,
+      "step": 120940
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.941917554200835e-05,
+      "loss": 0.0127,
+      "step": 120950
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.941912750226988e-05,
+      "loss": 0.0079,
+      "step": 120960
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.941907946253141e-05,
+      "loss": 0.009,
+      "step": 120970
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9419031422792934e-05,
+      "loss": 0.0103,
+      "step": 120980
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9418983383054464e-05,
+      "loss": 0.0109,
+      "step": 120990
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9418935343315993e-05,
+      "loss": 0.0136,
+      "step": 121000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.941888730357752e-05,
+      "loss": 0.0111,
+      "step": 121010
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9418839263839048e-05,
+      "loss": 0.0109,
+      "step": 121020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9418791224100577e-05,
+      "loss": 0.0102,
+      "step": 121030
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9418743184362107e-05,
+      "loss": 0.0089,
+      "step": 121040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9418695144623633e-05,
+      "loss": 0.0089,
+      "step": 121050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9418647104885162e-05,
+      "loss": 0.0132,
+      "step": 121060
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.941859906514669e-05,
+      "loss": 0.0108,
+      "step": 121070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.941855102540822e-05,
+      "loss": 0.0105,
+      "step": 121080
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9418502985669746e-05,
+      "loss": 0.0101,
+      "step": 121090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9418454945931275e-05,
+      "loss": 0.0099,
+      "step": 121100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9418406906192805e-05,
+      "loss": 0.0072,
+      "step": 121110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941835886645433e-05,
+      "loss": 0.0092,
+      "step": 121120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941831082671586e-05,
+      "loss": 0.0077,
+      "step": 121130
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941826278697739e-05,
+      "loss": 0.0081,
+      "step": 121140
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941821474723892e-05,
+      "loss": 0.0137,
+      "step": 121150
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9418166707500444e-05,
+      "loss": 0.0072,
+      "step": 121160
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9418118667761974e-05,
+      "loss": 0.0089,
+      "step": 121170
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9418070628023503e-05,
+      "loss": 0.0111,
+      "step": 121180
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941802258828503e-05,
+      "loss": 0.0133,
+      "step": 121190
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417974548546558e-05,
+      "loss": 0.0091,
+      "step": 121200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417926508808087e-05,
+      "loss": 0.0109,
+      "step": 121210
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417878469069617e-05,
+      "loss": 0.0091,
+      "step": 121220
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417830429331142e-05,
+      "loss": 0.0138,
+      "step": 121230
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417782389592672e-05,
+      "loss": 0.0085,
+      "step": 121240
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.94177343498542e-05,
+      "loss": 0.0119,
+      "step": 121250
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941768631011573e-05,
+      "loss": 0.0078,
+      "step": 121260
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417638270377256e-05,
+      "loss": 0.0084,
+      "step": 121270
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417590230638785e-05,
+      "loss": 0.0096,
+      "step": 121280
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417542190900315e-05,
+      "loss": 0.0123,
+      "step": 121290
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941749415116184e-05,
+      "loss": 0.0162,
+      "step": 121300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941744611142337e-05,
+      "loss": 0.0124,
+      "step": 121310
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.94173980716849e-05,
+      "loss": 0.0115,
+      "step": 121320
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941735003194643e-05,
+      "loss": 0.0107,
+      "step": 121330
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417301992207954e-05,
+      "loss": 0.0091,
+      "step": 121340
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417253952469484e-05,
+      "loss": 0.0108,
+      "step": 121350
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417205912731013e-05,
+      "loss": 0.0136,
+      "step": 121360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941715787299254e-05,
+      "loss": 0.0102,
+      "step": 121370
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417109833254068e-05,
+      "loss": 0.0104,
+      "step": 121380
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417061793515597e-05,
+      "loss": 0.0095,
+      "step": 121390
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9417013753777126e-05,
+      "loss": 0.0123,
+      "step": 121400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416965714038652e-05,
+      "loss": 0.0086,
+      "step": 121410
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941692247827403e-05,
+      "loss": 0.0137,
+      "step": 121420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416874438535556e-05,
+      "loss": 0.0134,
+      "step": 121430
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416826398797086e-05,
+      "loss": 0.0078,
+      "step": 121440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416778359058615e-05,
+      "loss": 0.0121,
+      "step": 121450
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416730319320144e-05,
+      "loss": 0.0113,
+      "step": 121460
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941668227958167e-05,
+      "loss": 0.0112,
+      "step": 121470
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.94166342398432e-05,
+      "loss": 0.0099,
+      "step": 121480
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941658620010473e-05,
+      "loss": 0.012,
+      "step": 121490
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416538160366258e-05,
+      "loss": 0.0083,
+      "step": 121500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416490120627784e-05,
+      "loss": 0.0112,
+      "step": 121510
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416442080889313e-05,
+      "loss": 0.0135,
+      "step": 121520
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416394041150842e-05,
+      "loss": 0.0112,
+      "step": 121530
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416346001412368e-05,
+      "loss": 0.0132,
+      "step": 121540
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416297961673897e-05,
+      "loss": 0.0081,
+      "step": 121550
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416249921935427e-05,
+      "loss": 0.0101,
+      "step": 121560
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416201882196956e-05,
+      "loss": 0.0129,
+      "step": 121570
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416153842458482e-05,
+      "loss": 0.0094,
+      "step": 121580
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941610580272001e-05,
+      "loss": 0.0095,
+      "step": 121590
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941605776298154e-05,
+      "loss": 0.0119,
+      "step": 121600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9416009723243066e-05,
+      "loss": 0.0098,
+      "step": 121610
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9415961683504595e-05,
+      "loss": 0.0071,
+      "step": 121620
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9415913643766125e-05,
+      "loss": 0.0099,
+      "step": 121630
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9415865604027654e-05,
+      "loss": 0.0073,
+      "step": 121640
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941581756428918e-05,
+      "loss": 0.0059,
+      "step": 121650
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941576952455071e-05,
+      "loss": 0.0172,
+      "step": 121660
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941572148481224e-05,
+      "loss": 0.0108,
+      "step": 121670
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9415673445073768e-05,
+      "loss": 0.0086,
+      "step": 121680
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9415625405335294e-05,
+      "loss": 0.0087,
+      "step": 121690
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9415577365596823e-05,
+      "loss": 0.0087,
+      "step": 121700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9415529325858352e-05,
+      "loss": 0.0095,
+      "step": 121710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9415481286119878e-05,
+      "loss": 0.0137,
+      "step": 121720
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9415433246381407e-05,
+      "loss": 0.0094,
+      "step": 121730
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9415385206642937e-05,
+      "loss": 0.0096,
+      "step": 121740
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9415337166904466e-05,
+      "loss": 0.015,
+      "step": 121750
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941528912716599e-05,
+      "loss": 0.0092,
+      "step": 121760
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.941524108742752e-05,
+      "loss": 0.0082,
+      "step": 121770
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941519304768905e-05,
+      "loss": 0.012,
+      "step": 121780
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9415145007950576e-05,
+      "loss": 0.0115,
+      "step": 121790
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9415096968212105e-05,
+      "loss": 0.01,
+      "step": 121800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9415048928473635e-05,
+      "loss": 0.0156,
+      "step": 121810
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9415000888735164e-05,
+      "loss": 0.011,
+      "step": 121820
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941495284899669e-05,
+      "loss": 0.009,
+      "step": 121830
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941490480925822e-05,
+      "loss": 0.0111,
+      "step": 121840
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941485676951975e-05,
+      "loss": 0.0115,
+      "step": 121850
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9414808729781278e-05,
+      "loss": 0.0109,
+      "step": 121860
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9414760690042803e-05,
+      "loss": 0.0107,
+      "step": 121870
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9414712650304333e-05,
+      "loss": 0.0091,
+      "step": 121880
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9414664610565862e-05,
+      "loss": 0.011,
+      "step": 121890
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9414616570827388e-05,
+      "loss": 0.0072,
+      "step": 121900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9414568531088917e-05,
+      "loss": 0.0098,
+      "step": 121910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9414520491350446e-05,
+      "loss": 0.0075,
+      "step": 121920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9414472451611976e-05,
+      "loss": 0.0083,
+      "step": 121930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.94144244118735e-05,
+      "loss": 0.0098,
+      "step": 121940
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941437637213503e-05,
+      "loss": 0.0088,
+      "step": 121950
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941432833239656e-05,
+      "loss": 0.0114,
+      "step": 121960
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9414280292658086e-05,
+      "loss": 0.0081,
+      "step": 121970
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9414232252919615e-05,
+      "loss": 0.0117,
+      "step": 121980
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9414184213181145e-05,
+      "loss": 0.0094,
+      "step": 121990
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9414136173442674e-05,
+      "loss": 0.0132,
+      "step": 122000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.94140881337042e-05,
+      "loss": 0.0084,
+      "step": 122010
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941404009396573e-05,
+      "loss": 0.0105,
+      "step": 122020
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413992054227258e-05,
+      "loss": 0.0118,
+      "step": 122030
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413944014488788e-05,
+      "loss": 0.0118,
+      "step": 122040
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413895974750313e-05,
+      "loss": 0.0152,
+      "step": 122050
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413847935011843e-05,
+      "loss": 0.0096,
+      "step": 122060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413799895273372e-05,
+      "loss": 0.0119,
+      "step": 122070
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413751855534898e-05,
+      "loss": 0.0121,
+      "step": 122080
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413703815796427e-05,
+      "loss": 0.0101,
+      "step": 122090
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413655776057956e-05,
+      "loss": 0.0118,
+      "step": 122100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413607736319486e-05,
+      "loss": 0.0106,
+      "step": 122110
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941355969658101e-05,
+      "loss": 0.0067,
+      "step": 122120
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941351165684254e-05,
+      "loss": 0.0084,
+      "step": 122130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941346361710407e-05,
+      "loss": 0.0073,
+      "step": 122140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413415577365596e-05,
+      "loss": 0.0119,
+      "step": 122150
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413367537627125e-05,
+      "loss": 0.0091,
+      "step": 122160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413319497888654e-05,
+      "loss": 0.0071,
+      "step": 122170
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413271458150184e-05,
+      "loss": 0.0142,
+      "step": 122180
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941322341841171e-05,
+      "loss": 0.0129,
+      "step": 122190
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941317537867324e-05,
+      "loss": 0.011,
+      "step": 122200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413127338934768e-05,
+      "loss": 0.0082,
+      "step": 122210
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413079299196297e-05,
+      "loss": 0.008,
+      "step": 122220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9413031259457823e-05,
+      "loss": 0.0106,
+      "step": 122230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412983219719353e-05,
+      "loss": 0.0088,
+      "step": 122240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412935179980882e-05,
+      "loss": 0.0106,
+      "step": 122250
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412887140242408e-05,
+      "loss": 0.0085,
+      "step": 122260
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412839100503937e-05,
+      "loss": 0.0091,
+      "step": 122270
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412791060765466e-05,
+      "loss": 0.0094,
+      "step": 122280
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412743021026996e-05,
+      "loss": 0.0133,
+      "step": 122290
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941269498128852e-05,
+      "loss": 0.0081,
+      "step": 122300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941264694155005e-05,
+      "loss": 0.0098,
+      "step": 122310
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941259890181158e-05,
+      "loss": 0.0062,
+      "step": 122320
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941255086207311e-05,
+      "loss": 0.0102,
+      "step": 122330
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941250282233464e-05,
+      "loss": 0.015,
+      "step": 122340
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412454782596168e-05,
+      "loss": 0.0064,
+      "step": 122350
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412406742857694e-05,
+      "loss": 0.0089,
+      "step": 122360
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412358703119223e-05,
+      "loss": 0.0089,
+      "step": 122370
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412310663380752e-05,
+      "loss": 0.0093,
+      "step": 122380
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412262623642278e-05,
+      "loss": 0.01,
+      "step": 122390
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412214583903807e-05,
+      "loss": 0.0135,
+      "step": 122400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412166544165337e-05,
+      "loss": 0.0081,
+      "step": 122410
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412118504426866e-05,
+      "loss": 0.0111,
+      "step": 122420
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9412070464688392e-05,
+      "loss": 0.0102,
+      "step": 122430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941202242494992e-05,
+      "loss": 0.0055,
+      "step": 122440
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.941197438521145e-05,
+      "loss": 0.0084,
+      "step": 122450
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9411926345472976e-05,
+      "loss": 0.008,
+      "step": 122460
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411878305734505e-05,
+      "loss": 0.0122,
+      "step": 122470
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411830265996035e-05,
+      "loss": 0.0145,
+      "step": 122480
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411782226257564e-05,
+      "loss": 0.0091,
+      "step": 122490
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.941173418651909e-05,
+      "loss": 0.0117,
+      "step": 122500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.941168614678062e-05,
+      "loss": 0.013,
+      "step": 122510
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.941163810704215e-05,
+      "loss": 0.0097,
+      "step": 122520
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411590067303678e-05,
+      "loss": 0.01,
+      "step": 122530
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411542027565204e-05,
+      "loss": 0.0093,
+      "step": 122540
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411493987826733e-05,
+      "loss": 0.0086,
+      "step": 122550
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411445948088262e-05,
+      "loss": 0.0085,
+      "step": 122560
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411397908349788e-05,
+      "loss": 0.0131,
+      "step": 122570
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411349868611317e-05,
+      "loss": 0.0094,
+      "step": 122580
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411301828872847e-05,
+      "loss": 0.0095,
+      "step": 122590
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411253789134376e-05,
+      "loss": 0.0114,
+      "step": 122600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.94112057493959e-05,
+      "loss": 0.0074,
+      "step": 122610
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.941115770965743e-05,
+      "loss": 0.0089,
+      "step": 122620
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.941110966991896e-05,
+      "loss": 0.0094,
+      "step": 122630
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411061630180486e-05,
+      "loss": 0.0134,
+      "step": 122640
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9411013590442015e-05,
+      "loss": 0.0059,
+      "step": 122650
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410965550703545e-05,
+      "loss": 0.0099,
+      "step": 122660
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410917510965074e-05,
+      "loss": 0.0081,
+      "step": 122670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.94108694712266e-05,
+      "loss": 0.0127,
+      "step": 122680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.941082143148813e-05,
+      "loss": 0.0108,
+      "step": 122690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.941077339174966e-05,
+      "loss": 0.0075,
+      "step": 122700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410725352011188e-05,
+      "loss": 0.0088,
+      "step": 122710
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410677312272713e-05,
+      "loss": 0.0068,
+      "step": 122720
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410629272534243e-05,
+      "loss": 0.0114,
+      "step": 122730
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410581232795772e-05,
+      "loss": 0.0077,
+      "step": 122740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410533193057298e-05,
+      "loss": 0.0108,
+      "step": 122750
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410485153318827e-05,
+      "loss": 0.007,
+      "step": 122760
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410437113580356e-05,
+      "loss": 0.0111,
+      "step": 122770
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410389073841886e-05,
+      "loss": 0.0076,
+      "step": 122780
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.941034103410341e-05,
+      "loss": 0.0112,
+      "step": 122790
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.941029299436494e-05,
+      "loss": 0.0068,
+      "step": 122800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.941024495462647e-05,
+      "loss": 0.0104,
+      "step": 122810
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410196914887996e-05,
+      "loss": 0.0106,
+      "step": 122820
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410148875149525e-05,
+      "loss": 0.0117,
+      "step": 122830
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410100835411055e-05,
+      "loss": 0.0099,
+      "step": 122840
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9410052795672584e-05,
+      "loss": 0.008,
+      "step": 122850
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.941000475593411e-05,
+      "loss": 0.0098,
+      "step": 122860
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.940995671619564e-05,
+      "loss": 0.0077,
+      "step": 122870
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409908676457168e-05,
+      "loss": 0.0116,
+      "step": 122880
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409860636718698e-05,
+      "loss": 0.0115,
+      "step": 122890
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409812596980223e-05,
+      "loss": 0.0138,
+      "step": 122900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409764557241753e-05,
+      "loss": 0.0104,
+      "step": 122910
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409716517503282e-05,
+      "loss": 0.0116,
+      "step": 122920
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409668477764808e-05,
+      "loss": 0.0104,
+      "step": 122930
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409620438026337e-05,
+      "loss": 0.0118,
+      "step": 122940
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409572398287866e-05,
+      "loss": 0.0129,
+      "step": 122950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409524358549396e-05,
+      "loss": 0.0095,
+      "step": 122960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.940947631881092e-05,
+      "loss": 0.0174,
+      "step": 122970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.940942827907245e-05,
+      "loss": 0.0094,
+      "step": 122980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.940938023933398e-05,
+      "loss": 0.0112,
+      "step": 122990
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409332199595506e-05,
+      "loss": 0.0107,
+      "step": 123000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409284159857035e-05,
+      "loss": 0.0122,
+      "step": 123010
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409236120118564e-05,
+      "loss": 0.0118,
+      "step": 123020
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409188080380094e-05,
+      "loss": 0.0096,
+      "step": 123030
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.940914004064162e-05,
+      "loss": 0.0102,
+      "step": 123040
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.940909200090315e-05,
+      "loss": 0.0098,
+      "step": 123050
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9409043961164678e-05,
+      "loss": 0.0067,
+      "step": 123060
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9408995921426207e-05,
+      "loss": 0.0103,
+      "step": 123070
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9408947881687733e-05,
+      "loss": 0.0089,
+      "step": 123080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9408899841949263e-05,
+      "loss": 0.0089,
+      "step": 123090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9408851802210792e-05,
+      "loss": 0.0087,
+      "step": 123100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9408803762472318e-05,
+      "loss": 0.0091,
+      "step": 123110
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9408755722733847e-05,
+      "loss": 0.0127,
+      "step": 123120
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9408707682995376e-05,
+      "loss": 0.0132,
+      "step": 123130
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9408659643256906e-05,
+      "loss": 0.0099,
+      "step": 123140
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.940861160351843e-05,
+      "loss": 0.0098,
+      "step": 123150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.940856356377996e-05,
+      "loss": 0.0106,
+      "step": 123160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940851552404149e-05,
+      "loss": 0.0103,
+      "step": 123170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9408467484303016e-05,
+      "loss": 0.0064,
+      "step": 123180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9408419444564545e-05,
+      "loss": 0.0092,
+      "step": 123190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9408371404826074e-05,
+      "loss": 0.0077,
+      "step": 123200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9408323365087604e-05,
+      "loss": 0.0106,
+      "step": 123210
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940827532534913e-05,
+      "loss": 0.0094,
+      "step": 123220
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940822728561066e-05,
+      "loss": 0.0124,
+      "step": 123230
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9408179245872188e-05,
+      "loss": 0.0098,
+      "step": 123240
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9408131206133717e-05,
+      "loss": 0.0067,
+      "step": 123250
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9408083166395243e-05,
+      "loss": 0.0104,
+      "step": 123260
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9408035126656772e-05,
+      "loss": 0.0087,
+      "step": 123270
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407987086918302e-05,
+      "loss": 0.0091,
+      "step": 123280
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407939047179828e-05,
+      "loss": 0.0087,
+      "step": 123290
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407891007441357e-05,
+      "loss": 0.0092,
+      "step": 123300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407842967702886e-05,
+      "loss": 0.008,
+      "step": 123310
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407794927964415e-05,
+      "loss": 0.0134,
+      "step": 123320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940774688822594e-05,
+      "loss": 0.0141,
+      "step": 123330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940769884848747e-05,
+      "loss": 0.0123,
+      "step": 123340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407650808749e-05,
+      "loss": 0.0113,
+      "step": 123350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407602769010526e-05,
+      "loss": 0.0111,
+      "step": 123360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407554729272055e-05,
+      "loss": 0.0125,
+      "step": 123370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407506689533584e-05,
+      "loss": 0.0114,
+      "step": 123380
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407458649795114e-05,
+      "loss": 0.0076,
+      "step": 123390
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940741061005664e-05,
+      "loss": 0.0104,
+      "step": 123400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940736257031817e-05,
+      "loss": 0.0098,
+      "step": 123410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407314530579698e-05,
+      "loss": 0.0102,
+      "step": 123420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407266490841227e-05,
+      "loss": 0.0153,
+      "step": 123430
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407218451102753e-05,
+      "loss": 0.008,
+      "step": 123440
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407170411364282e-05,
+      "loss": 0.007,
+      "step": 123450
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940712237162581e-05,
+      "loss": 0.0078,
+      "step": 123460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407074331887338e-05,
+      "loss": 0.0084,
+      "step": 123470
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9407026292148867e-05,
+      "loss": 0.0106,
+      "step": 123480
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406978252410396e-05,
+      "loss": 0.0129,
+      "step": 123490
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406930212671925e-05,
+      "loss": 0.0126,
+      "step": 123500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940688217293345e-05,
+      "loss": 0.0103,
+      "step": 123510
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940683413319498e-05,
+      "loss": 0.0122,
+      "step": 123520
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940678609345651e-05,
+      "loss": 0.0098,
+      "step": 123530
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406738053718036e-05,
+      "loss": 0.0079,
+      "step": 123540
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406690013979565e-05,
+      "loss": 0.0085,
+      "step": 123550
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406641974241094e-05,
+      "loss": 0.0118,
+      "step": 123560
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406593934502623e-05,
+      "loss": 0.0101,
+      "step": 123570
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940654589476415e-05,
+      "loss": 0.0108,
+      "step": 123580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940649785502568e-05,
+      "loss": 0.0095,
+      "step": 123590
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406449815287208e-05,
+      "loss": 0.0096,
+      "step": 123600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406401775548737e-05,
+      "loss": 0.0092,
+      "step": 123610
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406353735810263e-05,
+      "loss": 0.0103,
+      "step": 123620
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406305696071792e-05,
+      "loss": 0.007,
+      "step": 123630
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940625765633332e-05,
+      "loss": 0.0137,
+      "step": 123640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406209616594847e-05,
+      "loss": 0.0113,
+      "step": 123650
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406161576856377e-05,
+      "loss": 0.0124,
+      "step": 123660
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406113537117906e-05,
+      "loss": 0.0076,
+      "step": 123670
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9406065497379435e-05,
+      "loss": 0.0083,
+      "step": 123680
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940601745764096e-05,
+      "loss": 0.0121,
+      "step": 123690
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940596941790249e-05,
+      "loss": 0.009,
+      "step": 123700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940592137816402e-05,
+      "loss": 0.0066,
+      "step": 123710
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9405873338425546e-05,
+      "loss": 0.0124,
+      "step": 123720
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9405825298687075e-05,
+      "loss": 0.0116,
+      "step": 123730
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9405777258948604e-05,
+      "loss": 0.01,
+      "step": 123740
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9405729219210133e-05,
+      "loss": 0.0093,
+      "step": 123750
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940568117947166e-05,
+      "loss": 0.0101,
+      "step": 123760
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940563313973319e-05,
+      "loss": 0.011,
+      "step": 123770
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9405585099994718e-05,
+      "loss": 0.0099,
+      "step": 123780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9405537060256247e-05,
+      "loss": 0.0088,
+      "step": 123790
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9405489020517773e-05,
+      "loss": 0.0089,
+      "step": 123800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9405440980779302e-05,
+      "loss": 0.008,
+      "step": 123810
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.940539294104083e-05,
+      "loss": 0.0095,
+      "step": 123820
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9405344901302357e-05,
+      "loss": 0.0103,
+      "step": 123830
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9405296861563887e-05,
+      "loss": 0.0134,
+      "step": 123840
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9405248821825416e-05,
+      "loss": 0.0079,
+      "step": 123850
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9405200782086945e-05,
+      "loss": 0.012,
+      "step": 123860
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940515274234847e-05,
+      "loss": 0.0124,
+      "step": 123870
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940510470261e-05,
+      "loss": 0.0072,
+      "step": 123880
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940505666287153e-05,
+      "loss": 0.0084,
+      "step": 123890
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9405008623133056e-05,
+      "loss": 0.0103,
+      "step": 123900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404960583394585e-05,
+      "loss": 0.0134,
+      "step": 123910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404912543656114e-05,
+      "loss": 0.0109,
+      "step": 123920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404864503917643e-05,
+      "loss": 0.0106,
+      "step": 123930
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940481646417917e-05,
+      "loss": 0.0123,
+      "step": 123940
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.94047684244407e-05,
+      "loss": 0.0086,
+      "step": 123950
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404720384702228e-05,
+      "loss": 0.0149,
+      "step": 123960
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404672344963754e-05,
+      "loss": 0.0119,
+      "step": 123970
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404624305225283e-05,
+      "loss": 0.0105,
+      "step": 123980
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404576265486812e-05,
+      "loss": 0.0102,
+      "step": 123990
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940452822574834e-05,
+      "loss": 0.008,
+      "step": 124000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404480186009867e-05,
+      "loss": 0.009,
+      "step": 124010
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404432146271397e-05,
+      "loss": 0.0111,
+      "step": 124020
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404384106532926e-05,
+      "loss": 0.0127,
+      "step": 124030
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404336066794455e-05,
+      "loss": 0.0105,
+      "step": 124040
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940428802705598e-05,
+      "loss": 0.0081,
+      "step": 124050
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940423998731751e-05,
+      "loss": 0.0108,
+      "step": 124060
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940419194757904e-05,
+      "loss": 0.0099,
+      "step": 124070
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404143907840565e-05,
+      "loss": 0.0097,
+      "step": 124080
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404095868102095e-05,
+      "loss": 0.0102,
+      "step": 124090
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9404047828363624e-05,
+      "loss": 0.0101,
+      "step": 124100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403999788625153e-05,
+      "loss": 0.01,
+      "step": 124110
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940395174888668e-05,
+      "loss": 0.0151,
+      "step": 124120
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940390370914821e-05,
+      "loss": 0.0093,
+      "step": 124130
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403855669409738e-05,
+      "loss": 0.0142,
+      "step": 124140
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403807629671264e-05,
+      "loss": 0.0087,
+      "step": 124150
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403759589932793e-05,
+      "loss": 0.0072,
+      "step": 124160
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403711550194322e-05,
+      "loss": 0.0103,
+      "step": 124170
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940366351045585e-05,
+      "loss": 0.0108,
+      "step": 124180
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403615470717377e-05,
+      "loss": 0.0114,
+      "step": 124190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403567430978906e-05,
+      "loss": 0.011,
+      "step": 124200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403519391240436e-05,
+      "loss": 0.0083,
+      "step": 124210
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403471351501965e-05,
+      "loss": 0.0113,
+      "step": 124220
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940342331176349e-05,
+      "loss": 0.0085,
+      "step": 124230
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940337527202502e-05,
+      "loss": 0.0109,
+      "step": 124240
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940332723228655e-05,
+      "loss": 0.0097,
+      "step": 124250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403279192548075e-05,
+      "loss": 0.0114,
+      "step": 124260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403231152809605e-05,
+      "loss": 0.0095,
+      "step": 124270
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403183113071134e-05,
+      "loss": 0.0111,
+      "step": 124280
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403135073332663e-05,
+      "loss": 0.012,
+      "step": 124290
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940308703359419e-05,
+      "loss": 0.0155,
+      "step": 124300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9403038993855718e-05,
+      "loss": 0.0101,
+      "step": 124310
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402990954117248e-05,
+      "loss": 0.0091,
+      "step": 124320
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402942914378773e-05,
+      "loss": 0.0076,
+      "step": 124330
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402894874640303e-05,
+      "loss": 0.0109,
+      "step": 124340
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402846834901832e-05,
+      "loss": 0.009,
+      "step": 124350
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940279879516336e-05,
+      "loss": 0.0117,
+      "step": 124360
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402750755424887e-05,
+      "loss": 0.012,
+      "step": 124370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402702715686416e-05,
+      "loss": 0.0099,
+      "step": 124380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402654675947946e-05,
+      "loss": 0.0094,
+      "step": 124390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402606636209475e-05,
+      "loss": 0.0075,
+      "step": 124400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402558596471e-05,
+      "loss": 0.012,
+      "step": 124410
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940251055673253e-05,
+      "loss": 0.0102,
+      "step": 124420
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940246251699406e-05,
+      "loss": 0.0096,
+      "step": 124430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402414477255585e-05,
+      "loss": 0.0097,
+      "step": 124440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402366437517115e-05,
+      "loss": 0.0096,
+      "step": 124450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402318397778644e-05,
+      "loss": 0.0075,
+      "step": 124460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402270358040173e-05,
+      "loss": 0.0106,
+      "step": 124470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.94022223183017e-05,
+      "loss": 0.0089,
+      "step": 124480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402174278563228e-05,
+      "loss": 0.0078,
+      "step": 124490
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402126238824757e-05,
+      "loss": 0.0121,
+      "step": 124500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402078199086283e-05,
+      "loss": 0.0105,
+      "step": 124510
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9402030159347813e-05,
+      "loss": 0.0132,
+      "step": 124520
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9401982119609342e-05,
+      "loss": 0.0111,
+      "step": 124530
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.940193407987087e-05,
+      "loss": 0.0078,
+      "step": 124540
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401886040132397e-05,
+      "loss": 0.0119,
+      "step": 124550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401838000393926e-05,
+      "loss": 0.0101,
+      "step": 124560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401789960655456e-05,
+      "loss": 0.0099,
+      "step": 124570
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401741920916985e-05,
+      "loss": 0.0108,
+      "step": 124580
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.940169388117851e-05,
+      "loss": 0.0114,
+      "step": 124590
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.940164584144004e-05,
+      "loss": 0.01,
+      "step": 124600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.940159780170157e-05,
+      "loss": 0.0116,
+      "step": 124610
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401549761963095e-05,
+      "loss": 0.0084,
+      "step": 124620
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401501722224624e-05,
+      "loss": 0.0099,
+      "step": 124630
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401453682486154e-05,
+      "loss": 0.0074,
+      "step": 124640
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401405642747683e-05,
+      "loss": 0.0097,
+      "step": 124650
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.940135760300921e-05,
+      "loss": 0.011,
+      "step": 124660
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401309563270738e-05,
+      "loss": 0.0082,
+      "step": 124670
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401261523532267e-05,
+      "loss": 0.0102,
+      "step": 124680
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401213483793793e-05,
+      "loss": 0.0112,
+      "step": 124690
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401165444055323e-05,
+      "loss": 0.0088,
+      "step": 124700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401117404316852e-05,
+      "loss": 0.0081,
+      "step": 124710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.940106936457838e-05,
+      "loss": 0.01,
+      "step": 124720
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9401021324839907e-05,
+      "loss": 0.0127,
+      "step": 124730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400973285101436e-05,
+      "loss": 0.0101,
+      "step": 124740
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400925245362966e-05,
+      "loss": 0.0102,
+      "step": 124750
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400877205624495e-05,
+      "loss": 0.0091,
+      "step": 124760
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.940082916588602e-05,
+      "loss": 0.0095,
+      "step": 124770
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.940078112614755e-05,
+      "loss": 0.0072,
+      "step": 124780
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.940073308640908e-05,
+      "loss": 0.0097,
+      "step": 124790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400685046670605e-05,
+      "loss": 0.0086,
+      "step": 124800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400637006932134e-05,
+      "loss": 0.0115,
+      "step": 124810
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400588967193664e-05,
+      "loss": 0.0092,
+      "step": 124820
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400540927455193e-05,
+      "loss": 0.0112,
+      "step": 124830
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.940049288771672e-05,
+      "loss": 0.0105,
+      "step": 124840
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400444847978248e-05,
+      "loss": 0.0106,
+      "step": 124850
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400396808239777e-05,
+      "loss": 0.0122,
+      "step": 124860
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400348768501303e-05,
+      "loss": 0.0079,
+      "step": 124870
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400300728762832e-05,
+      "loss": 0.0085,
+      "step": 124880
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400252689024362e-05,
+      "loss": 0.0104,
+      "step": 124890
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.940020464928589e-05,
+      "loss": 0.0094,
+      "step": 124900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.940015660954742e-05,
+      "loss": 0.009,
+      "step": 124910
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.940010856980895e-05,
+      "loss": 0.0101,
+      "step": 124920
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400060530070475e-05,
+      "loss": 0.0063,
+      "step": 124930
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9400012490332005e-05,
+      "loss": 0.0104,
+      "step": 124940
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399964450593534e-05,
+      "loss": 0.0091,
+      "step": 124950
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399916410855063e-05,
+      "loss": 0.0121,
+      "step": 124960
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.939986837111659e-05,
+      "loss": 0.0096,
+      "step": 124970
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.939982033137812e-05,
+      "loss": 0.0093,
+      "step": 124980
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399772291639648e-05,
+      "loss": 0.01,
+      "step": 124990
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399724251901177e-05,
+      "loss": 0.0083,
+      "step": 125000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399676212162703e-05,
+      "loss": 0.0173,
+      "step": 125010
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399628172424232e-05,
+      "loss": 0.0106,
+      "step": 125020
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.939958013268576e-05,
+      "loss": 0.0108,
+      "step": 125030
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399532092947287e-05,
+      "loss": 0.0094,
+      "step": 125040
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399484053208816e-05,
+      "loss": 0.0128,
+      "step": 125050
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399436013470346e-05,
+      "loss": 0.0113,
+      "step": 125060
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399387973731875e-05,
+      "loss": 0.0091,
+      "step": 125070
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.93993399339934e-05,
+      "loss": 0.0101,
+      "step": 125080
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.939929189425493e-05,
+      "loss": 0.0147,
+      "step": 125090
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.939924385451646e-05,
+      "loss": 0.0078,
+      "step": 125100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399195814777985e-05,
+      "loss": 0.0084,
+      "step": 125110
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399147775039515e-05,
+      "loss": 0.0078,
+      "step": 125120
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399099735301044e-05,
+      "loss": 0.0095,
+      "step": 125130
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9399051695562573e-05,
+      "loss": 0.0098,
+      "step": 125140
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.93990036558241e-05,
+      "loss": 0.0076,
+      "step": 125150
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9398955616085628e-05,
+      "loss": 0.0144,
+      "step": 125160
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9398907576347158e-05,
+      "loss": 0.0083,
+      "step": 125170
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9398859536608687e-05,
+      "loss": 0.0101,
+      "step": 125180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9398811496870213e-05,
+      "loss": 0.011,
+      "step": 125190
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9398763457131742e-05,
+      "loss": 0.0103,
+      "step": 125200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.939871541739327e-05,
+      "loss": 0.0077,
+      "step": 125210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9398667377654797e-05,
+      "loss": 0.0096,
+      "step": 125220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9398619337916326e-05,
+      "loss": 0.0116,
+      "step": 125230
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9398571298177856e-05,
+      "loss": 0.012,
+      "step": 125240
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9398523258439385e-05,
+      "loss": 0.0078,
+      "step": 125250
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939847521870091e-05,
+      "loss": 0.0095,
+      "step": 125260
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939842717896244e-05,
+      "loss": 0.0083,
+      "step": 125270
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939837913922397e-05,
+      "loss": 0.0133,
+      "step": 125280
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9398331099485495e-05,
+      "loss": 0.0098,
+      "step": 125290
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9398283059747025e-05,
+      "loss": 0.0089,
+      "step": 125300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9398235020008554e-05,
+      "loss": 0.01,
+      "step": 125310
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9398186980270083e-05,
+      "loss": 0.0132,
+      "step": 125320
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939813894053161e-05,
+      "loss": 0.0084,
+      "step": 125330
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9398090900793138e-05,
+      "loss": 0.014,
+      "step": 125340
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9398042861054667e-05,
+      "loss": 0.0156,
+      "step": 125350
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397994821316193e-05,
+      "loss": 0.0086,
+      "step": 125360
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397946781577723e-05,
+      "loss": 0.0103,
+      "step": 125370
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397898741839252e-05,
+      "loss": 0.013,
+      "step": 125380
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939785070210078e-05,
+      "loss": 0.0114,
+      "step": 125390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397802662362307e-05,
+      "loss": 0.0075,
+      "step": 125400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397754622623836e-05,
+      "loss": 0.0108,
+      "step": 125410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397706582885366e-05,
+      "loss": 0.0141,
+      "step": 125420
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397658543146895e-05,
+      "loss": 0.0087,
+      "step": 125430
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939761050340842e-05,
+      "loss": 0.0133,
+      "step": 125440
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939756246366995e-05,
+      "loss": 0.0089,
+      "step": 125450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939751442393148e-05,
+      "loss": 0.0151,
+      "step": 125460
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397466384193005e-05,
+      "loss": 0.0068,
+      "step": 125470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397418344454534e-05,
+      "loss": 0.009,
+      "step": 125480
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397370304716064e-05,
+      "loss": 0.0098,
+      "step": 125490
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397322264977593e-05,
+      "loss": 0.0081,
+      "step": 125500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939727422523912e-05,
+      "loss": 0.0085,
+      "step": 125510
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397226185500648e-05,
+      "loss": 0.0125,
+      "step": 125520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397178145762177e-05,
+      "loss": 0.0133,
+      "step": 125530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397130106023703e-05,
+      "loss": 0.0123,
+      "step": 125540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397082066285233e-05,
+      "loss": 0.0131,
+      "step": 125550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9397034026546762e-05,
+      "loss": 0.0082,
+      "step": 125560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939698598680829e-05,
+      "loss": 0.0106,
+      "step": 125570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396937947069817e-05,
+      "loss": 0.0092,
+      "step": 125580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396889907331346e-05,
+      "loss": 0.01,
+      "step": 125590
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396841867592876e-05,
+      "loss": 0.0083,
+      "step": 125600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396793827854405e-05,
+      "loss": 0.0087,
+      "step": 125610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939674578811593e-05,
+      "loss": 0.0139,
+      "step": 125620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939669774837746e-05,
+      "loss": 0.0099,
+      "step": 125630
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939664970863899e-05,
+      "loss": 0.0097,
+      "step": 125640
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396601668900515e-05,
+      "loss": 0.0104,
+      "step": 125650
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396553629162044e-05,
+      "loss": 0.0167,
+      "step": 125660
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396505589423574e-05,
+      "loss": 0.0082,
+      "step": 125670
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396457549685103e-05,
+      "loss": 0.0088,
+      "step": 125680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939640950994663e-05,
+      "loss": 0.012,
+      "step": 125690
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396361470208158e-05,
+      "loss": 0.012,
+      "step": 125700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396313430469687e-05,
+      "loss": 0.0074,
+      "step": 125710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396265390731213e-05,
+      "loss": 0.009,
+      "step": 125720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396217350992742e-05,
+      "loss": 0.012,
+      "step": 125730
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396169311254272e-05,
+      "loss": 0.0074,
+      "step": 125740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.93961212715158e-05,
+      "loss": 0.0083,
+      "step": 125750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396073231777327e-05,
+      "loss": 0.0157,
+      "step": 125760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9396025192038856e-05,
+      "loss": 0.0097,
+      "step": 125770
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9395977152300385e-05,
+      "loss": 0.0123,
+      "step": 125780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9395929112561915e-05,
+      "loss": 0.0106,
+      "step": 125790
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939588107282344e-05,
+      "loss": 0.0096,
+      "step": 125800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939583303308497e-05,
+      "loss": 0.011,
+      "step": 125810
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.93957849933465e-05,
+      "loss": 0.01,
+      "step": 125820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9395736953608025e-05,
+      "loss": 0.0127,
+      "step": 125830
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9395688913869554e-05,
+      "loss": 0.0076,
+      "step": 125840
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9395640874131084e-05,
+      "loss": 0.011,
+      "step": 125850
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9395592834392613e-05,
+      "loss": 0.0085,
+      "step": 125860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939554479465414e-05,
+      "loss": 0.0088,
+      "step": 125870
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9395496754915668e-05,
+      "loss": 0.0121,
+      "step": 125880
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9395448715177197e-05,
+      "loss": 0.0091,
+      "step": 125890
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9395400675438723e-05,
+      "loss": 0.0102,
+      "step": 125900
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9395352635700252e-05,
+      "loss": 0.0083,
+      "step": 125910
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939530459596178e-05,
+      "loss": 0.0082,
+      "step": 125920
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.939525655622331e-05,
+      "loss": 0.0116,
+      "step": 125930
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9395208516484837e-05,
+      "loss": 0.0086,
+      "step": 125940
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9395160476746366e-05,
+      "loss": 0.0076,
+      "step": 125950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9395112437007895e-05,
+      "loss": 0.0086,
+      "step": 125960
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9395064397269425e-05,
+      "loss": 0.0087,
+      "step": 125970
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939501635753095e-05,
+      "loss": 0.0145,
+      "step": 125980
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939496831779248e-05,
+      "loss": 0.0082,
+      "step": 125990
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939492027805401e-05,
+      "loss": 0.009,
+      "step": 126000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394872238315535e-05,
+      "loss": 0.0091,
+      "step": 126010
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394824198577064e-05,
+      "loss": 0.0111,
+      "step": 126020
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394776158838593e-05,
+      "loss": 0.0096,
+      "step": 126030
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394728119100123e-05,
+      "loss": 0.0082,
+      "step": 126040
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939468007936165e-05,
+      "loss": 0.0108,
+      "step": 126050
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394632039623178e-05,
+      "loss": 0.0104,
+      "step": 126060
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394583999884707e-05,
+      "loss": 0.0094,
+      "step": 126070
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394535960146233e-05,
+      "loss": 0.008,
+      "step": 126080
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394487920407762e-05,
+      "loss": 0.0084,
+      "step": 126090
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939443988066929e-05,
+      "loss": 0.0075,
+      "step": 126100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939439184093082e-05,
+      "loss": 0.009,
+      "step": 126110
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394343801192347e-05,
+      "loss": 0.01,
+      "step": 126120
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394295761453876e-05,
+      "loss": 0.0107,
+      "step": 126130
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394247721715405e-05,
+      "loss": 0.0111,
+      "step": 126140
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394199681976935e-05,
+      "loss": 0.0081,
+      "step": 126150
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939415164223846e-05,
+      "loss": 0.0124,
+      "step": 126160
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939410360249999e-05,
+      "loss": 0.0126,
+      "step": 126170
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939405556276152e-05,
+      "loss": 0.011,
+      "step": 126180
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9394007523023045e-05,
+      "loss": 0.0097,
+      "step": 126190
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393959483284574e-05,
+      "loss": 0.0083,
+      "step": 126200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393911443546103e-05,
+      "loss": 0.0112,
+      "step": 126210
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393863403807633e-05,
+      "loss": 0.0102,
+      "step": 126220
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939381536406916e-05,
+      "loss": 0.0108,
+      "step": 126230
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393767324330688e-05,
+      "loss": 0.0096,
+      "step": 126240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393719284592217e-05,
+      "loss": 0.0115,
+      "step": 126250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393671244853743e-05,
+      "loss": 0.0092,
+      "step": 126260
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393623205115272e-05,
+      "loss": 0.0085,
+      "step": 126270
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.93935751653768e-05,
+      "loss": 0.0089,
+      "step": 126280
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939352712563833e-05,
+      "loss": 0.0118,
+      "step": 126290
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393479085899857e-05,
+      "loss": 0.0121,
+      "step": 126300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393431046161386e-05,
+      "loss": 0.0091,
+      "step": 126310
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393383006422915e-05,
+      "loss": 0.0099,
+      "step": 126320
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393334966684444e-05,
+      "loss": 0.011,
+      "step": 126330
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939328692694597e-05,
+      "loss": 0.0107,
+      "step": 126340
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.93932388872075e-05,
+      "loss": 0.0123,
+      "step": 126350
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939319084746903e-05,
+      "loss": 0.0138,
+      "step": 126360
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393142807730555e-05,
+      "loss": 0.008,
+      "step": 126370
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393094767992084e-05,
+      "loss": 0.0077,
+      "step": 126380
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9393046728253613e-05,
+      "loss": 0.0129,
+      "step": 126390
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392998688515143e-05,
+      "loss": 0.012,
+      "step": 126400
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939295064877667e-05,
+      "loss": 0.0089,
+      "step": 126410
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392902609038198e-05,
+      "loss": 0.009,
+      "step": 126420
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392854569299727e-05,
+      "loss": 0.01,
+      "step": 126430
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392806529561253e-05,
+      "loss": 0.0099,
+      "step": 126440
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392758489822782e-05,
+      "loss": 0.0104,
+      "step": 126450
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939271045008431e-05,
+      "loss": 0.0108,
+      "step": 126460
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939266241034584e-05,
+      "loss": 0.0077,
+      "step": 126470
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392614370607367e-05,
+      "loss": 0.0115,
+      "step": 126480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392566330868896e-05,
+      "loss": 0.0081,
+      "step": 126490
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392518291130425e-05,
+      "loss": 0.0106,
+      "step": 126500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392470251391954e-05,
+      "loss": 0.0109,
+      "step": 126510
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939242221165348e-05,
+      "loss": 0.0104,
+      "step": 126520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939237417191501e-05,
+      "loss": 0.0106,
+      "step": 126530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939232613217654e-05,
+      "loss": 0.013,
+      "step": 126540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392278092438065e-05,
+      "loss": 0.0079,
+      "step": 126550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392230052699594e-05,
+      "loss": 0.0112,
+      "step": 126560
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392182012961123e-05,
+      "loss": 0.0099,
+      "step": 126570
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392133973222652e-05,
+      "loss": 0.0124,
+      "step": 126580
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.939208593348418e-05,
+      "loss": 0.0144,
+      "step": 126590
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9392037893745708e-05,
+      "loss": 0.0099,
+      "step": 126600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9391989854007237e-05,
+      "loss": 0.0077,
+      "step": 126610
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9391941814268763e-05,
+      "loss": 0.01,
+      "step": 126620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9391893774530292e-05,
+      "loss": 0.0086,
+      "step": 126630
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939184573479182e-05,
+      "loss": 0.0136,
+      "step": 126640
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939179769505335e-05,
+      "loss": 0.0085,
+      "step": 126650
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9391749655314876e-05,
+      "loss": 0.0146,
+      "step": 126660
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9391701615576406e-05,
+      "loss": 0.0149,
+      "step": 126670
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9391653575837935e-05,
+      "loss": 0.0075,
+      "step": 126680
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9391605536099464e-05,
+      "loss": 0.0089,
+      "step": 126690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939155749636099e-05,
+      "loss": 0.015,
+      "step": 126700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939150945662252e-05,
+      "loss": 0.0116,
+      "step": 126710
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939146141688405e-05,
+      "loss": 0.0072,
+      "step": 126720
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9391413377145575e-05,
+      "loss": 0.0125,
+      "step": 126730
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9391365337407104e-05,
+      "loss": 0.009,
+      "step": 126740
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9391322101642482e-05,
+      "loss": 0.0089,
+      "step": 126750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9391274061904008e-05,
+      "loss": 0.0081,
+      "step": 126760
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9391226022165537e-05,
+      "loss": 0.0153,
+      "step": 126770
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9391177982427066e-05,
+      "loss": 0.0083,
+      "step": 126780
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9391129942688592e-05,
+      "loss": 0.0128,
+      "step": 126790
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939108190295012e-05,
+      "loss": 0.0116,
+      "step": 126800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939103386321165e-05,
+      "loss": 0.0078,
+      "step": 126810
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939098582347318e-05,
+      "loss": 0.0109,
+      "step": 126820
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390937783734706e-05,
+      "loss": 0.0102,
+      "step": 126830
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390889743996235e-05,
+      "loss": 0.0143,
+      "step": 126840
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390841704257764e-05,
+      "loss": 0.011,
+      "step": 126850
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939079366451929e-05,
+      "loss": 0.0132,
+      "step": 126860
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939074562478082e-05,
+      "loss": 0.0082,
+      "step": 126870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939069758504235e-05,
+      "loss": 0.0101,
+      "step": 126880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390649545303878e-05,
+      "loss": 0.0114,
+      "step": 126890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390601505565404e-05,
+      "loss": 0.0084,
+      "step": 126900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390553465826933e-05,
+      "loss": 0.0097,
+      "step": 126910
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390505426088463e-05,
+      "loss": 0.0099,
+      "step": 126920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939045738634999e-05,
+      "loss": 0.0129,
+      "step": 126930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390409346611518e-05,
+      "loss": 0.0098,
+      "step": 126940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390361306873047e-05,
+      "loss": 0.01,
+      "step": 126950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390313267134576e-05,
+      "loss": 0.0076,
+      "step": 126960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390265227396102e-05,
+      "loss": 0.0117,
+      "step": 126970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939021718765763e-05,
+      "loss": 0.0125,
+      "step": 126980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939016914791916e-05,
+      "loss": 0.0117,
+      "step": 126990
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.939012110818069e-05,
+      "loss": 0.0096,
+      "step": 127000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390073068442216e-05,
+      "loss": 0.01,
+      "step": 127010
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390025028703745e-05,
+      "loss": 0.0126,
+      "step": 127020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9389976988965274e-05,
+      "loss": 0.008,
+      "step": 127030
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.93899289492268e-05,
+      "loss": 0.0092,
+      "step": 127040
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.938988090948833e-05,
+      "loss": 0.013,
+      "step": 127050
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.938983286974986e-05,
+      "loss": 0.0081,
+      "step": 127060
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9389784830011388e-05,
+      "loss": 0.0092,
+      "step": 127070
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9389736790272914e-05,
+      "loss": 0.0107,
+      "step": 127080
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9389688750534443e-05,
+      "loss": 0.0118,
+      "step": 127090
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9389640710795972e-05,
+      "loss": 0.0068,
+      "step": 127100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.93895926710575e-05,
+      "loss": 0.0102,
+      "step": 127110
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9389544631319028e-05,
+      "loss": 0.0091,
+      "step": 127120
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9389496591580557e-05,
+      "loss": 0.0077,
+      "step": 127130
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9389448551842086e-05,
+      "loss": 0.0092,
+      "step": 127140
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9389400512103612e-05,
+      "loss": 0.0122,
+      "step": 127150
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.938935247236514e-05,
+      "loss": 0.0096,
+      "step": 127160
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.938930443262667e-05,
+      "loss": 0.0123,
+      "step": 127170
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.93892563928882e-05,
+      "loss": 0.0079,
+      "step": 127180
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9389208353149726e-05,
+      "loss": 0.007,
+      "step": 127190
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9389160313411255e-05,
+      "loss": 0.0109,
+      "step": 127200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9389112273672784e-05,
+      "loss": 0.0089,
+      "step": 127210
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.938906423393431e-05,
+      "loss": 0.0086,
+      "step": 127220
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.938901619419584e-05,
+      "loss": 0.0117,
+      "step": 127230
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.938896815445737e-05,
+      "loss": 0.0091,
+      "step": 127240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9388920114718898e-05,
+      "loss": 0.0107,
+      "step": 127250
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9388872074980424e-05,
+      "loss": 0.0081,
+      "step": 127260
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9388824035241953e-05,
+      "loss": 0.0079,
+      "step": 127270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9388775995503482e-05,
+      "loss": 0.0047,
+      "step": 127280
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9388727955765008e-05,
+      "loss": 0.0089,
+      "step": 127290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9388679916026537e-05,
+      "loss": 0.0118,
+      "step": 127300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9388631876288067e-05,
+      "loss": 0.0098,
+      "step": 127310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9388583836549596e-05,
+      "loss": 0.0085,
+      "step": 127320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9388535796811122e-05,
+      "loss": 0.0082,
+      "step": 127330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938848775707265e-05,
+      "loss": 0.0065,
+      "step": 127340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938843971733418e-05,
+      "loss": 0.0071,
+      "step": 127350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938839167759571e-05,
+      "loss": 0.0133,
+      "step": 127360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9388343637857236e-05,
+      "loss": 0.01,
+      "step": 127370
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9388295598118765e-05,
+      "loss": 0.0083,
+      "step": 127380
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9388247558380294e-05,
+      "loss": 0.0122,
+      "step": 127390
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938819951864182e-05,
+      "loss": 0.0075,
+      "step": 127400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938815147890335e-05,
+      "loss": 0.0096,
+      "step": 127410
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938810343916488e-05,
+      "loss": 0.007,
+      "step": 127420
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9388055399426408e-05,
+      "loss": 0.0066,
+      "step": 127430
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9388007359687934e-05,
+      "loss": 0.0098,
+      "step": 127440
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387959319949463e-05,
+      "loss": 0.0102,
+      "step": 127450
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387911280210992e-05,
+      "loss": 0.0104,
+      "step": 127460
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387863240472518e-05,
+      "loss": 0.0111,
+      "step": 127470
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387815200734047e-05,
+      "loss": 0.0102,
+      "step": 127480
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387767160995577e-05,
+      "loss": 0.0106,
+      "step": 127490
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387719121257106e-05,
+      "loss": 0.0094,
+      "step": 127500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387671081518632e-05,
+      "loss": 0.013,
+      "step": 127510
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938762304178016e-05,
+      "loss": 0.0081,
+      "step": 127520
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938757500204169e-05,
+      "loss": 0.0096,
+      "step": 127530
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938752696230322e-05,
+      "loss": 0.0083,
+      "step": 127540
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387478922564746e-05,
+      "loss": 0.0139,
+      "step": 127550
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387430882826275e-05,
+      "loss": 0.0135,
+      "step": 127560
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387382843087804e-05,
+      "loss": 0.0123,
+      "step": 127570
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938733480334933e-05,
+      "loss": 0.0089,
+      "step": 127580
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938728676361086e-05,
+      "loss": 0.0098,
+      "step": 127590
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938723872387239e-05,
+      "loss": 0.0091,
+      "step": 127600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387190684133918e-05,
+      "loss": 0.017,
+      "step": 127610
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387142644395444e-05,
+      "loss": 0.0138,
+      "step": 127620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387094604656973e-05,
+      "loss": 0.0109,
+      "step": 127630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9387046564918502e-05,
+      "loss": 0.0071,
+      "step": 127640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386998525180028e-05,
+      "loss": 0.0078,
+      "step": 127650
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386950485441557e-05,
+      "loss": 0.0103,
+      "step": 127660
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386902445703087e-05,
+      "loss": 0.0088,
+      "step": 127670
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386854405964616e-05,
+      "loss": 0.007,
+      "step": 127680
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386806366226142e-05,
+      "loss": 0.0099,
+      "step": 127690
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938675832648767e-05,
+      "loss": 0.0085,
+      "step": 127700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.93867102867492e-05,
+      "loss": 0.0101,
+      "step": 127710
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938666224701073e-05,
+      "loss": 0.0114,
+      "step": 127720
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386614207272255e-05,
+      "loss": 0.007,
+      "step": 127730
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386566167533785e-05,
+      "loss": 0.0116,
+      "step": 127740
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386518127795314e-05,
+      "loss": 0.009,
+      "step": 127750
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938647008805684e-05,
+      "loss": 0.0119,
+      "step": 127760
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938642204831837e-05,
+      "loss": 0.0114,
+      "step": 127770
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.93863740085799e-05,
+      "loss": 0.009,
+      "step": 127780
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386325968841428e-05,
+      "loss": 0.0139,
+      "step": 127790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386277929102954e-05,
+      "loss": 0.0112,
+      "step": 127800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386229889364483e-05,
+      "loss": 0.0092,
+      "step": 127810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386181849626012e-05,
+      "loss": 0.0081,
+      "step": 127820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386133809887538e-05,
+      "loss": 0.0114,
+      "step": 127830
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386085770149067e-05,
+      "loss": 0.0096,
+      "step": 127840
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9386037730410597e-05,
+      "loss": 0.0083,
+      "step": 127850
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9385989690672126e-05,
+      "loss": 0.0106,
+      "step": 127860
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938594165093365e-05,
+      "loss": 0.0111,
+      "step": 127870
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938589361119518e-05,
+      "loss": 0.0102,
+      "step": 127880
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938584557145671e-05,
+      "loss": 0.0132,
+      "step": 127890
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938579753171824e-05,
+      "loss": 0.0102,
+      "step": 127900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9385749491979765e-05,
+      "loss": 0.012,
+      "step": 127910
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9385701452241295e-05,
+      "loss": 0.0109,
+      "step": 127920
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9385653412502824e-05,
+      "loss": 0.0084,
+      "step": 127930
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938560537276435e-05,
+      "loss": 0.0089,
+      "step": 127940
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938555733302588e-05,
+      "loss": 0.0105,
+      "step": 127950
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938550929328741e-05,
+      "loss": 0.0099,
+      "step": 127960
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9385461253548938e-05,
+      "loss": 0.0088,
+      "step": 127970
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9385413213810467e-05,
+      "loss": 0.0089,
+      "step": 127980
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9385365174071996e-05,
+      "loss": 0.0096,
+      "step": 127990
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9385317134333522e-05,
+      "loss": 0.007,
+      "step": 128000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938526909459505e-05,
+      "loss": 0.0089,
+      "step": 128010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938522105485658e-05,
+      "loss": 0.0076,
+      "step": 128020
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938517301511811e-05,
+      "loss": 0.0127,
+      "step": 128030
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9385124975379636e-05,
+      "loss": 0.0103,
+      "step": 128040
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9385076935641165e-05,
+      "loss": 0.0138,
+      "step": 128050
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9385028895902694e-05,
+      "loss": 0.0077,
+      "step": 128060
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938498085616422e-05,
+      "loss": 0.0093,
+      "step": 128070
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938493281642575e-05,
+      "loss": 0.0107,
+      "step": 128080
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938488477668728e-05,
+      "loss": 0.0139,
+      "step": 128090
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9384836736948808e-05,
+      "loss": 0.0081,
+      "step": 128100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9384788697210334e-05,
+      "loss": 0.007,
+      "step": 128110
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9384740657471863e-05,
+      "loss": 0.0091,
+      "step": 128120
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9384692617733392e-05,
+      "loss": 0.0099,
+      "step": 128130
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938464457799492e-05,
+      "loss": 0.0122,
+      "step": 128140
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9384596538256448e-05,
+      "loss": 0.0093,
+      "step": 128150
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9384548498517977e-05,
+      "loss": 0.0083,
+      "step": 128160
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9384500458779506e-05,
+      "loss": 0.0135,
+      "step": 128170
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9384452419041032e-05,
+      "loss": 0.0093,
+      "step": 128180
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938440437930256e-05,
+      "loss": 0.0104,
+      "step": 128190
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938435633956409e-05,
+      "loss": 0.0128,
+      "step": 128200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938430829982562e-05,
+      "loss": 0.0077,
+      "step": 128210
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9384260260087146e-05,
+      "loss": 0.0074,
+      "step": 128220
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9384212220348675e-05,
+      "loss": 0.0088,
+      "step": 128230
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9384164180610204e-05,
+      "loss": 0.0089,
+      "step": 128240
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938411614087173e-05,
+      "loss": 0.0114,
+      "step": 128250
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938406810113326e-05,
+      "loss": 0.0101,
+      "step": 128260
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938402006139479e-05,
+      "loss": 0.0096,
+      "step": 128270
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383972021656318e-05,
+      "loss": 0.0077,
+      "step": 128280
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383923981917844e-05,
+      "loss": 0.0095,
+      "step": 128290
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383875942179373e-05,
+      "loss": 0.0093,
+      "step": 128300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383827902440902e-05,
+      "loss": 0.0075,
+      "step": 128310
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383779862702428e-05,
+      "loss": 0.0077,
+      "step": 128320
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383731822963957e-05,
+      "loss": 0.0125,
+      "step": 128330
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383683783225487e-05,
+      "loss": 0.0087,
+      "step": 128340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383635743487016e-05,
+      "loss": 0.009,
+      "step": 128350
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383587703748542e-05,
+      "loss": 0.0089,
+      "step": 128360
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938353966401007e-05,
+      "loss": 0.012,
+      "step": 128370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.93834916242716e-05,
+      "loss": 0.0085,
+      "step": 128380
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938344358453313e-05,
+      "loss": 0.0096,
+      "step": 128390
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383395544794656e-05,
+      "loss": 0.0151,
+      "step": 128400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383347505056185e-05,
+      "loss": 0.0091,
+      "step": 128410
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383299465317714e-05,
+      "loss": 0.009,
+      "step": 128420
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938325142557924e-05,
+      "loss": 0.0098,
+      "step": 128430
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938320338584077e-05,
+      "loss": 0.0106,
+      "step": 128440
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.93831553461023e-05,
+      "loss": 0.0098,
+      "step": 128450
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383107306363828e-05,
+      "loss": 0.0115,
+      "step": 128460
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383059266625354e-05,
+      "loss": 0.008,
+      "step": 128470
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9383011226886883e-05,
+      "loss": 0.0099,
+      "step": 128480
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382963187148412e-05,
+      "loss": 0.0157,
+      "step": 128490
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382915147409938e-05,
+      "loss": 0.0099,
+      "step": 128500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382867107671467e-05,
+      "loss": 0.008,
+      "step": 128510
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382819067932997e-05,
+      "loss": 0.0068,
+      "step": 128520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382771028194526e-05,
+      "loss": 0.0088,
+      "step": 128530
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382722988456052e-05,
+      "loss": 0.009,
+      "step": 128540
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938267494871758e-05,
+      "loss": 0.0075,
+      "step": 128550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938262690897911e-05,
+      "loss": 0.0104,
+      "step": 128560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938257886924064e-05,
+      "loss": 0.0108,
+      "step": 128570
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382530829502165e-05,
+      "loss": 0.0159,
+      "step": 128580
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382482789763695e-05,
+      "loss": 0.0118,
+      "step": 128590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382434750025224e-05,
+      "loss": 0.0073,
+      "step": 128600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938238671028675e-05,
+      "loss": 0.0071,
+      "step": 128610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938233867054828e-05,
+      "loss": 0.009,
+      "step": 128620
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382295434783657e-05,
+      "loss": 0.0137,
+      "step": 128630
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382247395045183e-05,
+      "loss": 0.0114,
+      "step": 128640
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382199355306712e-05,
+      "loss": 0.0125,
+      "step": 128650
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938215131556824e-05,
+      "loss": 0.0091,
+      "step": 128660
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382103275829767e-05,
+      "loss": 0.0077,
+      "step": 128670
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382055236091297e-05,
+      "loss": 0.0084,
+      "step": 128680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9382007196352826e-05,
+      "loss": 0.0095,
+      "step": 128690
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9381959156614355e-05,
+      "loss": 0.0072,
+      "step": 128700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.938191111687588e-05,
+      "loss": 0.0089,
+      "step": 128710
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938186307713741e-05,
+      "loss": 0.0093,
+      "step": 128720
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938181503739894e-05,
+      "loss": 0.0074,
+      "step": 128730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938177180163431e-05,
+      "loss": 0.011,
+      "step": 128740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938172376189584e-05,
+      "loss": 0.0086,
+      "step": 128750
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938167572215737e-05,
+      "loss": 0.0131,
+      "step": 128760
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.93816276824189e-05,
+      "loss": 0.0112,
+      "step": 128770
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9381579642680425e-05,
+      "loss": 0.0113,
+      "step": 128780
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9381531602941954e-05,
+      "loss": 0.0131,
+      "step": 128790
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9381483563203483e-05,
+      "loss": 0.0106,
+      "step": 128800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9381435523465012e-05,
+      "loss": 0.0132,
+      "step": 128810
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938138748372654e-05,
+      "loss": 0.0108,
+      "step": 128820
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9381339443988068e-05,
+      "loss": 0.0081,
+      "step": 128830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9381291404249597e-05,
+      "loss": 0.0098,
+      "step": 128840
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9381243364511123e-05,
+      "loss": 0.0094,
+      "step": 128850
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9381195324772652e-05,
+      "loss": 0.011,
+      "step": 128860
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938114728503418e-05,
+      "loss": 0.0102,
+      "step": 128870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938109924529571e-05,
+      "loss": 0.0092,
+      "step": 128880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9381051205557236e-05,
+      "loss": 0.0108,
+      "step": 128890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9381003165818766e-05,
+      "loss": 0.0119,
+      "step": 128900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9380955126080295e-05,
+      "loss": 0.0106,
+      "step": 128910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938090708634182e-05,
+      "loss": 0.0121,
+      "step": 128920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938085904660335e-05,
+      "loss": 0.0087,
+      "step": 128930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938081100686488e-05,
+      "loss": 0.0096,
+      "step": 128940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938076296712641e-05,
+      "loss": 0.0098,
+      "step": 128950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9380714927387935e-05,
+      "loss": 0.0091,
+      "step": 128960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9380666887649464e-05,
+      "loss": 0.0096,
+      "step": 128970
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9380618847910993e-05,
+      "loss": 0.0074,
+      "step": 128980
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9380570808172522e-05,
+      "loss": 0.0078,
+      "step": 128990
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938052276843405e-05,
+      "loss": 0.0089,
+      "step": 129000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938047472869558e-05,
+      "loss": 0.0073,
+      "step": 129010
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9380426688957107e-05,
+      "loss": 0.0087,
+      "step": 129020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9380378649218636e-05,
+      "loss": 0.0104,
+      "step": 129030
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9380330609480165e-05,
+      "loss": 0.0102,
+      "step": 129040
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9380282569741695e-05,
+      "loss": 0.0067,
+      "step": 129050
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938023453000322e-05,
+      "loss": 0.009,
+      "step": 129060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938018649026475e-05,
+      "loss": 0.009,
+      "step": 129070
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.938013845052628e-05,
+      "loss": 0.0188,
+      "step": 129080
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9380090410787805e-05,
+      "loss": 0.013,
+      "step": 129090
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9380042371049334e-05,
+      "loss": 0.0087,
+      "step": 129100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379994331310863e-05,
+      "loss": 0.0112,
+      "step": 129110
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379946291572393e-05,
+      "loss": 0.0058,
+      "step": 129120
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.937989825183392e-05,
+      "loss": 0.0117,
+      "step": 129130
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379850212095448e-05,
+      "loss": 0.0065,
+      "step": 129140
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379802172356977e-05,
+      "loss": 0.0094,
+      "step": 129150
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379754132618503e-05,
+      "loss": 0.0111,
+      "step": 129160
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379706092880032e-05,
+      "loss": 0.0112,
+      "step": 129170
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.937965805314156e-05,
+      "loss": 0.0084,
+      "step": 129180
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.937961001340309e-05,
+      "loss": 0.0085,
+      "step": 129190
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379561973664617e-05,
+      "loss": 0.0108,
+      "step": 129200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379513933926146e-05,
+      "loss": 0.0077,
+      "step": 129210
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379465894187675e-05,
+      "loss": 0.0068,
+      "step": 129220
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379417854449205e-05,
+      "loss": 0.009,
+      "step": 129230
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.937936981471073e-05,
+      "loss": 0.0054,
+      "step": 129240
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.937932177497226e-05,
+      "loss": 0.0097,
+      "step": 129250
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.937927373523379e-05,
+      "loss": 0.0075,
+      "step": 129260
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379225695495315e-05,
+      "loss": 0.0087,
+      "step": 129270
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379177655756844e-05,
+      "loss": 0.007,
+      "step": 129280
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379129616018373e-05,
+      "loss": 0.0114,
+      "step": 129290
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9379081576279903e-05,
+      "loss": 0.0116,
+      "step": 129300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.937903353654143e-05,
+      "loss": 0.0101,
+      "step": 129310
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9378985496802958e-05,
+      "loss": 0.0077,
+      "step": 129320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9378937457064487e-05,
+      "loss": 0.0067,
+      "step": 129330
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9378889417326013e-05,
+      "loss": 0.0147,
+      "step": 129340
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9378841377587542e-05,
+      "loss": 0.0102,
+      "step": 129350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.937879333784907e-05,
+      "loss": 0.0097,
+      "step": 129360
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.93787452981106e-05,
+      "loss": 0.0107,
+      "step": 129370
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9378697258372127e-05,
+      "loss": 0.0101,
+      "step": 129380
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9378649218633656e-05,
+      "loss": 0.0081,
+      "step": 129390
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9378601178895185e-05,
+      "loss": 0.0103,
+      "step": 129400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9378553139156714e-05,
+      "loss": 0.0083,
+      "step": 129410
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937850509941824e-05,
+      "loss": 0.0075,
+      "step": 129420
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937845705967977e-05,
+      "loss": 0.0066,
+      "step": 129430
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.93784090199413e-05,
+      "loss": 0.0157,
+      "step": 129440
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9378360980202825e-05,
+      "loss": 0.0061,
+      "step": 129450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9378312940464354e-05,
+      "loss": 0.0103,
+      "step": 129460
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9378264900725883e-05,
+      "loss": 0.0066,
+      "step": 129470
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9378216860987413e-05,
+      "loss": 0.0144,
+      "step": 129480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937816882124894e-05,
+      "loss": 0.013,
+      "step": 129490
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9378120781510468e-05,
+      "loss": 0.0127,
+      "step": 129500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9378072741771997e-05,
+      "loss": 0.0073,
+      "step": 129510
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9378024702033523e-05,
+      "loss": 0.0089,
+      "step": 129520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377976662295052e-05,
+      "loss": 0.0167,
+      "step": 129530
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937792862255658e-05,
+      "loss": 0.0147,
+      "step": 129540
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937788058281811e-05,
+      "loss": 0.0125,
+      "step": 129550
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377832543079637e-05,
+      "loss": 0.0095,
+      "step": 129560
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377784503341166e-05,
+      "loss": 0.0103,
+      "step": 129570
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377736463602695e-05,
+      "loss": 0.0127,
+      "step": 129580
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377688423864224e-05,
+      "loss": 0.0089,
+      "step": 129590
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937764038412575e-05,
+      "loss": 0.0101,
+      "step": 129600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937759234438728e-05,
+      "loss": 0.0103,
+      "step": 129610
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937754430464881e-05,
+      "loss": 0.0104,
+      "step": 129620
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377496264910335e-05,
+      "loss": 0.0084,
+      "step": 129630
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377448225171864e-05,
+      "loss": 0.0157,
+      "step": 129640
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377400185433393e-05,
+      "loss": 0.0076,
+      "step": 129650
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377352145694922e-05,
+      "loss": 0.0104,
+      "step": 129660
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937730410595645e-05,
+      "loss": 0.0082,
+      "step": 129670
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377256066217978e-05,
+      "loss": 0.0099,
+      "step": 129680
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377208026479507e-05,
+      "loss": 0.0104,
+      "step": 129690
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377159986741033e-05,
+      "loss": 0.0104,
+      "step": 129700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9377111947002562e-05,
+      "loss": 0.0116,
+      "step": 129710
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937706390726409e-05,
+      "loss": 0.008,
+      "step": 129720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937701586752562e-05,
+      "loss": 0.0081,
+      "step": 129730
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376967827787146e-05,
+      "loss": 0.0082,
+      "step": 129740
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376919788048676e-05,
+      "loss": 0.0099,
+      "step": 129750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376871748310205e-05,
+      "loss": 0.0127,
+      "step": 129760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937682370857173e-05,
+      "loss": 0.0078,
+      "step": 129770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937677566883326e-05,
+      "loss": 0.009,
+      "step": 129780
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937672762909479e-05,
+      "loss": 0.0093,
+      "step": 129790
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937667958935632e-05,
+      "loss": 0.0082,
+      "step": 129800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376631549617845e-05,
+      "loss": 0.0067,
+      "step": 129810
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376583509879374e-05,
+      "loss": 0.007,
+      "step": 129820
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376535470140903e-05,
+      "loss": 0.0078,
+      "step": 129830
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376487430402432e-05,
+      "loss": 0.0106,
+      "step": 129840
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376439390663958e-05,
+      "loss": 0.01,
+      "step": 129850
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376391350925488e-05,
+      "loss": 0.0108,
+      "step": 129860
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376343311187017e-05,
+      "loss": 0.0103,
+      "step": 129870
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376295271448543e-05,
+      "loss": 0.0106,
+      "step": 129880
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376247231710072e-05,
+      "loss": 0.0091,
+      "step": 129890
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.93761991919716e-05,
+      "loss": 0.0085,
+      "step": 129900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937615115223313e-05,
+      "loss": 0.0142,
+      "step": 129910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376103112494656e-05,
+      "loss": 0.0061,
+      "step": 129920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376055072756186e-05,
+      "loss": 0.0102,
+      "step": 129930
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9376007033017715e-05,
+      "loss": 0.0096,
+      "step": 129940
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937595899327924e-05,
+      "loss": 0.0053,
+      "step": 129950
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937591095354077e-05,
+      "loss": 0.0083,
+      "step": 129960
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.93758629138023e-05,
+      "loss": 0.0081,
+      "step": 129970
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937581487406383e-05,
+      "loss": 0.0096,
+      "step": 129980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9375766834325354e-05,
+      "loss": 0.0098,
+      "step": 129990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9375718794586884e-05,
+      "loss": 0.0107,
+      "step": 130000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9375670754848413e-05,
+      "loss": 0.0102,
+      "step": 130010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9375622715109942e-05,
+      "loss": 0.0075,
+      "step": 130020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9375574675371468e-05,
+      "loss": 0.0061,
+      "step": 130030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9375526635632997e-05,
+      "loss": 0.0102,
+      "step": 130040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9375478595894527e-05,
+      "loss": 0.0114,
+      "step": 130050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9375430556156053e-05,
+      "loss": 0.0154,
+      "step": 130060
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9375382516417582e-05,
+      "loss": 0.009,
+      "step": 130070
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937533447667911e-05,
+      "loss": 0.0064,
+      "step": 130080
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.937528643694064e-05,
+      "loss": 0.0092,
+      "step": 130090
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9375238397202166e-05,
+      "loss": 0.0091,
+      "step": 130100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9375190357463696e-05,
+      "loss": 0.0095,
+      "step": 130110
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9375142317725225e-05,
+      "loss": 0.0113,
+      "step": 130120
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937509427798675e-05,
+      "loss": 0.0072,
+      "step": 130130
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937504623824828e-05,
+      "loss": 0.0132,
+      "step": 130140
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937499819850981e-05,
+      "loss": 0.01,
+      "step": 130150
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937495015877134e-05,
+      "loss": 0.0109,
+      "step": 130160
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374902119032864e-05,
+      "loss": 0.0112,
+      "step": 130170
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374854079294394e-05,
+      "loss": 0.0107,
+      "step": 130180
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374806039555923e-05,
+      "loss": 0.009,
+      "step": 130190
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374757999817452e-05,
+      "loss": 0.0056,
+      "step": 130200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374709960078978e-05,
+      "loss": 0.0096,
+      "step": 130210
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374661920340507e-05,
+      "loss": 0.0069,
+      "step": 130220
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374613880602037e-05,
+      "loss": 0.0097,
+      "step": 130230
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374565840863563e-05,
+      "loss": 0.0116,
+      "step": 130240
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374517801125092e-05,
+      "loss": 0.0086,
+      "step": 130250
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937446976138662e-05,
+      "loss": 0.0091,
+      "step": 130260
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937442172164815e-05,
+      "loss": 0.011,
+      "step": 130270
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374373681909676e-05,
+      "loss": 0.012,
+      "step": 130280
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374325642171205e-05,
+      "loss": 0.0101,
+      "step": 130290
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374277602432735e-05,
+      "loss": 0.0079,
+      "step": 130300
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937422956269426e-05,
+      "loss": 0.0119,
+      "step": 130310
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937418152295579e-05,
+      "loss": 0.0101,
+      "step": 130320
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937413348321732e-05,
+      "loss": 0.0091,
+      "step": 130330
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937408544347885e-05,
+      "loss": 0.0087,
+      "step": 130340
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374037403740374e-05,
+      "loss": 0.0117,
+      "step": 130350
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373989364001904e-05,
+      "loss": 0.007,
+      "step": 130360
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373941324263433e-05,
+      "loss": 0.0109,
+      "step": 130370
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373893284524962e-05,
+      "loss": 0.0074,
+      "step": 130380
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373845244786488e-05,
+      "loss": 0.0092,
+      "step": 130390
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373797205048017e-05,
+      "loss": 0.0087,
+      "step": 130400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373749165309547e-05,
+      "loss": 0.0117,
+      "step": 130410
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373701125571072e-05,
+      "loss": 0.009,
+      "step": 130420
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373653085832602e-05,
+      "loss": 0.0083,
+      "step": 130430
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937360504609413e-05,
+      "loss": 0.0107,
+      "step": 130440
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937355700635566e-05,
+      "loss": 0.008,
+      "step": 130450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373508966617186e-05,
+      "loss": 0.0087,
+      "step": 130460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373460926878715e-05,
+      "loss": 0.0084,
+      "step": 130470
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373412887140245e-05,
+      "loss": 0.0093,
+      "step": 130480
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937336484740177e-05,
+      "loss": 0.0117,
+      "step": 130490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.93733168076633e-05,
+      "loss": 0.0083,
+      "step": 130500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937326876792483e-05,
+      "loss": 0.0088,
+      "step": 130510
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937322072818636e-05,
+      "loss": 0.012,
+      "step": 130520
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373172688447884e-05,
+      "loss": 0.0082,
+      "step": 130530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373124648709414e-05,
+      "loss": 0.0078,
+      "step": 130540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373076608970943e-05,
+      "loss": 0.012,
+      "step": 130550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9373028569232472e-05,
+      "loss": 0.009,
+      "step": 130560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372980529493998e-05,
+      "loss": 0.0129,
+      "step": 130570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372932489755527e-05,
+      "loss": 0.008,
+      "step": 130580
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372884450017056e-05,
+      "loss": 0.0165,
+      "step": 130590
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372836410278582e-05,
+      "loss": 0.0099,
+      "step": 130600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937278837054011e-05,
+      "loss": 0.0105,
+      "step": 130610
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937274033080164e-05,
+      "loss": 0.0098,
+      "step": 130620
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937269229106317e-05,
+      "loss": 0.0093,
+      "step": 130630
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372644251324696e-05,
+      "loss": 0.0083,
+      "step": 130640
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372596211586225e-05,
+      "loss": 0.0127,
+      "step": 130650
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372548171847755e-05,
+      "loss": 0.0093,
+      "step": 130660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937250013210928e-05,
+      "loss": 0.0066,
+      "step": 130670
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937245209237081e-05,
+      "loss": 0.0102,
+      "step": 130680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937240405263234e-05,
+      "loss": 0.0069,
+      "step": 130690
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372356012893868e-05,
+      "loss": 0.0154,
+      "step": 130700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372307973155394e-05,
+      "loss": 0.0089,
+      "step": 130710
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372259933416923e-05,
+      "loss": 0.0078,
+      "step": 130720
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372211893678453e-05,
+      "loss": 0.012,
+      "step": 130730
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372163853939982e-05,
+      "loss": 0.0094,
+      "step": 130740
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372115814201508e-05,
+      "loss": 0.0074,
+      "step": 130750
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372067774463037e-05,
+      "loss": 0.0106,
+      "step": 130760
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9372019734724566e-05,
+      "loss": 0.0088,
+      "step": 130770
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9371971694986092e-05,
+      "loss": 0.0119,
+      "step": 130780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.937192365524762e-05,
+      "loss": 0.0112,
+      "step": 130790
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.937187561550915e-05,
+      "loss": 0.0104,
+      "step": 130800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.937182757577068e-05,
+      "loss": 0.0096,
+      "step": 130810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9371779536032206e-05,
+      "loss": 0.0147,
+      "step": 130820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9371731496293735e-05,
+      "loss": 0.0096,
+      "step": 130830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9371683456555264e-05,
+      "loss": 0.0128,
+      "step": 130840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.937163541681679e-05,
+      "loss": 0.0119,
+      "step": 130850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.937158737707832e-05,
+      "loss": 0.0143,
+      "step": 130860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.937153933733985e-05,
+      "loss": 0.0104,
+      "step": 130870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9371491297601378e-05,
+      "loss": 0.0102,
+      "step": 130880
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9371443257862904e-05,
+      "loss": 0.0084,
+      "step": 130890
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9371395218124433e-05,
+      "loss": 0.0092,
+      "step": 130900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9371347178385963e-05,
+      "loss": 0.0112,
+      "step": 130910
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9371299138647492e-05,
+      "loss": 0.0095,
+      "step": 130920
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9371251098909018e-05,
+      "loss": 0.0076,
+      "step": 130930
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9371203059170547e-05,
+      "loss": 0.0103,
+      "step": 130940
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9371155019432076e-05,
+      "loss": 0.0085,
+      "step": 130950
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9371106979693602e-05,
+      "loss": 0.0107,
+      "step": 130960
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.937105893995513e-05,
+      "loss": 0.01,
+      "step": 130970
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.937101090021666e-05,
+      "loss": 0.0094,
+      "step": 130980
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.937096286047819e-05,
+      "loss": 0.0108,
+      "step": 130990
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370914820739716e-05,
+      "loss": 0.0087,
+      "step": 131000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370866781001245e-05,
+      "loss": 0.0103,
+      "step": 131010
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370818741262774e-05,
+      "loss": 0.0098,
+      "step": 131020
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.93707707015243e-05,
+      "loss": 0.0086,
+      "step": 131030
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.937072266178583e-05,
+      "loss": 0.0091,
+      "step": 131040
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.937067462204736e-05,
+      "loss": 0.0145,
+      "step": 131050
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370626582308888e-05,
+      "loss": 0.0104,
+      "step": 131060
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370578542570414e-05,
+      "loss": 0.009,
+      "step": 131070
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370530502831943e-05,
+      "loss": 0.0074,
+      "step": 131080
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370482463093473e-05,
+      "loss": 0.0078,
+      "step": 131090
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370434423355002e-05,
+      "loss": 0.0104,
+      "step": 131100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370386383616528e-05,
+      "loss": 0.0061,
+      "step": 131110
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370338343878057e-05,
+      "loss": 0.0126,
+      "step": 131120
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370290304139586e-05,
+      "loss": 0.0126,
+      "step": 131130
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370242264401112e-05,
+      "loss": 0.0098,
+      "step": 131140
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.937019422466264e-05,
+      "loss": 0.0051,
+      "step": 131150
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.937014618492417e-05,
+      "loss": 0.0114,
+      "step": 131160
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.93700981451857e-05,
+      "loss": 0.0123,
+      "step": 131170
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370050105447226e-05,
+      "loss": 0.0079,
+      "step": 131180
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9370002065708755e-05,
+      "loss": 0.0111,
+      "step": 131190
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369954025970284e-05,
+      "loss": 0.01,
+      "step": 131200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.936990598623181e-05,
+      "loss": 0.0101,
+      "step": 131210
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.936985794649334e-05,
+      "loss": 0.0099,
+      "step": 131220
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.936980990675487e-05,
+      "loss": 0.0055,
+      "step": 131230
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369761867016398e-05,
+      "loss": 0.0076,
+      "step": 131240
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369713827277924e-05,
+      "loss": 0.0083,
+      "step": 131250
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369665787539453e-05,
+      "loss": 0.0109,
+      "step": 131260
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369617747800982e-05,
+      "loss": 0.0056,
+      "step": 131270
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369569708062512e-05,
+      "loss": 0.0127,
+      "step": 131280
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369521668324038e-05,
+      "loss": 0.0106,
+      "step": 131290
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369473628585567e-05,
+      "loss": 0.0087,
+      "step": 131300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369425588847096e-05,
+      "loss": 0.0113,
+      "step": 131310
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369377549108622e-05,
+      "loss": 0.0084,
+      "step": 131320
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.936932950937015e-05,
+      "loss": 0.0132,
+      "step": 131330
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.936928146963168e-05,
+      "loss": 0.0109,
+      "step": 131340
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.936923342989321e-05,
+      "loss": 0.0093,
+      "step": 131350
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369185390154736e-05,
+      "loss": 0.011,
+      "step": 131360
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369137350416265e-05,
+      "loss": 0.009,
+      "step": 131370
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9369089310677794e-05,
+      "loss": 0.0047,
+      "step": 131380
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.936904127093932e-05,
+      "loss": 0.0135,
+      "step": 131390
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.936899323120085e-05,
+      "loss": 0.0087,
+      "step": 131400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.936894519146238e-05,
+      "loss": 0.0149,
+      "step": 131410
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9368897151723908e-05,
+      "loss": 0.0102,
+      "step": 131420
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9368849111985434e-05,
+      "loss": 0.0108,
+      "step": 131430
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9368801072246963e-05,
+      "loss": 0.0091,
+      "step": 131440
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9368753032508492e-05,
+      "loss": 0.0092,
+      "step": 131450
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9368704992770018e-05,
+      "loss": 0.013,
+      "step": 131460
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9368656953031547e-05,
+      "loss": 0.008,
+      "step": 131470
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9368608913293077e-05,
+      "loss": 0.0093,
+      "step": 131480
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9368560873554606e-05,
+      "loss": 0.0091,
+      "step": 131490
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9368512833816132e-05,
+      "loss": 0.0096,
+      "step": 131500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936846479407766e-05,
+      "loss": 0.009,
+      "step": 131510
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936841675433919e-05,
+      "loss": 0.0096,
+      "step": 131520
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936836871460072e-05,
+      "loss": 0.0085,
+      "step": 131530
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9368320674862246e-05,
+      "loss": 0.0098,
+      "step": 131540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9368272635123775e-05,
+      "loss": 0.0113,
+      "step": 131550
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9368224595385304e-05,
+      "loss": 0.0084,
+      "step": 131560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9368176555646833e-05,
+      "loss": 0.014,
+      "step": 131570
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9368128515908363e-05,
+      "loss": 0.0075,
+      "step": 131580
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9368080476169892e-05,
+      "loss": 0.0106,
+      "step": 131590
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9368032436431418e-05,
+      "loss": 0.0104,
+      "step": 131600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367984396692947e-05,
+      "loss": 0.0088,
+      "step": 131610
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367936356954476e-05,
+      "loss": 0.01,
+      "step": 131620
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367888317216002e-05,
+      "loss": 0.012,
+      "step": 131630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936784027747753e-05,
+      "loss": 0.0098,
+      "step": 131640
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936779223773906e-05,
+      "loss": 0.0088,
+      "step": 131650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936774419800059e-05,
+      "loss": 0.008,
+      "step": 131660
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367696158262116e-05,
+      "loss": 0.0085,
+      "step": 131670
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367648118523645e-05,
+      "loss": 0.0108,
+      "step": 131680
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367600078785174e-05,
+      "loss": 0.0117,
+      "step": 131690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.93675520390467e-05,
+      "loss": 0.0109,
+      "step": 131700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936750399930823e-05,
+      "loss": 0.0063,
+      "step": 131710
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936745595956976e-05,
+      "loss": 0.009,
+      "step": 131720
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367407919831288e-05,
+      "loss": 0.0087,
+      "step": 131730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367359880092814e-05,
+      "loss": 0.0081,
+      "step": 131740
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367311840354343e-05,
+      "loss": 0.0124,
+      "step": 131750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367263800615873e-05,
+      "loss": 0.0106,
+      "step": 131760
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367215760877402e-05,
+      "loss": 0.0093,
+      "step": 131770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367167721138928e-05,
+      "loss": 0.0061,
+      "step": 131780
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367119681400457e-05,
+      "loss": 0.0101,
+      "step": 131790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367071641661986e-05,
+      "loss": 0.0077,
+      "step": 131800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9367023601923512e-05,
+      "loss": 0.0106,
+      "step": 131810
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936697556218504e-05,
+      "loss": 0.0103,
+      "step": 131820
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936692752244657e-05,
+      "loss": 0.0099,
+      "step": 131830
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.93668794827081e-05,
+      "loss": 0.0082,
+      "step": 131840
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9366831442969626e-05,
+      "loss": 0.0086,
+      "step": 131850
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9366783403231155e-05,
+      "loss": 0.0118,
+      "step": 131860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9366735363492684e-05,
+      "loss": 0.0082,
+      "step": 131870
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936668732375421e-05,
+      "loss": 0.0069,
+      "step": 131880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936663928401574e-05,
+      "loss": 0.0099,
+      "step": 131890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936659124427727e-05,
+      "loss": 0.0099,
+      "step": 131900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9366543204538798e-05,
+      "loss": 0.0098,
+      "step": 131910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9366495164800324e-05,
+      "loss": 0.0087,
+      "step": 131920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9366447125061853e-05,
+      "loss": 0.0104,
+      "step": 131930
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9366399085323383e-05,
+      "loss": 0.008,
+      "step": 131940
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9366351045584912e-05,
+      "loss": 0.0101,
+      "step": 131950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9366303005846438e-05,
+      "loss": 0.0111,
+      "step": 131960
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9366254966107967e-05,
+      "loss": 0.0079,
+      "step": 131970
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9366206926369496e-05,
+      "loss": 0.0091,
+      "step": 131980
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9366158886631022e-05,
+      "loss": 0.0061,
+      "step": 131990
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936611084689255e-05,
+      "loss": 0.0221,
+      "step": 132000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936606280715408e-05,
+      "loss": 0.0081,
+      "step": 132010
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936601476741561e-05,
+      "loss": 0.012,
+      "step": 132020
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9365966727677136e-05,
+      "loss": 0.0147,
+      "step": 132030
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9365918687938665e-05,
+      "loss": 0.013,
+      "step": 132040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9365870648200194e-05,
+      "loss": 0.008,
+      "step": 132050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936582260846172e-05,
+      "loss": 0.0085,
+      "step": 132060
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936577456872325e-05,
+      "loss": 0.0169,
+      "step": 132070
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936572652898478e-05,
+      "loss": 0.008,
+      "step": 132080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9365678489246308e-05,
+      "loss": 0.0111,
+      "step": 132090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9365630449507834e-05,
+      "loss": 0.01,
+      "step": 132100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9365582409769363e-05,
+      "loss": 0.0083,
+      "step": 132110
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9365534370030892e-05,
+      "loss": 0.0092,
+      "step": 132120
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9365486330292422e-05,
+      "loss": 0.0114,
+      "step": 132130
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9365438290553948e-05,
+      "loss": 0.0095,
+      "step": 132140
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9365390250815477e-05,
+      "loss": 0.0142,
+      "step": 132150
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9365342211077006e-05,
+      "loss": 0.0095,
+      "step": 132160
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9365294171338532e-05,
+      "loss": 0.0111,
+      "step": 132170
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.936524613160006e-05,
+      "loss": 0.0091,
+      "step": 132180
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936519809186159e-05,
+      "loss": 0.0098,
+      "step": 132190
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936515005212312e-05,
+      "loss": 0.0088,
+      "step": 132200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9365102012384646e-05,
+      "loss": 0.0085,
+      "step": 132210
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9365053972646175e-05,
+      "loss": 0.0097,
+      "step": 132220
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9365005932907704e-05,
+      "loss": 0.0079,
+      "step": 132230
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936495789316923e-05,
+      "loss": 0.0101,
+      "step": 132240
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936490985343076e-05,
+      "loss": 0.0109,
+      "step": 132250
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936486181369229e-05,
+      "loss": 0.0098,
+      "step": 132260
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9364813773953818e-05,
+      "loss": 0.0084,
+      "step": 132270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9364765734215344e-05,
+      "loss": 0.0103,
+      "step": 132280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9364717694476873e-05,
+      "loss": 0.013,
+      "step": 132290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9364669654738402e-05,
+      "loss": 0.0142,
+      "step": 132300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936462161499993e-05,
+      "loss": 0.0122,
+      "step": 132310
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9364573575261457e-05,
+      "loss": 0.0087,
+      "step": 132320
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9364525535522987e-05,
+      "loss": 0.0098,
+      "step": 132330
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9364477495784516e-05,
+      "loss": 0.0109,
+      "step": 132340
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9364429456046042e-05,
+      "loss": 0.0119,
+      "step": 132350
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936438141630757e-05,
+      "loss": 0.0109,
+      "step": 132360
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.93643333765691e-05,
+      "loss": 0.0074,
+      "step": 132370
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936428533683063e-05,
+      "loss": 0.0117,
+      "step": 132380
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9364237297092156e-05,
+      "loss": 0.0131,
+      "step": 132390
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9364189257353685e-05,
+      "loss": 0.0076,
+      "step": 132400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9364141217615214e-05,
+      "loss": 0.0091,
+      "step": 132410
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936409317787674e-05,
+      "loss": 0.0104,
+      "step": 132420
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936404513813827e-05,
+      "loss": 0.0106,
+      "step": 132430
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.93639970983998e-05,
+      "loss": 0.0082,
+      "step": 132440
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363949058661328e-05,
+      "loss": 0.0101,
+      "step": 132450
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363901018922854e-05,
+      "loss": 0.0075,
+      "step": 132460
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363852979184383e-05,
+      "loss": 0.0094,
+      "step": 132470
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363804939445912e-05,
+      "loss": 0.0059,
+      "step": 132480
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936375689970744e-05,
+      "loss": 0.0104,
+      "step": 132490
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363708859968967e-05,
+      "loss": 0.0108,
+      "step": 132500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363660820230497e-05,
+      "loss": 0.0098,
+      "step": 132510
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363612780492026e-05,
+      "loss": 0.0089,
+      "step": 132520
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363564740753552e-05,
+      "loss": 0.0092,
+      "step": 132530
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936351670101508e-05,
+      "loss": 0.0085,
+      "step": 132540
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936346866127661e-05,
+      "loss": 0.0071,
+      "step": 132550
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936342062153814e-05,
+      "loss": 0.0112,
+      "step": 132560
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363372581799666e-05,
+      "loss": 0.0118,
+      "step": 132570
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363324542061195e-05,
+      "loss": 0.008,
+      "step": 132580
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363276502322724e-05,
+      "loss": 0.0096,
+      "step": 132590
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936322846258425e-05,
+      "loss": 0.009,
+      "step": 132600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936318042284578e-05,
+      "loss": 0.009,
+      "step": 132610
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936313238310731e-05,
+      "loss": 0.0091,
+      "step": 132620
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363084343368838e-05,
+      "loss": 0.0179,
+      "step": 132630
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9363036303630364e-05,
+      "loss": 0.0094,
+      "step": 132640
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362988263891893e-05,
+      "loss": 0.0084,
+      "step": 132650
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362940224153422e-05,
+      "loss": 0.0116,
+      "step": 132660
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936289218441495e-05,
+      "loss": 0.0108,
+      "step": 132670
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362844144676477e-05,
+      "loss": 0.0131,
+      "step": 132680
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362796104938007e-05,
+      "loss": 0.0079,
+      "step": 132690
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362748065199536e-05,
+      "loss": 0.0073,
+      "step": 132700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362700025461062e-05,
+      "loss": 0.0059,
+      "step": 132710
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936265198572259e-05,
+      "loss": 0.0086,
+      "step": 132720
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936260394598412e-05,
+      "loss": 0.0101,
+      "step": 132730
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936255590624565e-05,
+      "loss": 0.0097,
+      "step": 132740
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362507866507175e-05,
+      "loss": 0.0107,
+      "step": 132750
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362459826768705e-05,
+      "loss": 0.0085,
+      "step": 132760
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362411787030234e-05,
+      "loss": 0.0112,
+      "step": 132770
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936236374729176e-05,
+      "loss": 0.0117,
+      "step": 132780
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936231570755329e-05,
+      "loss": 0.008,
+      "step": 132790
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936226766781482e-05,
+      "loss": 0.0069,
+      "step": 132800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362219628076348e-05,
+      "loss": 0.0072,
+      "step": 132810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362171588337874e-05,
+      "loss": 0.012,
+      "step": 132820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362123548599403e-05,
+      "loss": 0.0113,
+      "step": 132830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9362075508860932e-05,
+      "loss": 0.0129,
+      "step": 132840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.936202746912246e-05,
+      "loss": 0.0092,
+      "step": 132850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9361979429383987e-05,
+      "loss": 0.011,
+      "step": 132860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9361931389645517e-05,
+      "loss": 0.0091,
+      "step": 132870
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9361883349907046e-05,
+      "loss": 0.0135,
+      "step": 132880
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936183531016857e-05,
+      "loss": 0.0122,
+      "step": 132890
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.93617872704301e-05,
+      "loss": 0.009,
+      "step": 132900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936173923069163e-05,
+      "loss": 0.0087,
+      "step": 132910
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936169119095316e-05,
+      "loss": 0.0118,
+      "step": 132920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9361643151214685e-05,
+      "loss": 0.0107,
+      "step": 132930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9361595111476215e-05,
+      "loss": 0.0085,
+      "step": 132940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9361547071737744e-05,
+      "loss": 0.0083,
+      "step": 132950
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936149903199927e-05,
+      "loss": 0.0094,
+      "step": 132960
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.93614509922608e-05,
+      "loss": 0.0129,
+      "step": 132970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936140295252233e-05,
+      "loss": 0.0088,
+      "step": 132980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9361354912783858e-05,
+      "loss": 0.0091,
+      "step": 132990
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9361306873045383e-05,
+      "loss": 0.0105,
+      "step": 133000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9361258833306913e-05,
+      "loss": 0.0135,
+      "step": 133010
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9361210793568442e-05,
+      "loss": 0.008,
+      "step": 133020
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9361162753829968e-05,
+      "loss": 0.0071,
+      "step": 133030
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9361114714091497e-05,
+      "loss": 0.0121,
+      "step": 133040
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9361066674353026e-05,
+      "loss": 0.0149,
+      "step": 133050
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9361018634614556e-05,
+      "loss": 0.0098,
+      "step": 133060
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936097059487608e-05,
+      "loss": 0.0109,
+      "step": 133070
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936092255513761e-05,
+      "loss": 0.0089,
+      "step": 133080
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936087451539914e-05,
+      "loss": 0.0112,
+      "step": 133090
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936082647566067e-05,
+      "loss": 0.0117,
+      "step": 133100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9360778435922195e-05,
+      "loss": 0.0076,
+      "step": 133110
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9360730396183725e-05,
+      "loss": 0.0101,
+      "step": 133120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9360682356445254e-05,
+      "loss": 0.0084,
+      "step": 133130
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936063431670678e-05,
+      "loss": 0.0081,
+      "step": 133140
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936058627696831e-05,
+      "loss": 0.0086,
+      "step": 133150
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9360538237229838e-05,
+      "loss": 0.0084,
+      "step": 133160
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9360490197491367e-05,
+      "loss": 0.0135,
+      "step": 133170
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9360442157752893e-05,
+      "loss": 0.0146,
+      "step": 133180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9360394118014423e-05,
+      "loss": 0.008,
+      "step": 133190
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9360346078275952e-05,
+      "loss": 0.0064,
+      "step": 133200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9360298038537478e-05,
+      "loss": 0.0085,
+      "step": 133210
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9360249998799007e-05,
+      "loss": 0.0119,
+      "step": 133220
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9360201959060536e-05,
+      "loss": 0.0064,
+      "step": 133230
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9360153919322066e-05,
+      "loss": 0.01,
+      "step": 133240
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936010587958359e-05,
+      "loss": 0.0132,
+      "step": 133250
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936005783984512e-05,
+      "loss": 0.0082,
+      "step": 133260
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.936000980010665e-05,
+      "loss": 0.0104,
+      "step": 133270
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.935996176036818e-05,
+      "loss": 0.0147,
+      "step": 133280
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359913720629705e-05,
+      "loss": 0.0103,
+      "step": 133290
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359865680891234e-05,
+      "loss": 0.0101,
+      "step": 133300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359817641152764e-05,
+      "loss": 0.0121,
+      "step": 133310
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.935976960141429e-05,
+      "loss": 0.0089,
+      "step": 133320
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.935972156167582e-05,
+      "loss": 0.0088,
+      "step": 133330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359673521937348e-05,
+      "loss": 0.005,
+      "step": 133340
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359625482198877e-05,
+      "loss": 0.0134,
+      "step": 133350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359577442460403e-05,
+      "loss": 0.0105,
+      "step": 133360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359529402721933e-05,
+      "loss": 0.0087,
+      "step": 133370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359481362983462e-05,
+      "loss": 0.0107,
+      "step": 133380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359433323244988e-05,
+      "loss": 0.01,
+      "step": 133390
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359385283506517e-05,
+      "loss": 0.012,
+      "step": 133400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359337243768046e-05,
+      "loss": 0.0108,
+      "step": 133410
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359289204029576e-05,
+      "loss": 0.0086,
+      "step": 133420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.93592411642911e-05,
+      "loss": 0.0064,
+      "step": 133430
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.935919312455263e-05,
+      "loss": 0.0085,
+      "step": 133440
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.935914508481416e-05,
+      "loss": 0.0079,
+      "step": 133450
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.935909704507569e-05,
+      "loss": 0.0096,
+      "step": 133460
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359049005337215e-05,
+      "loss": 0.0086,
+      "step": 133470
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9359000965598744e-05,
+      "loss": 0.0091,
+      "step": 133480
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9358952925860274e-05,
+      "loss": 0.0109,
+      "step": 133490
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.93589048861218e-05,
+      "loss": 0.0094,
+      "step": 133500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.935885684638333e-05,
+      "loss": 0.0088,
+      "step": 133510
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9358808806644858e-05,
+      "loss": 0.0089,
+      "step": 133520
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9358760766906387e-05,
+      "loss": 0.0082,
+      "step": 133530
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9358712727167913e-05,
+      "loss": 0.0114,
+      "step": 133540
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9358664687429442e-05,
+      "loss": 0.0077,
+      "step": 133550
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9358616647690972e-05,
+      "loss": 0.0092,
+      "step": 133560
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9358568607952498e-05,
+      "loss": 0.0084,
+      "step": 133570
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9358520568214027e-05,
+      "loss": 0.0101,
+      "step": 133580
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9358472528475556e-05,
+      "loss": 0.0089,
+      "step": 133590
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9358424488737085e-05,
+      "loss": 0.008,
+      "step": 133600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935837644899861e-05,
+      "loss": 0.0113,
+      "step": 133610
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935832840926014e-05,
+      "loss": 0.0097,
+      "step": 133620
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935828036952167e-05,
+      "loss": 0.0094,
+      "step": 133630
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.93582323297832e-05,
+      "loss": 0.0064,
+      "step": 133640
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9358184290044725e-05,
+      "loss": 0.0129,
+      "step": 133650
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9358136250306254e-05,
+      "loss": 0.0081,
+      "step": 133660
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9358088210567784e-05,
+      "loss": 0.0084,
+      "step": 133670
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935804017082931e-05,
+      "loss": 0.0112,
+      "step": 133680
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935799213109084e-05,
+      "loss": 0.0152,
+      "step": 133690
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357944091352368e-05,
+      "loss": 0.0076,
+      "step": 133700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357896051613897e-05,
+      "loss": 0.0085,
+      "step": 133710
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357848011875423e-05,
+      "loss": 0.0084,
+      "step": 133720
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357799972136952e-05,
+      "loss": 0.0101,
+      "step": 133730
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935775193239848e-05,
+      "loss": 0.011,
+      "step": 133740
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357703892660008e-05,
+      "loss": 0.0106,
+      "step": 133750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357655852921537e-05,
+      "loss": 0.0111,
+      "step": 133760
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357607813183066e-05,
+      "loss": 0.0107,
+      "step": 133770
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357559773444595e-05,
+      "loss": 0.0092,
+      "step": 133780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935751173370612e-05,
+      "loss": 0.0075,
+      "step": 133790
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935746369396765e-05,
+      "loss": 0.0124,
+      "step": 133800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935741565422918e-05,
+      "loss": 0.0141,
+      "step": 133810
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935736761449071e-05,
+      "loss": 0.0107,
+      "step": 133820
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357319574752235e-05,
+      "loss": 0.0118,
+      "step": 133830
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357271535013764e-05,
+      "loss": 0.01,
+      "step": 133840
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357223495275293e-05,
+      "loss": 0.0083,
+      "step": 133850
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935717545553682e-05,
+      "loss": 0.0095,
+      "step": 133860
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935712741579835e-05,
+      "loss": 0.0092,
+      "step": 133870
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357079376059878e-05,
+      "loss": 0.0071,
+      "step": 133880
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9357031336321407e-05,
+      "loss": 0.011,
+      "step": 133890
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356983296582933e-05,
+      "loss": 0.0095,
+      "step": 133900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356935256844462e-05,
+      "loss": 0.0111,
+      "step": 133910
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935688721710599e-05,
+      "loss": 0.0076,
+      "step": 133920
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356839177367517e-05,
+      "loss": 0.011,
+      "step": 133930
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356791137629047e-05,
+      "loss": 0.011,
+      "step": 133940
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356743097890576e-05,
+      "loss": 0.0108,
+      "step": 133950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356695058152105e-05,
+      "loss": 0.0093,
+      "step": 133960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935664701841363e-05,
+      "loss": 0.0168,
+      "step": 133970
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935659897867516e-05,
+      "loss": 0.0067,
+      "step": 133980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935655093893669e-05,
+      "loss": 0.0099,
+      "step": 133990
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935650289919822e-05,
+      "loss": 0.011,
+      "step": 134000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356454859459745e-05,
+      "loss": 0.0107,
+      "step": 134010
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356406819721274e-05,
+      "loss": 0.0087,
+      "step": 134020
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356358779982803e-05,
+      "loss": 0.0091,
+      "step": 134030
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935631074024433e-05,
+      "loss": 0.0077,
+      "step": 134040
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935626270050586e-05,
+      "loss": 0.0085,
+      "step": 134050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356214660767388e-05,
+      "loss": 0.0104,
+      "step": 134060
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356166621028917e-05,
+      "loss": 0.0078,
+      "step": 134070
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356118581290443e-05,
+      "loss": 0.0077,
+      "step": 134080
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9356070541551972e-05,
+      "loss": 0.0111,
+      "step": 134090
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.93560225018135e-05,
+      "loss": 0.0078,
+      "step": 134100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9355974462075027e-05,
+      "loss": 0.0116,
+      "step": 134110
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9355926422336557e-05,
+      "loss": 0.0114,
+      "step": 134120
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935587838259809e-05,
+      "loss": 0.0104,
+      "step": 134130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9355830342859615e-05,
+      "loss": 0.0096,
+      "step": 134140
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9355782303121144e-05,
+      "loss": 0.0099,
+      "step": 134150
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9355734263382674e-05,
+      "loss": 0.0099,
+      "step": 134160
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.93556862236442e-05,
+      "loss": 0.0108,
+      "step": 134170
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.935563818390573e-05,
+      "loss": 0.0094,
+      "step": 134180
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9355590144167258e-05,
+      "loss": 0.0102,
+      "step": 134190
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9355542104428787e-05,
+      "loss": 0.0102,
+      "step": 134200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9355494064690313e-05,
+      "loss": 0.0119,
+      "step": 134210
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9355446024951843e-05,
+      "loss": 0.0076,
+      "step": 134220
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9355397985213372e-05,
+      "loss": 0.0118,
+      "step": 134230
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.93553499454749e-05,
+      "loss": 0.0104,
+      "step": 134240
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9355301905736427e-05,
+      "loss": 0.0097,
+      "step": 134250
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9355253865997956e-05,
+      "loss": 0.0139,
+      "step": 134260
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9355205826259486e-05,
+      "loss": 0.0101,
+      "step": 134270
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935515778652101e-05,
+      "loss": 0.0076,
+      "step": 134280
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935510974678254e-05,
+      "loss": 0.0113,
+      "step": 134290
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935506170704407e-05,
+      "loss": 0.012,
+      "step": 134300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.93550136673056e-05,
+      "loss": 0.011,
+      "step": 134310
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354965627567125e-05,
+      "loss": 0.0092,
+      "step": 134320
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354917587828654e-05,
+      "loss": 0.0115,
+      "step": 134330
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354869548090184e-05,
+      "loss": 0.0081,
+      "step": 134340
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935482150835171e-05,
+      "loss": 0.0081,
+      "step": 134350
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935477346861324e-05,
+      "loss": 0.008,
+      "step": 134360
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354725428874768e-05,
+      "loss": 0.0092,
+      "step": 134370
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354677389136297e-05,
+      "loss": 0.0123,
+      "step": 134380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354629349397823e-05,
+      "loss": 0.0179,
+      "step": 134390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354581309659352e-05,
+      "loss": 0.009,
+      "step": 134400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354533269920882e-05,
+      "loss": 0.0125,
+      "step": 134410
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354485230182408e-05,
+      "loss": 0.0147,
+      "step": 134420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354437190443937e-05,
+      "loss": 0.0116,
+      "step": 134430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354389150705466e-05,
+      "loss": 0.0103,
+      "step": 134440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354341110966995e-05,
+      "loss": 0.0106,
+      "step": 134450
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935429307122852e-05,
+      "loss": 0.0096,
+      "step": 134460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935424503149005e-05,
+      "loss": 0.0102,
+      "step": 134470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935419699175158e-05,
+      "loss": 0.008,
+      "step": 134480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935414895201311e-05,
+      "loss": 0.0082,
+      "step": 134490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354100912274635e-05,
+      "loss": 0.0102,
+      "step": 134500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354052872536164e-05,
+      "loss": 0.0097,
+      "step": 134510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9354004832797694e-05,
+      "loss": 0.0103,
+      "step": 134520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935395679305922e-05,
+      "loss": 0.0095,
+      "step": 134530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935390875332075e-05,
+      "loss": 0.0054,
+      "step": 134540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9353860713582278e-05,
+      "loss": 0.0073,
+      "step": 134550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9353812673843807e-05,
+      "loss": 0.0122,
+      "step": 134560
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9353764634105333e-05,
+      "loss": 0.0097,
+      "step": 134570
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9353716594366862e-05,
+      "loss": 0.0091,
+      "step": 134580
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935366855462839e-05,
+      "loss": 0.0076,
+      "step": 134590
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9353620514889918e-05,
+      "loss": 0.0074,
+      "step": 134600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9353572475151447e-05,
+      "loss": 0.01,
+      "step": 134610
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9353524435412976e-05,
+      "loss": 0.0126,
+      "step": 134620
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9353476395674505e-05,
+      "loss": 0.0123,
+      "step": 134630
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935342835593603e-05,
+      "loss": 0.0084,
+      "step": 134640
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935338031619756e-05,
+      "loss": 0.0114,
+      "step": 134650
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935333227645909e-05,
+      "loss": 0.0101,
+      "step": 134660
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935328423672062e-05,
+      "loss": 0.0122,
+      "step": 134670
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9353236196982145e-05,
+      "loss": 0.0086,
+      "step": 134680
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9353188157243674e-05,
+      "loss": 0.0067,
+      "step": 134690
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9353140117505203e-05,
+      "loss": 0.0071,
+      "step": 134700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935309207776673e-05,
+      "loss": 0.0074,
+      "step": 134710
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935304403802826e-05,
+      "loss": 0.0127,
+      "step": 134720
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352995998289788e-05,
+      "loss": 0.0098,
+      "step": 134730
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352947958551317e-05,
+      "loss": 0.0088,
+      "step": 134740
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352899918812843e-05,
+      "loss": 0.0095,
+      "step": 134750
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352851879074372e-05,
+      "loss": 0.0067,
+      "step": 134760
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.93528038393359e-05,
+      "loss": 0.0075,
+      "step": 134770
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352755799597427e-05,
+      "loss": 0.0093,
+      "step": 134780
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352707759858957e-05,
+      "loss": 0.0104,
+      "step": 134790
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352659720120486e-05,
+      "loss": 0.0105,
+      "step": 134800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352611680382015e-05,
+      "loss": 0.0094,
+      "step": 134810
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935256364064354e-05,
+      "loss": 0.0099,
+      "step": 134820
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935251560090507e-05,
+      "loss": 0.0078,
+      "step": 134830
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.93524675611666e-05,
+      "loss": 0.009,
+      "step": 134840
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935241952142813e-05,
+      "loss": 0.0104,
+      "step": 134850
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352371481689655e-05,
+      "loss": 0.0076,
+      "step": 134860
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352323441951184e-05,
+      "loss": 0.0092,
+      "step": 134870
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352275402212713e-05,
+      "loss": 0.0104,
+      "step": 134880
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935222736247424e-05,
+      "loss": 0.0099,
+      "step": 134890
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935217932273577e-05,
+      "loss": 0.0089,
+      "step": 134900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352131282997298e-05,
+      "loss": 0.0073,
+      "step": 134910
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352083243258827e-05,
+      "loss": 0.0114,
+      "step": 134920
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9352035203520353e-05,
+      "loss": 0.0099,
+      "step": 134930
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9351987163781882e-05,
+      "loss": 0.011,
+      "step": 134940
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935193912404341e-05,
+      "loss": 0.0094,
+      "step": 134950
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9351891084304937e-05,
+      "loss": 0.0126,
+      "step": 134960
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9351843044566467e-05,
+      "loss": 0.0089,
+      "step": 134970
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9351795004827996e-05,
+      "loss": 0.0089,
+      "step": 134980
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9351746965089525e-05,
+      "loss": 0.0108,
+      "step": 134990
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935169892535105e-05,
+      "loss": 0.0078,
+      "step": 135000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935165088561258e-05,
+      "loss": 0.0083,
+      "step": 135010
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935160284587411e-05,
+      "loss": 0.0106,
+      "step": 135020
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935155480613564e-05,
+      "loss": 0.0067,
+      "step": 135030
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9351506766397165e-05,
+      "loss": 0.0082,
+      "step": 135040
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9351458726658694e-05,
+      "loss": 0.0079,
+      "step": 135050
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9351410686920223e-05,
+      "loss": 0.0098,
+      "step": 135060
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935136264718175e-05,
+      "loss": 0.0087,
+      "step": 135070
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935131460744328e-05,
+      "loss": 0.0108,
+      "step": 135080
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9351266567704808e-05,
+      "loss": 0.0092,
+      "step": 135090
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9351218527966337e-05,
+      "loss": 0.0094,
+      "step": 135100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9351170488227863e-05,
+      "loss": 0.0116,
+      "step": 135110
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9351122448489392e-05,
+      "loss": 0.0087,
+      "step": 135120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935107440875092e-05,
+      "loss": 0.0076,
+      "step": 135130
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9351026369012447e-05,
+      "loss": 0.0093,
+      "step": 135140
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350978329273977e-05,
+      "loss": 0.0071,
+      "step": 135150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350930289535506e-05,
+      "loss": 0.0081,
+      "step": 135160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350882249797035e-05,
+      "loss": 0.0083,
+      "step": 135170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935083421005856e-05,
+      "loss": 0.008,
+      "step": 135180
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935078617032009e-05,
+      "loss": 0.0101,
+      "step": 135190
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935073813058162e-05,
+      "loss": 0.0119,
+      "step": 135200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935069009084315e-05,
+      "loss": 0.0105,
+      "step": 135210
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350642051104675e-05,
+      "loss": 0.008,
+      "step": 135220
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350594011366204e-05,
+      "loss": 0.0079,
+      "step": 135230
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350545971627733e-05,
+      "loss": 0.0093,
+      "step": 135240
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935049793188926e-05,
+      "loss": 0.0104,
+      "step": 135250
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935044989215079e-05,
+      "loss": 0.0084,
+      "step": 135260
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350401852412318e-05,
+      "loss": 0.0093,
+      "step": 135270
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350353812673847e-05,
+      "loss": 0.0068,
+      "step": 135280
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350305772935373e-05,
+      "loss": 0.0115,
+      "step": 135290
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350257733196902e-05,
+      "loss": 0.0104,
+      "step": 135300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935020969345843e-05,
+      "loss": 0.0101,
+      "step": 135310
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350161653719957e-05,
+      "loss": 0.0122,
+      "step": 135320
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350113613981486e-05,
+      "loss": 0.0116,
+      "step": 135330
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350065574243016e-05,
+      "loss": 0.009,
+      "step": 135340
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9350017534504545e-05,
+      "loss": 0.0093,
+      "step": 135350
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.934996949476607e-05,
+      "loss": 0.0074,
+      "step": 135360
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.93499214550276e-05,
+      "loss": 0.0073,
+      "step": 135370
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.934987341528913e-05,
+      "loss": 0.0099,
+      "step": 135380
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.934982537555066e-05,
+      "loss": 0.0096,
+      "step": 135390
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9349777335812185e-05,
+      "loss": 0.0104,
+      "step": 135400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9349729296073714e-05,
+      "loss": 0.0081,
+      "step": 135410
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9349681256335243e-05,
+      "loss": 0.0067,
+      "step": 135420
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.934963321659677e-05,
+      "loss": 0.0119,
+      "step": 135430
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9349585176858298e-05,
+      "loss": 0.0086,
+      "step": 135440
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9349537137119828e-05,
+      "loss": 0.0119,
+      "step": 135450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9349489097381357e-05,
+      "loss": 0.0093,
+      "step": 135460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9349441057642883e-05,
+      "loss": 0.0089,
+      "step": 135470
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9349393017904412e-05,
+      "loss": 0.0101,
+      "step": 135480
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.934934497816594e-05,
+      "loss": 0.0101,
+      "step": 135490
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9349296938427467e-05,
+      "loss": 0.0114,
+      "step": 135500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9349248898688996e-05,
+      "loss": 0.0104,
+      "step": 135510
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9349200858950526e-05,
+      "loss": 0.0101,
+      "step": 135520
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9349152819212055e-05,
+      "loss": 0.0108,
+      "step": 135530
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.934910477947358e-05,
+      "loss": 0.0094,
+      "step": 135540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.934905673973511e-05,
+      "loss": 0.0093,
+      "step": 135550
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.934900869999664e-05,
+      "loss": 0.0114,
+      "step": 135560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.934896066025817e-05,
+      "loss": 0.0102,
+      "step": 135570
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9348912620519695e-05,
+      "loss": 0.0091,
+      "step": 135580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9348864580781224e-05,
+      "loss": 0.0094,
+      "step": 135590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9348816541042753e-05,
+      "loss": 0.0079,
+      "step": 135600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.934876850130428e-05,
+      "loss": 0.0138,
+      "step": 135610
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9348720461565808e-05,
+      "loss": 0.0077,
+      "step": 135620
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9348672421827337e-05,
+      "loss": 0.0078,
+      "step": 135630
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9348624382088867e-05,
+      "loss": 0.0094,
+      "step": 135640
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9348576342350393e-05,
+      "loss": 0.0106,
+      "step": 135650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9348528302611922e-05,
+      "loss": 0.0084,
+      "step": 135660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934848026287345e-05,
+      "loss": 0.0121,
+      "step": 135670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9348432223134977e-05,
+      "loss": 0.0115,
+      "step": 135680
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9348388987370355e-05,
+      "loss": 0.0102,
+      "step": 135690
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9348340947631884e-05,
+      "loss": 0.01,
+      "step": 135700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934829290789341e-05,
+      "loss": 0.0115,
+      "step": 135710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934824486815494e-05,
+      "loss": 0.0132,
+      "step": 135720
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934819682841647e-05,
+      "loss": 0.0105,
+      "step": 135730
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9348148788677995e-05,
+      "loss": 0.009,
+      "step": 135740
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9348100748939524e-05,
+      "loss": 0.0096,
+      "step": 135750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9348052709201053e-05,
+      "loss": 0.0108,
+      "step": 135760
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9348004669462582e-05,
+      "loss": 0.0081,
+      "step": 135770
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934795662972411e-05,
+      "loss": 0.0099,
+      "step": 135780
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347908589985638e-05,
+      "loss": 0.0076,
+      "step": 135790
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347860550247167e-05,
+      "loss": 0.0078,
+      "step": 135800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347812510508696e-05,
+      "loss": 0.0085,
+      "step": 135810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347764470770222e-05,
+      "loss": 0.0076,
+      "step": 135820
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934771643103175e-05,
+      "loss": 0.0126,
+      "step": 135830
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934766839129328e-05,
+      "loss": 0.008,
+      "step": 135840
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347620351554806e-05,
+      "loss": 0.0097,
+      "step": 135850
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347572311816336e-05,
+      "loss": 0.0094,
+      "step": 135860
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347524272077865e-05,
+      "loss": 0.0099,
+      "step": 135870
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347476232339394e-05,
+      "loss": 0.0065,
+      "step": 135880
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934742819260092e-05,
+      "loss": 0.0086,
+      "step": 135890
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934738015286245e-05,
+      "loss": 0.0105,
+      "step": 135900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934733211312398e-05,
+      "loss": 0.0057,
+      "step": 135910
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347284073385505e-05,
+      "loss": 0.0092,
+      "step": 135920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347236033647034e-05,
+      "loss": 0.01,
+      "step": 135930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347187993908563e-05,
+      "loss": 0.0111,
+      "step": 135940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347139954170092e-05,
+      "loss": 0.0094,
+      "step": 135950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347091914431618e-05,
+      "loss": 0.0076,
+      "step": 135960
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9347043874693148e-05,
+      "loss": 0.0115,
+      "step": 135970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346995834954677e-05,
+      "loss": 0.0112,
+      "step": 135980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346947795216203e-05,
+      "loss": 0.0097,
+      "step": 135990
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346899755477732e-05,
+      "loss": 0.0091,
+      "step": 136000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934685171573926e-05,
+      "loss": 0.0104,
+      "step": 136010
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934680367600079e-05,
+      "loss": 0.0094,
+      "step": 136020
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346755636262316e-05,
+      "loss": 0.0072,
+      "step": 136030
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346707596523846e-05,
+      "loss": 0.0087,
+      "step": 136040
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346659556785375e-05,
+      "loss": 0.0118,
+      "step": 136050
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346611517046904e-05,
+      "loss": 0.0097,
+      "step": 136060
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934656347730843e-05,
+      "loss": 0.0067,
+      "step": 136070
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934651543756996e-05,
+      "loss": 0.0083,
+      "step": 136080
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934646739783149e-05,
+      "loss": 0.0107,
+      "step": 136090
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346419358093014e-05,
+      "loss": 0.0094,
+      "step": 136100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346371318354544e-05,
+      "loss": 0.0073,
+      "step": 136110
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346323278616073e-05,
+      "loss": 0.0097,
+      "step": 136120
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346275238877602e-05,
+      "loss": 0.0136,
+      "step": 136130
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346227199139128e-05,
+      "loss": 0.0103,
+      "step": 136140
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346179159400657e-05,
+      "loss": 0.011,
+      "step": 136150
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346131119662187e-05,
+      "loss": 0.0115,
+      "step": 136160
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346083079923713e-05,
+      "loss": 0.0065,
+      "step": 136170
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9346035040185242e-05,
+      "loss": 0.0109,
+      "step": 136180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934598700044677e-05,
+      "loss": 0.0087,
+      "step": 136190
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.93459389607083e-05,
+      "loss": 0.0057,
+      "step": 136200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9345890920969826e-05,
+      "loss": 0.0091,
+      "step": 136210
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9345842881231356e-05,
+      "loss": 0.0125,
+      "step": 136220
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9345794841492885e-05,
+      "loss": 0.01,
+      "step": 136230
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9345746801754414e-05,
+      "loss": 0.0066,
+      "step": 136240
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934569876201594e-05,
+      "loss": 0.0088,
+      "step": 136250
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.934565072227747e-05,
+      "loss": 0.0086,
+      "step": 136260
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9345602682539e-05,
+      "loss": 0.0073,
+      "step": 136270
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9345554642800524e-05,
+      "loss": 0.0104,
+      "step": 136280
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9345506603062054e-05,
+      "loss": 0.0086,
+      "step": 136290
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9345458563323583e-05,
+      "loss": 0.0129,
+      "step": 136300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9345410523585112e-05,
+      "loss": 0.0088,
+      "step": 136310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9345362483846638e-05,
+      "loss": 0.0078,
+      "step": 136320
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9345314444108167e-05,
+      "loss": 0.0092,
+      "step": 136330
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9345266404369697e-05,
+      "loss": 0.0072,
+      "step": 136340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9345218364631223e-05,
+      "loss": 0.0109,
+      "step": 136350
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9345170324892752e-05,
+      "loss": 0.0116,
+      "step": 136360
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934512228515428e-05,
+      "loss": 0.0078,
+      "step": 136370
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934507424541581e-05,
+      "loss": 0.009,
+      "step": 136380
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9345026205677336e-05,
+      "loss": 0.0065,
+      "step": 136390
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344978165938865e-05,
+      "loss": 0.0115,
+      "step": 136400
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344930126200395e-05,
+      "loss": 0.0124,
+      "step": 136410
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344882086461924e-05,
+      "loss": 0.0114,
+      "step": 136420
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934483404672345e-05,
+      "loss": 0.0123,
+      "step": 136430
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934478600698498e-05,
+      "loss": 0.0121,
+      "step": 136440
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934473796724651e-05,
+      "loss": 0.0085,
+      "step": 136450
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344689927508034e-05,
+      "loss": 0.0143,
+      "step": 136460
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344641887769564e-05,
+      "loss": 0.0079,
+      "step": 136470
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344593848031093e-05,
+      "loss": 0.0101,
+      "step": 136480
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344545808292622e-05,
+      "loss": 0.0071,
+      "step": 136490
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344497768554148e-05,
+      "loss": 0.0094,
+      "step": 136500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344449728815677e-05,
+      "loss": 0.0227,
+      "step": 136510
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344401689077207e-05,
+      "loss": 0.0089,
+      "step": 136520
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344353649338732e-05,
+      "loss": 0.0084,
+      "step": 136530
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934430560960026e-05,
+      "loss": 0.0082,
+      "step": 136540
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934425756986179e-05,
+      "loss": 0.0124,
+      "step": 136550
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934420953012332e-05,
+      "loss": 0.0095,
+      "step": 136560
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344161490384846e-05,
+      "loss": 0.0074,
+      "step": 136570
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344113450646375e-05,
+      "loss": 0.0082,
+      "step": 136580
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344065410907905e-05,
+      "loss": 0.011,
+      "step": 136590
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9344017371169434e-05,
+      "loss": 0.0125,
+      "step": 136600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934396933143096e-05,
+      "loss": 0.0108,
+      "step": 136610
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934392129169249e-05,
+      "loss": 0.008,
+      "step": 136620
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934387325195402e-05,
+      "loss": 0.0082,
+      "step": 136630
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343825212215544e-05,
+      "loss": 0.0073,
+      "step": 136640
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343777172477073e-05,
+      "loss": 0.0105,
+      "step": 136650
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343729132738603e-05,
+      "loss": 0.0073,
+      "step": 136660
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343681093000132e-05,
+      "loss": 0.0098,
+      "step": 136670
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343633053261658e-05,
+      "loss": 0.0061,
+      "step": 136680
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343585013523187e-05,
+      "loss": 0.0118,
+      "step": 136690
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343536973784716e-05,
+      "loss": 0.011,
+      "step": 136700
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343488934046242e-05,
+      "loss": 0.0075,
+      "step": 136710
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934344089430777e-05,
+      "loss": 0.0081,
+      "step": 136720
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.93433928545693e-05,
+      "loss": 0.0094,
+      "step": 136730
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934334481483083e-05,
+      "loss": 0.0086,
+      "step": 136740
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343296775092356e-05,
+      "loss": 0.0088,
+      "step": 136750
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343248735353885e-05,
+      "loss": 0.0101,
+      "step": 136760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343200695615415e-05,
+      "loss": 0.0096,
+      "step": 136770
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343152655876944e-05,
+      "loss": 0.0097,
+      "step": 136780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934310461613847e-05,
+      "loss": 0.0056,
+      "step": 136790
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.93430565764e-05,
+      "loss": 0.0091,
+      "step": 136800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9343008536661528e-05,
+      "loss": 0.0119,
+      "step": 136810
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342960496923054e-05,
+      "loss": 0.0061,
+      "step": 136820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342912457184583e-05,
+      "loss": 0.0073,
+      "step": 136830
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342864417446113e-05,
+      "loss": 0.0059,
+      "step": 136840
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342816377707642e-05,
+      "loss": 0.0075,
+      "step": 136850
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342768337969168e-05,
+      "loss": 0.0089,
+      "step": 136860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342720298230697e-05,
+      "loss": 0.0128,
+      "step": 136870
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342672258492226e-05,
+      "loss": 0.0105,
+      "step": 136880
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342624218753752e-05,
+      "loss": 0.0089,
+      "step": 136890
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934257617901528e-05,
+      "loss": 0.008,
+      "step": 136900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934252813927681e-05,
+      "loss": 0.0076,
+      "step": 136910
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934248009953834e-05,
+      "loss": 0.0129,
+      "step": 136920
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342432059799866e-05,
+      "loss": 0.0119,
+      "step": 136930
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342384020061395e-05,
+      "loss": 0.0127,
+      "step": 136940
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342335980322924e-05,
+      "loss": 0.0132,
+      "step": 136950
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342287940584454e-05,
+      "loss": 0.0081,
+      "step": 136960
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934223990084598e-05,
+      "loss": 0.013,
+      "step": 136970
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.934219186110751e-05,
+      "loss": 0.0096,
+      "step": 136980
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342143821369038e-05,
+      "loss": 0.0067,
+      "step": 136990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342095781630564e-05,
+      "loss": 0.0103,
+      "step": 137000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9342047741892093e-05,
+      "loss": 0.0101,
+      "step": 137010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9341999702153623e-05,
+      "loss": 0.0125,
+      "step": 137020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9341951662415152e-05,
+      "loss": 0.01,
+      "step": 137030
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341903622676678e-05,
+      "loss": 0.0149,
+      "step": 137040
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341855582938207e-05,
+      "loss": 0.0081,
+      "step": 137050
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341807543199736e-05,
+      "loss": 0.0096,
+      "step": 137060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341759503461262e-05,
+      "loss": 0.009,
+      "step": 137070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.934171146372279e-05,
+      "loss": 0.0103,
+      "step": 137080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.934166342398432e-05,
+      "loss": 0.0098,
+      "step": 137090
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.934161538424585e-05,
+      "loss": 0.0107,
+      "step": 137100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341567344507376e-05,
+      "loss": 0.0132,
+      "step": 137110
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341519304768905e-05,
+      "loss": 0.0104,
+      "step": 137120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341471265030434e-05,
+      "loss": 0.0095,
+      "step": 137130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341423225291964e-05,
+      "loss": 0.0073,
+      "step": 137140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.934137518555349e-05,
+      "loss": 0.0079,
+      "step": 137150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.934132714581502e-05,
+      "loss": 0.0147,
+      "step": 137160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341279106076548e-05,
+      "loss": 0.0135,
+      "step": 137170
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341231066338074e-05,
+      "loss": 0.0079,
+      "step": 137180
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341183026599603e-05,
+      "loss": 0.008,
+      "step": 137190
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341134986861133e-05,
+      "loss": 0.0099,
+      "step": 137200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9341086947122662e-05,
+      "loss": 0.011,
+      "step": 137210
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.934103890738419e-05,
+      "loss": 0.0093,
+      "step": 137220
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.934099086764572e-05,
+      "loss": 0.0071,
+      "step": 137230
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340942827907246e-05,
+      "loss": 0.0084,
+      "step": 137240
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340894788168775e-05,
+      "loss": 0.0052,
+      "step": 137250
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340846748430305e-05,
+      "loss": 0.0079,
+      "step": 137260
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340798708691834e-05,
+      "loss": 0.0104,
+      "step": 137270
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.934075066895336e-05,
+      "loss": 0.0082,
+      "step": 137280
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.934070262921489e-05,
+      "loss": 0.0081,
+      "step": 137290
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.934065458947642e-05,
+      "loss": 0.0102,
+      "step": 137300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340606549737944e-05,
+      "loss": 0.012,
+      "step": 137310
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340558509999474e-05,
+      "loss": 0.0114,
+      "step": 137320
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340510470261003e-05,
+      "loss": 0.0097,
+      "step": 137330
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340462430522532e-05,
+      "loss": 0.0103,
+      "step": 137340
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340414390784058e-05,
+      "loss": 0.0081,
+      "step": 137350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340366351045587e-05,
+      "loss": 0.0089,
+      "step": 137360
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340318311307117e-05,
+      "loss": 0.0095,
+      "step": 137370
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340270271568646e-05,
+      "loss": 0.007,
+      "step": 137380
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.934022223183017e-05,
+      "loss": 0.0117,
+      "step": 137390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.93401741920917e-05,
+      "loss": 0.0111,
+      "step": 137400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.934012615235323e-05,
+      "loss": 0.0111,
+      "step": 137410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340078112614756e-05,
+      "loss": 0.011,
+      "step": 137420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9340030072876285e-05,
+      "loss": 0.0073,
+      "step": 137430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339982033137815e-05,
+      "loss": 0.009,
+      "step": 137440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339933993399344e-05,
+      "loss": 0.0091,
+      "step": 137450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.933988595366087e-05,
+      "loss": 0.0089,
+      "step": 137460
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.93398379139224e-05,
+      "loss": 0.0097,
+      "step": 137470
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.933978987418393e-05,
+      "loss": 0.0106,
+      "step": 137480
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339741834445454e-05,
+      "loss": 0.0083,
+      "step": 137490
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339693794706983e-05,
+      "loss": 0.0107,
+      "step": 137500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339645754968513e-05,
+      "loss": 0.0113,
+      "step": 137510
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339597715230042e-05,
+      "loss": 0.0071,
+      "step": 137520
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339549675491568e-05,
+      "loss": 0.0099,
+      "step": 137530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339501635753097e-05,
+      "loss": 0.009,
+      "step": 137540
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339453596014626e-05,
+      "loss": 0.0093,
+      "step": 137550
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339405556276152e-05,
+      "loss": 0.0111,
+      "step": 137560
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.933935751653768e-05,
+      "loss": 0.0084,
+      "step": 137570
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.933930947679921e-05,
+      "loss": 0.0072,
+      "step": 137580
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.933926143706074e-05,
+      "loss": 0.0152,
+      "step": 137590
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339213397322266e-05,
+      "loss": 0.008,
+      "step": 137600
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339165357583795e-05,
+      "loss": 0.0085,
+      "step": 137610
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339117317845325e-05,
+      "loss": 0.0097,
+      "step": 137620
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9339069278106854e-05,
+      "loss": 0.0077,
+      "step": 137630
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.933902123836838e-05,
+      "loss": 0.0107,
+      "step": 137640
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.933897319862991e-05,
+      "loss": 0.0083,
+      "step": 137650
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9338925158891438e-05,
+      "loss": 0.0108,
+      "step": 137660
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9338877119152964e-05,
+      "loss": 0.0104,
+      "step": 137670
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9338829079414493e-05,
+      "loss": 0.0095,
+      "step": 137680
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9338781039676023e-05,
+      "loss": 0.0083,
+      "step": 137690
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9338732999937552e-05,
+      "loss": 0.0085,
+      "step": 137700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9338684960199078e-05,
+      "loss": 0.0072,
+      "step": 137710
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9338636920460607e-05,
+      "loss": 0.0087,
+      "step": 137720
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9338588880722136e-05,
+      "loss": 0.0118,
+      "step": 137730
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9338540840983662e-05,
+      "loss": 0.0091,
+      "step": 137740
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933849280124519e-05,
+      "loss": 0.0161,
+      "step": 137750
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933844476150672e-05,
+      "loss": 0.0061,
+      "step": 137760
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933839672176825e-05,
+      "loss": 0.0083,
+      "step": 137770
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9338348682029776e-05,
+      "loss": 0.0113,
+      "step": 137780
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9338300642291305e-05,
+      "loss": 0.0076,
+      "step": 137790
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9338252602552834e-05,
+      "loss": 0.0093,
+      "step": 137800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9338204562814364e-05,
+      "loss": 0.0084,
+      "step": 137810
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933815652307589e-05,
+      "loss": 0.0072,
+      "step": 137820
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933810848333742e-05,
+      "loss": 0.0076,
+      "step": 137830
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9338060443598948e-05,
+      "loss": 0.0073,
+      "step": 137840
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9338012403860474e-05,
+      "loss": 0.0079,
+      "step": 137850
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337964364122003e-05,
+      "loss": 0.0089,
+      "step": 137860
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337916324383533e-05,
+      "loss": 0.0116,
+      "step": 137870
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337868284645062e-05,
+      "loss": 0.0099,
+      "step": 137880
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337820244906588e-05,
+      "loss": 0.0089,
+      "step": 137890
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337772205168117e-05,
+      "loss": 0.0095,
+      "step": 137900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337724165429646e-05,
+      "loss": 0.0084,
+      "step": 137910
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337676125691172e-05,
+      "loss": 0.0091,
+      "step": 137920
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.93376280859527e-05,
+      "loss": 0.0073,
+      "step": 137930
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933758004621423e-05,
+      "loss": 0.0135,
+      "step": 137940
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933753200647576e-05,
+      "loss": 0.0085,
+      "step": 137950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337483966737286e-05,
+      "loss": 0.0096,
+      "step": 137960
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337435926998815e-05,
+      "loss": 0.01,
+      "step": 137970
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337387887260344e-05,
+      "loss": 0.0106,
+      "step": 137980
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337339847521874e-05,
+      "loss": 0.0095,
+      "step": 137990
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.93372918077834e-05,
+      "loss": 0.007,
+      "step": 138000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933724376804493e-05,
+      "loss": 0.0152,
+      "step": 138010
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337195728306458e-05,
+      "loss": 0.0074,
+      "step": 138020
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337147688567984e-05,
+      "loss": 0.0086,
+      "step": 138030
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337099648829513e-05,
+      "loss": 0.0066,
+      "step": 138040
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337051609091043e-05,
+      "loss": 0.0104,
+      "step": 138050
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337003569352572e-05,
+      "loss": 0.0047,
+      "step": 138060
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336955529614098e-05,
+      "loss": 0.0083,
+      "step": 138070
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336907489875627e-05,
+      "loss": 0.013,
+      "step": 138080
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336859450137156e-05,
+      "loss": 0.0094,
+      "step": 138090
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336811410398682e-05,
+      "loss": 0.0105,
+      "step": 138100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933676337066021e-05,
+      "loss": 0.0089,
+      "step": 138110
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933671533092174e-05,
+      "loss": 0.01,
+      "step": 138120
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933666729118327e-05,
+      "loss": 0.0079,
+      "step": 138130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336619251444796e-05,
+      "loss": 0.0114,
+      "step": 138140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336571211706325e-05,
+      "loss": 0.0085,
+      "step": 138150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336523171967854e-05,
+      "loss": 0.0128,
+      "step": 138160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336475132229384e-05,
+      "loss": 0.0063,
+      "step": 138170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933642709249091e-05,
+      "loss": 0.0115,
+      "step": 138180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933637905275244e-05,
+      "loss": 0.0092,
+      "step": 138190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336331013013968e-05,
+      "loss": 0.0057,
+      "step": 138200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336282973275494e-05,
+      "loss": 0.0072,
+      "step": 138210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336234933537023e-05,
+      "loss": 0.0123,
+      "step": 138220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336186893798552e-05,
+      "loss": 0.0097,
+      "step": 138230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933613885406008e-05,
+      "loss": 0.0084,
+      "step": 138240
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336090814321608e-05,
+      "loss": 0.0076,
+      "step": 138250
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9336042774583137e-05,
+      "loss": 0.0123,
+      "step": 138260
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9335994734844666e-05,
+      "loss": 0.0085,
+      "step": 138270
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9335946695106192e-05,
+      "loss": 0.0105,
+      "step": 138280
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933589865536772e-05,
+      "loss": 0.0099,
+      "step": 138290
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933585061562925e-05,
+      "loss": 0.0099,
+      "step": 138300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933580257589078e-05,
+      "loss": 0.0111,
+      "step": 138310
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9335754536152306e-05,
+      "loss": 0.0097,
+      "step": 138320
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9335706496413835e-05,
+      "loss": 0.0097,
+      "step": 138330
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9335658456675364e-05,
+      "loss": 0.0116,
+      "step": 138340
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9335610416936893e-05,
+      "loss": 0.009,
+      "step": 138350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933556237719842e-05,
+      "loss": 0.0076,
+      "step": 138360
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933551433745995e-05,
+      "loss": 0.0061,
+      "step": 138370
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9335466297721478e-05,
+      "loss": 0.0114,
+      "step": 138380
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9335418257983004e-05,
+      "loss": 0.012,
+      "step": 138390
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9335370218244533e-05,
+      "loss": 0.0093,
+      "step": 138400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9335322178506062e-05,
+      "loss": 0.0085,
+      "step": 138410
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.933527413876759e-05,
+      "loss": 0.0079,
+      "step": 138420
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9335226099029117e-05,
+      "loss": 0.0076,
+      "step": 138430
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9335178059290647e-05,
+      "loss": 0.0125,
+      "step": 138440
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9335130019552176e-05,
+      "loss": 0.0101,
+      "step": 138450
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9335081979813702e-05,
+      "loss": 0.0111,
+      "step": 138460
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933503394007523e-05,
+      "loss": 0.0102,
+      "step": 138470
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933498590033676e-05,
+      "loss": 0.0103,
+      "step": 138480
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933493786059829e-05,
+      "loss": 0.0096,
+      "step": 138490
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334889820859816e-05,
+      "loss": 0.012,
+      "step": 138500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334841781121345e-05,
+      "loss": 0.0068,
+      "step": 138510
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334793741382874e-05,
+      "loss": 0.0139,
+      "step": 138520
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334745701644403e-05,
+      "loss": 0.009,
+      "step": 138530
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933469766190593e-05,
+      "loss": 0.0124,
+      "step": 138540
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933464962216746e-05,
+      "loss": 0.0084,
+      "step": 138550
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334601582428988e-05,
+      "loss": 0.0067,
+      "step": 138560
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334553542690514e-05,
+      "loss": 0.0072,
+      "step": 138570
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334505502952043e-05,
+      "loss": 0.0141,
+      "step": 138580
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334457463213572e-05,
+      "loss": 0.0095,
+      "step": 138590
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.93344094234751e-05,
+      "loss": 0.0094,
+      "step": 138600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334361383736627e-05,
+      "loss": 0.0085,
+      "step": 138610
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334313343998157e-05,
+      "loss": 0.0081,
+      "step": 138620
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334265304259686e-05,
+      "loss": 0.0099,
+      "step": 138630
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334217264521212e-05,
+      "loss": 0.0081,
+      "step": 138640
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933416922478274e-05,
+      "loss": 0.0073,
+      "step": 138650
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933412118504427e-05,
+      "loss": 0.0089,
+      "step": 138660
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.93340731453058e-05,
+      "loss": 0.0066,
+      "step": 138670
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9334025105567326e-05,
+      "loss": 0.0095,
+      "step": 138680
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333977065828855e-05,
+      "loss": 0.008,
+      "step": 138690
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333929026090384e-05,
+      "loss": 0.0083,
+      "step": 138700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333880986351913e-05,
+      "loss": 0.0054,
+      "step": 138710
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933383294661344e-05,
+      "loss": 0.0108,
+      "step": 138720
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933378490687497e-05,
+      "loss": 0.0079,
+      "step": 138730
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333736867136498e-05,
+      "loss": 0.0074,
+      "step": 138740
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333688827398024e-05,
+      "loss": 0.0073,
+      "step": 138750
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333640787659553e-05,
+      "loss": 0.0108,
+      "step": 138760
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333592747921082e-05,
+      "loss": 0.0063,
+      "step": 138770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933354470818261e-05,
+      "loss": 0.0091,
+      "step": 138780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333496668444137e-05,
+      "loss": 0.0072,
+      "step": 138790
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333448628705667e-05,
+      "loss": 0.006,
+      "step": 138800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333400588967196e-05,
+      "loss": 0.0124,
+      "step": 138810
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333352549228722e-05,
+      "loss": 0.0091,
+      "step": 138820
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933330450949025e-05,
+      "loss": 0.0079,
+      "step": 138830
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933325646975178e-05,
+      "loss": 0.0062,
+      "step": 138840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933320843001331e-05,
+      "loss": 0.0084,
+      "step": 138850
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333160390274835e-05,
+      "loss": 0.007,
+      "step": 138860
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333112350536365e-05,
+      "loss": 0.0067,
+      "step": 138870
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333064310797894e-05,
+      "loss": 0.0108,
+      "step": 138880
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9333016271059423e-05,
+      "loss": 0.0055,
+      "step": 138890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933296823132095e-05,
+      "loss": 0.0078,
+      "step": 138900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933292019158248e-05,
+      "loss": 0.0099,
+      "step": 138910
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332872151844008e-05,
+      "loss": 0.0099,
+      "step": 138920
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332824112105534e-05,
+      "loss": 0.0088,
+      "step": 138930
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332776072367063e-05,
+      "loss": 0.0093,
+      "step": 138940
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332728032628592e-05,
+      "loss": 0.0066,
+      "step": 138950
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933267999289012e-05,
+      "loss": 0.0065,
+      "step": 138960
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332631953151647e-05,
+      "loss": 0.0087,
+      "step": 138970
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332583913413177e-05,
+      "loss": 0.0103,
+      "step": 138980
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332535873674706e-05,
+      "loss": 0.0071,
+      "step": 138990
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933248783393623e-05,
+      "loss": 0.0078,
+      "step": 139000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933243979419776e-05,
+      "loss": 0.0069,
+      "step": 139010
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933239175445929e-05,
+      "loss": 0.0082,
+      "step": 139020
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933234371472082e-05,
+      "loss": 0.0059,
+      "step": 139030
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332295674982345e-05,
+      "loss": 0.0103,
+      "step": 139040
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332247635243875e-05,
+      "loss": 0.0092,
+      "step": 139050
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332199595505404e-05,
+      "loss": 0.0065,
+      "step": 139060
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332151555766933e-05,
+      "loss": 0.0078,
+      "step": 139070
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933210351602846e-05,
+      "loss": 0.0077,
+      "step": 139080
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332055476289988e-05,
+      "loss": 0.0085,
+      "step": 139090
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9332007436551518e-05,
+      "loss": 0.0101,
+      "step": 139100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9331959396813043e-05,
+      "loss": 0.0075,
+      "step": 139110
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9331911357074573e-05,
+      "loss": 0.0048,
+      "step": 139120
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9331863317336102e-05,
+      "loss": 0.0069,
+      "step": 139130
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933181527759763e-05,
+      "loss": 0.0056,
+      "step": 139140
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9331767237859157e-05,
+      "loss": 0.0054,
+      "step": 139150
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9331719198120686e-05,
+      "loss": 0.0085,
+      "step": 139160
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9331671158382216e-05,
+      "loss": 0.0084,
+      "step": 139170
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933162311864374e-05,
+      "loss": 0.0094,
+      "step": 139180
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933157507890527e-05,
+      "loss": 0.0058,
+      "step": 139190
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.93315270391668e-05,
+      "loss": 0.0077,
+      "step": 139200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933147899942833e-05,
+      "loss": 0.0069,
+      "step": 139210
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9331430959689855e-05,
+      "loss": 0.0066,
+      "step": 139220
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9331382919951385e-05,
+      "loss": 0.006,
+      "step": 139230
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9331334880212914e-05,
+      "loss": 0.006,
+      "step": 139240
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933128684047444e-05,
+      "loss": 0.0076,
+      "step": 139250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933123880073597e-05,
+      "loss": 0.0101,
+      "step": 139260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9331190760997498e-05,
+      "loss": 0.0076,
+      "step": 139270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9331142721259027e-05,
+      "loss": 0.0092,
+      "step": 139280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9331094681520553e-05,
+      "loss": 0.0084,
+      "step": 139290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9331046641782083e-05,
+      "loss": 0.007,
+      "step": 139300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330998602043612e-05,
+      "loss": 0.0092,
+      "step": 139310
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933095056230514e-05,
+      "loss": 0.0064,
+      "step": 139320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330902522566667e-05,
+      "loss": 0.0069,
+      "step": 139330
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330854482828196e-05,
+      "loss": 0.007,
+      "step": 139340
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330806443089726e-05,
+      "loss": 0.0101,
+      "step": 139350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933075840335125e-05,
+      "loss": 0.0083,
+      "step": 139360
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933071036361278e-05,
+      "loss": 0.008,
+      "step": 139370
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933066232387431e-05,
+      "loss": 0.0072,
+      "step": 139380
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933061428413584e-05,
+      "loss": 0.0091,
+      "step": 139390
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330566244397365e-05,
+      "loss": 0.0089,
+      "step": 139400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330518204658894e-05,
+      "loss": 0.0056,
+      "step": 139410
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330470164920424e-05,
+      "loss": 0.0064,
+      "step": 139420
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933042212518195e-05,
+      "loss": 0.0061,
+      "step": 139430
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933037408544348e-05,
+      "loss": 0.0134,
+      "step": 139440
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330326045705008e-05,
+      "loss": 0.0064,
+      "step": 139450
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330278005966537e-05,
+      "loss": 0.0125,
+      "step": 139460
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330229966228063e-05,
+      "loss": 0.0069,
+      "step": 139470
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330181926489593e-05,
+      "loss": 0.0093,
+      "step": 139480
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330133886751122e-05,
+      "loss": 0.0059,
+      "step": 139490
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.933008584701265e-05,
+      "loss": 0.0089,
+      "step": 139500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9330037807274177e-05,
+      "loss": 0.0063,
+      "step": 139510
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329989767535706e-05,
+      "loss": 0.0112,
+      "step": 139520
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329941727797236e-05,
+      "loss": 0.0085,
+      "step": 139530
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.932989368805876e-05,
+      "loss": 0.0077,
+      "step": 139540
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.932984564832029e-05,
+      "loss": 0.0053,
+      "step": 139550
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.932979760858182e-05,
+      "loss": 0.0076,
+      "step": 139560
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.932974956884335e-05,
+      "loss": 0.0074,
+      "step": 139570
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329701529104875e-05,
+      "loss": 0.0057,
+      "step": 139580
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329653489366404e-05,
+      "loss": 0.0086,
+      "step": 139590
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329605449627934e-05,
+      "loss": 0.0081,
+      "step": 139600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.932955740988946e-05,
+      "loss": 0.0055,
+      "step": 139610
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.932950937015099e-05,
+      "loss": 0.0058,
+      "step": 139620
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329461330412518e-05,
+      "loss": 0.0064,
+      "step": 139630
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329413290674047e-05,
+      "loss": 0.0051,
+      "step": 139640
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329365250935573e-05,
+      "loss": 0.0076,
+      "step": 139650
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329317211197102e-05,
+      "loss": 0.0064,
+      "step": 139660
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329269171458632e-05,
+      "loss": 0.0077,
+      "step": 139670
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.932922113172016e-05,
+      "loss": 0.0085,
+      "step": 139680
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329173091981687e-05,
+      "loss": 0.0097,
+      "step": 139690
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329125052243216e-05,
+      "loss": 0.0086,
+      "step": 139700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9329077012504745e-05,
+      "loss": 0.0058,
+      "step": 139710
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.932902897276627e-05,
+      "loss": 0.0097,
+      "step": 139720
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.93289809330278e-05,
+      "loss": 0.0049,
+      "step": 139730
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.932893289328933e-05,
+      "loss": 0.0059,
+      "step": 139740
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.932888485355086e-05,
+      "loss": 0.008,
+      "step": 139750
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9328836813812385e-05,
+      "loss": 0.0092,
+      "step": 139760
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9328788774073918e-05,
+      "loss": 0.0082,
+      "step": 139770
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9328740734335444e-05,
+      "loss": 0.008,
+      "step": 139780
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9328692694596973e-05,
+      "loss": 0.008,
+      "step": 139790
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9328644654858502e-05,
+      "loss": 0.0059,
+      "step": 139800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.932859661512003e-05,
+      "loss": 0.0083,
+      "step": 139810
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9328548575381557e-05,
+      "loss": 0.0063,
+      "step": 139820
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9328500535643087e-05,
+      "loss": 0.0096,
+      "step": 139830
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9328452495904616e-05,
+      "loss": 0.0107,
+      "step": 139840
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932840445616614e-05,
+      "loss": 0.0077,
+      "step": 139850
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932835641642767e-05,
+      "loss": 0.0069,
+      "step": 139860
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.93283083766892e-05,
+      "loss": 0.0039,
+      "step": 139870
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932826033695073e-05,
+      "loss": 0.0078,
+      "step": 139880
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9328212297212255e-05,
+      "loss": 0.0088,
+      "step": 139890
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9328164257473785e-05,
+      "loss": 0.0115,
+      "step": 139900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9328116217735314e-05,
+      "loss": 0.0064,
+      "step": 139910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9328068177996843e-05,
+      "loss": 0.0071,
+      "step": 139920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932802013825837e-05,
+      "loss": 0.0084,
+      "step": 139930
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.93279720985199e-05,
+      "loss": 0.0061,
+      "step": 139940
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9327924058781428e-05,
+      "loss": 0.0071,
+      "step": 139950
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9327876019042953e-05,
+      "loss": 0.0105,
+      "step": 139960
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9327827979304483e-05,
+      "loss": 0.0097,
+      "step": 139970
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9327779939566012e-05,
+      "loss": 0.0084,
+      "step": 139980
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932773189982754e-05,
+      "loss": 0.0078,
+      "step": 139990
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9327683860089067e-05,
+      "loss": 0.0058,
+      "step": 140000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9327635820350596e-05,
+      "loss": 0.0072,
+      "step": 140010
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9327587780612126e-05,
+      "loss": 0.0069,
+      "step": 140020
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932753974087365e-05,
+      "loss": 0.0085,
+      "step": 140030
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932749170113518e-05,
+      "loss": 0.0066,
+      "step": 140040
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932744366139671e-05,
+      "loss": 0.0077,
+      "step": 140050
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932739562165824e-05,
+      "loss": 0.0051,
+      "step": 140060
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9327347581919765e-05,
+      "loss": 0.0113,
+      "step": 140070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932730434615514e-05,
+      "loss": 0.0102,
+      "step": 140080
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932725630641667e-05,
+      "loss": 0.0087,
+      "step": 140090
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.93272082666782e-05,
+      "loss": 0.0083,
+      "step": 140100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9327160226939724e-05,
+      "loss": 0.0085,
+      "step": 140110
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9327112187201254e-05,
+      "loss": 0.0072,
+      "step": 140120
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9327064147462783e-05,
+      "loss": 0.0073,
+      "step": 140130
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932701610772431e-05,
+      "loss": 0.0072,
+      "step": 140140
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326968067985838e-05,
+      "loss": 0.0118,
+      "step": 140150
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326920028247367e-05,
+      "loss": 0.0086,
+      "step": 140160
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326871988508897e-05,
+      "loss": 0.0055,
+      "step": 140170
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326823948770422e-05,
+      "loss": 0.0086,
+      "step": 140180
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326775909031952e-05,
+      "loss": 0.008,
+      "step": 140190
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932672786929348e-05,
+      "loss": 0.0092,
+      "step": 140200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326679829555007e-05,
+      "loss": 0.0087,
+      "step": 140210
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326631789816536e-05,
+      "loss": 0.0048,
+      "step": 140220
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326583750078065e-05,
+      "loss": 0.0078,
+      "step": 140230
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326535710339595e-05,
+      "loss": 0.0076,
+      "step": 140240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932648767060112e-05,
+      "loss": 0.0057,
+      "step": 140250
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932643963086265e-05,
+      "loss": 0.0088,
+      "step": 140260
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932639159112418e-05,
+      "loss": 0.0082,
+      "step": 140270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932634355138571e-05,
+      "loss": 0.0086,
+      "step": 140280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326295511647238e-05,
+      "loss": 0.0071,
+      "step": 140290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326247471908767e-05,
+      "loss": 0.0085,
+      "step": 140300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326199432170293e-05,
+      "loss": 0.0098,
+      "step": 140310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326151392431822e-05,
+      "loss": 0.0056,
+      "step": 140320
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932610335269335e-05,
+      "loss": 0.0053,
+      "step": 140330
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932605531295488e-05,
+      "loss": 0.0105,
+      "step": 140340
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9326007273216406e-05,
+      "loss": 0.0065,
+      "step": 140350
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9325959233477936e-05,
+      "loss": 0.0067,
+      "step": 140360
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9325911193739465e-05,
+      "loss": 0.0078,
+      "step": 140370
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932586315400099e-05,
+      "loss": 0.0072,
+      "step": 140380
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9325819918236366e-05,
+      "loss": 0.0116,
+      "step": 140390
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9325771878497895e-05,
+      "loss": 0.0098,
+      "step": 140400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9325723838759424e-05,
+      "loss": 0.008,
+      "step": 140410
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932567579902095e-05,
+      "loss": 0.0075,
+      "step": 140420
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932562775928248e-05,
+      "loss": 0.0091,
+      "step": 140430
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.932557971954401e-05,
+      "loss": 0.0073,
+      "step": 140440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9325531679805534e-05,
+      "loss": 0.0097,
+      "step": 140450
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9325483640067064e-05,
+      "loss": 0.0077,
+      "step": 140460
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9325435600328593e-05,
+      "loss": 0.0072,
+      "step": 140470
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9325387560590122e-05,
+      "loss": 0.0058,
+      "step": 140480
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9325339520851648e-05,
+      "loss": 0.0072,
+      "step": 140490
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9325291481113177e-05,
+      "loss": 0.008,
+      "step": 140500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9325243441374707e-05,
+      "loss": 0.0088,
+      "step": 140510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9325195401636236e-05,
+      "loss": 0.0111,
+      "step": 140520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9325147361897762e-05,
+      "loss": 0.0061,
+      "step": 140530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932509932215929e-05,
+      "loss": 0.0075,
+      "step": 140540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932505128242082e-05,
+      "loss": 0.0067,
+      "step": 140550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9325003242682346e-05,
+      "loss": 0.0077,
+      "step": 140560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324955202943875e-05,
+      "loss": 0.0097,
+      "step": 140570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324907163205405e-05,
+      "loss": 0.007,
+      "step": 140580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324859123466934e-05,
+      "loss": 0.0064,
+      "step": 140590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932481108372846e-05,
+      "loss": 0.007,
+      "step": 140600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932476304398999e-05,
+      "loss": 0.0083,
+      "step": 140610
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932471500425152e-05,
+      "loss": 0.0087,
+      "step": 140620
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324666964513044e-05,
+      "loss": 0.0121,
+      "step": 140630
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324618924774574e-05,
+      "loss": 0.0093,
+      "step": 140640
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324570885036103e-05,
+      "loss": 0.0085,
+      "step": 140650
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324522845297632e-05,
+      "loss": 0.0069,
+      "step": 140660
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324474805559158e-05,
+      "loss": 0.0069,
+      "step": 140670
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324426765820687e-05,
+      "loss": 0.0089,
+      "step": 140680
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324378726082217e-05,
+      "loss": 0.0082,
+      "step": 140690
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324330686343742e-05,
+      "loss": 0.007,
+      "step": 140700
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932428264660527e-05,
+      "loss": 0.0078,
+      "step": 140710
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.93242346068668e-05,
+      "loss": 0.0055,
+      "step": 140720
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932418656712833e-05,
+      "loss": 0.0077,
+      "step": 140730
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324138527389856e-05,
+      "loss": 0.0065,
+      "step": 140740
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324090487651385e-05,
+      "loss": 0.0093,
+      "step": 140750
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9324042447912915e-05,
+      "loss": 0.01,
+      "step": 140760
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323994408174444e-05,
+      "loss": 0.0066,
+      "step": 140770
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932394636843597e-05,
+      "loss": 0.0082,
+      "step": 140780
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.93238983286975e-05,
+      "loss": 0.0085,
+      "step": 140790
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932385028895903e-05,
+      "loss": 0.0069,
+      "step": 140800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323802249220558e-05,
+      "loss": 0.0125,
+      "step": 140810
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323754209482087e-05,
+      "loss": 0.0088,
+      "step": 140820
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323706169743616e-05,
+      "loss": 0.0073,
+      "step": 140830
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323658130005142e-05,
+      "loss": 0.0081,
+      "step": 140840
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932361009026667e-05,
+      "loss": 0.0062,
+      "step": 140850
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.93235620505282e-05,
+      "loss": 0.0095,
+      "step": 140860
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323514010789726e-05,
+      "loss": 0.01,
+      "step": 140870
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323465971051256e-05,
+      "loss": 0.007,
+      "step": 140880
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323417931312785e-05,
+      "loss": 0.0082,
+      "step": 140890
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323369891574314e-05,
+      "loss": 0.0087,
+      "step": 140900
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932332185183584e-05,
+      "loss": 0.008,
+      "step": 140910
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932327381209737e-05,
+      "loss": 0.0088,
+      "step": 140920
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.93232257723589e-05,
+      "loss": 0.0061,
+      "step": 140930
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323177732620425e-05,
+      "loss": 0.0094,
+      "step": 140940
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323129692881954e-05,
+      "loss": 0.0096,
+      "step": 140950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323081653143483e-05,
+      "loss": 0.0098,
+      "step": 140960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9323033613405012e-05,
+      "loss": 0.0075,
+      "step": 140970
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322985573666538e-05,
+      "loss": 0.009,
+      "step": 140980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322937533928068e-05,
+      "loss": 0.0092,
+      "step": 140990
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322889494189597e-05,
+      "loss": 0.0083,
+      "step": 141000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322841454451126e-05,
+      "loss": 0.0064,
+      "step": 141010
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322793414712652e-05,
+      "loss": 0.007,
+      "step": 141020
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932274537497418e-05,
+      "loss": 0.0121,
+      "step": 141030
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932269733523571e-05,
+      "loss": 0.0083,
+      "step": 141040
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322649295497236e-05,
+      "loss": 0.0118,
+      "step": 141050
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322601255758766e-05,
+      "loss": 0.0088,
+      "step": 141060
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322553216020295e-05,
+      "loss": 0.0078,
+      "step": 141070
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322505176281824e-05,
+      "loss": 0.0082,
+      "step": 141080
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932245713654335e-05,
+      "loss": 0.0081,
+      "step": 141090
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932240909680488e-05,
+      "loss": 0.0097,
+      "step": 141100
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.932236105706641e-05,
+      "loss": 0.0122,
+      "step": 141110
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322313017327934e-05,
+      "loss": 0.0107,
+      "step": 141120
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322264977589464e-05,
+      "loss": 0.0067,
+      "step": 141130
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322216937850993e-05,
+      "loss": 0.0046,
+      "step": 141140
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322168898112522e-05,
+      "loss": 0.0103,
+      "step": 141150
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322120858374048e-05,
+      "loss": 0.0064,
+      "step": 141160
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322072818635577e-05,
+      "loss": 0.0063,
+      "step": 141170
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9322024778897107e-05,
+      "loss": 0.0074,
+      "step": 141180
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9321976739158636e-05,
+      "loss": 0.0105,
+      "step": 141190
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9321928699420162e-05,
+      "loss": 0.008,
+      "step": 141200
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.932188065968169e-05,
+      "loss": 0.0062,
+      "step": 141210
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.932183261994322e-05,
+      "loss": 0.007,
+      "step": 141220
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321784580204746e-05,
+      "loss": 0.0062,
+      "step": 141230
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321736540466276e-05,
+      "loss": 0.0109,
+      "step": 141240
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321688500727805e-05,
+      "loss": 0.0069,
+      "step": 141250
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321640460989334e-05,
+      "loss": 0.0099,
+      "step": 141260
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.932159242125086e-05,
+      "loss": 0.0086,
+      "step": 141270
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.932154438151239e-05,
+      "loss": 0.0063,
+      "step": 141280
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.932149634177392e-05,
+      "loss": 0.0095,
+      "step": 141290
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321448302035444e-05,
+      "loss": 0.0078,
+      "step": 141300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321400262296974e-05,
+      "loss": 0.0085,
+      "step": 141310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321352222558503e-05,
+      "loss": 0.0077,
+      "step": 141320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321304182820032e-05,
+      "loss": 0.008,
+      "step": 141330
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321256143081558e-05,
+      "loss": 0.0088,
+      "step": 141340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321208103343087e-05,
+      "loss": 0.0071,
+      "step": 141350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321160063604617e-05,
+      "loss": 0.0085,
+      "step": 141360
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321112023866146e-05,
+      "loss": 0.0079,
+      "step": 141370
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9321063984127672e-05,
+      "loss": 0.0081,
+      "step": 141380
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.93210159443892e-05,
+      "loss": 0.0078,
+      "step": 141390
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.932096790465073e-05,
+      "loss": 0.0097,
+      "step": 141400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320919864912256e-05,
+      "loss": 0.0075,
+      "step": 141410
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320871825173785e-05,
+      "loss": 0.0079,
+      "step": 141420
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320823785435315e-05,
+      "loss": 0.007,
+      "step": 141430
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320775745696844e-05,
+      "loss": 0.0063,
+      "step": 141440
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.932072770595837e-05,
+      "loss": 0.0058,
+      "step": 141450
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.93206796662199e-05,
+      "loss": 0.0101,
+      "step": 141460
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.932063162648143e-05,
+      "loss": 0.0067,
+      "step": 141470
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320583586742954e-05,
+      "loss": 0.0085,
+      "step": 141480
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320535547004484e-05,
+      "loss": 0.0077,
+      "step": 141490
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320487507266013e-05,
+      "loss": 0.0064,
+      "step": 141500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320439467527542e-05,
+      "loss": 0.0074,
+      "step": 141510
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320391427789068e-05,
+      "loss": 0.0075,
+      "step": 141520
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320343388050597e-05,
+      "loss": 0.0071,
+      "step": 141530
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320295348312127e-05,
+      "loss": 0.0054,
+      "step": 141540
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320247308573656e-05,
+      "loss": 0.0082,
+      "step": 141550
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.932019926883518e-05,
+      "loss": 0.0087,
+      "step": 141560
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.932015122909671e-05,
+      "loss": 0.0083,
+      "step": 141570
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.932010318935824e-05,
+      "loss": 0.008,
+      "step": 141580
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320055149619766e-05,
+      "loss": 0.0121,
+      "step": 141590
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9320007109881295e-05,
+      "loss": 0.0054,
+      "step": 141600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319959070142825e-05,
+      "loss": 0.0054,
+      "step": 141610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319911030404354e-05,
+      "loss": 0.0074,
+      "step": 141620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.931986299066588e-05,
+      "loss": 0.0053,
+      "step": 141630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.931981495092741e-05,
+      "loss": 0.0089,
+      "step": 141640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.931976691118894e-05,
+      "loss": 0.005,
+      "step": 141650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319718871450464e-05,
+      "loss": 0.0096,
+      "step": 141660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319670831711993e-05,
+      "loss": 0.0059,
+      "step": 141670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319622791973523e-05,
+      "loss": 0.0063,
+      "step": 141680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319574752235052e-05,
+      "loss": 0.0066,
+      "step": 141690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319526712496578e-05,
+      "loss": 0.0085,
+      "step": 141700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319478672758107e-05,
+      "loss": 0.0059,
+      "step": 141710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319430633019636e-05,
+      "loss": 0.0063,
+      "step": 141720
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319382593281166e-05,
+      "loss": 0.012,
+      "step": 141730
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.931933455354269e-05,
+      "loss": 0.0076,
+      "step": 141740
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.931928651380422e-05,
+      "loss": 0.0064,
+      "step": 141750
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.931923847406575e-05,
+      "loss": 0.0087,
+      "step": 141760
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319190434327276e-05,
+      "loss": 0.0107,
+      "step": 141770
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319142394588805e-05,
+      "loss": 0.0093,
+      "step": 141780
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319094354850335e-05,
+      "loss": 0.0056,
+      "step": 141790
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9319046315111864e-05,
+      "loss": 0.0071,
+      "step": 141800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.931899827537339e-05,
+      "loss": 0.0055,
+      "step": 141810
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.931895023563492e-05,
+      "loss": 0.009,
+      "step": 141820
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9318902195896448e-05,
+      "loss": 0.0065,
+      "step": 141830
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9318854156157974e-05,
+      "loss": 0.0095,
+      "step": 141840
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9318806116419503e-05,
+      "loss": 0.0087,
+      "step": 141850
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9318758076681033e-05,
+      "loss": 0.0077,
+      "step": 141860
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9318710036942562e-05,
+      "loss": 0.0088,
+      "step": 141870
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9318661997204088e-05,
+      "loss": 0.0061,
+      "step": 141880
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9318613957465617e-05,
+      "loss": 0.0086,
+      "step": 141890
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9318565917727146e-05,
+      "loss": 0.008,
+      "step": 141900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9318517877988676e-05,
+      "loss": 0.0053,
+      "step": 141910
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.93184698382502e-05,
+      "loss": 0.0082,
+      "step": 141920
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931842179851173e-05,
+      "loss": 0.0095,
+      "step": 141930
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931837375877326e-05,
+      "loss": 0.0065,
+      "step": 141940
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9318325719034786e-05,
+      "loss": 0.0079,
+      "step": 141950
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9318277679296315e-05,
+      "loss": 0.0064,
+      "step": 141960
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9318229639557844e-05,
+      "loss": 0.006,
+      "step": 141970
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9318181599819374e-05,
+      "loss": 0.0072,
+      "step": 141980
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.93181335600809e-05,
+      "loss": 0.0086,
+      "step": 141990
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931808552034243e-05,
+      "loss": 0.0088,
+      "step": 142000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9318037480603958e-05,
+      "loss": 0.0085,
+      "step": 142010
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317989440865484e-05,
+      "loss": 0.0069,
+      "step": 142020
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317941401127013e-05,
+      "loss": 0.0065,
+      "step": 142030
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317893361388543e-05,
+      "loss": 0.0065,
+      "step": 142040
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317845321650072e-05,
+      "loss": 0.011,
+      "step": 142050
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317797281911598e-05,
+      "loss": 0.0059,
+      "step": 142060
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317749242173127e-05,
+      "loss": 0.0062,
+      "step": 142070
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317701202434656e-05,
+      "loss": 0.0087,
+      "step": 142080
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317653162696182e-05,
+      "loss": 0.0074,
+      "step": 142090
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931760512295771e-05,
+      "loss": 0.0105,
+      "step": 142100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931755708321924e-05,
+      "loss": 0.0091,
+      "step": 142110
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931750904348077e-05,
+      "loss": 0.0083,
+      "step": 142120
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317461003742296e-05,
+      "loss": 0.0056,
+      "step": 142130
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317412964003825e-05,
+      "loss": 0.0045,
+      "step": 142140
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317364924265354e-05,
+      "loss": 0.0086,
+      "step": 142150
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317316884526884e-05,
+      "loss": 0.007,
+      "step": 142160
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931726884478841e-05,
+      "loss": 0.0059,
+      "step": 142170
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931722080504994e-05,
+      "loss": 0.0083,
+      "step": 142180
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317172765311468e-05,
+      "loss": 0.008,
+      "step": 142190
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317124725572994e-05,
+      "loss": 0.0089,
+      "step": 142200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317076685834523e-05,
+      "loss": 0.0052,
+      "step": 142210
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9317028646096053e-05,
+      "loss": 0.0093,
+      "step": 142220
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316980606357582e-05,
+      "loss": 0.0075,
+      "step": 142230
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316932566619108e-05,
+      "loss": 0.0063,
+      "step": 142240
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316884526880637e-05,
+      "loss": 0.0083,
+      "step": 142250
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316836487142166e-05,
+      "loss": 0.0093,
+      "step": 142260
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316788447403692e-05,
+      "loss": 0.0072,
+      "step": 142270
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931674040766522e-05,
+      "loss": 0.0084,
+      "step": 142280
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931669236792675e-05,
+      "loss": 0.0075,
+      "step": 142290
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931664432818828e-05,
+      "loss": 0.0095,
+      "step": 142300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316596288449806e-05,
+      "loss": 0.0072,
+      "step": 142310
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316548248711335e-05,
+      "loss": 0.006,
+      "step": 142320
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316500208972864e-05,
+      "loss": 0.0079,
+      "step": 142330
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316452169234394e-05,
+      "loss": 0.0057,
+      "step": 142340
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931640412949592e-05,
+      "loss": 0.0121,
+      "step": 142350
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931635608975745e-05,
+      "loss": 0.0057,
+      "step": 142360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316308050018978e-05,
+      "loss": 0.0089,
+      "step": 142370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316260010280504e-05,
+      "loss": 0.0054,
+      "step": 142380
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316211970542033e-05,
+      "loss": 0.0096,
+      "step": 142390
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316163930803562e-05,
+      "loss": 0.0069,
+      "step": 142400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931611589106509e-05,
+      "loss": 0.0075,
+      "step": 142410
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316067851326618e-05,
+      "loss": 0.0083,
+      "step": 142420
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9316019811588147e-05,
+      "loss": 0.008,
+      "step": 142430
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9315971771849676e-05,
+      "loss": 0.0098,
+      "step": 142440
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9315923732111202e-05,
+      "loss": 0.0082,
+      "step": 142450
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931587569237273e-05,
+      "loss": 0.0128,
+      "step": 142460
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931582765263426e-05,
+      "loss": 0.0077,
+      "step": 142470
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931577961289579e-05,
+      "loss": 0.0066,
+      "step": 142480
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9315731573157316e-05,
+      "loss": 0.0106,
+      "step": 142490
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9315683533418845e-05,
+      "loss": 0.0071,
+      "step": 142500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9315635493680374e-05,
+      "loss": 0.0059,
+      "step": 142510
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9315587453941903e-05,
+      "loss": 0.0086,
+      "step": 142520
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931553941420343e-05,
+      "loss": 0.0059,
+      "step": 142530
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.931549137446496e-05,
+      "loss": 0.0102,
+      "step": 142540
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9315443334726488e-05,
+      "loss": 0.0098,
+      "step": 142550
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9315395294988014e-05,
+      "loss": 0.0059,
+      "step": 142560
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9315347255249543e-05,
+      "loss": 0.0088,
+      "step": 142570
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9315299215511072e-05,
+      "loss": 0.005,
+      "step": 142580
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.93152511757726e-05,
+      "loss": 0.0082,
+      "step": 142590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9315203136034127e-05,
+      "loss": 0.0049,
+      "step": 142600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9315155096295657e-05,
+      "loss": 0.0064,
+      "step": 142610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9315107056557186e-05,
+      "loss": 0.0044,
+      "step": 142620
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9315059016818712e-05,
+      "loss": 0.0117,
+      "step": 142630
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931501097708024e-05,
+      "loss": 0.0081,
+      "step": 142640
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931496293734177e-05,
+      "loss": 0.0118,
+      "step": 142650
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.93149148976033e-05,
+      "loss": 0.0067,
+      "step": 142660
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314866857864826e-05,
+      "loss": 0.0062,
+      "step": 142670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314818818126355e-05,
+      "loss": 0.0056,
+      "step": 142680
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314770778387884e-05,
+      "loss": 0.0062,
+      "step": 142690
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314722738649413e-05,
+      "loss": 0.0087,
+      "step": 142700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931467469891094e-05,
+      "loss": 0.0094,
+      "step": 142710
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931462665917247e-05,
+      "loss": 0.0052,
+      "step": 142720
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314578619433998e-05,
+      "loss": 0.0052,
+      "step": 142730
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314530579695524e-05,
+      "loss": 0.0095,
+      "step": 142740
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314482539957053e-05,
+      "loss": 0.0067,
+      "step": 142750
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314434500218582e-05,
+      "loss": 0.0048,
+      "step": 142760
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931438646048011e-05,
+      "loss": 0.0071,
+      "step": 142770
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314338420741637e-05,
+      "loss": 0.009,
+      "step": 142780
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314290381003167e-05,
+      "loss": 0.0085,
+      "step": 142790
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314242341264696e-05,
+      "loss": 0.0067,
+      "step": 142800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314194301526222e-05,
+      "loss": 0.0093,
+      "step": 142810
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931414626178775e-05,
+      "loss": 0.0078,
+      "step": 142820
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931409822204928e-05,
+      "loss": 0.0086,
+      "step": 142830
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931405018231081e-05,
+      "loss": 0.0091,
+      "step": 142840
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9314002142572336e-05,
+      "loss": 0.0081,
+      "step": 142850
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313954102833865e-05,
+      "loss": 0.0076,
+      "step": 142860
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313906063095394e-05,
+      "loss": 0.0067,
+      "step": 142870
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313858023356923e-05,
+      "loss": 0.0049,
+      "step": 142880
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931380998361845e-05,
+      "loss": 0.0091,
+      "step": 142890
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931376194387998e-05,
+      "loss": 0.008,
+      "step": 142900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313713904141508e-05,
+      "loss": 0.006,
+      "step": 142910
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313665864403034e-05,
+      "loss": 0.0085,
+      "step": 142920
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313617824664563e-05,
+      "loss": 0.0082,
+      "step": 142930
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313569784926092e-05,
+      "loss": 0.0143,
+      "step": 142940
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931352174518762e-05,
+      "loss": 0.0065,
+      "step": 142950
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313473705449147e-05,
+      "loss": 0.0123,
+      "step": 142960
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313425665710677e-05,
+      "loss": 0.0062,
+      "step": 142970
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313377625972206e-05,
+      "loss": 0.0125,
+      "step": 142980
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313329586233732e-05,
+      "loss": 0.0071,
+      "step": 142990
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931328154649526e-05,
+      "loss": 0.0085,
+      "step": 143000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931323350675679e-05,
+      "loss": 0.0089,
+      "step": 143010
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931318546701832e-05,
+      "loss": 0.0088,
+      "step": 143020
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313137427279845e-05,
+      "loss": 0.0055,
+      "step": 143030
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313089387541375e-05,
+      "loss": 0.0095,
+      "step": 143040
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9313041347802904e-05,
+      "loss": 0.0092,
+      "step": 143050
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312993308064433e-05,
+      "loss": 0.0056,
+      "step": 143060
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931294526832596e-05,
+      "loss": 0.0103,
+      "step": 143070
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931289722858749e-05,
+      "loss": 0.0074,
+      "step": 143080
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312849188849018e-05,
+      "loss": 0.0082,
+      "step": 143090
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312801149110544e-05,
+      "loss": 0.0052,
+      "step": 143100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312753109372073e-05,
+      "loss": 0.0068,
+      "step": 143110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312705069633602e-05,
+      "loss": 0.0061,
+      "step": 143120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931265702989513e-05,
+      "loss": 0.0093,
+      "step": 143130
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312608990156657e-05,
+      "loss": 0.0075,
+      "step": 143140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312560950418186e-05,
+      "loss": 0.0073,
+      "step": 143150
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312512910679716e-05,
+      "loss": 0.0065,
+      "step": 143160
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931246487094124e-05,
+      "loss": 0.0074,
+      "step": 143170
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931241683120277e-05,
+      "loss": 0.0092,
+      "step": 143180
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.93123687914643e-05,
+      "loss": 0.0098,
+      "step": 143190
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931232075172583e-05,
+      "loss": 0.006,
+      "step": 143200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312272711987355e-05,
+      "loss": 0.0081,
+      "step": 143210
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312224672248885e-05,
+      "loss": 0.0051,
+      "step": 143220
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312176632510414e-05,
+      "loss": 0.0066,
+      "step": 143230
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312128592771943e-05,
+      "loss": 0.0082,
+      "step": 143240
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.931208055303347e-05,
+      "loss": 0.0067,
+      "step": 143250
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9312032513294998e-05,
+      "loss": 0.0084,
+      "step": 143260
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9311984473556528e-05,
+      "loss": 0.0077,
+      "step": 143270
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9311936433818053e-05,
+      "loss": 0.008,
+      "step": 143280
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311888394079583e-05,
+      "loss": 0.009,
+      "step": 143290
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311840354341112e-05,
+      "loss": 0.0101,
+      "step": 143300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931179231460264e-05,
+      "loss": 0.0073,
+      "step": 143310
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311744274864167e-05,
+      "loss": 0.0063,
+      "step": 143320
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311696235125696e-05,
+      "loss": 0.0082,
+      "step": 143330
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311648195387226e-05,
+      "loss": 0.009,
+      "step": 143340
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931160015564875e-05,
+      "loss": 0.0089,
+      "step": 143350
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311552115910284e-05,
+      "loss": 0.0068,
+      "step": 143360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311504076171813e-05,
+      "loss": 0.0094,
+      "step": 143370
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931145603643334e-05,
+      "loss": 0.0075,
+      "step": 143380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931140799669487e-05,
+      "loss": 0.0058,
+      "step": 143390
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311359956956398e-05,
+      "loss": 0.0078,
+      "step": 143400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311311917217924e-05,
+      "loss": 0.0087,
+      "step": 143410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311263877479453e-05,
+      "loss": 0.0086,
+      "step": 143420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311215837740982e-05,
+      "loss": 0.0081,
+      "step": 143430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931116779800251e-05,
+      "loss": 0.0059,
+      "step": 143440
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311119758264037e-05,
+      "loss": 0.007,
+      "step": 143450
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311071718525567e-05,
+      "loss": 0.0071,
+      "step": 143460
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9311023678787096e-05,
+      "loss": 0.01,
+      "step": 143470
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310975639048622e-05,
+      "loss": 0.008,
+      "step": 143480
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931092759931015e-05,
+      "loss": 0.0078,
+      "step": 143490
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931087955957168e-05,
+      "loss": 0.0069,
+      "step": 143500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931083151983321e-05,
+      "loss": 0.0094,
+      "step": 143510
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310783480094736e-05,
+      "loss": 0.0064,
+      "step": 143520
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310735440356265e-05,
+      "loss": 0.0074,
+      "step": 143530
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310687400617794e-05,
+      "loss": 0.0074,
+      "step": 143540
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310639360879323e-05,
+      "loss": 0.0084,
+      "step": 143550
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931059132114085e-05,
+      "loss": 0.0093,
+      "step": 143560
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931054328140238e-05,
+      "loss": 0.0089,
+      "step": 143570
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310495241663908e-05,
+      "loss": 0.0084,
+      "step": 143580
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310447201925434e-05,
+      "loss": 0.0057,
+      "step": 143590
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310399162186963e-05,
+      "loss": 0.0051,
+      "step": 143600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310351122448492e-05,
+      "loss": 0.0087,
+      "step": 143610
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931030308271002e-05,
+      "loss": 0.0063,
+      "step": 143620
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310255042971547e-05,
+      "loss": 0.0086,
+      "step": 143630
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310207003233077e-05,
+      "loss": 0.0079,
+      "step": 143640
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310158963494606e-05,
+      "loss": 0.0065,
+      "step": 143650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9310110923756132e-05,
+      "loss": 0.0063,
+      "step": 143660
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931006288401766e-05,
+      "loss": 0.0097,
+      "step": 143670
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.931001484427919e-05,
+      "loss": 0.0085,
+      "step": 143680
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.930996680454072e-05,
+      "loss": 0.0134,
+      "step": 143690
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309918764802246e-05,
+      "loss": 0.0095,
+      "step": 143700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309870725063775e-05,
+      "loss": 0.0063,
+      "step": 143710
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309822685325304e-05,
+      "loss": 0.0086,
+      "step": 143720
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309774645586833e-05,
+      "loss": 0.0081,
+      "step": 143730
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.930972660584836e-05,
+      "loss": 0.0066,
+      "step": 143740
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.930967856610989e-05,
+      "loss": 0.0095,
+      "step": 143750
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309630526371418e-05,
+      "loss": 0.0066,
+      "step": 143760
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309582486632944e-05,
+      "loss": 0.0066,
+      "step": 143770
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309534446894473e-05,
+      "loss": 0.0082,
+      "step": 143780
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309486407156002e-05,
+      "loss": 0.0073,
+      "step": 143790
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.930943836741753e-05,
+      "loss": 0.0121,
+      "step": 143800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309390327679057e-05,
+      "loss": 0.0055,
+      "step": 143810
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309342287940587e-05,
+      "loss": 0.0068,
+      "step": 143820
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309294248202116e-05,
+      "loss": 0.0066,
+      "step": 143830
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309246208463642e-05,
+      "loss": 0.0062,
+      "step": 143840
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309202972699016e-05,
+      "loss": 0.0051,
+      "step": 143850
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309154932960546e-05,
+      "loss": 0.0085,
+      "step": 143860
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309106893222075e-05,
+      "loss": 0.0073,
+      "step": 143870
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309058853483604e-05,
+      "loss": 0.0071,
+      "step": 143880
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9309010813745133e-05,
+      "loss": 0.0072,
+      "step": 143890
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.930896277400666e-05,
+      "loss": 0.0101,
+      "step": 143900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.930891473426819e-05,
+      "loss": 0.0088,
+      "step": 143910
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9308866694529718e-05,
+      "loss": 0.0078,
+      "step": 143920
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9308818654791247e-05,
+      "loss": 0.0072,
+      "step": 143930
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9308770615052773e-05,
+      "loss": 0.0063,
+      "step": 143940
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9308722575314302e-05,
+      "loss": 0.0074,
+      "step": 143950
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.930867453557583e-05,
+      "loss": 0.0086,
+      "step": 143960
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.930862649583736e-05,
+      "loss": 0.0065,
+      "step": 143970
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9308578456098887e-05,
+      "loss": 0.0053,
+      "step": 143980
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9308530416360416e-05,
+      "loss": 0.0067,
+      "step": 143990
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9308482376621945e-05,
+      "loss": 0.0065,
+      "step": 144000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930843433688347e-05,
+      "loss": 0.0067,
+      "step": 144010
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9308386297145e-05,
+      "loss": 0.008,
+      "step": 144020
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930833825740653e-05,
+      "loss": 0.0065,
+      "step": 144030
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930829021766806e-05,
+      "loss": 0.0056,
+      "step": 144040
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9308242177929585e-05,
+      "loss": 0.0087,
+      "step": 144050
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9308194138191114e-05,
+      "loss": 0.0078,
+      "step": 144060
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9308146098452643e-05,
+      "loss": 0.0063,
+      "step": 144070
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930809805871417e-05,
+      "loss": 0.0103,
+      "step": 144080
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.93080500189757e-05,
+      "loss": 0.0073,
+      "step": 144090
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9308001979237228e-05,
+      "loss": 0.0077,
+      "step": 144100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9307953939498757e-05,
+      "loss": 0.007,
+      "step": 144110
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9307905899760283e-05,
+      "loss": 0.0079,
+      "step": 144120
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9307857860021812e-05,
+      "loss": 0.006,
+      "step": 144130
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930780982028334e-05,
+      "loss": 0.0082,
+      "step": 144140
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930776178054487e-05,
+      "loss": 0.0095,
+      "step": 144150
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9307713740806397e-05,
+      "loss": 0.0065,
+      "step": 144160
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9307665701067926e-05,
+      "loss": 0.0071,
+      "step": 144170
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9307617661329455e-05,
+      "loss": 0.0066,
+      "step": 144180
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930756962159098e-05,
+      "loss": 0.0094,
+      "step": 144190
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930752158185251e-05,
+      "loss": 0.0081,
+      "step": 144200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930747354211404e-05,
+      "loss": 0.0069,
+      "step": 144210
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930742550237557e-05,
+      "loss": 0.0096,
+      "step": 144220
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9307377462637095e-05,
+      "loss": 0.0072,
+      "step": 144230
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9307329422898624e-05,
+      "loss": 0.0087,
+      "step": 144240
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9307281383160153e-05,
+      "loss": 0.0056,
+      "step": 144250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930723334342168e-05,
+      "loss": 0.0056,
+      "step": 144260
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930718530368321e-05,
+      "loss": 0.0083,
+      "step": 144270
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9307137263944738e-05,
+      "loss": 0.0114,
+      "step": 144280
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9307089224206267e-05,
+      "loss": 0.0062,
+      "step": 144290
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9307041184467793e-05,
+      "loss": 0.0085,
+      "step": 144300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9306993144729322e-05,
+      "loss": 0.0074,
+      "step": 144310
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930694510499085e-05,
+      "loss": 0.0058,
+      "step": 144320
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930689706525238e-05,
+      "loss": 0.0101,
+      "step": 144330
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9306849025513907e-05,
+      "loss": 0.0108,
+      "step": 144340
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9306800985775436e-05,
+      "loss": 0.0063,
+      "step": 144350
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9306752946036965e-05,
+      "loss": 0.0073,
+      "step": 144360
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930670490629849e-05,
+      "loss": 0.0067,
+      "step": 144370
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930665686656002e-05,
+      "loss": 0.0053,
+      "step": 144380
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930660882682155e-05,
+      "loss": 0.0112,
+      "step": 144390
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930656078708308e-05,
+      "loss": 0.0107,
+      "step": 144400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9306512747344605e-05,
+      "loss": 0.0054,
+      "step": 144410
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9306464707606134e-05,
+      "loss": 0.0093,
+      "step": 144420
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9306416667867663e-05,
+      "loss": 0.0083,
+      "step": 144430
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930636862812919e-05,
+      "loss": 0.0096,
+      "step": 144440
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930632058839072e-05,
+      "loss": 0.01,
+      "step": 144450
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9306272548652248e-05,
+      "loss": 0.0085,
+      "step": 144460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9306224508913777e-05,
+      "loss": 0.0073,
+      "step": 144470
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9306176469175303e-05,
+      "loss": 0.0095,
+      "step": 144480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9306128429436832e-05,
+      "loss": 0.0108,
+      "step": 144490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930608038969836e-05,
+      "loss": 0.006,
+      "step": 144500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930603234995989e-05,
+      "loss": 0.0066,
+      "step": 144510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9305984310221416e-05,
+      "loss": 0.0061,
+      "step": 144520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9305936270482946e-05,
+      "loss": 0.0077,
+      "step": 144530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9305888230744475e-05,
+      "loss": 0.0091,
+      "step": 144540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9305840191006e-05,
+      "loss": 0.0076,
+      "step": 144550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930579215126753e-05,
+      "loss": 0.0095,
+      "step": 144560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930574411152906e-05,
+      "loss": 0.0085,
+      "step": 144570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.930569607179059e-05,
+      "loss": 0.0064,
+      "step": 144580
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9305648032052115e-05,
+      "loss": 0.0071,
+      "step": 144590
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9305599992313644e-05,
+      "loss": 0.0083,
+      "step": 144600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9305551952575173e-05,
+      "loss": 0.0076,
+      "step": 144610
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.93055039128367e-05,
+      "loss": 0.0081,
+      "step": 144620
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9305455873098228e-05,
+      "loss": 0.0059,
+      "step": 144630
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9305407833359758e-05,
+      "loss": 0.0078,
+      "step": 144640
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9305359793621287e-05,
+      "loss": 0.008,
+      "step": 144650
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9305311753882813e-05,
+      "loss": 0.008,
+      "step": 144660
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9305263714144342e-05,
+      "loss": 0.0069,
+      "step": 144670
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930521567440587e-05,
+      "loss": 0.005,
+      "step": 144680
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.93051676346674e-05,
+      "loss": 0.0074,
+      "step": 144690
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9305119594928926e-05,
+      "loss": 0.0052,
+      "step": 144700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9305071555190456e-05,
+      "loss": 0.0081,
+      "step": 144710
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9305023515451985e-05,
+      "loss": 0.0049,
+      "step": 144720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930497547571351e-05,
+      "loss": 0.0112,
+      "step": 144730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930492743597504e-05,
+      "loss": 0.0088,
+      "step": 144740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930487939623657e-05,
+      "loss": 0.0117,
+      "step": 144750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.93048313564981e-05,
+      "loss": 0.007,
+      "step": 144760
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9304783316759624e-05,
+      "loss": 0.0109,
+      "step": 144770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9304735277021154e-05,
+      "loss": 0.0083,
+      "step": 144780
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9304687237282683e-05,
+      "loss": 0.0094,
+      "step": 144790
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930463919754421e-05,
+      "loss": 0.009,
+      "step": 144800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9304591157805738e-05,
+      "loss": 0.0041,
+      "step": 144810
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9304543118067267e-05,
+      "loss": 0.0074,
+      "step": 144820
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9304495078328797e-05,
+      "loss": 0.0063,
+      "step": 144830
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9304447038590323e-05,
+      "loss": 0.0093,
+      "step": 144840
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9304398998851852e-05,
+      "loss": 0.01,
+      "step": 144850
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930435095911338e-05,
+      "loss": 0.0073,
+      "step": 144860
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930430291937491e-05,
+      "loss": 0.0079,
+      "step": 144870
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9304254879636436e-05,
+      "loss": 0.0064,
+      "step": 144880
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9304206839897966e-05,
+      "loss": 0.0076,
+      "step": 144890
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9304158800159495e-05,
+      "loss": 0.0058,
+      "step": 144900
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930411076042102e-05,
+      "loss": 0.0082,
+      "step": 144910
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930406272068255e-05,
+      "loss": 0.0075,
+      "step": 144920
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930401468094408e-05,
+      "loss": 0.0071,
+      "step": 144930
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930396664120561e-05,
+      "loss": 0.0086,
+      "step": 144940
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303918601467134e-05,
+      "loss": 0.0067,
+      "step": 144950
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303870561728664e-05,
+      "loss": 0.0074,
+      "step": 144960
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303822521990193e-05,
+      "loss": 0.008,
+      "step": 144970
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930377448225172e-05,
+      "loss": 0.0085,
+      "step": 144980
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303726442513248e-05,
+      "loss": 0.0091,
+      "step": 144990
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303678402774777e-05,
+      "loss": 0.0112,
+      "step": 145000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303630363036307e-05,
+      "loss": 0.0064,
+      "step": 145010
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303582323297833e-05,
+      "loss": 0.0099,
+      "step": 145020
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303534283559362e-05,
+      "loss": 0.0086,
+      "step": 145030
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930348624382089e-05,
+      "loss": 0.0086,
+      "step": 145040
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303438204082417e-05,
+      "loss": 0.0067,
+      "step": 145050
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303390164343946e-05,
+      "loss": 0.0065,
+      "step": 145060
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303342124605475e-05,
+      "loss": 0.0089,
+      "step": 145070
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303294084867005e-05,
+      "loss": 0.0076,
+      "step": 145080
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930324604512853e-05,
+      "loss": 0.0055,
+      "step": 145090
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930319800539006e-05,
+      "loss": 0.0064,
+      "step": 145100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930314996565159e-05,
+      "loss": 0.0064,
+      "step": 145110
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930310192591312e-05,
+      "loss": 0.0079,
+      "step": 145120
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303053886174644e-05,
+      "loss": 0.0069,
+      "step": 145130
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9303005846436174e-05,
+      "loss": 0.0086,
+      "step": 145140
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302957806697703e-05,
+      "loss": 0.0044,
+      "step": 145150
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930290976695923e-05,
+      "loss": 0.0078,
+      "step": 145160
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302861727220758e-05,
+      "loss": 0.0047,
+      "step": 145170
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302813687482287e-05,
+      "loss": 0.0073,
+      "step": 145180
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302765647743817e-05,
+      "loss": 0.0074,
+      "step": 145190
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302717608005342e-05,
+      "loss": 0.0103,
+      "step": 145200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302669568266872e-05,
+      "loss": 0.0069,
+      "step": 145210
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.93026215285284e-05,
+      "loss": 0.0085,
+      "step": 145220
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302573488789927e-05,
+      "loss": 0.0088,
+      "step": 145230
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302525449051456e-05,
+      "loss": 0.0081,
+      "step": 145240
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302477409312985e-05,
+      "loss": 0.0095,
+      "step": 145250
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302429369574515e-05,
+      "loss": 0.0086,
+      "step": 145260
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930238132983604e-05,
+      "loss": 0.0117,
+      "step": 145270
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930233329009757e-05,
+      "loss": 0.0085,
+      "step": 145280
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.93022852503591e-05,
+      "loss": 0.0099,
+      "step": 145290
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930223721062063e-05,
+      "loss": 0.0094,
+      "step": 145300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302189170882154e-05,
+      "loss": 0.0069,
+      "step": 145310
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302141131143684e-05,
+      "loss": 0.0076,
+      "step": 145320
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9302093091405213e-05,
+      "loss": 0.0064,
+      "step": 145330
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.930204505166674e-05,
+      "loss": 0.0086,
+      "step": 145340
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9301997011928268e-05,
+      "loss": 0.0122,
+      "step": 145350
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9301948972189797e-05,
+      "loss": 0.0077,
+      "step": 145360
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301900932451326e-05,
+      "loss": 0.0096,
+      "step": 145370
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301852892712852e-05,
+      "loss": 0.0085,
+      "step": 145380
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930180485297438e-05,
+      "loss": 0.0087,
+      "step": 145390
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930175681323591e-05,
+      "loss": 0.0101,
+      "step": 145400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301708773497437e-05,
+      "loss": 0.0093,
+      "step": 145410
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301660733758966e-05,
+      "loss": 0.0069,
+      "step": 145420
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301612694020495e-05,
+      "loss": 0.0079,
+      "step": 145430
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301564654282025e-05,
+      "loss": 0.0091,
+      "step": 145440
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930151661454355e-05,
+      "loss": 0.0082,
+      "step": 145450
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930146857480508e-05,
+      "loss": 0.0096,
+      "step": 145460
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930142053506661e-05,
+      "loss": 0.0094,
+      "step": 145470
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301372495328138e-05,
+      "loss": 0.0066,
+      "step": 145480
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301324455589664e-05,
+      "loss": 0.0058,
+      "step": 145490
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301276415851193e-05,
+      "loss": 0.008,
+      "step": 145500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301228376112723e-05,
+      "loss": 0.0088,
+      "step": 145510
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930118033637425e-05,
+      "loss": 0.008,
+      "step": 145520
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301132296635778e-05,
+      "loss": 0.0065,
+      "step": 145530
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301084256897307e-05,
+      "loss": 0.0099,
+      "step": 145540
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9301036217158836e-05,
+      "loss": 0.0072,
+      "step": 145550
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300988177420362e-05,
+      "loss": 0.0096,
+      "step": 145560
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930094013768189e-05,
+      "loss": 0.0055,
+      "step": 145570
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930089209794342e-05,
+      "loss": 0.0103,
+      "step": 145580
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300844058204947e-05,
+      "loss": 0.0063,
+      "step": 145590
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300796018466476e-05,
+      "loss": 0.0088,
+      "step": 145600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300747978728005e-05,
+      "loss": 0.0069,
+      "step": 145610
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300699938989535e-05,
+      "loss": 0.0099,
+      "step": 145620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930065189925106e-05,
+      "loss": 0.0075,
+      "step": 145630
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930060385951259e-05,
+      "loss": 0.0081,
+      "step": 145640
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930055581977412e-05,
+      "loss": 0.0088,
+      "step": 145650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300507780035648e-05,
+      "loss": 0.0053,
+      "step": 145660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300459740297174e-05,
+      "loss": 0.0044,
+      "step": 145670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300411700558703e-05,
+      "loss": 0.0101,
+      "step": 145680
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300363660820233e-05,
+      "loss": 0.0082,
+      "step": 145690
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930031562108176e-05,
+      "loss": 0.0069,
+      "step": 145700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300267581343288e-05,
+      "loss": 0.0065,
+      "step": 145710
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300219541604817e-05,
+      "loss": 0.0115,
+      "step": 145720
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300171501866346e-05,
+      "loss": 0.0065,
+      "step": 145730
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9300123462127872e-05,
+      "loss": 0.0083,
+      "step": 145740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.93000754223894e-05,
+      "loss": 0.0087,
+      "step": 145750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.930002738265093e-05,
+      "loss": 0.0126,
+      "step": 145760
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299979342912457e-05,
+      "loss": 0.0053,
+      "step": 145770
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299931303173986e-05,
+      "loss": 0.008,
+      "step": 145780
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299883263435515e-05,
+      "loss": 0.0056,
+      "step": 145790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299835223697044e-05,
+      "loss": 0.0054,
+      "step": 145800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.929978718395857e-05,
+      "loss": 0.0082,
+      "step": 145810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.92997391442201e-05,
+      "loss": 0.0088,
+      "step": 145820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.929969110448163e-05,
+      "loss": 0.0095,
+      "step": 145830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299643064743158e-05,
+      "loss": 0.0069,
+      "step": 145840
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299595025004684e-05,
+      "loss": 0.0062,
+      "step": 145850
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299546985266213e-05,
+      "loss": 0.0077,
+      "step": 145860
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299498945527743e-05,
+      "loss": 0.0101,
+      "step": 145870
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.929945090578927e-05,
+      "loss": 0.0083,
+      "step": 145880
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299402866050798e-05,
+      "loss": 0.0073,
+      "step": 145890
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299354826312327e-05,
+      "loss": 0.0072,
+      "step": 145900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299306786573856e-05,
+      "loss": 0.01,
+      "step": 145910
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299258746835382e-05,
+      "loss": 0.0066,
+      "step": 145920
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.929921070709691e-05,
+      "loss": 0.005,
+      "step": 145930
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.929916266735844e-05,
+      "loss": 0.0085,
+      "step": 145940
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299114627619967e-05,
+      "loss": 0.0071,
+      "step": 145950
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299066587881496e-05,
+      "loss": 0.0072,
+      "step": 145960
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9299018548143025e-05,
+      "loss": 0.0082,
+      "step": 145970
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9298970508404554e-05,
+      "loss": 0.0084,
+      "step": 145980
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.929892246866608e-05,
+      "loss": 0.0116,
+      "step": 145990
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.929887442892761e-05,
+      "loss": 0.0096,
+      "step": 146000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.929882638918914e-05,
+      "loss": 0.0096,
+      "step": 146010
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9298778349450668e-05,
+      "loss": 0.0076,
+      "step": 146020
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9298730309712194e-05,
+      "loss": 0.0086,
+      "step": 146030
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9298682269973723e-05,
+      "loss": 0.0075,
+      "step": 146040
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9298634230235252e-05,
+      "loss": 0.0103,
+      "step": 146050
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929858619049678e-05,
+      "loss": 0.0095,
+      "step": 146060
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9298538150758308e-05,
+      "loss": 0.0086,
+      "step": 146070
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9298490111019837e-05,
+      "loss": 0.0087,
+      "step": 146080
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9298442071281366e-05,
+      "loss": 0.0082,
+      "step": 146090
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9298394031542892e-05,
+      "loss": 0.0088,
+      "step": 146100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929834599180442e-05,
+      "loss": 0.0075,
+      "step": 146110
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929829795206595e-05,
+      "loss": 0.0068,
+      "step": 146120
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9298249912327476e-05,
+      "loss": 0.0104,
+      "step": 146130
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9298201872589006e-05,
+      "loss": 0.0084,
+      "step": 146140
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9298153832850535e-05,
+      "loss": 0.0068,
+      "step": 146150
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9298105793112064e-05,
+      "loss": 0.0064,
+      "step": 146160
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929805775337359e-05,
+      "loss": 0.0107,
+      "step": 146170
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929800971363512e-05,
+      "loss": 0.008,
+      "step": 146180
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929796167389665e-05,
+      "loss": 0.0076,
+      "step": 146190
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297913634158178e-05,
+      "loss": 0.0083,
+      "step": 146200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297865594419704e-05,
+      "loss": 0.0065,
+      "step": 146210
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297817554681233e-05,
+      "loss": 0.0079,
+      "step": 146220
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297769514942762e-05,
+      "loss": 0.0058,
+      "step": 146230
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297721475204288e-05,
+      "loss": 0.0073,
+      "step": 146240
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297673435465818e-05,
+      "loss": 0.0057,
+      "step": 146250
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297625395727347e-05,
+      "loss": 0.0073,
+      "step": 146260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297577355988876e-05,
+      "loss": 0.0096,
+      "step": 146270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297529316250402e-05,
+      "loss": 0.0048,
+      "step": 146280
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929748127651193e-05,
+      "loss": 0.0059,
+      "step": 146290
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929743323677346e-05,
+      "loss": 0.0085,
+      "step": 146300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297385197034986e-05,
+      "loss": 0.0071,
+      "step": 146310
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297337157296516e-05,
+      "loss": 0.0068,
+      "step": 146320
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297289117558045e-05,
+      "loss": 0.0088,
+      "step": 146330
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297241077819574e-05,
+      "loss": 0.0085,
+      "step": 146340
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.92971930380811e-05,
+      "loss": 0.0063,
+      "step": 146350
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929714499834263e-05,
+      "loss": 0.0093,
+      "step": 146360
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929709695860416e-05,
+      "loss": 0.0073,
+      "step": 146370
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297048918865688e-05,
+      "loss": 0.0072,
+      "step": 146380
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9297000879127214e-05,
+      "loss": 0.0074,
+      "step": 146390
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296952839388743e-05,
+      "loss": 0.0078,
+      "step": 146400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296904799650272e-05,
+      "loss": 0.0057,
+      "step": 146410
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296856759911798e-05,
+      "loss": 0.0049,
+      "step": 146420
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296808720173327e-05,
+      "loss": 0.0084,
+      "step": 146430
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929676068043486e-05,
+      "loss": 0.0093,
+      "step": 146440
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296712640696386e-05,
+      "loss": 0.0112,
+      "step": 146450
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296664600957915e-05,
+      "loss": 0.0107,
+      "step": 146460
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296616561219445e-05,
+      "loss": 0.0046,
+      "step": 146470
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929656852148097e-05,
+      "loss": 0.0078,
+      "step": 146480
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.92965204817425e-05,
+      "loss": 0.0075,
+      "step": 146490
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929647244200403e-05,
+      "loss": 0.0115,
+      "step": 146500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296424402265558e-05,
+      "loss": 0.0089,
+      "step": 146510
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296376362527084e-05,
+      "loss": 0.0118,
+      "step": 146520
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296328322788613e-05,
+      "loss": 0.0061,
+      "step": 146530
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296280283050143e-05,
+      "loss": 0.0068,
+      "step": 146540
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929623224331167e-05,
+      "loss": 0.0079,
+      "step": 146550
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296184203573198e-05,
+      "loss": 0.0085,
+      "step": 146560
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296136163834727e-05,
+      "loss": 0.0091,
+      "step": 146570
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296088124096256e-05,
+      "loss": 0.0095,
+      "step": 146580
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9296040084357782e-05,
+      "loss": 0.0065,
+      "step": 146590
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929599204461931e-05,
+      "loss": 0.012,
+      "step": 146600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929594400488084e-05,
+      "loss": 0.0117,
+      "step": 146610
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9295895965142367e-05,
+      "loss": 0.0063,
+      "step": 146620
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9295847925403896e-05,
+      "loss": 0.0059,
+      "step": 146630
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9295799885665425e-05,
+      "loss": 0.0102,
+      "step": 146640
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9295751845926954e-05,
+      "loss": 0.0089,
+      "step": 146650
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929570380618848e-05,
+      "loss": 0.0081,
+      "step": 146660
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929565576645001e-05,
+      "loss": 0.0069,
+      "step": 146670
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9295612530685384e-05,
+      "loss": 0.0087,
+      "step": 146680
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9295564490946913e-05,
+      "loss": 0.0086,
+      "step": 146690
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929551645120844e-05,
+      "loss": 0.0059,
+      "step": 146700
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.929546841146997e-05,
+      "loss": 0.0053,
+      "step": 146710
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9295420371731498e-05,
+      "loss": 0.0086,
+      "step": 146720
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9295372331993024e-05,
+      "loss": 0.0066,
+      "step": 146730
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9295324292254553e-05,
+      "loss": 0.0072,
+      "step": 146740
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9295276252516082e-05,
+      "loss": 0.0051,
+      "step": 146750
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929522821277761e-05,
+      "loss": 0.008,
+      "step": 146760
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9295180173039137e-05,
+      "loss": 0.0066,
+      "step": 146770
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9295132133300667e-05,
+      "loss": 0.0068,
+      "step": 146780
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9295084093562196e-05,
+      "loss": 0.0094,
+      "step": 146790
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9295036053823722e-05,
+      "loss": 0.0066,
+      "step": 146800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929498801408525e-05,
+      "loss": 0.0077,
+      "step": 146810
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929493997434678e-05,
+      "loss": 0.0084,
+      "step": 146820
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929489193460831e-05,
+      "loss": 0.008,
+      "step": 146830
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9294843894869836e-05,
+      "loss": 0.0223,
+      "step": 146840
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9294795855131365e-05,
+      "loss": 0.0087,
+      "step": 146850
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9294747815392894e-05,
+      "loss": 0.0054,
+      "step": 146860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9294699775654423e-05,
+      "loss": 0.0066,
+      "step": 146870
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929465173591595e-05,
+      "loss": 0.006,
+      "step": 146880
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929460369617748e-05,
+      "loss": 0.0106,
+      "step": 146890
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9294555656439008e-05,
+      "loss": 0.0109,
+      "step": 146900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9294507616700534e-05,
+      "loss": 0.011,
+      "step": 146910
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9294459576962063e-05,
+      "loss": 0.0079,
+      "step": 146920
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9294411537223592e-05,
+      "loss": 0.0079,
+      "step": 146930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929436349748512e-05,
+      "loss": 0.0077,
+      "step": 146940
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929431545774665e-05,
+      "loss": 0.0073,
+      "step": 146950
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929426741800818e-05,
+      "loss": 0.0084,
+      "step": 146960
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9294219378269706e-05,
+      "loss": 0.0055,
+      "step": 146970
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9294171338531235e-05,
+      "loss": 0.008,
+      "step": 146980
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9294123298792764e-05,
+      "loss": 0.008,
+      "step": 146990
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9294075259054294e-05,
+      "loss": 0.0122,
+      "step": 147000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929402721931582e-05,
+      "loss": 0.0119,
+      "step": 147010
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929397917957735e-05,
+      "loss": 0.0126,
+      "step": 147020
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293931139838878e-05,
+      "loss": 0.01,
+      "step": 147030
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293883100100404e-05,
+      "loss": 0.0052,
+      "step": 147040
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293835060361933e-05,
+      "loss": 0.0087,
+      "step": 147050
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293787020623463e-05,
+      "loss": 0.0113,
+      "step": 147060
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293738980884992e-05,
+      "loss": 0.0057,
+      "step": 147070
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293690941146518e-05,
+      "loss": 0.0072,
+      "step": 147080
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293642901408047e-05,
+      "loss": 0.0084,
+      "step": 147090
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293594861669576e-05,
+      "loss": 0.007,
+      "step": 147100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293546821931106e-05,
+      "loss": 0.0083,
+      "step": 147110
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929349878219263e-05,
+      "loss": 0.0083,
+      "step": 147120
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929345074245416e-05,
+      "loss": 0.0065,
+      "step": 147130
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929340270271569e-05,
+      "loss": 0.0071,
+      "step": 147140
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293354662977216e-05,
+      "loss": 0.0095,
+      "step": 147150
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293306623238745e-05,
+      "loss": 0.0072,
+      "step": 147160
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293258583500274e-05,
+      "loss": 0.0098,
+      "step": 147170
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293210543761804e-05,
+      "loss": 0.0095,
+      "step": 147180
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929316250402333e-05,
+      "loss": 0.009,
+      "step": 147190
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929311446428486e-05,
+      "loss": 0.0103,
+      "step": 147200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293066424546388e-05,
+      "loss": 0.0094,
+      "step": 147210
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9293018384807914e-05,
+      "loss": 0.006,
+      "step": 147220
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292970345069443e-05,
+      "loss": 0.007,
+      "step": 147230
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292922305330973e-05,
+      "loss": 0.0062,
+      "step": 147240
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292874265592502e-05,
+      "loss": 0.0075,
+      "step": 147250
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292826225854028e-05,
+      "loss": 0.0085,
+      "step": 147260
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292778186115557e-05,
+      "loss": 0.0063,
+      "step": 147270
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292730146377086e-05,
+      "loss": 0.0082,
+      "step": 147280
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292682106638615e-05,
+      "loss": 0.0096,
+      "step": 147290
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929263406690014e-05,
+      "loss": 0.0069,
+      "step": 147300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929258602716167e-05,
+      "loss": 0.0071,
+      "step": 147310
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.92925379874232e-05,
+      "loss": 0.0065,
+      "step": 147320
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292489947684726e-05,
+      "loss": 0.008,
+      "step": 147330
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292441907946255e-05,
+      "loss": 0.014,
+      "step": 147340
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292393868207784e-05,
+      "loss": 0.0086,
+      "step": 147350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292345828469314e-05,
+      "loss": 0.0103,
+      "step": 147360
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929229778873084e-05,
+      "loss": 0.0052,
+      "step": 147370
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929224974899237e-05,
+      "loss": 0.0084,
+      "step": 147380
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292201709253898e-05,
+      "loss": 0.0083,
+      "step": 147390
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292153669515424e-05,
+      "loss": 0.0081,
+      "step": 147400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292105629776953e-05,
+      "loss": 0.0106,
+      "step": 147410
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9292057590038482e-05,
+      "loss": 0.0099,
+      "step": 147420
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.929200955030001e-05,
+      "loss": 0.0082,
+      "step": 147430
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9291961510561538e-05,
+      "loss": 0.0082,
+      "step": 147440
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291913470823067e-05,
+      "loss": 0.0067,
+      "step": 147450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291865431084596e-05,
+      "loss": 0.0064,
+      "step": 147460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291817391346125e-05,
+      "loss": 0.0104,
+      "step": 147470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929176935160765e-05,
+      "loss": 0.0083,
+      "step": 147480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929172131186918e-05,
+      "loss": 0.0062,
+      "step": 147490
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929167327213071e-05,
+      "loss": 0.0104,
+      "step": 147500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291625232392236e-05,
+      "loss": 0.008,
+      "step": 147510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291577192653765e-05,
+      "loss": 0.0137,
+      "step": 147520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291529152915294e-05,
+      "loss": 0.007,
+      "step": 147530
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291481113176823e-05,
+      "loss": 0.007,
+      "step": 147540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929143307343835e-05,
+      "loss": 0.0091,
+      "step": 147550
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929138503369988e-05,
+      "loss": 0.0056,
+      "step": 147560
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291336993961408e-05,
+      "loss": 0.0086,
+      "step": 147570
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291288954222934e-05,
+      "loss": 0.0119,
+      "step": 147580
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291240914484463e-05,
+      "loss": 0.0099,
+      "step": 147590
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291192874745992e-05,
+      "loss": 0.0092,
+      "step": 147600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929114483500752e-05,
+      "loss": 0.0095,
+      "step": 147610
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291096795269047e-05,
+      "loss": 0.0082,
+      "step": 147620
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291048755530577e-05,
+      "loss": 0.0068,
+      "step": 147630
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9291000715792106e-05,
+      "loss": 0.0074,
+      "step": 147640
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290952676053635e-05,
+      "loss": 0.0066,
+      "step": 147650
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929090463631516e-05,
+      "loss": 0.0071,
+      "step": 147660
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929085659657669e-05,
+      "loss": 0.0088,
+      "step": 147670
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929080855683822e-05,
+      "loss": 0.0086,
+      "step": 147680
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290760517099746e-05,
+      "loss": 0.0074,
+      "step": 147690
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290712477361275e-05,
+      "loss": 0.0108,
+      "step": 147700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290664437622804e-05,
+      "loss": 0.009,
+      "step": 147710
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290616397884333e-05,
+      "loss": 0.008,
+      "step": 147720
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929056835814586e-05,
+      "loss": 0.0094,
+      "step": 147730
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929052031840739e-05,
+      "loss": 0.0089,
+      "step": 147740
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290472278668918e-05,
+      "loss": 0.0108,
+      "step": 147750
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290424238930444e-05,
+      "loss": 0.0057,
+      "step": 147760
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290376199191973e-05,
+      "loss": 0.0064,
+      "step": 147770
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290328159453502e-05,
+      "loss": 0.0075,
+      "step": 147780
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929028011971503e-05,
+      "loss": 0.0079,
+      "step": 147790
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290232079976557e-05,
+      "loss": 0.007,
+      "step": 147800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290184040238087e-05,
+      "loss": 0.0054,
+      "step": 147810
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290136000499616e-05,
+      "loss": 0.0064,
+      "step": 147820
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9290087960761145e-05,
+      "loss": 0.0092,
+      "step": 147830
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.929003992102267e-05,
+      "loss": 0.0059,
+      "step": 147840
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.92899918812842e-05,
+      "loss": 0.0104,
+      "step": 147850
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.928994384154573e-05,
+      "loss": 0.0105,
+      "step": 147860
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289895801807256e-05,
+      "loss": 0.0105,
+      "step": 147870
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289847762068785e-05,
+      "loss": 0.0069,
+      "step": 147880
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289799722330314e-05,
+      "loss": 0.0077,
+      "step": 147890
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289751682591843e-05,
+      "loss": 0.0074,
+      "step": 147900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.928970364285337e-05,
+      "loss": 0.0071,
+      "step": 147910
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.92896556031149e-05,
+      "loss": 0.0062,
+      "step": 147920
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289607563376428e-05,
+      "loss": 0.0056,
+      "step": 147930
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289559523637954e-05,
+      "loss": 0.0086,
+      "step": 147940
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289511483899483e-05,
+      "loss": 0.0075,
+      "step": 147950
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289463444161012e-05,
+      "loss": 0.0102,
+      "step": 147960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.928941540442254e-05,
+      "loss": 0.0039,
+      "step": 147970
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289367364684067e-05,
+      "loss": 0.011,
+      "step": 147980
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289319324945597e-05,
+      "loss": 0.0067,
+      "step": 147990
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289271285207126e-05,
+      "loss": 0.0077,
+      "step": 148000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289223245468655e-05,
+      "loss": 0.0078,
+      "step": 148010
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.928917520573018e-05,
+      "loss": 0.0065,
+      "step": 148020
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.928912716599171e-05,
+      "loss": 0.0073,
+      "step": 148030
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.928907912625324e-05,
+      "loss": 0.0103,
+      "step": 148040
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9289031086514765e-05,
+      "loss": 0.0079,
+      "step": 148050
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9288983046776295e-05,
+      "loss": 0.0075,
+      "step": 148060
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9288935007037824e-05,
+      "loss": 0.0049,
+      "step": 148070
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9288886967299353e-05,
+      "loss": 0.0064,
+      "step": 148080
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.928883892756088e-05,
+      "loss": 0.0042,
+      "step": 148090
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.928879088782241e-05,
+      "loss": 0.0093,
+      "step": 148100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9288742848083938e-05,
+      "loss": 0.0082,
+      "step": 148110
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9288694808345464e-05,
+      "loss": 0.0105,
+      "step": 148120
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9288646768606993e-05,
+      "loss": 0.0065,
+      "step": 148130
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9288598728868522e-05,
+      "loss": 0.0106,
+      "step": 148140
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928855068913005e-05,
+      "loss": 0.0089,
+      "step": 148150
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9288502649391577e-05,
+      "loss": 0.01,
+      "step": 148160
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9288454609653106e-05,
+      "loss": 0.0075,
+      "step": 148170
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9288406569914636e-05,
+      "loss": 0.0052,
+      "step": 148180
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928835853017616e-05,
+      "loss": 0.0106,
+      "step": 148190
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928831049043769e-05,
+      "loss": 0.0085,
+      "step": 148200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928826245069922e-05,
+      "loss": 0.008,
+      "step": 148210
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928821441096075e-05,
+      "loss": 0.0065,
+      "step": 148220
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9288166371222275e-05,
+      "loss": 0.0073,
+      "step": 148230
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9288118331483805e-05,
+      "loss": 0.0073,
+      "step": 148240
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9288070291745334e-05,
+      "loss": 0.008,
+      "step": 148250
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9288022252006863e-05,
+      "loss": 0.006,
+      "step": 148260
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928797421226839e-05,
+      "loss": 0.0088,
+      "step": 148270
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287926172529918e-05,
+      "loss": 0.007,
+      "step": 148280
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287878132791448e-05,
+      "loss": 0.0076,
+      "step": 148290
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287830093052973e-05,
+      "loss": 0.0087,
+      "step": 148300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287782053314503e-05,
+      "loss": 0.0077,
+      "step": 148310
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287734013576032e-05,
+      "loss": 0.0097,
+      "step": 148320
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928768597383756e-05,
+      "loss": 0.0096,
+      "step": 148330
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287637934099087e-05,
+      "loss": 0.0067,
+      "step": 148340
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287589894360616e-05,
+      "loss": 0.0086,
+      "step": 148350
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287541854622146e-05,
+      "loss": 0.0094,
+      "step": 148360
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928749381488367e-05,
+      "loss": 0.009,
+      "step": 148370
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.92874457751452e-05,
+      "loss": 0.0117,
+      "step": 148380
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928739773540673e-05,
+      "loss": 0.0106,
+      "step": 148390
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928734969566826e-05,
+      "loss": 0.0069,
+      "step": 148400
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287301655929785e-05,
+      "loss": 0.0052,
+      "step": 148410
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287253616191315e-05,
+      "loss": 0.0089,
+      "step": 148420
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287205576452844e-05,
+      "loss": 0.0084,
+      "step": 148430
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287157536714373e-05,
+      "loss": 0.0043,
+      "step": 148440
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.92871094969759e-05,
+      "loss": 0.0089,
+      "step": 148450
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287061457237428e-05,
+      "loss": 0.0115,
+      "step": 148460
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9287013417498957e-05,
+      "loss": 0.0108,
+      "step": 148470
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286965377760483e-05,
+      "loss": 0.0065,
+      "step": 148480
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286917338022013e-05,
+      "loss": 0.0092,
+      "step": 148490
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286869298283542e-05,
+      "loss": 0.008,
+      "step": 148500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928682125854507e-05,
+      "loss": 0.0063,
+      "step": 148510
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286773218806597e-05,
+      "loss": 0.0072,
+      "step": 148520
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286725179068126e-05,
+      "loss": 0.0105,
+      "step": 148530
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286677139329656e-05,
+      "loss": 0.0074,
+      "step": 148540
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928662909959118e-05,
+      "loss": 0.0073,
+      "step": 148550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928658105985271e-05,
+      "loss": 0.0054,
+      "step": 148560
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928653302011424e-05,
+      "loss": 0.0107,
+      "step": 148570
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928648498037577e-05,
+      "loss": 0.0073,
+      "step": 148580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286436940637295e-05,
+      "loss": 0.0103,
+      "step": 148590
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286388900898824e-05,
+      "loss": 0.0081,
+      "step": 148600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286340861160354e-05,
+      "loss": 0.0087,
+      "step": 148610
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286292821421883e-05,
+      "loss": 0.0057,
+      "step": 148620
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928624478168341e-05,
+      "loss": 0.0081,
+      "step": 148630
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286196741944938e-05,
+      "loss": 0.012,
+      "step": 148640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286148702206467e-05,
+      "loss": 0.007,
+      "step": 148650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286100662467993e-05,
+      "loss": 0.0062,
+      "step": 148660
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286052622729523e-05,
+      "loss": 0.0114,
+      "step": 148670
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9286004582991052e-05,
+      "loss": 0.0081,
+      "step": 148680
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928595654325258e-05,
+      "loss": 0.0066,
+      "step": 148690
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9285908503514107e-05,
+      "loss": 0.0078,
+      "step": 148700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9285860463775636e-05,
+      "loss": 0.0088,
+      "step": 148710
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9285812424037166e-05,
+      "loss": 0.0059,
+      "step": 148720
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928576918827254e-05,
+      "loss": 0.0094,
+      "step": 148730
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928572114853407e-05,
+      "loss": 0.0081,
+      "step": 148740
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.92856731087956e-05,
+      "loss": 0.0092,
+      "step": 148750
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9285625069057125e-05,
+      "loss": 0.0112,
+      "step": 148760
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9285577029318654e-05,
+      "loss": 0.0072,
+      "step": 148770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9285528989580183e-05,
+      "loss": 0.0061,
+      "step": 148780
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.928548094984171e-05,
+      "loss": 0.0069,
+      "step": 148790
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9285432910103238e-05,
+      "loss": 0.0085,
+      "step": 148800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9285384870364768e-05,
+      "loss": 0.0051,
+      "step": 148810
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9285336830626297e-05,
+      "loss": 0.0101,
+      "step": 148820
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9285288790887823e-05,
+      "loss": 0.0057,
+      "step": 148830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9285240751149352e-05,
+      "loss": 0.0102,
+      "step": 148840
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928519271141088e-05,
+      "loss": 0.0087,
+      "step": 148850
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928514467167241e-05,
+      "loss": 0.0083,
+      "step": 148860
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9285096631933936e-05,
+      "loss": 0.0069,
+      "step": 148870
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9285048592195466e-05,
+      "loss": 0.0074,
+      "step": 148880
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9285000552456995e-05,
+      "loss": 0.0077,
+      "step": 148890
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928495251271852e-05,
+      "loss": 0.0098,
+      "step": 148900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928490447298005e-05,
+      "loss": 0.0086,
+      "step": 148910
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928485643324158e-05,
+      "loss": 0.0107,
+      "step": 148920
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928480839350311e-05,
+      "loss": 0.008,
+      "step": 148930
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9284760353764634e-05,
+      "loss": 0.0084,
+      "step": 148940
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9284712314026164e-05,
+      "loss": 0.0049,
+      "step": 148950
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9284664274287693e-05,
+      "loss": 0.0096,
+      "step": 148960
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928461623454922e-05,
+      "loss": 0.0087,
+      "step": 148970
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9284568194810748e-05,
+      "loss": 0.0087,
+      "step": 148980
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9284520155072277e-05,
+      "loss": 0.0068,
+      "step": 148990
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9284472115333807e-05,
+      "loss": 0.0104,
+      "step": 149000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9284424075595333e-05,
+      "loss": 0.0071,
+      "step": 149010
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9284376035856862e-05,
+      "loss": 0.0095,
+      "step": 149020
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928432799611839e-05,
+      "loss": 0.0068,
+      "step": 149030
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928427995637992e-05,
+      "loss": 0.0085,
+      "step": 149040
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9284231916641446e-05,
+      "loss": 0.0076,
+      "step": 149050
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9284183876902976e-05,
+      "loss": 0.0097,
+      "step": 149060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9284135837164505e-05,
+      "loss": 0.01,
+      "step": 149070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928408779742603e-05,
+      "loss": 0.0075,
+      "step": 149080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928403975768756e-05,
+      "loss": 0.0073,
+      "step": 149090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928399171794909e-05,
+      "loss": 0.0081,
+      "step": 149100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928394367821062e-05,
+      "loss": 0.0059,
+      "step": 149110
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9283895638472144e-05,
+      "loss": 0.006,
+      "step": 149120
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9283847598733674e-05,
+      "loss": 0.006,
+      "step": 149130
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9283799558995203e-05,
+      "loss": 0.008,
+      "step": 149140
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928375151925673e-05,
+      "loss": 0.0056,
+      "step": 149150
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9283703479518258e-05,
+      "loss": 0.0046,
+      "step": 149160
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9283655439779787e-05,
+      "loss": 0.0076,
+      "step": 149170
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9283607400041317e-05,
+      "loss": 0.0085,
+      "step": 149180
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9283559360302843e-05,
+      "loss": 0.0095,
+      "step": 149190
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9283511320564372e-05,
+      "loss": 0.0069,
+      "step": 149200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.92834632808259e-05,
+      "loss": 0.0084,
+      "step": 149210
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928341524108743e-05,
+      "loss": 0.0067,
+      "step": 149220
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9283367201348956e-05,
+      "loss": 0.0069,
+      "step": 149230
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9283319161610485e-05,
+      "loss": 0.0093,
+      "step": 149240
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9283271121872015e-05,
+      "loss": 0.0104,
+      "step": 149250
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928322308213354e-05,
+      "loss": 0.0069,
+      "step": 149260
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928317504239507e-05,
+      "loss": 0.0091,
+      "step": 149270
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.92831270026566e-05,
+      "loss": 0.0081,
+      "step": 149280
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928307896291813e-05,
+      "loss": 0.0082,
+      "step": 149290
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9283030923179654e-05,
+      "loss": 0.0072,
+      "step": 149300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282982883441184e-05,
+      "loss": 0.0059,
+      "step": 149310
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282934843702713e-05,
+      "loss": 0.01,
+      "step": 149320
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928288680396424e-05,
+      "loss": 0.0067,
+      "step": 149330
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282838764225768e-05,
+      "loss": 0.0068,
+      "step": 149340
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282790724487297e-05,
+      "loss": 0.0102,
+      "step": 149350
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282742684748827e-05,
+      "loss": 0.0103,
+      "step": 149360
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282694645010352e-05,
+      "loss": 0.0053,
+      "step": 149370
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282646605271882e-05,
+      "loss": 0.0084,
+      "step": 149380
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928259856553341e-05,
+      "loss": 0.0079,
+      "step": 149390
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928255052579494e-05,
+      "loss": 0.0089,
+      "step": 149400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282502486056466e-05,
+      "loss": 0.0075,
+      "step": 149410
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282454446317995e-05,
+      "loss": 0.0079,
+      "step": 149420
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282406406579525e-05,
+      "loss": 0.0081,
+      "step": 149430
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928235836684105e-05,
+      "loss": 0.009,
+      "step": 149440
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928231032710258e-05,
+      "loss": 0.0076,
+      "step": 149450
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928226228736411e-05,
+      "loss": 0.0064,
+      "step": 149460
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928221424762564e-05,
+      "loss": 0.011,
+      "step": 149470
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282166207887164e-05,
+      "loss": 0.0088,
+      "step": 149480
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282118168148694e-05,
+      "loss": 0.0067,
+      "step": 149490
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9282070128410223e-05,
+      "loss": 0.0094,
+      "step": 149500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.928202208867175e-05,
+      "loss": 0.0096,
+      "step": 149510
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9281974048933278e-05,
+      "loss": 0.0073,
+      "step": 149520
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281926009194807e-05,
+      "loss": 0.0106,
+      "step": 149530
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281877969456336e-05,
+      "loss": 0.0072,
+      "step": 149540
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281829929717862e-05,
+      "loss": 0.0071,
+      "step": 149550
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928178188997939e-05,
+      "loss": 0.0082,
+      "step": 149560
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928173385024092e-05,
+      "loss": 0.0079,
+      "step": 149570
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928168581050245e-05,
+      "loss": 0.0055,
+      "step": 149580
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281637770763976e-05,
+      "loss": 0.0067,
+      "step": 149590
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281589731025505e-05,
+      "loss": 0.0103,
+      "step": 149600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281541691287035e-05,
+      "loss": 0.0091,
+      "step": 149610
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928149365154856e-05,
+      "loss": 0.007,
+      "step": 149620
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928144561181009e-05,
+      "loss": 0.008,
+      "step": 149630
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928139757207162e-05,
+      "loss": 0.0083,
+      "step": 149640
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281349532333148e-05,
+      "loss": 0.0116,
+      "step": 149650
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281301492594674e-05,
+      "loss": 0.0081,
+      "step": 149660
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281253452856203e-05,
+      "loss": 0.0072,
+      "step": 149670
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281205413117733e-05,
+      "loss": 0.0072,
+      "step": 149680
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928115737337926e-05,
+      "loss": 0.0072,
+      "step": 149690
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281109333640788e-05,
+      "loss": 0.0069,
+      "step": 149700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281061293902317e-05,
+      "loss": 0.0087,
+      "step": 149710
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9281013254163846e-05,
+      "loss": 0.0094,
+      "step": 149720
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280965214425372e-05,
+      "loss": 0.0064,
+      "step": 149730
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.92809171746869e-05,
+      "loss": 0.0088,
+      "step": 149740
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928086913494843e-05,
+      "loss": 0.0088,
+      "step": 149750
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280821095209957e-05,
+      "loss": 0.0082,
+      "step": 149760
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280773055471486e-05,
+      "loss": 0.0076,
+      "step": 149770
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280725015733015e-05,
+      "loss": 0.0109,
+      "step": 149780
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280676975994544e-05,
+      "loss": 0.0106,
+      "step": 149790
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928062893625607e-05,
+      "loss": 0.0082,
+      "step": 149800
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.92805808965176e-05,
+      "loss": 0.0083,
+      "step": 149810
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928053285677913e-05,
+      "loss": 0.0081,
+      "step": 149820
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280484817040658e-05,
+      "loss": 0.0091,
+      "step": 149830
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280436777302184e-05,
+      "loss": 0.0074,
+      "step": 149840
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280388737563713e-05,
+      "loss": 0.0079,
+      "step": 149850
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280340697825243e-05,
+      "loss": 0.0074,
+      "step": 149860
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928029265808677e-05,
+      "loss": 0.0098,
+      "step": 149870
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280244618348298e-05,
+      "loss": 0.0086,
+      "step": 149880
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280196578609827e-05,
+      "loss": 0.0098,
+      "step": 149890
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280148538871356e-05,
+      "loss": 0.007,
+      "step": 149900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9280100499132882e-05,
+      "loss": 0.0089,
+      "step": 149910
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928005245939441e-05,
+      "loss": 0.007,
+      "step": 149920
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.928000441965594e-05,
+      "loss": 0.0057,
+      "step": 149930
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9279956379917467e-05,
+      "loss": 0.0093,
+      "step": 149940
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9279908340178996e-05,
+      "loss": 0.0054,
+      "step": 149950
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9279860300440525e-05,
+      "loss": 0.008,
+      "step": 149960
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9279812260702054e-05,
+      "loss": 0.0057,
+      "step": 149970
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.927976422096358e-05,
+      "loss": 0.0095,
+      "step": 149980
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.927971618122511e-05,
+      "loss": 0.0064,
+      "step": 149990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.927966814148664e-05,
+      "loss": 0.0088,
+      "step": 150000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9279620101748168e-05,
+      "loss": 0.006,
+      "step": 150010
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9279572062009694e-05,
+      "loss": 0.0079,
+      "step": 150020
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9279524022271227e-05,
+      "loss": 0.0062,
+      "step": 150030
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9279475982532753e-05,
+      "loss": 0.0075,
+      "step": 150040
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9279427942794282e-05,
+      "loss": 0.0072,
+      "step": 150050
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.927937990305581e-05,
+      "loss": 0.009,
+      "step": 150060
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.927933186331734e-05,
+      "loss": 0.0086,
+      "step": 150070
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9279283823578866e-05,
+      "loss": 0.0057,
+      "step": 150080
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9279235783840395e-05,
+      "loss": 0.0093,
+      "step": 150090
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9279187744101925e-05,
+      "loss": 0.0072,
+      "step": 150100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.927913970436345e-05,
+      "loss": 0.0069,
+      "step": 150110
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.927909166462498e-05,
+      "loss": 0.0095,
+      "step": 150120
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.927904362488651e-05,
+      "loss": 0.0086,
+      "step": 150130
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.927899558514804e-05,
+      "loss": 0.0097,
+      "step": 150140
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9278947545409564e-05,
+      "loss": 0.0101,
+      "step": 150150
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9278899505671094e-05,
+      "loss": 0.0117,
+      "step": 150160
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9278851465932623e-05,
+      "loss": 0.0089,
+      "step": 150170
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.927880342619415e-05,
+      "loss": 0.0071,
+      "step": 150180
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9278755386455678e-05,
+      "loss": 0.0099,
+      "step": 150190
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9278707346717207e-05,
+      "loss": 0.0076,
+      "step": 150200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9278659306978737e-05,
+      "loss": 0.0105,
+      "step": 150210
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9278611267240262e-05,
+      "loss": 0.0066,
+      "step": 150220
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9278563227501792e-05,
+      "loss": 0.007,
+      "step": 150230
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927851518776332e-05,
+      "loss": 0.0081,
+      "step": 150240
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927846714802485e-05,
+      "loss": 0.009,
+      "step": 150250
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9278419108286376e-05,
+      "loss": 0.0075,
+      "step": 150260
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9278371068547905e-05,
+      "loss": 0.0057,
+      "step": 150270
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9278323028809435e-05,
+      "loss": 0.0111,
+      "step": 150280
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927827498907096e-05,
+      "loss": 0.0088,
+      "step": 150290
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927822694933249e-05,
+      "loss": 0.0078,
+      "step": 150300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927817890959402e-05,
+      "loss": 0.008,
+      "step": 150310
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927813086985555e-05,
+      "loss": 0.0094,
+      "step": 150320
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9278082830117074e-05,
+      "loss": 0.0074,
+      "step": 150330
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9278034790378604e-05,
+      "loss": 0.0059,
+      "step": 150340
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277986750640133e-05,
+      "loss": 0.0065,
+      "step": 150350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927793871090166e-05,
+      "loss": 0.0098,
+      "step": 150360
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277890671163188e-05,
+      "loss": 0.0076,
+      "step": 150370
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277842631424717e-05,
+      "loss": 0.0072,
+      "step": 150380
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277794591686246e-05,
+      "loss": 0.0067,
+      "step": 150390
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277746551947772e-05,
+      "loss": 0.0064,
+      "step": 150400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.92776985122093e-05,
+      "loss": 0.01,
+      "step": 150410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927765047247083e-05,
+      "loss": 0.0086,
+      "step": 150420
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927760243273236e-05,
+      "loss": 0.0091,
+      "step": 150430
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277554392993886e-05,
+      "loss": 0.0074,
+      "step": 150440
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277506353255415e-05,
+      "loss": 0.0056,
+      "step": 150450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277458313516945e-05,
+      "loss": 0.0088,
+      "step": 150460
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927741027377847e-05,
+      "loss": 0.0088,
+      "step": 150470
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927736223404e-05,
+      "loss": 0.0061,
+      "step": 150480
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927731419430153e-05,
+      "loss": 0.006,
+      "step": 150490
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277266154563058e-05,
+      "loss": 0.0072,
+      "step": 150500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277218114824584e-05,
+      "loss": 0.0079,
+      "step": 150510
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277170075086113e-05,
+      "loss": 0.006,
+      "step": 150520
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277122035347643e-05,
+      "loss": 0.0054,
+      "step": 150530
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927707399560917e-05,
+      "loss": 0.0108,
+      "step": 150540
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9277025955870698e-05,
+      "loss": 0.0057,
+      "step": 150550
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276977916132227e-05,
+      "loss": 0.0073,
+      "step": 150560
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276929876393756e-05,
+      "loss": 0.0082,
+      "step": 150570
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276881836655282e-05,
+      "loss": 0.0052,
+      "step": 150580
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927683379691681e-05,
+      "loss": 0.0074,
+      "step": 150590
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927678575717834e-05,
+      "loss": 0.0078,
+      "step": 150600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927673771743987e-05,
+      "loss": 0.0073,
+      "step": 150610
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276689677701396e-05,
+      "loss": 0.0069,
+      "step": 150620
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276641637962925e-05,
+      "loss": 0.0096,
+      "step": 150630
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276593598224454e-05,
+      "loss": 0.0099,
+      "step": 150640
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927654555848598e-05,
+      "loss": 0.0102,
+      "step": 150650
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927649751874751e-05,
+      "loss": 0.01,
+      "step": 150660
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927644947900904e-05,
+      "loss": 0.0065,
+      "step": 150670
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276401439270568e-05,
+      "loss": 0.0086,
+      "step": 150680
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276353399532094e-05,
+      "loss": 0.0076,
+      "step": 150690
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276305359793623e-05,
+      "loss": 0.0067,
+      "step": 150700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276257320055153e-05,
+      "loss": 0.0078,
+      "step": 150710
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927620928031668e-05,
+      "loss": 0.0081,
+      "step": 150720
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276161240578208e-05,
+      "loss": 0.0056,
+      "step": 150730
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276113200839737e-05,
+      "loss": 0.0064,
+      "step": 150740
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276065161101266e-05,
+      "loss": 0.0064,
+      "step": 150750
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9276017121362792e-05,
+      "loss": 0.007,
+      "step": 150760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927596908162432e-05,
+      "loss": 0.0072,
+      "step": 150770
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927592104188585e-05,
+      "loss": 0.0105,
+      "step": 150780
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927587300214738e-05,
+      "loss": 0.0054,
+      "step": 150790
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9275824962408906e-05,
+      "loss": 0.0106,
+      "step": 150800
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9275776922670435e-05,
+      "loss": 0.0059,
+      "step": 150810
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9275728882931964e-05,
+      "loss": 0.0076,
+      "step": 150820
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927568084319349e-05,
+      "loss": 0.0063,
+      "step": 150830
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927563280345502e-05,
+      "loss": 0.0069,
+      "step": 150840
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927558476371655e-05,
+      "loss": 0.0075,
+      "step": 150850
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9275536723978078e-05,
+      "loss": 0.0067,
+      "step": 150860
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9275488684239604e-05,
+      "loss": 0.0078,
+      "step": 150870
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9275440644501133e-05,
+      "loss": 0.0091,
+      "step": 150880
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9275392604762663e-05,
+      "loss": 0.0081,
+      "step": 150890
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.927534456502419e-05,
+      "loss": 0.011,
+      "step": 150900
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9275296525285718e-05,
+      "loss": 0.0065,
+      "step": 150910
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9275248485547247e-05,
+      "loss": 0.0085,
+      "step": 150920
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9275200445808776e-05,
+      "loss": 0.0086,
+      "step": 150930
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9275152406070302e-05,
+      "loss": 0.0055,
+      "step": 150940
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927510436633183e-05,
+      "loss": 0.0056,
+      "step": 150950
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927505632659336e-05,
+      "loss": 0.011,
+      "step": 150960
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927500828685489e-05,
+      "loss": 0.0049,
+      "step": 150970
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274960247116416e-05,
+      "loss": 0.0068,
+      "step": 150980
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274912207377945e-05,
+      "loss": 0.0091,
+      "step": 150990
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274864167639474e-05,
+      "loss": 0.0067,
+      "step": 151000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274816127901e-05,
+      "loss": 0.0064,
+      "step": 151010
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927476808816253e-05,
+      "loss": 0.0082,
+      "step": 151020
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927472004842406e-05,
+      "loss": 0.0064,
+      "step": 151030
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274672008685588e-05,
+      "loss": 0.0081,
+      "step": 151040
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274623968947114e-05,
+      "loss": 0.0075,
+      "step": 151050
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274575929208643e-05,
+      "loss": 0.0065,
+      "step": 151060
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274527889470172e-05,
+      "loss": 0.0108,
+      "step": 151070
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.92744798497317e-05,
+      "loss": 0.0074,
+      "step": 151080
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274431809993228e-05,
+      "loss": 0.008,
+      "step": 151090
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274383770254757e-05,
+      "loss": 0.0097,
+      "step": 151100
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274335730516286e-05,
+      "loss": 0.0115,
+      "step": 151110
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274287690777812e-05,
+      "loss": 0.0059,
+      "step": 151120
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927423965103934e-05,
+      "loss": 0.0081,
+      "step": 151130
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927419161130087e-05,
+      "loss": 0.0072,
+      "step": 151140
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274143571562396e-05,
+      "loss": 0.0079,
+      "step": 151150
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274095531823926e-05,
+      "loss": 0.011,
+      "step": 151160
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9274047492085455e-05,
+      "loss": 0.0057,
+      "step": 151170
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273999452346984e-05,
+      "loss": 0.0094,
+      "step": 151180
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927395141260851e-05,
+      "loss": 0.0065,
+      "step": 151190
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927390337287004e-05,
+      "loss": 0.0092,
+      "step": 151200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927385533313157e-05,
+      "loss": 0.0105,
+      "step": 151210
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273807293393098e-05,
+      "loss": 0.009,
+      "step": 151220
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273759253654624e-05,
+      "loss": 0.0083,
+      "step": 151230
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273711213916153e-05,
+      "loss": 0.0097,
+      "step": 151240
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273663174177682e-05,
+      "loss": 0.0073,
+      "step": 151250
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273615134439208e-05,
+      "loss": 0.0064,
+      "step": 151260
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273567094700738e-05,
+      "loss": 0.0074,
+      "step": 151270
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273519054962267e-05,
+      "loss": 0.0099,
+      "step": 151280
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273471015223796e-05,
+      "loss": 0.0067,
+      "step": 151290
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273422975485322e-05,
+      "loss": 0.0063,
+      "step": 151300
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927337493574685e-05,
+      "loss": 0.0071,
+      "step": 151310
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927332689600838e-05,
+      "loss": 0.0075,
+      "step": 151320
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273278856269906e-05,
+      "loss": 0.0079,
+      "step": 151330
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273230816531436e-05,
+      "loss": 0.0102,
+      "step": 151340
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273182776792965e-05,
+      "loss": 0.0074,
+      "step": 151350
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9273134737054494e-05,
+      "loss": 0.0091,
+      "step": 151360
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927308669731602e-05,
+      "loss": 0.0094,
+      "step": 151370
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927303865757755e-05,
+      "loss": 0.0091,
+      "step": 151380
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927299061783908e-05,
+      "loss": 0.006,
+      "step": 151390
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272942578100608e-05,
+      "loss": 0.0133,
+      "step": 151400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272894538362134e-05,
+      "loss": 0.007,
+      "step": 151410
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272846498623663e-05,
+      "loss": 0.0063,
+      "step": 151420
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272798458885192e-05,
+      "loss": 0.0089,
+      "step": 151430
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272750419146718e-05,
+      "loss": 0.0078,
+      "step": 151440
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272702379408247e-05,
+      "loss": 0.0046,
+      "step": 151450
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272654339669777e-05,
+      "loss": 0.0064,
+      "step": 151460
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272606299931306e-05,
+      "loss": 0.0106,
+      "step": 151470
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272558260192832e-05,
+      "loss": 0.0108,
+      "step": 151480
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927251022045436e-05,
+      "loss": 0.01,
+      "step": 151490
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927246218071589e-05,
+      "loss": 0.0091,
+      "step": 151500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272414140977416e-05,
+      "loss": 0.0079,
+      "step": 151510
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272366101238946e-05,
+      "loss": 0.0086,
+      "step": 151520
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272318061500475e-05,
+      "loss": 0.0055,
+      "step": 151530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272270021762004e-05,
+      "loss": 0.0089,
+      "step": 151540
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927222198202353e-05,
+      "loss": 0.007,
+      "step": 151550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927217394228506e-05,
+      "loss": 0.0097,
+      "step": 151560
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.927212590254659e-05,
+      "loss": 0.0053,
+      "step": 151570
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272077862808118e-05,
+      "loss": 0.0067,
+      "step": 151580
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9272029823069644e-05,
+      "loss": 0.0067,
+      "step": 151590
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9271981783331173e-05,
+      "loss": 0.0082,
+      "step": 151600
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9271933743592702e-05,
+      "loss": 0.0088,
+      "step": 151610
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271885703854228e-05,
+      "loss": 0.009,
+      "step": 151620
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271837664115757e-05,
+      "loss": 0.0085,
+      "step": 151630
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271789624377287e-05,
+      "loss": 0.0097,
+      "step": 151640
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271741584638816e-05,
+      "loss": 0.0078,
+      "step": 151650
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271693544900342e-05,
+      "loss": 0.0083,
+      "step": 151660
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.927164550516187e-05,
+      "loss": 0.0074,
+      "step": 151670
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.92715974654234e-05,
+      "loss": 0.0069,
+      "step": 151680
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271549425684926e-05,
+      "loss": 0.0158,
+      "step": 151690
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271501385946455e-05,
+      "loss": 0.0074,
+      "step": 151700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271453346207985e-05,
+      "loss": 0.0109,
+      "step": 151710
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271405306469514e-05,
+      "loss": 0.0097,
+      "step": 151720
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.927135726673104e-05,
+      "loss": 0.0088,
+      "step": 151730
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.927130922699257e-05,
+      "loss": 0.007,
+      "step": 151740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.92712611872541e-05,
+      "loss": 0.0089,
+      "step": 151750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271213147515628e-05,
+      "loss": 0.0058,
+      "step": 151760
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271165107777154e-05,
+      "loss": 0.0074,
+      "step": 151770
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271117068038683e-05,
+      "loss": 0.0111,
+      "step": 151780
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271069028300212e-05,
+      "loss": 0.0074,
+      "step": 151790
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9271020988561738e-05,
+      "loss": 0.0062,
+      "step": 151800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270972948823267e-05,
+      "loss": 0.0073,
+      "step": 151810
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270924909084797e-05,
+      "loss": 0.01,
+      "step": 151820
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270876869346326e-05,
+      "loss": 0.0053,
+      "step": 151830
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.927082882960785e-05,
+      "loss": 0.0107,
+      "step": 151840
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.927078078986938e-05,
+      "loss": 0.0094,
+      "step": 151850
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.927073275013091e-05,
+      "loss": 0.0054,
+      "step": 151860
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270684710392436e-05,
+      "loss": 0.0085,
+      "step": 151870
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270636670653965e-05,
+      "loss": 0.0085,
+      "step": 151880
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270588630915495e-05,
+      "loss": 0.0089,
+      "step": 151890
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270540591177024e-05,
+      "loss": 0.0087,
+      "step": 151900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.927049255143855e-05,
+      "loss": 0.0059,
+      "step": 151910
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.927044451170008e-05,
+      "loss": 0.0103,
+      "step": 151920
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.927039647196161e-05,
+      "loss": 0.0074,
+      "step": 151930
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270348432223138e-05,
+      "loss": 0.0108,
+      "step": 151940
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270300392484663e-05,
+      "loss": 0.0048,
+      "step": 151950
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270252352746193e-05,
+      "loss": 0.0095,
+      "step": 151960
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270204313007722e-05,
+      "loss": 0.0078,
+      "step": 151970
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270156273269248e-05,
+      "loss": 0.009,
+      "step": 151980
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270108233530777e-05,
+      "loss": 0.008,
+      "step": 151990
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270060193792306e-05,
+      "loss": 0.0066,
+      "step": 152000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9270012154053836e-05,
+      "loss": 0.0082,
+      "step": 152010
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.926996411431536e-05,
+      "loss": 0.0075,
+      "step": 152020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.926991607457689e-05,
+      "loss": 0.0064,
+      "step": 152030
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.926986803483842e-05,
+      "loss": 0.0077,
+      "step": 152040
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269819995099946e-05,
+      "loss": 0.0069,
+      "step": 152050
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269771955361475e-05,
+      "loss": 0.0086,
+      "step": 152060
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269723915623005e-05,
+      "loss": 0.0078,
+      "step": 152070
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269675875884534e-05,
+      "loss": 0.0065,
+      "step": 152080
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.926962783614606e-05,
+      "loss": 0.0068,
+      "step": 152090
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.926957979640759e-05,
+      "loss": 0.0084,
+      "step": 152100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269531756669118e-05,
+      "loss": 0.0063,
+      "step": 152110
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269483716930648e-05,
+      "loss": 0.0064,
+      "step": 152120
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269435677192173e-05,
+      "loss": 0.0087,
+      "step": 152130
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269387637453703e-05,
+      "loss": 0.007,
+      "step": 152140
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269339597715232e-05,
+      "loss": 0.0093,
+      "step": 152150
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269291557976758e-05,
+      "loss": 0.0051,
+      "step": 152160
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269243518238287e-05,
+      "loss": 0.0088,
+      "step": 152170
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269195478499816e-05,
+      "loss": 0.0072,
+      "step": 152180
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9269147438761346e-05,
+      "loss": 0.0077,
+      "step": 152190
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.926909939902287e-05,
+      "loss": 0.0062,
+      "step": 152200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.92690513592844e-05,
+      "loss": 0.0071,
+      "step": 152210
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.926900331954593e-05,
+      "loss": 0.0069,
+      "step": 152220
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9268955279807456e-05,
+      "loss": 0.0087,
+      "step": 152230
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9268907240068985e-05,
+      "loss": 0.0078,
+      "step": 152240
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9268859200330514e-05,
+      "loss": 0.0064,
+      "step": 152250
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9268811160592044e-05,
+      "loss": 0.0071,
+      "step": 152260
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.926876312085357e-05,
+      "loss": 0.008,
+      "step": 152270
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.92687150811151e-05,
+      "loss": 0.0091,
+      "step": 152280
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9268667041376628e-05,
+      "loss": 0.011,
+      "step": 152290
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9268619001638157e-05,
+      "loss": 0.0063,
+      "step": 152300
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9268570961899683e-05,
+      "loss": 0.0073,
+      "step": 152310
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9268522922161213e-05,
+      "loss": 0.0052,
+      "step": 152320
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9268474882422742e-05,
+      "loss": 0.0087,
+      "step": 152330
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9268426842684268e-05,
+      "loss": 0.0072,
+      "step": 152340
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9268378802945797e-05,
+      "loss": 0.0097,
+      "step": 152350
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9268330763207326e-05,
+      "loss": 0.0067,
+      "step": 152360
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9268282723468856e-05,
+      "loss": 0.0062,
+      "step": 152370
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926823468373038e-05,
+      "loss": 0.0082,
+      "step": 152380
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926818664399191e-05,
+      "loss": 0.0095,
+      "step": 152390
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926813860425344e-05,
+      "loss": 0.0063,
+      "step": 152400
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9268090564514966e-05,
+      "loss": 0.0087,
+      "step": 152410
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9268042524776495e-05,
+      "loss": 0.0131,
+      "step": 152420
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267994485038024e-05,
+      "loss": 0.0082,
+      "step": 152430
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267946445299554e-05,
+      "loss": 0.0092,
+      "step": 152440
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926789840556108e-05,
+      "loss": 0.0067,
+      "step": 152450
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926785036582261e-05,
+      "loss": 0.0093,
+      "step": 152460
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267802326084138e-05,
+      "loss": 0.0064,
+      "step": 152470
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267754286345667e-05,
+      "loss": 0.0071,
+      "step": 152480
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267706246607193e-05,
+      "loss": 0.0114,
+      "step": 152490
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267658206868722e-05,
+      "loss": 0.0084,
+      "step": 152500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267610167130252e-05,
+      "loss": 0.0077,
+      "step": 152510
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267562127391778e-05,
+      "loss": 0.0062,
+      "step": 152520
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267514087653307e-05,
+      "loss": 0.0072,
+      "step": 152530
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267466047914836e-05,
+      "loss": 0.0103,
+      "step": 152540
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267418008176365e-05,
+      "loss": 0.0048,
+      "step": 152550
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926736996843789e-05,
+      "loss": 0.0089,
+      "step": 152560
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926732192869942e-05,
+      "loss": 0.0106,
+      "step": 152570
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926727388896095e-05,
+      "loss": 0.0099,
+      "step": 152580
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926722584922248e-05,
+      "loss": 0.0079,
+      "step": 152590
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926717780948401e-05,
+      "loss": 0.0089,
+      "step": 152600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267129769745538e-05,
+      "loss": 0.0088,
+      "step": 152610
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267081730007064e-05,
+      "loss": 0.0068,
+      "step": 152620
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9267033690268593e-05,
+      "loss": 0.0105,
+      "step": 152630
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266985650530122e-05,
+      "loss": 0.0057,
+      "step": 152640
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266937610791648e-05,
+      "loss": 0.0075,
+      "step": 152650
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266889571053177e-05,
+      "loss": 0.007,
+      "step": 152660
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266846335288552e-05,
+      "loss": 0.0057,
+      "step": 152670
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926679829555008e-05,
+      "loss": 0.0074,
+      "step": 152680
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266750255811607e-05,
+      "loss": 0.0082,
+      "step": 152690
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266702216073136e-05,
+      "loss": 0.0073,
+      "step": 152700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266654176334666e-05,
+      "loss": 0.0097,
+      "step": 152710
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926660613659619e-05,
+      "loss": 0.01,
+      "step": 152720
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926655809685772e-05,
+      "loss": 0.005,
+      "step": 152730
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926651005711925e-05,
+      "loss": 0.0063,
+      "step": 152740
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926646201738078e-05,
+      "loss": 0.0075,
+      "step": 152750
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266413977642305e-05,
+      "loss": 0.005,
+      "step": 152760
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266365937903834e-05,
+      "loss": 0.0079,
+      "step": 152770
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266317898165364e-05,
+      "loss": 0.0072,
+      "step": 152780
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266269858426893e-05,
+      "loss": 0.0059,
+      "step": 152790
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926622181868842e-05,
+      "loss": 0.0097,
+      "step": 152800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266173778949948e-05,
+      "loss": 0.0063,
+      "step": 152810
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266125739211477e-05,
+      "loss": 0.01,
+      "step": 152820
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266077699473003e-05,
+      "loss": 0.0071,
+      "step": 152830
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9266029659734533e-05,
+      "loss": 0.0073,
+      "step": 152840
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9265981619996062e-05,
+      "loss": 0.0077,
+      "step": 152850
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926593358025759e-05,
+      "loss": 0.0136,
+      "step": 152860
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9265885540519117e-05,
+      "loss": 0.0077,
+      "step": 152870
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9265837500780646e-05,
+      "loss": 0.0076,
+      "step": 152880
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9265789461042176e-05,
+      "loss": 0.0111,
+      "step": 152890
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.92657414213037e-05,
+      "loss": 0.0096,
+      "step": 152900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926569338156523e-05,
+      "loss": 0.0068,
+      "step": 152910
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926564534182676e-05,
+      "loss": 0.0075,
+      "step": 152920
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926559730208829e-05,
+      "loss": 0.0061,
+      "step": 152930
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9265549262349815e-05,
+      "loss": 0.0078,
+      "step": 152940
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9265501222611344e-05,
+      "loss": 0.0101,
+      "step": 152950
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9265453182872874e-05,
+      "loss": 0.0055,
+      "step": 152960
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9265405143134403e-05,
+      "loss": 0.0091,
+      "step": 152970
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.926535710339593e-05,
+      "loss": 0.0065,
+      "step": 152980
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9265309063657458e-05,
+      "loss": 0.008,
+      "step": 152990
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9265261023918987e-05,
+      "loss": 0.0072,
+      "step": 153000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9265212984180513e-05,
+      "loss": 0.0079,
+      "step": 153010
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9265164944442042e-05,
+      "loss": 0.0083,
+      "step": 153020
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9265116904703572e-05,
+      "loss": 0.0064,
+      "step": 153030
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.92650688649651e-05,
+      "loss": 0.0097,
+      "step": 153040
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9265020825226627e-05,
+      "loss": 0.0092,
+      "step": 153050
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9264972785488156e-05,
+      "loss": 0.0085,
+      "step": 153060
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9264924745749685e-05,
+      "loss": 0.0084,
+      "step": 153070
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926487670601121e-05,
+      "loss": 0.0068,
+      "step": 153080
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926482866627274e-05,
+      "loss": 0.0049,
+      "step": 153090
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926478062653427e-05,
+      "loss": 0.0055,
+      "step": 153100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.92647325867958e-05,
+      "loss": 0.0056,
+      "step": 153110
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926468454705733e-05,
+      "loss": 0.0079,
+      "step": 153120
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9264636507318858e-05,
+      "loss": 0.0067,
+      "step": 153130
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9264588467580384e-05,
+      "loss": 0.0064,
+      "step": 153140
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9264540427841913e-05,
+      "loss": 0.0088,
+      "step": 153150
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9264492388103442e-05,
+      "loss": 0.0079,
+      "step": 153160
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926444434836497e-05,
+      "loss": 0.0086,
+      "step": 153170
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9264396308626497e-05,
+      "loss": 0.0068,
+      "step": 153180
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9264348268888026e-05,
+      "loss": 0.0052,
+      "step": 153190
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9264300229149556e-05,
+      "loss": 0.0083,
+      "step": 153200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9264252189411085e-05,
+      "loss": 0.0059,
+      "step": 153210
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926420414967261e-05,
+      "loss": 0.0072,
+      "step": 153220
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926415610993414e-05,
+      "loss": 0.0127,
+      "step": 153230
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926410807019567e-05,
+      "loss": 0.006,
+      "step": 153240
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9264060030457195e-05,
+      "loss": 0.0065,
+      "step": 153250
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9264011990718725e-05,
+      "loss": 0.0078,
+      "step": 153260
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263963950980254e-05,
+      "loss": 0.0076,
+      "step": 153270
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263915911241783e-05,
+      "loss": 0.0086,
+      "step": 153280
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926386787150331e-05,
+      "loss": 0.0098,
+      "step": 153290
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263819831764838e-05,
+      "loss": 0.0068,
+      "step": 153300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263771792026368e-05,
+      "loss": 0.0061,
+      "step": 153310
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263723752287893e-05,
+      "loss": 0.0091,
+      "step": 153320
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263675712549423e-05,
+      "loss": 0.0043,
+      "step": 153330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263627672810952e-05,
+      "loss": 0.0067,
+      "step": 153340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926357963307248e-05,
+      "loss": 0.006,
+      "step": 153350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263531593334007e-05,
+      "loss": 0.0098,
+      "step": 153360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263483553595536e-05,
+      "loss": 0.0048,
+      "step": 153370
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263435513857066e-05,
+      "loss": 0.0107,
+      "step": 153380
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263387474118595e-05,
+      "loss": 0.0109,
+      "step": 153390
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926333943438012e-05,
+      "loss": 0.0071,
+      "step": 153400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926329139464165e-05,
+      "loss": 0.0078,
+      "step": 153410
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926324335490318e-05,
+      "loss": 0.0105,
+      "step": 153420
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263195315164705e-05,
+      "loss": 0.0088,
+      "step": 153430
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263147275426235e-05,
+      "loss": 0.0096,
+      "step": 153440
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263099235687764e-05,
+      "loss": 0.0107,
+      "step": 153450
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9263051195949293e-05,
+      "loss": 0.0132,
+      "step": 153460
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926300315621082e-05,
+      "loss": 0.0061,
+      "step": 153470
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262955116472348e-05,
+      "loss": 0.0091,
+      "step": 153480
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262907076733877e-05,
+      "loss": 0.0077,
+      "step": 153490
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262859036995403e-05,
+      "loss": 0.0097,
+      "step": 153500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262810997256933e-05,
+      "loss": 0.004,
+      "step": 153510
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262762957518462e-05,
+      "loss": 0.0095,
+      "step": 153520
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926271491777999e-05,
+      "loss": 0.0079,
+      "step": 153530
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262666878041517e-05,
+      "loss": 0.0056,
+      "step": 153540
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262618838303046e-05,
+      "loss": 0.0084,
+      "step": 153550
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262570798564576e-05,
+      "loss": 0.0094,
+      "step": 153560
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262522758826105e-05,
+      "loss": 0.0119,
+      "step": 153570
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926247471908763e-05,
+      "loss": 0.006,
+      "step": 153580
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926242667934916e-05,
+      "loss": 0.0079,
+      "step": 153590
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926237863961069e-05,
+      "loss": 0.008,
+      "step": 153600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262330599872215e-05,
+      "loss": 0.009,
+      "step": 153610
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262282560133744e-05,
+      "loss": 0.0052,
+      "step": 153620
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262234520395274e-05,
+      "loss": 0.0084,
+      "step": 153630
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262186480656803e-05,
+      "loss": 0.0047,
+      "step": 153640
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.926213844091833e-05,
+      "loss": 0.0065,
+      "step": 153650
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262090401179858e-05,
+      "loss": 0.0076,
+      "step": 153660
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9262042361441387e-05,
+      "loss": 0.0068,
+      "step": 153670
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9261994321702913e-05,
+      "loss": 0.0076,
+      "step": 153680
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9261946281964443e-05,
+      "loss": 0.0081,
+      "step": 153690
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261898242225972e-05,
+      "loss": 0.0088,
+      "step": 153700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.92618502024875e-05,
+      "loss": 0.0122,
+      "step": 153710
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261802162749027e-05,
+      "loss": 0.0081,
+      "step": 153720
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261754123010556e-05,
+      "loss": 0.0111,
+      "step": 153730
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261706083272086e-05,
+      "loss": 0.0083,
+      "step": 153740
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261658043533615e-05,
+      "loss": 0.0066,
+      "step": 153750
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.926161000379514e-05,
+      "loss": 0.0084,
+      "step": 153760
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.926156196405667e-05,
+      "loss": 0.0112,
+      "step": 153770
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.92615139243182e-05,
+      "loss": 0.0066,
+      "step": 153780
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261465884579725e-05,
+      "loss": 0.0068,
+      "step": 153790
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261417844841254e-05,
+      "loss": 0.0064,
+      "step": 153800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261369805102784e-05,
+      "loss": 0.0079,
+      "step": 153810
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261321765364313e-05,
+      "loss": 0.0079,
+      "step": 153820
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.926127372562584e-05,
+      "loss": 0.0073,
+      "step": 153830
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261225685887368e-05,
+      "loss": 0.0094,
+      "step": 153840
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261177646148897e-05,
+      "loss": 0.0066,
+      "step": 153850
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261129606410423e-05,
+      "loss": 0.0085,
+      "step": 153860
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261081566671952e-05,
+      "loss": 0.0074,
+      "step": 153870
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9261033526933482e-05,
+      "loss": 0.0103,
+      "step": 153880
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.926098548719501e-05,
+      "loss": 0.006,
+      "step": 153890
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260937447456537e-05,
+      "loss": 0.0058,
+      "step": 153900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260889407718066e-05,
+      "loss": 0.0071,
+      "step": 153910
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260841367979595e-05,
+      "loss": 0.0064,
+      "step": 153920
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260793328241125e-05,
+      "loss": 0.0084,
+      "step": 153930
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.926074528850265e-05,
+      "loss": 0.0085,
+      "step": 153940
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.926069724876418e-05,
+      "loss": 0.0071,
+      "step": 153950
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.926064920902571e-05,
+      "loss": 0.0086,
+      "step": 153960
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260601169287235e-05,
+      "loss": 0.007,
+      "step": 153970
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260553129548764e-05,
+      "loss": 0.0054,
+      "step": 153980
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260505089810294e-05,
+      "loss": 0.0114,
+      "step": 153990
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260457050071823e-05,
+      "loss": 0.0071,
+      "step": 154000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.926040901033335e-05,
+      "loss": 0.0052,
+      "step": 154010
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260360970594878e-05,
+      "loss": 0.0089,
+      "step": 154020
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260312930856407e-05,
+      "loss": 0.0078,
+      "step": 154030
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260264891117933e-05,
+      "loss": 0.0078,
+      "step": 154040
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260216851379462e-05,
+      "loss": 0.0077,
+      "step": 154050
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.926016881164099e-05,
+      "loss": 0.0082,
+      "step": 154060
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.926012077190252e-05,
+      "loss": 0.011,
+      "step": 154070
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260072732164047e-05,
+      "loss": 0.0072,
+      "step": 154080
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9260024692425576e-05,
+      "loss": 0.0061,
+      "step": 154090
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259976652687105e-05,
+      "loss": 0.0101,
+      "step": 154100
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259928612948635e-05,
+      "loss": 0.0081,
+      "step": 154110
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.925988057321016e-05,
+      "loss": 0.01,
+      "step": 154120
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.925983253347169e-05,
+      "loss": 0.0097,
+      "step": 154130
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.925978449373322e-05,
+      "loss": 0.0097,
+      "step": 154140
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259736453994745e-05,
+      "loss": 0.0062,
+      "step": 154150
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259688414256274e-05,
+      "loss": 0.0072,
+      "step": 154160
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259640374517803e-05,
+      "loss": 0.0076,
+      "step": 154170
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259592334779333e-05,
+      "loss": 0.0085,
+      "step": 154180
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.925954429504086e-05,
+      "loss": 0.0098,
+      "step": 154190
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259496255302388e-05,
+      "loss": 0.0072,
+      "step": 154200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259448215563917e-05,
+      "loss": 0.0062,
+      "step": 154210
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259400175825443e-05,
+      "loss": 0.0052,
+      "step": 154220
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259352136086972e-05,
+      "loss": 0.0058,
+      "step": 154230
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.92593040963485e-05,
+      "loss": 0.0066,
+      "step": 154240
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.925925605661003e-05,
+      "loss": 0.0079,
+      "step": 154250
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259208016871557e-05,
+      "loss": 0.0103,
+      "step": 154260
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259159977133086e-05,
+      "loss": 0.0099,
+      "step": 154270
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9259111937394615e-05,
+      "loss": 0.0067,
+      "step": 154280
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.925906389765614e-05,
+      "loss": 0.0085,
+      "step": 154290
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.925901585791767e-05,
+      "loss": 0.0078,
+      "step": 154300
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.92589678181792e-05,
+      "loss": 0.0083,
+      "step": 154310
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.925891977844073e-05,
+      "loss": 0.0083,
+      "step": 154320
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9258871738702255e-05,
+      "loss": 0.0081,
+      "step": 154330
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9258823698963784e-05,
+      "loss": 0.0064,
+      "step": 154340
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9258775659225313e-05,
+      "loss": 0.009,
+      "step": 154350
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9258727619486843e-05,
+      "loss": 0.0072,
+      "step": 154360
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.925867957974837e-05,
+      "loss": 0.008,
+      "step": 154370
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9258631540009898e-05,
+      "loss": 0.0066,
+      "step": 154380
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9258583500271427e-05,
+      "loss": 0.0095,
+      "step": 154390
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9258535460532953e-05,
+      "loss": 0.0076,
+      "step": 154400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9258487420794482e-05,
+      "loss": 0.0086,
+      "step": 154410
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925843938105601e-05,
+      "loss": 0.0063,
+      "step": 154420
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925839134131754e-05,
+      "loss": 0.0079,
+      "step": 154430
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9258343301579067e-05,
+      "loss": 0.0083,
+      "step": 154440
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9258295261840596e-05,
+      "loss": 0.0085,
+      "step": 154450
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9258247222102125e-05,
+      "loss": 0.004,
+      "step": 154460
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925819918236365e-05,
+      "loss": 0.01,
+      "step": 154470
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925815114262518e-05,
+      "loss": 0.0078,
+      "step": 154480
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925810310288671e-05,
+      "loss": 0.0073,
+      "step": 154490
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925805506314824e-05,
+      "loss": 0.0071,
+      "step": 154500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9258007023409765e-05,
+      "loss": 0.0108,
+      "step": 154510
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257958983671294e-05,
+      "loss": 0.008,
+      "step": 154520
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257910943932823e-05,
+      "loss": 0.0099,
+      "step": 154530
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257862904194353e-05,
+      "loss": 0.0046,
+      "step": 154540
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925781486445588e-05,
+      "loss": 0.0112,
+      "step": 154550
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257766824717408e-05,
+      "loss": 0.0084,
+      "step": 154560
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257718784978937e-05,
+      "loss": 0.006,
+      "step": 154570
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257670745240463e-05,
+      "loss": 0.0094,
+      "step": 154580
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257622705501992e-05,
+      "loss": 0.0071,
+      "step": 154590
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925757466576352e-05,
+      "loss": 0.0062,
+      "step": 154600
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925752662602505e-05,
+      "loss": 0.0103,
+      "step": 154610
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257478586286577e-05,
+      "loss": 0.0069,
+      "step": 154620
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257430546548106e-05,
+      "loss": 0.0089,
+      "step": 154630
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257382506809635e-05,
+      "loss": 0.0059,
+      "step": 154640
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925733446707116e-05,
+      "loss": 0.0084,
+      "step": 154650
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925728642733269e-05,
+      "loss": 0.009,
+      "step": 154660
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925723838759422e-05,
+      "loss": 0.0067,
+      "step": 154670
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925719034785575e-05,
+      "loss": 0.0054,
+      "step": 154680
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257142308117275e-05,
+      "loss": 0.0086,
+      "step": 154690
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257094268378804e-05,
+      "loss": 0.0062,
+      "step": 154700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9257046228640333e-05,
+      "loss": 0.0066,
+      "step": 154710
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256998188901862e-05,
+      "loss": 0.0084,
+      "step": 154720
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925695014916339e-05,
+      "loss": 0.007,
+      "step": 154730
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256902109424918e-05,
+      "loss": 0.0089,
+      "step": 154740
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256854069686447e-05,
+      "loss": 0.0078,
+      "step": 154750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256806029947973e-05,
+      "loss": 0.007,
+      "step": 154760
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256757990209502e-05,
+      "loss": 0.0096,
+      "step": 154770
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925670995047103e-05,
+      "loss": 0.0067,
+      "step": 154780
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925666191073256e-05,
+      "loss": 0.0088,
+      "step": 154790
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256613870994086e-05,
+      "loss": 0.0087,
+      "step": 154800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256565831255616e-05,
+      "loss": 0.0063,
+      "step": 154810
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256517791517145e-05,
+      "loss": 0.0088,
+      "step": 154820
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925646975177867e-05,
+      "loss": 0.0058,
+      "step": 154830
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.92564217120402e-05,
+      "loss": 0.0072,
+      "step": 154840
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925637367230173e-05,
+      "loss": 0.0065,
+      "step": 154850
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925632563256326e-05,
+      "loss": 0.0113,
+      "step": 154860
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256277592824785e-05,
+      "loss": 0.0067,
+      "step": 154870
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256229553086314e-05,
+      "loss": 0.0051,
+      "step": 154880
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256181513347843e-05,
+      "loss": 0.0098,
+      "step": 154890
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256133473609372e-05,
+      "loss": 0.0102,
+      "step": 154900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256085433870898e-05,
+      "loss": 0.0071,
+      "step": 154910
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9256037394132428e-05,
+      "loss": 0.0105,
+      "step": 154920
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9255989354393957e-05,
+      "loss": 0.0064,
+      "step": 154930
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9255941314655483e-05,
+      "loss": 0.0068,
+      "step": 154940
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9255893274917012e-05,
+      "loss": 0.0062,
+      "step": 154950
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925584523517854e-05,
+      "loss": 0.0095,
+      "step": 154960
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925579719544007e-05,
+      "loss": 0.0077,
+      "step": 154970
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9255749155701596e-05,
+      "loss": 0.0061,
+      "step": 154980
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9255701115963126e-05,
+      "loss": 0.0076,
+      "step": 154990
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9255653076224655e-05,
+      "loss": 0.0081,
+      "step": 155000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925560503648618e-05,
+      "loss": 0.0102,
+      "step": 155010
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925555699674771e-05,
+      "loss": 0.0112,
+      "step": 155020
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925550895700924e-05,
+      "loss": 0.0063,
+      "step": 155030
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.925546091727077e-05,
+      "loss": 0.0076,
+      "step": 155040
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9255412877532294e-05,
+      "loss": 0.0097,
+      "step": 155050
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9255364837793824e-05,
+      "loss": 0.0079,
+      "step": 155060
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9255316798055353e-05,
+      "loss": 0.0094,
+      "step": 155070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9255268758316882e-05,
+      "loss": 0.0071,
+      "step": 155080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9255220718578408e-05,
+      "loss": 0.011,
+      "step": 155090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9255172678839937e-05,
+      "loss": 0.0064,
+      "step": 155100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9255124639101467e-05,
+      "loss": 0.0099,
+      "step": 155110
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9255076599362993e-05,
+      "loss": 0.0064,
+      "step": 155120
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9255028559624522e-05,
+      "loss": 0.0062,
+      "step": 155130
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925498051988605e-05,
+      "loss": 0.0083,
+      "step": 155140
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925493248014758e-05,
+      "loss": 0.0108,
+      "step": 155150
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254884440409106e-05,
+      "loss": 0.007,
+      "step": 155160
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254836400670636e-05,
+      "loss": 0.0068,
+      "step": 155170
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254788360932165e-05,
+      "loss": 0.0061,
+      "step": 155180
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925474032119369e-05,
+      "loss": 0.0076,
+      "step": 155190
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925469228145522e-05,
+      "loss": 0.0056,
+      "step": 155200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925464424171675e-05,
+      "loss": 0.0064,
+      "step": 155210
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925459620197828e-05,
+      "loss": 0.0095,
+      "step": 155220
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254548162239804e-05,
+      "loss": 0.0092,
+      "step": 155230
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254500122501334e-05,
+      "loss": 0.0052,
+      "step": 155240
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254452082762863e-05,
+      "loss": 0.0101,
+      "step": 155250
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254404043024392e-05,
+      "loss": 0.0068,
+      "step": 155260
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254356003285918e-05,
+      "loss": 0.0091,
+      "step": 155270
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254307963547447e-05,
+      "loss": 0.0065,
+      "step": 155280
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254259923808977e-05,
+      "loss": 0.0064,
+      "step": 155290
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254211884070503e-05,
+      "loss": 0.01,
+      "step": 155300
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254163844332032e-05,
+      "loss": 0.0099,
+      "step": 155310
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925411580459356e-05,
+      "loss": 0.0096,
+      "step": 155320
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925406776485509e-05,
+      "loss": 0.0075,
+      "step": 155330
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9254019725116616e-05,
+      "loss": 0.0117,
+      "step": 155340
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253971685378145e-05,
+      "loss": 0.0085,
+      "step": 155350
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253923645639675e-05,
+      "loss": 0.0072,
+      "step": 155360
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.92538756059012e-05,
+      "loss": 0.0082,
+      "step": 155370
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925382756616273e-05,
+      "loss": 0.0074,
+      "step": 155380
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925377952642426e-05,
+      "loss": 0.0075,
+      "step": 155390
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925373148668579e-05,
+      "loss": 0.0078,
+      "step": 155400
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253683446947314e-05,
+      "loss": 0.0077,
+      "step": 155410
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253635407208844e-05,
+      "loss": 0.0078,
+      "step": 155420
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253587367470373e-05,
+      "loss": 0.0065,
+      "step": 155430
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253539327731902e-05,
+      "loss": 0.0088,
+      "step": 155440
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253491287993428e-05,
+      "loss": 0.0079,
+      "step": 155450
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253443248254957e-05,
+      "loss": 0.0073,
+      "step": 155460
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253395208516487e-05,
+      "loss": 0.0081,
+      "step": 155470
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253347168778012e-05,
+      "loss": 0.0088,
+      "step": 155480
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253299129039542e-05,
+      "loss": 0.0069,
+      "step": 155490
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925325108930107e-05,
+      "loss": 0.0065,
+      "step": 155500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.92532030495626e-05,
+      "loss": 0.0108,
+      "step": 155510
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253155009824126e-05,
+      "loss": 0.0051,
+      "step": 155520
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253106970085655e-05,
+      "loss": 0.0131,
+      "step": 155530
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9253058930347185e-05,
+      "loss": 0.0084,
+      "step": 155540
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925301089060871e-05,
+      "loss": 0.009,
+      "step": 155550
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925296285087024e-05,
+      "loss": 0.0087,
+      "step": 155560
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925291481113177e-05,
+      "loss": 0.0093,
+      "step": 155570
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.92528667713933e-05,
+      "loss": 0.0068,
+      "step": 155580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9252818731654824e-05,
+      "loss": 0.0057,
+      "step": 155590
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9252770691916354e-05,
+      "loss": 0.0064,
+      "step": 155600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9252722652177883e-05,
+      "loss": 0.006,
+      "step": 155610
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9252674612439412e-05,
+      "loss": 0.0077,
+      "step": 155620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9252626572700938e-05,
+      "loss": 0.0058,
+      "step": 155630
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9252578532962467e-05,
+      "loss": 0.01,
+      "step": 155640
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9252530493223996e-05,
+      "loss": 0.0067,
+      "step": 155650
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9252482453485522e-05,
+      "loss": 0.0057,
+      "step": 155660
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9252434413747055e-05,
+      "loss": 0.0074,
+      "step": 155670
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925238637400858e-05,
+      "loss": 0.0056,
+      "step": 155680
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925233833427011e-05,
+      "loss": 0.0081,
+      "step": 155690
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925229029453164e-05,
+      "loss": 0.0064,
+      "step": 155700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9252247058767014e-05,
+      "loss": 0.0071,
+      "step": 155710
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925219901902854e-05,
+      "loss": 0.0086,
+      "step": 155720
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.925215097929007e-05,
+      "loss": 0.0122,
+      "step": 155730
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.92521029395516e-05,
+      "loss": 0.0082,
+      "step": 155740
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9252054899813128e-05,
+      "loss": 0.0095,
+      "step": 155750
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9252006860074654e-05,
+      "loss": 0.0083,
+      "step": 155760
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9251958820336183e-05,
+      "loss": 0.0063,
+      "step": 155770
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251910780597712e-05,
+      "loss": 0.0065,
+      "step": 155780
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251862740859238e-05,
+      "loss": 0.006,
+      "step": 155790
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251814701120767e-05,
+      "loss": 0.0066,
+      "step": 155800
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251766661382297e-05,
+      "loss": 0.0115,
+      "step": 155810
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251718621643826e-05,
+      "loss": 0.0093,
+      "step": 155820
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251670581905352e-05,
+      "loss": 0.01,
+      "step": 155830
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925162254216688e-05,
+      "loss": 0.008,
+      "step": 155840
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925157450242841e-05,
+      "loss": 0.0075,
+      "step": 155850
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251526462689936e-05,
+      "loss": 0.0066,
+      "step": 155860
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251478422951465e-05,
+      "loss": 0.0091,
+      "step": 155870
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251430383212995e-05,
+      "loss": 0.0059,
+      "step": 155880
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251382343474524e-05,
+      "loss": 0.0065,
+      "step": 155890
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925133430373605e-05,
+      "loss": 0.0047,
+      "step": 155900
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925128626399758e-05,
+      "loss": 0.0082,
+      "step": 155910
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925123822425911e-05,
+      "loss": 0.0088,
+      "step": 155920
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251190184520638e-05,
+      "loss": 0.0104,
+      "step": 155930
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251142144782164e-05,
+      "loss": 0.0078,
+      "step": 155940
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251094105043693e-05,
+      "loss": 0.0087,
+      "step": 155950
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251046065305222e-05,
+      "loss": 0.0097,
+      "step": 155960
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250998025566748e-05,
+      "loss": 0.0043,
+      "step": 155970
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250949985828277e-05,
+      "loss": 0.0069,
+      "step": 155980
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250901946089807e-05,
+      "loss": 0.0087,
+      "step": 155990
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250853906351336e-05,
+      "loss": 0.0087,
+      "step": 156000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925080586661286e-05,
+      "loss": 0.0109,
+      "step": 156010
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925075782687439e-05,
+      "loss": 0.0087,
+      "step": 156020
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925070978713592e-05,
+      "loss": 0.0066,
+      "step": 156030
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250661747397446e-05,
+      "loss": 0.0096,
+      "step": 156040
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250613707658975e-05,
+      "loss": 0.0073,
+      "step": 156050
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250565667920505e-05,
+      "loss": 0.0085,
+      "step": 156060
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250517628182034e-05,
+      "loss": 0.0058,
+      "step": 156070
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925046958844356e-05,
+      "loss": 0.0089,
+      "step": 156080
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925042154870509e-05,
+      "loss": 0.006,
+      "step": 156090
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925037350896662e-05,
+      "loss": 0.009,
+      "step": 156100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250325469228148e-05,
+      "loss": 0.0086,
+      "step": 156110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250277429489673e-05,
+      "loss": 0.0078,
+      "step": 156120
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250229389751203e-05,
+      "loss": 0.0073,
+      "step": 156130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250181350012732e-05,
+      "loss": 0.0055,
+      "step": 156140
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250138114248107e-05,
+      "loss": 0.0077,
+      "step": 156150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250090074509636e-05,
+      "loss": 0.007,
+      "step": 156160
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9250042034771165e-05,
+      "loss": 0.0079,
+      "step": 156170
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.924999399503269e-05,
+      "loss": 0.009,
+      "step": 156180
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.924994595529422e-05,
+      "loss": 0.0069,
+      "step": 156190
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.924989791555575e-05,
+      "loss": 0.0057,
+      "step": 156200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9249849875817276e-05,
+      "loss": 0.0078,
+      "step": 156210
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9249801836078805e-05,
+      "loss": 0.0076,
+      "step": 156220
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9249753796340334e-05,
+      "loss": 0.0078,
+      "step": 156230
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9249705756601863e-05,
+      "loss": 0.0113,
+      "step": 156240
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.924965771686339e-05,
+      "loss": 0.0078,
+      "step": 156250
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.924960967712492e-05,
+      "loss": 0.0107,
+      "step": 156260
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9249561637386448e-05,
+      "loss": 0.0084,
+      "step": 156270
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9249513597647974e-05,
+      "loss": 0.0039,
+      "step": 156280
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9249465557909503e-05,
+      "loss": 0.0047,
+      "step": 156290
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9249417518171032e-05,
+      "loss": 0.0085,
+      "step": 156300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.924936947843256e-05,
+      "loss": 0.0086,
+      "step": 156310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9249321438694087e-05,
+      "loss": 0.0067,
+      "step": 156320
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9249273398955617e-05,
+      "loss": 0.0061,
+      "step": 156330
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9249225359217146e-05,
+      "loss": 0.0074,
+      "step": 156340
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9249177319478675e-05,
+      "loss": 0.0069,
+      "step": 156350
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.92491292797402e-05,
+      "loss": 0.0071,
+      "step": 156360
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.924908124000173e-05,
+      "loss": 0.0065,
+      "step": 156370
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.924903320026326e-05,
+      "loss": 0.0093,
+      "step": 156380
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9248985160524785e-05,
+      "loss": 0.0081,
+      "step": 156390
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9248937120786315e-05,
+      "loss": 0.0097,
+      "step": 156400
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9248889081047844e-05,
+      "loss": 0.0064,
+      "step": 156410
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9248841041309373e-05,
+      "loss": 0.0087,
+      "step": 156420
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.92487930015709e-05,
+      "loss": 0.0065,
+      "step": 156430
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.924874496183243e-05,
+      "loss": 0.0051,
+      "step": 156440
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9248696922093958e-05,
+      "loss": 0.0067,
+      "step": 156450
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9248648882355484e-05,
+      "loss": 0.0074,
+      "step": 156460
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9248600842617013e-05,
+      "loss": 0.0074,
+      "step": 156470
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9248552802878542e-05,
+      "loss": 0.0088,
+      "step": 156480
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924850476314007e-05,
+      "loss": 0.0085,
+      "step": 156490
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9248456723401597e-05,
+      "loss": 0.0064,
+      "step": 156500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9248408683663126e-05,
+      "loss": 0.0061,
+      "step": 156510
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9248360643924656e-05,
+      "loss": 0.0085,
+      "step": 156520
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9248312604186185e-05,
+      "loss": 0.0068,
+      "step": 156530
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924826456444771e-05,
+      "loss": 0.0101,
+      "step": 156540
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924821652470924e-05,
+      "loss": 0.0074,
+      "step": 156550
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924816848497077e-05,
+      "loss": 0.0065,
+      "step": 156560
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9248120445232295e-05,
+      "loss": 0.0063,
+      "step": 156570
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9248072405493825e-05,
+      "loss": 0.0087,
+      "step": 156580
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9248024365755354e-05,
+      "loss": 0.0065,
+      "step": 156590
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247976326016883e-05,
+      "loss": 0.0082,
+      "step": 156600
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924792828627841e-05,
+      "loss": 0.0059,
+      "step": 156610
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247880246539938e-05,
+      "loss": 0.0077,
+      "step": 156620
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247832206801468e-05,
+      "loss": 0.0103,
+      "step": 156630
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247784167062993e-05,
+      "loss": 0.0078,
+      "step": 156640
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247736127324523e-05,
+      "loss": 0.0064,
+      "step": 156650
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247688087586052e-05,
+      "loss": 0.0098,
+      "step": 156660
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924764004784758e-05,
+      "loss": 0.0068,
+      "step": 156670
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247592008109107e-05,
+      "loss": 0.0079,
+      "step": 156680
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247543968370636e-05,
+      "loss": 0.009,
+      "step": 156690
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247495928632166e-05,
+      "loss": 0.0088,
+      "step": 156700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247447888893695e-05,
+      "loss": 0.0103,
+      "step": 156710
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247399849155224e-05,
+      "loss": 0.0093,
+      "step": 156720
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247351809416753e-05,
+      "loss": 0.0087,
+      "step": 156730
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924730376967828e-05,
+      "loss": 0.0092,
+      "step": 156740
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924725572993981e-05,
+      "loss": 0.0084,
+      "step": 156750
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247207690201338e-05,
+      "loss": 0.0066,
+      "step": 156760
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247159650462867e-05,
+      "loss": 0.0073,
+      "step": 156770
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247111610724393e-05,
+      "loss": 0.0107,
+      "step": 156780
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9247063570985922e-05,
+      "loss": 0.0106,
+      "step": 156790
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924701553124745e-05,
+      "loss": 0.0076,
+      "step": 156800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246967491508977e-05,
+      "loss": 0.0083,
+      "step": 156810
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246919451770507e-05,
+      "loss": 0.0073,
+      "step": 156820
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246871412032036e-05,
+      "loss": 0.0066,
+      "step": 156830
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246823372293565e-05,
+      "loss": 0.0077,
+      "step": 156840
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924677533255509e-05,
+      "loss": 0.007,
+      "step": 156850
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924672729281662e-05,
+      "loss": 0.0119,
+      "step": 156860
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924667925307815e-05,
+      "loss": 0.012,
+      "step": 156870
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246631213339676e-05,
+      "loss": 0.0078,
+      "step": 156880
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246583173601205e-05,
+      "loss": 0.0079,
+      "step": 156890
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246535133862734e-05,
+      "loss": 0.0085,
+      "step": 156900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246487094124263e-05,
+      "loss": 0.0107,
+      "step": 156910
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924643905438579e-05,
+      "loss": 0.0084,
+      "step": 156920
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924639101464732e-05,
+      "loss": 0.0079,
+      "step": 156930
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246342974908848e-05,
+      "loss": 0.009,
+      "step": 156940
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246294935170374e-05,
+      "loss": 0.0099,
+      "step": 156950
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246246895431903e-05,
+      "loss": 0.0074,
+      "step": 156960
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246198855693432e-05,
+      "loss": 0.0096,
+      "step": 156970
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924615081595496e-05,
+      "loss": 0.0071,
+      "step": 156980
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246102776216487e-05,
+      "loss": 0.0082,
+      "step": 156990
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246054736478017e-05,
+      "loss": 0.0068,
+      "step": 157000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246006696739546e-05,
+      "loss": 0.0046,
+      "step": 157010
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9245958657001075e-05,
+      "loss": 0.0061,
+      "step": 157020
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.92459106172626e-05,
+      "loss": 0.0075,
+      "step": 157030
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924586257752413e-05,
+      "loss": 0.0115,
+      "step": 157040
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924581453778566e-05,
+      "loss": 0.0082,
+      "step": 157050
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9245766498047186e-05,
+      "loss": 0.0061,
+      "step": 157060
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9245718458308715e-05,
+      "loss": 0.0093,
+      "step": 157070
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9245670418570244e-05,
+      "loss": 0.0073,
+      "step": 157080
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9245622378831773e-05,
+      "loss": 0.0071,
+      "step": 157090
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.92455743390933e-05,
+      "loss": 0.0072,
+      "step": 157100
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924552629935483e-05,
+      "loss": 0.0047,
+      "step": 157110
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9245478259616358e-05,
+      "loss": 0.0094,
+      "step": 157120
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9245430219877884e-05,
+      "loss": 0.006,
+      "step": 157130
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9245382180139413e-05,
+      "loss": 0.0079,
+      "step": 157140
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9245334140400942e-05,
+      "loss": 0.0075,
+      "step": 157150
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924528610066247e-05,
+      "loss": 0.0093,
+      "step": 157160
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9245238060923997e-05,
+      "loss": 0.0066,
+      "step": 157170
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9245190021185527e-05,
+      "loss": 0.0048,
+      "step": 157180
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9245141981447056e-05,
+      "loss": 0.0079,
+      "step": 157190
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9245093941708585e-05,
+      "loss": 0.0083,
+      "step": 157200
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924504590197011e-05,
+      "loss": 0.0081,
+      "step": 157210
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924499786223164e-05,
+      "loss": 0.0086,
+      "step": 157220
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924494982249317e-05,
+      "loss": 0.0075,
+      "step": 157230
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244901782754695e-05,
+      "loss": 0.011,
+      "step": 157240
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244853743016225e-05,
+      "loss": 0.0087,
+      "step": 157250
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244805703277754e-05,
+      "loss": 0.008,
+      "step": 157260
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244757663539283e-05,
+      "loss": 0.0102,
+      "step": 157270
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924470962380081e-05,
+      "loss": 0.0097,
+      "step": 157280
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924466158406234e-05,
+      "loss": 0.0068,
+      "step": 157290
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244613544323868e-05,
+      "loss": 0.0079,
+      "step": 157300
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244565504585394e-05,
+      "loss": 0.0085,
+      "step": 157310
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244517464846923e-05,
+      "loss": 0.0082,
+      "step": 157320
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244469425108452e-05,
+      "loss": 0.0075,
+      "step": 157330
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924442138536998e-05,
+      "loss": 0.008,
+      "step": 157340
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244373345631507e-05,
+      "loss": 0.0058,
+      "step": 157350
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244325305893036e-05,
+      "loss": 0.008,
+      "step": 157360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244277266154566e-05,
+      "loss": 0.013,
+      "step": 157370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244229226416095e-05,
+      "loss": 0.0085,
+      "step": 157380
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924418118667762e-05,
+      "loss": 0.008,
+      "step": 157390
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924413314693915e-05,
+      "loss": 0.0087,
+      "step": 157400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924408510720068e-05,
+      "loss": 0.0098,
+      "step": 157410
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244037067462205e-05,
+      "loss": 0.0084,
+      "step": 157420
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243989027723735e-05,
+      "loss": 0.0213,
+      "step": 157430
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243940987985264e-05,
+      "loss": 0.0112,
+      "step": 157440
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243892948246793e-05,
+      "loss": 0.0075,
+      "step": 157450
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924384490850832e-05,
+      "loss": 0.0089,
+      "step": 157460
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243796868769848e-05,
+      "loss": 0.0066,
+      "step": 157470
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243748829031378e-05,
+      "loss": 0.0071,
+      "step": 157480
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243700789292903e-05,
+      "loss": 0.0097,
+      "step": 157490
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243652749554433e-05,
+      "loss": 0.008,
+      "step": 157500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243604709815962e-05,
+      "loss": 0.0062,
+      "step": 157510
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924355667007749e-05,
+      "loss": 0.0074,
+      "step": 157520
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243508630339017e-05,
+      "loss": 0.0104,
+      "step": 157530
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243460590600546e-05,
+      "loss": 0.0067,
+      "step": 157540
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243412550862076e-05,
+      "loss": 0.0074,
+      "step": 157550
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243364511123605e-05,
+      "loss": 0.008,
+      "step": 157560
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924331647138513e-05,
+      "loss": 0.0073,
+      "step": 157570
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924326843164666e-05,
+      "loss": 0.0096,
+      "step": 157580
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924322039190819e-05,
+      "loss": 0.0077,
+      "step": 157590
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243172352169715e-05,
+      "loss": 0.0129,
+      "step": 157600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243124312431245e-05,
+      "loss": 0.0065,
+      "step": 157610
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243076272692774e-05,
+      "loss": 0.0074,
+      "step": 157620
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9243028232954303e-05,
+      "loss": 0.0075,
+      "step": 157630
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924298019321583e-05,
+      "loss": 0.0094,
+      "step": 157640
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242932153477358e-05,
+      "loss": 0.0095,
+      "step": 157650
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242884113738887e-05,
+      "loss": 0.0077,
+      "step": 157660
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242836074000413e-05,
+      "loss": 0.0072,
+      "step": 157670
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242788034261943e-05,
+      "loss": 0.007,
+      "step": 157680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242739994523472e-05,
+      "loss": 0.0046,
+      "step": 157690
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242691954785e-05,
+      "loss": 0.0051,
+      "step": 157700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242643915046527e-05,
+      "loss": 0.0083,
+      "step": 157710
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242595875308056e-05,
+      "loss": 0.0078,
+      "step": 157720
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242547835569586e-05,
+      "loss": 0.0087,
+      "step": 157730
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242499795831115e-05,
+      "loss": 0.0073,
+      "step": 157740
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924245175609264e-05,
+      "loss": 0.0076,
+      "step": 157750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924240371635417e-05,
+      "loss": 0.0052,
+      "step": 157760
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.92423556766157e-05,
+      "loss": 0.0097,
+      "step": 157770
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242307636877225e-05,
+      "loss": 0.0077,
+      "step": 157780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242259597138754e-05,
+      "loss": 0.0047,
+      "step": 157790
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242211557400284e-05,
+      "loss": 0.0073,
+      "step": 157800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242163517661813e-05,
+      "loss": 0.0096,
+      "step": 157810
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924211547792334e-05,
+      "loss": 0.01,
+      "step": 157820
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242067438184868e-05,
+      "loss": 0.007,
+      "step": 157830
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9242019398446397e-05,
+      "loss": 0.0077,
+      "step": 157840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9241971358707923e-05,
+      "loss": 0.0089,
+      "step": 157850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241923318969453e-05,
+      "loss": 0.0061,
+      "step": 157860
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241875279230982e-05,
+      "loss": 0.0093,
+      "step": 157870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.924182723949251e-05,
+      "loss": 0.0079,
+      "step": 157880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241779199754037e-05,
+      "loss": 0.008,
+      "step": 157890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241731160015566e-05,
+      "loss": 0.0108,
+      "step": 157900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241683120277096e-05,
+      "loss": 0.0083,
+      "step": 157910
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241635080538625e-05,
+      "loss": 0.0078,
+      "step": 157920
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.924158704080015e-05,
+      "loss": 0.0089,
+      "step": 157930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.924153900106168e-05,
+      "loss": 0.0064,
+      "step": 157940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.924149096132321e-05,
+      "loss": 0.0101,
+      "step": 157950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241442921584735e-05,
+      "loss": 0.0086,
+      "step": 157960
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241394881846264e-05,
+      "loss": 0.0056,
+      "step": 157970
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241346842107794e-05,
+      "loss": 0.0061,
+      "step": 157980
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241298802369323e-05,
+      "loss": 0.011,
+      "step": 157990
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.924125076263085e-05,
+      "loss": 0.008,
+      "step": 158000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241202722892378e-05,
+      "loss": 0.0097,
+      "step": 158010
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241154683153907e-05,
+      "loss": 0.0086,
+      "step": 158020
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241106643415433e-05,
+      "loss": 0.0096,
+      "step": 158030
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241058603676962e-05,
+      "loss": 0.0075,
+      "step": 158040
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9241010563938492e-05,
+      "loss": 0.0157,
+      "step": 158050
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.924096252420002e-05,
+      "loss": 0.0109,
+      "step": 158060
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240914484461547e-05,
+      "loss": 0.0083,
+      "step": 158070
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240866444723076e-05,
+      "loss": 0.0052,
+      "step": 158080
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240818404984605e-05,
+      "loss": 0.0066,
+      "step": 158090
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240770365246135e-05,
+      "loss": 0.0059,
+      "step": 158100
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.924072232550766e-05,
+      "loss": 0.0111,
+      "step": 158110
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.924067428576919e-05,
+      "loss": 0.0079,
+      "step": 158120
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.924062624603072e-05,
+      "loss": 0.0077,
+      "step": 158130
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240578206292245e-05,
+      "loss": 0.0069,
+      "step": 158140
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240530166553774e-05,
+      "loss": 0.0079,
+      "step": 158150
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240482126815304e-05,
+      "loss": 0.0098,
+      "step": 158160
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240434087076833e-05,
+      "loss": 0.0063,
+      "step": 158170
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.924038604733836e-05,
+      "loss": 0.0056,
+      "step": 158180
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240338007599888e-05,
+      "loss": 0.0062,
+      "step": 158190
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240289967861417e-05,
+      "loss": 0.0082,
+      "step": 158200
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240241928122943e-05,
+      "loss": 0.0088,
+      "step": 158210
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240193888384472e-05,
+      "loss": 0.012,
+      "step": 158220
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240145848646e-05,
+      "loss": 0.0065,
+      "step": 158230
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.924009780890753e-05,
+      "loss": 0.0101,
+      "step": 158240
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240049769169057e-05,
+      "loss": 0.0074,
+      "step": 158250
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9240001729430586e-05,
+      "loss": 0.0113,
+      "step": 158260
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239953689692115e-05,
+      "loss": 0.0085,
+      "step": 158270
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239905649953645e-05,
+      "loss": 0.0057,
+      "step": 158280
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.923985761021517e-05,
+      "loss": 0.0073,
+      "step": 158290
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.92398095704767e-05,
+      "loss": 0.013,
+      "step": 158300
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.923976153073823e-05,
+      "loss": 0.0083,
+      "step": 158310
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239713490999755e-05,
+      "loss": 0.0068,
+      "step": 158320
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239665451261284e-05,
+      "loss": 0.0089,
+      "step": 158330
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239617411522813e-05,
+      "loss": 0.0051,
+      "step": 158340
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239569371784343e-05,
+      "loss": 0.0069,
+      "step": 158350
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.923952133204587e-05,
+      "loss": 0.0093,
+      "step": 158360
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239473292307398e-05,
+      "loss": 0.0064,
+      "step": 158370
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239425252568927e-05,
+      "loss": 0.0063,
+      "step": 158380
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239377212830453e-05,
+      "loss": 0.0086,
+      "step": 158390
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239329173091982e-05,
+      "loss": 0.008,
+      "step": 158400
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.923928113335351e-05,
+      "loss": 0.0086,
+      "step": 158410
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.923923309361504e-05,
+      "loss": 0.0045,
+      "step": 158420
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239185053876567e-05,
+      "loss": 0.0067,
+      "step": 158430
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239137014138096e-05,
+      "loss": 0.0087,
+      "step": 158440
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239088974399625e-05,
+      "loss": 0.0061,
+      "step": 158450
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9239040934661155e-05,
+      "loss": 0.007,
+      "step": 158460
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.923899289492268e-05,
+      "loss": 0.0082,
+      "step": 158470
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.923894485518421e-05,
+      "loss": 0.0097,
+      "step": 158480
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.923889681544574e-05,
+      "loss": 0.0095,
+      "step": 158490
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9238848775707265e-05,
+      "loss": 0.0062,
+      "step": 158500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9238800735968794e-05,
+      "loss": 0.0101,
+      "step": 158510
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9238752696230323e-05,
+      "loss": 0.0065,
+      "step": 158520
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9238704656491853e-05,
+      "loss": 0.0096,
+      "step": 158530
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.923865661675338e-05,
+      "loss": 0.0065,
+      "step": 158540
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9238608577014908e-05,
+      "loss": 0.0093,
+      "step": 158550
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9238560537276437e-05,
+      "loss": 0.0062,
+      "step": 158560
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9238512497537963e-05,
+      "loss": 0.0093,
+      "step": 158570
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9238464457799492e-05,
+      "loss": 0.0078,
+      "step": 158580
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923841641806102e-05,
+      "loss": 0.0054,
+      "step": 158590
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923836837832255e-05,
+      "loss": 0.0093,
+      "step": 158600
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9238320338584077e-05,
+      "loss": 0.0084,
+      "step": 158610
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9238272298845606e-05,
+      "loss": 0.0097,
+      "step": 158620
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9238224259107135e-05,
+      "loss": 0.0045,
+      "step": 158630
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9238176219368664e-05,
+      "loss": 0.0081,
+      "step": 158640
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923812817963019e-05,
+      "loss": 0.0072,
+      "step": 158650
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923808013989172e-05,
+      "loss": 0.0073,
+      "step": 158660
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923803210015325e-05,
+      "loss": 0.007,
+      "step": 158670
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237984060414775e-05,
+      "loss": 0.0074,
+      "step": 158680
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237936020676304e-05,
+      "loss": 0.0097,
+      "step": 158690
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237887980937833e-05,
+      "loss": 0.0078,
+      "step": 158700
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237839941199363e-05,
+      "loss": 0.0091,
+      "step": 158710
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923779190146089e-05,
+      "loss": 0.0093,
+      "step": 158720
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237743861722418e-05,
+      "loss": 0.0103,
+      "step": 158730
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237695821983947e-05,
+      "loss": 0.0068,
+      "step": 158740
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237647782245473e-05,
+      "loss": 0.0067,
+      "step": 158750
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237599742507002e-05,
+      "loss": 0.0084,
+      "step": 158760
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923755170276853e-05,
+      "loss": 0.0075,
+      "step": 158770
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923750366303006e-05,
+      "loss": 0.0091,
+      "step": 158780
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237455623291587e-05,
+      "loss": 0.0108,
+      "step": 158790
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237407583553116e-05,
+      "loss": 0.0068,
+      "step": 158800
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237359543814645e-05,
+      "loss": 0.008,
+      "step": 158810
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923731150407617e-05,
+      "loss": 0.0092,
+      "step": 158820
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.92372634643377e-05,
+      "loss": 0.0076,
+      "step": 158830
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923721542459923e-05,
+      "loss": 0.0059,
+      "step": 158840
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923716738486076e-05,
+      "loss": 0.0099,
+      "step": 158850
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237119345122285e-05,
+      "loss": 0.0117,
+      "step": 158860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237071305383814e-05,
+      "loss": 0.0081,
+      "step": 158870
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9237023265645343e-05,
+      "loss": 0.0087,
+      "step": 158880
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236975225906872e-05,
+      "loss": 0.0076,
+      "step": 158890
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.92369271861684e-05,
+      "loss": 0.0058,
+      "step": 158900
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236879146429928e-05,
+      "loss": 0.0069,
+      "step": 158910
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236831106691457e-05,
+      "loss": 0.0072,
+      "step": 158920
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236783066952983e-05,
+      "loss": 0.0068,
+      "step": 158930
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236735027214512e-05,
+      "loss": 0.0061,
+      "step": 158940
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923668698747604e-05,
+      "loss": 0.0067,
+      "step": 158950
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923663894773757e-05,
+      "loss": 0.0135,
+      "step": 158960
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236590907999096e-05,
+      "loss": 0.0083,
+      "step": 158970
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236542868260626e-05,
+      "loss": 0.0124,
+      "step": 158980
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236494828522155e-05,
+      "loss": 0.0084,
+      "step": 158990
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923644678878368e-05,
+      "loss": 0.0068,
+      "step": 159000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923639874904521e-05,
+      "loss": 0.0062,
+      "step": 159010
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923635070930674e-05,
+      "loss": 0.0092,
+      "step": 159020
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923630266956827e-05,
+      "loss": 0.0097,
+      "step": 159030
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236254629829795e-05,
+      "loss": 0.0109,
+      "step": 159040
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236206590091324e-05,
+      "loss": 0.0111,
+      "step": 159050
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236158550352853e-05,
+      "loss": 0.0081,
+      "step": 159060
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236110510614382e-05,
+      "loss": 0.0048,
+      "step": 159070
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236062470875908e-05,
+      "loss": 0.0067,
+      "step": 159080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9236014431137438e-05,
+      "loss": 0.0073,
+      "step": 159090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9235966391398967e-05,
+      "loss": 0.0216,
+      "step": 159100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9235918351660493e-05,
+      "loss": 0.0059,
+      "step": 159110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9235870311922022e-05,
+      "loss": 0.007,
+      "step": 159120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923582227218355e-05,
+      "loss": 0.0067,
+      "step": 159130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923577423244508e-05,
+      "loss": 0.0101,
+      "step": 159140
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9235726192706606e-05,
+      "loss": 0.006,
+      "step": 159150
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9235678152968136e-05,
+      "loss": 0.0085,
+      "step": 159160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9235630113229665e-05,
+      "loss": 0.0095,
+      "step": 159170
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923558207349119e-05,
+      "loss": 0.0092,
+      "step": 159180
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923553403375272e-05,
+      "loss": 0.0083,
+      "step": 159190
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923548599401425e-05,
+      "loss": 0.0064,
+      "step": 159200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.923543795427578e-05,
+      "loss": 0.0051,
+      "step": 159210
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9235389914537304e-05,
+      "loss": 0.0097,
+      "step": 159220
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9235341874798834e-05,
+      "loss": 0.0075,
+      "step": 159230
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9235293835060363e-05,
+      "loss": 0.0076,
+      "step": 159240
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9235245795321892e-05,
+      "loss": 0.0047,
+      "step": 159250
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923519775558342e-05,
+      "loss": 0.0077,
+      "step": 159260
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923514971584495e-05,
+      "loss": 0.0076,
+      "step": 159270
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9235101676106477e-05,
+      "loss": 0.0121,
+      "step": 159280
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9235053636368006e-05,
+      "loss": 0.0084,
+      "step": 159290
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9235005596629535e-05,
+      "loss": 0.0049,
+      "step": 159300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234957556891065e-05,
+      "loss": 0.0072,
+      "step": 159310
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923490951715259e-05,
+      "loss": 0.0077,
+      "step": 159320
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923486147741412e-05,
+      "loss": 0.0067,
+      "step": 159330
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923481343767565e-05,
+      "loss": 0.0089,
+      "step": 159340
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234765397937175e-05,
+      "loss": 0.0089,
+      "step": 159350
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234717358198704e-05,
+      "loss": 0.0063,
+      "step": 159360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234669318460233e-05,
+      "loss": 0.0095,
+      "step": 159370
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234621278721763e-05,
+      "loss": 0.0073,
+      "step": 159380
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923457323898329e-05,
+      "loss": 0.0061,
+      "step": 159390
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234525199244818e-05,
+      "loss": 0.01,
+      "step": 159400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234477159506347e-05,
+      "loss": 0.0076,
+      "step": 159410
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234429119767873e-05,
+      "loss": 0.0073,
+      "step": 159420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234381080029402e-05,
+      "loss": 0.007,
+      "step": 159430
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923433304029093e-05,
+      "loss": 0.008,
+      "step": 159440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923428500055246e-05,
+      "loss": 0.0055,
+      "step": 159450
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234236960813987e-05,
+      "loss": 0.0056,
+      "step": 159460
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234188921075516e-05,
+      "loss": 0.0059,
+      "step": 159470
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234140881337045e-05,
+      "loss": 0.0081,
+      "step": 159480
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9234092841598574e-05,
+      "loss": 0.0063,
+      "step": 159490
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.92340448018601e-05,
+      "loss": 0.0053,
+      "step": 159500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923399676212163e-05,
+      "loss": 0.0122,
+      "step": 159510
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923394872238316e-05,
+      "loss": 0.0095,
+      "step": 159520
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233900682644685e-05,
+      "loss": 0.0117,
+      "step": 159530
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233852642906214e-05,
+      "loss": 0.0088,
+      "step": 159540
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233804603167743e-05,
+      "loss": 0.0078,
+      "step": 159550
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233756563429273e-05,
+      "loss": 0.0091,
+      "step": 159560
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.92337085236908e-05,
+      "loss": 0.0079,
+      "step": 159570
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233660483952328e-05,
+      "loss": 0.0051,
+      "step": 159580
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233612444213857e-05,
+      "loss": 0.007,
+      "step": 159590
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233564404475383e-05,
+      "loss": 0.0065,
+      "step": 159600
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233516364736912e-05,
+      "loss": 0.0086,
+      "step": 159610
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923346832499844e-05,
+      "loss": 0.0102,
+      "step": 159620
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923342028525997e-05,
+      "loss": 0.0075,
+      "step": 159630
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233372245521497e-05,
+      "loss": 0.0058,
+      "step": 159640
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233324205783026e-05,
+      "loss": 0.0061,
+      "step": 159650
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233276166044555e-05,
+      "loss": 0.0108,
+      "step": 159660
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233228126306084e-05,
+      "loss": 0.0081,
+      "step": 159670
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923318008656761e-05,
+      "loss": 0.0073,
+      "step": 159680
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923313204682914e-05,
+      "loss": 0.0084,
+      "step": 159690
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923308400709067e-05,
+      "loss": 0.0079,
+      "step": 159700
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9233035967352195e-05,
+      "loss": 0.0063,
+      "step": 159710
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232987927613724e-05,
+      "loss": 0.0084,
+      "step": 159720
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232939887875253e-05,
+      "loss": 0.0092,
+      "step": 159730
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232891848136782e-05,
+      "loss": 0.0092,
+      "step": 159740
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923284380839831e-05,
+      "loss": 0.0108,
+      "step": 159750
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232795768659838e-05,
+      "loss": 0.01,
+      "step": 159760
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232747728921367e-05,
+      "loss": 0.0093,
+      "step": 159770
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232699689182893e-05,
+      "loss": 0.006,
+      "step": 159780
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232651649444422e-05,
+      "loss": 0.005,
+      "step": 159790
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923260360970595e-05,
+      "loss": 0.0075,
+      "step": 159800
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923255556996748e-05,
+      "loss": 0.0084,
+      "step": 159810
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232507530229006e-05,
+      "loss": 0.0101,
+      "step": 159820
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232459490490536e-05,
+      "loss": 0.0081,
+      "step": 159830
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232411450752065e-05,
+      "loss": 0.0067,
+      "step": 159840
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232363411013594e-05,
+      "loss": 0.0083,
+      "step": 159850
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923231537127512e-05,
+      "loss": 0.0067,
+      "step": 159860
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923226733153665e-05,
+      "loss": 0.0075,
+      "step": 159870
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.923221929179818e-05,
+      "loss": 0.0058,
+      "step": 159880
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232171252059705e-05,
+      "loss": 0.0082,
+      "step": 159890
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232123212321234e-05,
+      "loss": 0.008,
+      "step": 159900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232075172582763e-05,
+      "loss": 0.0083,
+      "step": 159910
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9232027132844292e-05,
+      "loss": 0.0081,
+      "step": 159920
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9231979093105818e-05,
+      "loss": 0.0084,
+      "step": 159930
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231931053367348e-05,
+      "loss": 0.0081,
+      "step": 159940
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231883013628877e-05,
+      "loss": 0.0082,
+      "step": 159950
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231834973890403e-05,
+      "loss": 0.0058,
+      "step": 159960
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231786934151932e-05,
+      "loss": 0.0079,
+      "step": 159970
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.923173889441346e-05,
+      "loss": 0.0093,
+      "step": 159980
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.923169085467499e-05,
+      "loss": 0.0075,
+      "step": 159990
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231642814936516e-05,
+      "loss": 0.007,
+      "step": 160000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231594775198046e-05,
+      "loss": 0.0085,
+      "step": 160010
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231546735459575e-05,
+      "loss": 0.0071,
+      "step": 160020
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231498695721104e-05,
+      "loss": 0.007,
+      "step": 160030
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.923145065598263e-05,
+      "loss": 0.0089,
+      "step": 160040
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.923140261624416e-05,
+      "loss": 0.0096,
+      "step": 160050
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.923135457650569e-05,
+      "loss": 0.0067,
+      "step": 160060
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231306536767214e-05,
+      "loss": 0.0082,
+      "step": 160070
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231258497028744e-05,
+      "loss": 0.0096,
+      "step": 160080
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231210457290273e-05,
+      "loss": 0.0059,
+      "step": 160090
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231162417551802e-05,
+      "loss": 0.0069,
+      "step": 160100
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231114377813328e-05,
+      "loss": 0.0092,
+      "step": 160110
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231066338074857e-05,
+      "loss": 0.0079,
+      "step": 160120
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9231018298336387e-05,
+      "loss": 0.0077,
+      "step": 160130
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230970258597913e-05,
+      "loss": 0.008,
+      "step": 160140
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230922218859442e-05,
+      "loss": 0.0124,
+      "step": 160150
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.923087417912097e-05,
+      "loss": 0.0099,
+      "step": 160160
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.92308261393825e-05,
+      "loss": 0.0079,
+      "step": 160170
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230778099644026e-05,
+      "loss": 0.0061,
+      "step": 160180
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230730059905556e-05,
+      "loss": 0.0084,
+      "step": 160190
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230682020167085e-05,
+      "loss": 0.01,
+      "step": 160200
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230633980428614e-05,
+      "loss": 0.0074,
+      "step": 160210
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.923058594069014e-05,
+      "loss": 0.0061,
+      "step": 160220
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.923053790095167e-05,
+      "loss": 0.0068,
+      "step": 160230
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.92304898612132e-05,
+      "loss": 0.0078,
+      "step": 160240
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230441821474724e-05,
+      "loss": 0.0057,
+      "step": 160250
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230393781736254e-05,
+      "loss": 0.0058,
+      "step": 160260
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230345741997783e-05,
+      "loss": 0.006,
+      "step": 160270
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230297702259312e-05,
+      "loss": 0.011,
+      "step": 160280
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230249662520838e-05,
+      "loss": 0.0062,
+      "step": 160290
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230201622782367e-05,
+      "loss": 0.0073,
+      "step": 160300
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230153583043897e-05,
+      "loss": 0.0087,
+      "step": 160310
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230105543305423e-05,
+      "loss": 0.0095,
+      "step": 160320
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9230057503566952e-05,
+      "loss": 0.0069,
+      "step": 160330
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.923000946382848e-05,
+      "loss": 0.0072,
+      "step": 160340
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.922996142409001e-05,
+      "loss": 0.0101,
+      "step": 160350
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229913384351536e-05,
+      "loss": 0.0081,
+      "step": 160360
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229865344613065e-05,
+      "loss": 0.0091,
+      "step": 160370
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229817304874595e-05,
+      "loss": 0.0107,
+      "step": 160380
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.922976926513612e-05,
+      "loss": 0.0096,
+      "step": 160390
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.922972122539765e-05,
+      "loss": 0.0117,
+      "step": 160400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.922967318565918e-05,
+      "loss": 0.0096,
+      "step": 160410
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.922962514592071e-05,
+      "loss": 0.0065,
+      "step": 160420
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229577106182234e-05,
+      "loss": 0.0085,
+      "step": 160430
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229529066443764e-05,
+      "loss": 0.0083,
+      "step": 160440
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229481026705293e-05,
+      "loss": 0.0088,
+      "step": 160450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229432986966822e-05,
+      "loss": 0.0092,
+      "step": 160460
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229384947228348e-05,
+      "loss": 0.0075,
+      "step": 160470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229336907489877e-05,
+      "loss": 0.01,
+      "step": 160480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229288867751407e-05,
+      "loss": 0.0077,
+      "step": 160490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229240828012932e-05,
+      "loss": 0.0063,
+      "step": 160500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229192788274462e-05,
+      "loss": 0.0084,
+      "step": 160510
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.922914474853599e-05,
+      "loss": 0.0048,
+      "step": 160520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.922909670879752e-05,
+      "loss": 0.0063,
+      "step": 160530
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229048669059046e-05,
+      "loss": 0.0064,
+      "step": 160540
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9229000629320575e-05,
+      "loss": 0.0071,
+      "step": 160550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9228952589582105e-05,
+      "loss": 0.007,
+      "step": 160560
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.922890454984363e-05,
+      "loss": 0.0084,
+      "step": 160570
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.922885651010516e-05,
+      "loss": 0.0058,
+      "step": 160580
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.922880847036669e-05,
+      "loss": 0.0082,
+      "step": 160590
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.922876043062822e-05,
+      "loss": 0.0095,
+      "step": 160600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9228712390889744e-05,
+      "loss": 0.0072,
+      "step": 160610
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9228664351151273e-05,
+      "loss": 0.0092,
+      "step": 160620
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9228616311412803e-05,
+      "loss": 0.0068,
+      "step": 160630
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9228568271674332e-05,
+      "loss": 0.0098,
+      "step": 160640
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9228520231935858e-05,
+      "loss": 0.0084,
+      "step": 160650
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9228472192197387e-05,
+      "loss": 0.0063,
+      "step": 160660
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9228424152458916e-05,
+      "loss": 0.0088,
+      "step": 160670
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9228376112720442e-05,
+      "loss": 0.0099,
+      "step": 160680
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922832807298197e-05,
+      "loss": 0.0056,
+      "step": 160690
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.92282800332435e-05,
+      "loss": 0.0057,
+      "step": 160700
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922823199350503e-05,
+      "loss": 0.0099,
+      "step": 160710
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9228183953766556e-05,
+      "loss": 0.0049,
+      "step": 160720
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9228135914028085e-05,
+      "loss": 0.0068,
+      "step": 160730
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9228087874289615e-05,
+      "loss": 0.0091,
+      "step": 160740
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922803983455114e-05,
+      "loss": 0.0063,
+      "step": 160750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922799179481267e-05,
+      "loss": 0.0062,
+      "step": 160760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.92279437550742e-05,
+      "loss": 0.0112,
+      "step": 160770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227895715335728e-05,
+      "loss": 0.0106,
+      "step": 160780
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227847675597254e-05,
+      "loss": 0.0087,
+      "step": 160790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227799635858783e-05,
+      "loss": 0.0057,
+      "step": 160800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227751596120313e-05,
+      "loss": 0.0056,
+      "step": 160810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227703556381842e-05,
+      "loss": 0.0059,
+      "step": 160820
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227655516643368e-05,
+      "loss": 0.0049,
+      "step": 160830
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227607476904897e-05,
+      "loss": 0.0051,
+      "step": 160840
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227559437166426e-05,
+      "loss": 0.0049,
+      "step": 160850
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227511397427952e-05,
+      "loss": 0.0093,
+      "step": 160860
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922746335768948e-05,
+      "loss": 0.0083,
+      "step": 160870
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922741531795101e-05,
+      "loss": 0.0079,
+      "step": 160880
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922736727821254e-05,
+      "loss": 0.0076,
+      "step": 160890
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227319238474066e-05,
+      "loss": 0.0083,
+      "step": 160900
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227271198735595e-05,
+      "loss": 0.0068,
+      "step": 160910
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227223158997124e-05,
+      "loss": 0.0081,
+      "step": 160920
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922717511925865e-05,
+      "loss": 0.0076,
+      "step": 160930
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922712707952018e-05,
+      "loss": 0.0062,
+      "step": 160940
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922707903978171e-05,
+      "loss": 0.0077,
+      "step": 160950
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9227031000043238e-05,
+      "loss": 0.0055,
+      "step": 160960
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226982960304764e-05,
+      "loss": 0.0063,
+      "step": 160970
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226934920566293e-05,
+      "loss": 0.0075,
+      "step": 160980
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226886880827823e-05,
+      "loss": 0.0072,
+      "step": 160990
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226838841089352e-05,
+      "loss": 0.0063,
+      "step": 161000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226790801350878e-05,
+      "loss": 0.0082,
+      "step": 161010
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226742761612407e-05,
+      "loss": 0.0073,
+      "step": 161020
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226694721873936e-05,
+      "loss": 0.0069,
+      "step": 161030
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226646682135462e-05,
+      "loss": 0.0079,
+      "step": 161040
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922659864239699e-05,
+      "loss": 0.0097,
+      "step": 161050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922655060265852e-05,
+      "loss": 0.0054,
+      "step": 161060
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922650256292005e-05,
+      "loss": 0.0055,
+      "step": 161070
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226454523181576e-05,
+      "loss": 0.0071,
+      "step": 161080
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226406483443105e-05,
+      "loss": 0.0101,
+      "step": 161090
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226358443704634e-05,
+      "loss": 0.0061,
+      "step": 161100
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922631040396616e-05,
+      "loss": 0.0079,
+      "step": 161110
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922626236422769e-05,
+      "loss": 0.0085,
+      "step": 161120
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922621432448922e-05,
+      "loss": 0.0073,
+      "step": 161130
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226166284750748e-05,
+      "loss": 0.0099,
+      "step": 161140
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226118245012274e-05,
+      "loss": 0.0071,
+      "step": 161150
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226070205273803e-05,
+      "loss": 0.0099,
+      "step": 161160
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9226022165535333e-05,
+      "loss": 0.0082,
+      "step": 161170
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9225974125796862e-05,
+      "loss": 0.0059,
+      "step": 161180
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9225926086058388e-05,
+      "loss": 0.0063,
+      "step": 161190
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9225878046319917e-05,
+      "loss": 0.0075,
+      "step": 161200
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9225830006581446e-05,
+      "loss": 0.0072,
+      "step": 161210
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9225781966842972e-05,
+      "loss": 0.0076,
+      "step": 161220
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.92257339271045e-05,
+      "loss": 0.0056,
+      "step": 161230
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922568588736603e-05,
+      "loss": 0.0087,
+      "step": 161240
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922563784762756e-05,
+      "loss": 0.0082,
+      "step": 161250
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9225589807889086e-05,
+      "loss": 0.0065,
+      "step": 161260
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9225541768150615e-05,
+      "loss": 0.007,
+      "step": 161270
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9225493728412144e-05,
+      "loss": 0.0139,
+      "step": 161280
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922544568867367e-05,
+      "loss": 0.0071,
+      "step": 161290
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.92253976489352e-05,
+      "loss": 0.0106,
+      "step": 161300
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.922534960919673e-05,
+      "loss": 0.0099,
+      "step": 161310
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9225301569458258e-05,
+      "loss": 0.0078,
+      "step": 161320
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9225253529719784e-05,
+      "loss": 0.0076,
+      "step": 161330
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9225205489981313e-05,
+      "loss": 0.0069,
+      "step": 161340
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9225157450242842e-05,
+      "loss": 0.0079,
+      "step": 161350
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9225109410504372e-05,
+      "loss": 0.009,
+      "step": 161360
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9225061370765898e-05,
+      "loss": 0.0107,
+      "step": 161370
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9225013331027427e-05,
+      "loss": 0.0059,
+      "step": 161380
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224965291288956e-05,
+      "loss": 0.0079,
+      "step": 161390
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224917251550482e-05,
+      "loss": 0.0082,
+      "step": 161400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922486921181201e-05,
+      "loss": 0.0167,
+      "step": 161410
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922482117207354e-05,
+      "loss": 0.0095,
+      "step": 161420
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922477313233507e-05,
+      "loss": 0.0074,
+      "step": 161430
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224725092596596e-05,
+      "loss": 0.0063,
+      "step": 161440
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224677052858125e-05,
+      "loss": 0.0077,
+      "step": 161450
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224629013119654e-05,
+      "loss": 0.0069,
+      "step": 161460
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922458577735503e-05,
+      "loss": 0.0077,
+      "step": 161470
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224537737616558e-05,
+      "loss": 0.0055,
+      "step": 161480
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224489697878087e-05,
+      "loss": 0.0098,
+      "step": 161490
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224441658139613e-05,
+      "loss": 0.0077,
+      "step": 161500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224393618401143e-05,
+      "loss": 0.0072,
+      "step": 161510
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224345578662672e-05,
+      "loss": 0.0086,
+      "step": 161520
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224297538924198e-05,
+      "loss": 0.0072,
+      "step": 161530
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224249499185727e-05,
+      "loss": 0.0062,
+      "step": 161540
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224201459447256e-05,
+      "loss": 0.0082,
+      "step": 161550
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9224153419708786e-05,
+      "loss": 0.0081,
+      "step": 161560
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922410537997031e-05,
+      "loss": 0.0097,
+      "step": 161570
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922405734023184e-05,
+      "loss": 0.0084,
+      "step": 161580
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922400930049337e-05,
+      "loss": 0.0064,
+      "step": 161590
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.92239612607549e-05,
+      "loss": 0.0062,
+      "step": 161600
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223913221016425e-05,
+      "loss": 0.0081,
+      "step": 161610
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223865181277954e-05,
+      "loss": 0.0069,
+      "step": 161620
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223817141539484e-05,
+      "loss": 0.0075,
+      "step": 161630
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922376910180101e-05,
+      "loss": 0.0091,
+      "step": 161640
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922372106206254e-05,
+      "loss": 0.0097,
+      "step": 161650
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223673022324068e-05,
+      "loss": 0.0066,
+      "step": 161660
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223624982585597e-05,
+      "loss": 0.007,
+      "step": 161670
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223576942847123e-05,
+      "loss": 0.0055,
+      "step": 161680
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223528903108652e-05,
+      "loss": 0.0103,
+      "step": 161690
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223480863370182e-05,
+      "loss": 0.0063,
+      "step": 161700
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223432823631708e-05,
+      "loss": 0.0081,
+      "step": 161710
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223384783893237e-05,
+      "loss": 0.0095,
+      "step": 161720
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223336744154766e-05,
+      "loss": 0.0056,
+      "step": 161730
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223288704416295e-05,
+      "loss": 0.0069,
+      "step": 161740
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922324066467782e-05,
+      "loss": 0.0079,
+      "step": 161750
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922319262493935e-05,
+      "loss": 0.0055,
+      "step": 161760
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922314458520088e-05,
+      "loss": 0.0063,
+      "step": 161770
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922309654546241e-05,
+      "loss": 0.0068,
+      "step": 161780
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223048505723935e-05,
+      "loss": 0.0057,
+      "step": 161790
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9223000465985464e-05,
+      "loss": 0.0087,
+      "step": 161800
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222952426246994e-05,
+      "loss": 0.011,
+      "step": 161810
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922290438650852e-05,
+      "loss": 0.0055,
+      "step": 161820
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922285634677005e-05,
+      "loss": 0.0066,
+      "step": 161830
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222808307031578e-05,
+      "loss": 0.006,
+      "step": 161840
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222760267293107e-05,
+      "loss": 0.0079,
+      "step": 161850
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222712227554633e-05,
+      "loss": 0.0046,
+      "step": 161860
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222664187816162e-05,
+      "loss": 0.0067,
+      "step": 161870
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922261614807769e-05,
+      "loss": 0.0072,
+      "step": 161880
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222568108339218e-05,
+      "loss": 0.0055,
+      "step": 161890
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222520068600747e-05,
+      "loss": 0.0057,
+      "step": 161900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222472028862276e-05,
+      "loss": 0.0064,
+      "step": 161910
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222423989123805e-05,
+      "loss": 0.0096,
+      "step": 161920
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922237594938533e-05,
+      "loss": 0.0057,
+      "step": 161930
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922232790964686e-05,
+      "loss": 0.011,
+      "step": 161940
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922227986990839e-05,
+      "loss": 0.0055,
+      "step": 161950
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222231830169916e-05,
+      "loss": 0.0102,
+      "step": 161960
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222183790431445e-05,
+      "loss": 0.0083,
+      "step": 161970
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222135750692974e-05,
+      "loss": 0.0085,
+      "step": 161980
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9222087710954503e-05,
+      "loss": 0.0093,
+      "step": 161990
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922203967121603e-05,
+      "loss": 0.0073,
+      "step": 162000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.922199163147756e-05,
+      "loss": 0.0056,
+      "step": 162010
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221943591739088e-05,
+      "loss": 0.0085,
+      "step": 162020
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221895552000617e-05,
+      "loss": 0.0048,
+      "step": 162030
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221847512262143e-05,
+      "loss": 0.0056,
+      "step": 162040
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221799472523672e-05,
+      "loss": 0.008,
+      "step": 162050
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.92217514327852e-05,
+      "loss": 0.0079,
+      "step": 162060
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221703393046727e-05,
+      "loss": 0.0092,
+      "step": 162070
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221655353308257e-05,
+      "loss": 0.0061,
+      "step": 162080
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221607313569786e-05,
+      "loss": 0.0063,
+      "step": 162090
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221559273831315e-05,
+      "loss": 0.0061,
+      "step": 162100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.922151123409284e-05,
+      "loss": 0.01,
+      "step": 162110
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.922146319435437e-05,
+      "loss": 0.0101,
+      "step": 162120
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.92214151546159e-05,
+      "loss": 0.0086,
+      "step": 162130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221367114877426e-05,
+      "loss": 0.0071,
+      "step": 162140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221319075138955e-05,
+      "loss": 0.0069,
+      "step": 162150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221271035400484e-05,
+      "loss": 0.0094,
+      "step": 162160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221222995662013e-05,
+      "loss": 0.0047,
+      "step": 162170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.922117495592354e-05,
+      "loss": 0.0064,
+      "step": 162180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.922112691618507e-05,
+      "loss": 0.0082,
+      "step": 162190
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221078876446598e-05,
+      "loss": 0.0102,
+      "step": 162200
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9221030836708127e-05,
+      "loss": 0.0087,
+      "step": 162210
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220982796969653e-05,
+      "loss": 0.0082,
+      "step": 162220
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220934757231182e-05,
+      "loss": 0.0077,
+      "step": 162230
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.922088671749271e-05,
+      "loss": 0.0138,
+      "step": 162240
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220838677754237e-05,
+      "loss": 0.0074,
+      "step": 162250
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220790638015767e-05,
+      "loss": 0.0099,
+      "step": 162260
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220742598277296e-05,
+      "loss": 0.0068,
+      "step": 162270
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220694558538825e-05,
+      "loss": 0.0073,
+      "step": 162280
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.922064651880035e-05,
+      "loss": 0.006,
+      "step": 162290
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.922059847906188e-05,
+      "loss": 0.0115,
+      "step": 162300
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.922055043932341e-05,
+      "loss": 0.0082,
+      "step": 162310
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220502399584935e-05,
+      "loss": 0.0079,
+      "step": 162320
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220454359846465e-05,
+      "loss": 0.0079,
+      "step": 162330
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220406320107997e-05,
+      "loss": 0.01,
+      "step": 162340
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220358280369523e-05,
+      "loss": 0.0081,
+      "step": 162350
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220310240631053e-05,
+      "loss": 0.012,
+      "step": 162360
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220262200892582e-05,
+      "loss": 0.0087,
+      "step": 162370
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220214161154108e-05,
+      "loss": 0.0095,
+      "step": 162380
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220166121415637e-05,
+      "loss": 0.0079,
+      "step": 162390
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220118081677166e-05,
+      "loss": 0.0079,
+      "step": 162400
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9220070041938696e-05,
+      "loss": 0.0082,
+      "step": 162410
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.922002200220022e-05,
+      "loss": 0.007,
+      "step": 162420
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.921997396246175e-05,
+      "loss": 0.0064,
+      "step": 162430
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.921992592272328e-05,
+      "loss": 0.0074,
+      "step": 162440
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.921987788298481e-05,
+      "loss": 0.0076,
+      "step": 162450
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9219829843246335e-05,
+      "loss": 0.0063,
+      "step": 162460
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9219781803507864e-05,
+      "loss": 0.0069,
+      "step": 162470
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9219733763769394e-05,
+      "loss": 0.0072,
+      "step": 162480
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.921968572403092e-05,
+      "loss": 0.006,
+      "step": 162490
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.921963768429245e-05,
+      "loss": 0.009,
+      "step": 162500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9219589644553978e-05,
+      "loss": 0.0074,
+      "step": 162510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9219541604815507e-05,
+      "loss": 0.0049,
+      "step": 162520
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9219493565077033e-05,
+      "loss": 0.0079,
+      "step": 162530
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9219445525338562e-05,
+      "loss": 0.008,
+      "step": 162540
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9219397485600092e-05,
+      "loss": 0.0094,
+      "step": 162550
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9219349445861618e-05,
+      "loss": 0.007,
+      "step": 162560
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9219301406123147e-05,
+      "loss": 0.008,
+      "step": 162570
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9219253366384676e-05,
+      "loss": 0.0087,
+      "step": 162580
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9219205326646205e-05,
+      "loss": 0.0103,
+      "step": 162590
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.921915728690773e-05,
+      "loss": 0.0073,
+      "step": 162600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.921910924716926e-05,
+      "loss": 0.0078,
+      "step": 162610
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.921906120743079e-05,
+      "loss": 0.0063,
+      "step": 162620
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.921901316769232e-05,
+      "loss": 0.0045,
+      "step": 162630
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9218965127953845e-05,
+      "loss": 0.0086,
+      "step": 162640
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9218917088215374e-05,
+      "loss": 0.0061,
+      "step": 162650
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9218869048476904e-05,
+      "loss": 0.0077,
+      "step": 162660
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.921882100873843e-05,
+      "loss": 0.0097,
+      "step": 162670
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.921877296899996e-05,
+      "loss": 0.0084,
+      "step": 162680
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9218724929261488e-05,
+      "loss": 0.0063,
+      "step": 162690
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9218676889523017e-05,
+      "loss": 0.0075,
+      "step": 162700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9218628849784543e-05,
+      "loss": 0.0076,
+      "step": 162710
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9218580810046072e-05,
+      "loss": 0.0075,
+      "step": 162720
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.92185327703076e-05,
+      "loss": 0.0049,
+      "step": 162730
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9218484730569128e-05,
+      "loss": 0.0063,
+      "step": 162740
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9218436690830657e-05,
+      "loss": 0.0084,
+      "step": 162750
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9218388651092186e-05,
+      "loss": 0.0088,
+      "step": 162760
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9218340611353715e-05,
+      "loss": 0.0092,
+      "step": 162770
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921829257161524e-05,
+      "loss": 0.0068,
+      "step": 162780
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921824453187677e-05,
+      "loss": 0.0088,
+      "step": 162790
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.92181964921383e-05,
+      "loss": 0.0065,
+      "step": 162800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921814845239983e-05,
+      "loss": 0.0079,
+      "step": 162810
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9218100412661355e-05,
+      "loss": 0.0076,
+      "step": 162820
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9218052372922884e-05,
+      "loss": 0.0068,
+      "step": 162830
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9218004333184413e-05,
+      "loss": 0.0066,
+      "step": 162840
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921795629344594e-05,
+      "loss": 0.0103,
+      "step": 162850
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921790825370747e-05,
+      "loss": 0.0074,
+      "step": 162860
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9217860213968998e-05,
+      "loss": 0.0047,
+      "step": 162870
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9217812174230527e-05,
+      "loss": 0.0067,
+      "step": 162880
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9217764134492053e-05,
+      "loss": 0.0076,
+      "step": 162890
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9217716094753582e-05,
+      "loss": 0.0073,
+      "step": 162900
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921766805501511e-05,
+      "loss": 0.0065,
+      "step": 162910
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9217620015276637e-05,
+      "loss": 0.0045,
+      "step": 162920
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9217571975538167e-05,
+      "loss": 0.0058,
+      "step": 162930
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9217523935799696e-05,
+      "loss": 0.0075,
+      "step": 162940
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9217475896061225e-05,
+      "loss": 0.0048,
+      "step": 162950
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921742785632275e-05,
+      "loss": 0.0084,
+      "step": 162960
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921737981658428e-05,
+      "loss": 0.0032,
+      "step": 162970
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921733177684581e-05,
+      "loss": 0.0096,
+      "step": 162980
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921728373710734e-05,
+      "loss": 0.007,
+      "step": 162990
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9217235697368865e-05,
+      "loss": 0.0084,
+      "step": 163000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9217187657630394e-05,
+      "loss": 0.0089,
+      "step": 163010
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9217139617891923e-05,
+      "loss": 0.009,
+      "step": 163020
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921709157815345e-05,
+      "loss": 0.0062,
+      "step": 163030
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921704353841498e-05,
+      "loss": 0.0081,
+      "step": 163040
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216995498676508e-05,
+      "loss": 0.0122,
+      "step": 163050
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216947458938037e-05,
+      "loss": 0.0068,
+      "step": 163060
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216899419199563e-05,
+      "loss": 0.009,
+      "step": 163070
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216851379461092e-05,
+      "loss": 0.0065,
+      "step": 163080
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921680333972262e-05,
+      "loss": 0.0062,
+      "step": 163090
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216755299984147e-05,
+      "loss": 0.0097,
+      "step": 163100
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216707260245677e-05,
+      "loss": 0.0084,
+      "step": 163110
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216659220507206e-05,
+      "loss": 0.0079,
+      "step": 163120
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216611180768735e-05,
+      "loss": 0.0075,
+      "step": 163130
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921656314103026e-05,
+      "loss": 0.0066,
+      "step": 163140
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921651510129179e-05,
+      "loss": 0.0074,
+      "step": 163150
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921646706155332e-05,
+      "loss": 0.0082,
+      "step": 163160
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921641902181485e-05,
+      "loss": 0.007,
+      "step": 163170
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216370982076375e-05,
+      "loss": 0.0093,
+      "step": 163180
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216322942337904e-05,
+      "loss": 0.0077,
+      "step": 163190
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921627970657328e-05,
+      "loss": 0.0092,
+      "step": 163200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216231666834808e-05,
+      "loss": 0.0079,
+      "step": 163210
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216183627096337e-05,
+      "loss": 0.0064,
+      "step": 163220
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216135587357863e-05,
+      "loss": 0.0075,
+      "step": 163230
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9216087547619392e-05,
+      "loss": 0.0101,
+      "step": 163240
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921603950788092e-05,
+      "loss": 0.0091,
+      "step": 163250
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921599146814245e-05,
+      "loss": 0.0079,
+      "step": 163260
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9215943428403977e-05,
+      "loss": 0.0056,
+      "step": 163270
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9215895388665506e-05,
+      "loss": 0.0081,
+      "step": 163280
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9215847348927035e-05,
+      "loss": 0.0059,
+      "step": 163290
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9215799309188565e-05,
+      "loss": 0.0077,
+      "step": 163300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921575126945009e-05,
+      "loss": 0.0096,
+      "step": 163310
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921570322971162e-05,
+      "loss": 0.0093,
+      "step": 163320
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921565518997315e-05,
+      "loss": 0.0079,
+      "step": 163330
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9215607150234675e-05,
+      "loss": 0.0075,
+      "step": 163340
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9215559110496204e-05,
+      "loss": 0.0076,
+      "step": 163350
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9215511070757733e-05,
+      "loss": 0.0065,
+      "step": 163360
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9215463031019263e-05,
+      "loss": 0.0073,
+      "step": 163370
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.921541499128079e-05,
+      "loss": 0.0063,
+      "step": 163380
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9215366951542318e-05,
+      "loss": 0.0075,
+      "step": 163390
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9215318911803847e-05,
+      "loss": 0.0066,
+      "step": 163400
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9215270872065373e-05,
+      "loss": 0.007,
+      "step": 163410
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9215222832326902e-05,
+      "loss": 0.0073,
+      "step": 163420
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921517479258843e-05,
+      "loss": 0.0048,
+      "step": 163430
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921512675284996e-05,
+      "loss": 0.0061,
+      "step": 163440
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9215078713111487e-05,
+      "loss": 0.0051,
+      "step": 163450
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9215030673373016e-05,
+      "loss": 0.0063,
+      "step": 163460
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214982633634545e-05,
+      "loss": 0.0113,
+      "step": 163470
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214934593896075e-05,
+      "loss": 0.0065,
+      "step": 163480
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.92148865541576e-05,
+      "loss": 0.009,
+      "step": 163490
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921483851441913e-05,
+      "loss": 0.0059,
+      "step": 163500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921479047468066e-05,
+      "loss": 0.006,
+      "step": 163510
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214742434942185e-05,
+      "loss": 0.0077,
+      "step": 163520
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214694395203714e-05,
+      "loss": 0.0076,
+      "step": 163530
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214646355465243e-05,
+      "loss": 0.0101,
+      "step": 163540
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214598315726773e-05,
+      "loss": 0.008,
+      "step": 163550
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.92145502759883e-05,
+      "loss": 0.009,
+      "step": 163560
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214502236249828e-05,
+      "loss": 0.0108,
+      "step": 163570
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214454196511357e-05,
+      "loss": 0.0084,
+      "step": 163580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214406156772883e-05,
+      "loss": 0.0062,
+      "step": 163590
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214358117034412e-05,
+      "loss": 0.0086,
+      "step": 163600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921431007729594e-05,
+      "loss": 0.0075,
+      "step": 163610
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921426203755747e-05,
+      "loss": 0.0071,
+      "step": 163620
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214213997818997e-05,
+      "loss": 0.0062,
+      "step": 163630
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214165958080526e-05,
+      "loss": 0.0047,
+      "step": 163640
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214117918342055e-05,
+      "loss": 0.0097,
+      "step": 163650
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9214069878603584e-05,
+      "loss": 0.0072,
+      "step": 163660
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921402183886511e-05,
+      "loss": 0.0062,
+      "step": 163670
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921397379912664e-05,
+      "loss": 0.0077,
+      "step": 163680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921392575938817e-05,
+      "loss": 0.0062,
+      "step": 163690
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213877719649695e-05,
+      "loss": 0.0089,
+      "step": 163700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213829679911224e-05,
+      "loss": 0.0081,
+      "step": 163710
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213781640172753e-05,
+      "loss": 0.0064,
+      "step": 163720
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213733600434283e-05,
+      "loss": 0.0092,
+      "step": 163730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921368556069581e-05,
+      "loss": 0.0077,
+      "step": 163740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213637520957338e-05,
+      "loss": 0.0057,
+      "step": 163750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213589481218867e-05,
+      "loss": 0.0076,
+      "step": 163760
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213541441480393e-05,
+      "loss": 0.0052,
+      "step": 163770
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213493401741922e-05,
+      "loss": 0.0072,
+      "step": 163780
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921344536200345e-05,
+      "loss": 0.0058,
+      "step": 163790
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921339732226498e-05,
+      "loss": 0.0087,
+      "step": 163800
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213349282526507e-05,
+      "loss": 0.0076,
+      "step": 163810
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213301242788036e-05,
+      "loss": 0.0097,
+      "step": 163820
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213253203049565e-05,
+      "loss": 0.0067,
+      "step": 163830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213205163311094e-05,
+      "loss": 0.0088,
+      "step": 163840
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921315712357262e-05,
+      "loss": 0.0077,
+      "step": 163850
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921310908383415e-05,
+      "loss": 0.0091,
+      "step": 163860
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921306104409568e-05,
+      "loss": 0.0073,
+      "step": 163870
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9213013004357205e-05,
+      "loss": 0.0075,
+      "step": 163880
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212964964618734e-05,
+      "loss": 0.0094,
+      "step": 163890
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212916924880263e-05,
+      "loss": 0.008,
+      "step": 163900
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212868885141792e-05,
+      "loss": 0.0094,
+      "step": 163910
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921282084540332e-05,
+      "loss": 0.0099,
+      "step": 163920
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212772805664848e-05,
+      "loss": 0.007,
+      "step": 163930
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212724765926377e-05,
+      "loss": 0.0069,
+      "step": 163940
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212676726187903e-05,
+      "loss": 0.0068,
+      "step": 163950
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212628686449432e-05,
+      "loss": 0.0073,
+      "step": 163960
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921258064671096e-05,
+      "loss": 0.0064,
+      "step": 163970
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921253260697249e-05,
+      "loss": 0.0048,
+      "step": 163980
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212484567234016e-05,
+      "loss": 0.0059,
+      "step": 163990
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212436527495546e-05,
+      "loss": 0.0092,
+      "step": 164000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212388487757075e-05,
+      "loss": 0.0087,
+      "step": 164010
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212340448018604e-05,
+      "loss": 0.0061,
+      "step": 164020
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921229240828013e-05,
+      "loss": 0.0065,
+      "step": 164030
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921224436854166e-05,
+      "loss": 0.0072,
+      "step": 164040
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921219632880319e-05,
+      "loss": 0.008,
+      "step": 164050
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212148289064715e-05,
+      "loss": 0.0072,
+      "step": 164060
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212100249326244e-05,
+      "loss": 0.0062,
+      "step": 164070
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212052209587773e-05,
+      "loss": 0.0069,
+      "step": 164080
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9212004169849302e-05,
+      "loss": 0.0073,
+      "step": 164090
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9211956130110828e-05,
+      "loss": 0.0065,
+      "step": 164100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211908090372358e-05,
+      "loss": 0.0104,
+      "step": 164110
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211860050633887e-05,
+      "loss": 0.0071,
+      "step": 164120
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211812010895413e-05,
+      "loss": 0.0055,
+      "step": 164130
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211763971156942e-05,
+      "loss": 0.0063,
+      "step": 164140
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.921171593141847e-05,
+      "loss": 0.0086,
+      "step": 164150
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.921166789168e-05,
+      "loss": 0.0073,
+      "step": 164160
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211619851941526e-05,
+      "loss": 0.0083,
+      "step": 164170
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211571812203056e-05,
+      "loss": 0.0075,
+      "step": 164180
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211523772464585e-05,
+      "loss": 0.0074,
+      "step": 164190
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211475732726114e-05,
+      "loss": 0.0069,
+      "step": 164200
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.921142769298764e-05,
+      "loss": 0.0084,
+      "step": 164210
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.921137965324917e-05,
+      "loss": 0.0083,
+      "step": 164220
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.92113316135107e-05,
+      "loss": 0.0061,
+      "step": 164230
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211283573772224e-05,
+      "loss": 0.0093,
+      "step": 164240
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211235534033754e-05,
+      "loss": 0.0053,
+      "step": 164250
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211187494295283e-05,
+      "loss": 0.0087,
+      "step": 164260
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211139454556812e-05,
+      "loss": 0.0053,
+      "step": 164270
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211091414818338e-05,
+      "loss": 0.009,
+      "step": 164280
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9211043375079867e-05,
+      "loss": 0.0066,
+      "step": 164290
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210995335341397e-05,
+      "loss": 0.0059,
+      "step": 164300
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210947295602923e-05,
+      "loss": 0.0057,
+      "step": 164310
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210899255864452e-05,
+      "loss": 0.0041,
+      "step": 164320
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.921085121612598e-05,
+      "loss": 0.0092,
+      "step": 164330
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.921080317638751e-05,
+      "loss": 0.0092,
+      "step": 164340
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210755136649036e-05,
+      "loss": 0.009,
+      "step": 164350
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210707096910566e-05,
+      "loss": 0.0098,
+      "step": 164360
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210659057172095e-05,
+      "loss": 0.0081,
+      "step": 164370
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210611017433624e-05,
+      "loss": 0.006,
+      "step": 164380
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.921056297769515e-05,
+      "loss": 0.0084,
+      "step": 164390
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.921051493795668e-05,
+      "loss": 0.007,
+      "step": 164400
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.921046689821821e-05,
+      "loss": 0.0047,
+      "step": 164410
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210418858479734e-05,
+      "loss": 0.005,
+      "step": 164420
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210370818741264e-05,
+      "loss": 0.0079,
+      "step": 164430
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210322779002793e-05,
+      "loss": 0.0052,
+      "step": 164440
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210274739264322e-05,
+      "loss": 0.0053,
+      "step": 164450
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210226699525848e-05,
+      "loss": 0.0059,
+      "step": 164460
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210178659787377e-05,
+      "loss": 0.0122,
+      "step": 164470
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210130620048907e-05,
+      "loss": 0.0116,
+      "step": 164480
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210082580310433e-05,
+      "loss": 0.0064,
+      "step": 164490
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9210034540571962e-05,
+      "loss": 0.0079,
+      "step": 164500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.920998650083349e-05,
+      "loss": 0.007,
+      "step": 164510
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.920993846109502e-05,
+      "loss": 0.004,
+      "step": 164520
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209890421356546e-05,
+      "loss": 0.0076,
+      "step": 164530
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209842381618075e-05,
+      "loss": 0.0079,
+      "step": 164540
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209794341879605e-05,
+      "loss": 0.0069,
+      "step": 164550
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209746302141134e-05,
+      "loss": 0.0072,
+      "step": 164560
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.920969826240266e-05,
+      "loss": 0.0076,
+      "step": 164570
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.920965022266419e-05,
+      "loss": 0.009,
+      "step": 164580
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.920960218292572e-05,
+      "loss": 0.0092,
+      "step": 164590
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209554143187244e-05,
+      "loss": 0.0085,
+      "step": 164600
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209506103448774e-05,
+      "loss": 0.0091,
+      "step": 164610
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209458063710303e-05,
+      "loss": 0.0067,
+      "step": 164620
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209410023971832e-05,
+      "loss": 0.0077,
+      "step": 164630
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209361984233358e-05,
+      "loss": 0.0075,
+      "step": 164640
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209313944494887e-05,
+      "loss": 0.0073,
+      "step": 164650
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209265904756417e-05,
+      "loss": 0.0093,
+      "step": 164660
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209217865017942e-05,
+      "loss": 0.0062,
+      "step": 164670
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.920916982527947e-05,
+      "loss": 0.0073,
+      "step": 164680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209121785541e-05,
+      "loss": 0.0044,
+      "step": 164690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.920907374580253e-05,
+      "loss": 0.0054,
+      "step": 164700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9209025706064056e-05,
+      "loss": 0.0093,
+      "step": 164710
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9208977666325585e-05,
+      "loss": 0.0071,
+      "step": 164720
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9208929626587115e-05,
+      "loss": 0.0072,
+      "step": 164730
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9208881586848644e-05,
+      "loss": 0.0058,
+      "step": 164740
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.920883354711017e-05,
+      "loss": 0.0071,
+      "step": 164750
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.92087855073717e-05,
+      "loss": 0.0059,
+      "step": 164760
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.920873746763323e-05,
+      "loss": 0.0079,
+      "step": 164770
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9208689427894754e-05,
+      "loss": 0.0079,
+      "step": 164780
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9208641388156283e-05,
+      "loss": 0.012,
+      "step": 164790
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9208593348417813e-05,
+      "loss": 0.0146,
+      "step": 164800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9208545308679342e-05,
+      "loss": 0.0078,
+      "step": 164810
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9208497268940868e-05,
+      "loss": 0.0082,
+      "step": 164820
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9208449229202397e-05,
+      "loss": 0.0046,
+      "step": 164830
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9208401189463926e-05,
+      "loss": 0.0079,
+      "step": 164840
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9208353149725452e-05,
+      "loss": 0.0062,
+      "step": 164850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920830510998698e-05,
+      "loss": 0.0087,
+      "step": 164860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920825707024851e-05,
+      "loss": 0.0101,
+      "step": 164870
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920820903051004e-05,
+      "loss": 0.0046,
+      "step": 164880
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9208160990771566e-05,
+      "loss": 0.0046,
+      "step": 164890
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9208112951033095e-05,
+      "loss": 0.0088,
+      "step": 164900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9208064911294625e-05,
+      "loss": 0.0064,
+      "step": 164910
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920801687155615e-05,
+      "loss": 0.0079,
+      "step": 164920
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920796883181768e-05,
+      "loss": 0.0083,
+      "step": 164930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920792079207921e-05,
+      "loss": 0.0069,
+      "step": 164940
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207872752340738e-05,
+      "loss": 0.0089,
+      "step": 164950
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207824712602264e-05,
+      "loss": 0.0067,
+      "step": 164960
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207776672863793e-05,
+      "loss": 0.0126,
+      "step": 164970
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207728633125323e-05,
+      "loss": 0.0086,
+      "step": 164980
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207680593386852e-05,
+      "loss": 0.0109,
+      "step": 164990
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207632553648378e-05,
+      "loss": 0.0074,
+      "step": 165000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207584513909907e-05,
+      "loss": 0.0073,
+      "step": 165010
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207536474171436e-05,
+      "loss": 0.0077,
+      "step": 165020
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207488434432962e-05,
+      "loss": 0.0053,
+      "step": 165030
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920744039469449e-05,
+      "loss": 0.0088,
+      "step": 165040
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920739235495602e-05,
+      "loss": 0.007,
+      "step": 165050
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920734431521755e-05,
+      "loss": 0.0089,
+      "step": 165060
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207296275479076e-05,
+      "loss": 0.0073,
+      "step": 165070
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207248235740605e-05,
+      "loss": 0.0059,
+      "step": 165080
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207200196002134e-05,
+      "loss": 0.0057,
+      "step": 165090
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920715215626366e-05,
+      "loss": 0.0045,
+      "step": 165100
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920710411652519e-05,
+      "loss": 0.005,
+      "step": 165110
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920705607678672e-05,
+      "loss": 0.0094,
+      "step": 165120
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9207008037048248e-05,
+      "loss": 0.0063,
+      "step": 165130
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206959997309774e-05,
+      "loss": 0.0153,
+      "step": 165140
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206911957571303e-05,
+      "loss": 0.0113,
+      "step": 165150
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206863917832833e-05,
+      "loss": 0.0072,
+      "step": 165160
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206815878094362e-05,
+      "loss": 0.0054,
+      "step": 165170
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206767838355888e-05,
+      "loss": 0.0045,
+      "step": 165180
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206719798617417e-05,
+      "loss": 0.0097,
+      "step": 165190
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206671758878946e-05,
+      "loss": 0.007,
+      "step": 165200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206623719140472e-05,
+      "loss": 0.0056,
+      "step": 165210
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206575679402e-05,
+      "loss": 0.0063,
+      "step": 165220
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920652763966353e-05,
+      "loss": 0.009,
+      "step": 165230
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920647959992506e-05,
+      "loss": 0.006,
+      "step": 165240
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206431560186586e-05,
+      "loss": 0.0053,
+      "step": 165250
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206383520448115e-05,
+      "loss": 0.0068,
+      "step": 165260
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206335480709644e-05,
+      "loss": 0.0057,
+      "step": 165270
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920628744097117e-05,
+      "loss": 0.0091,
+      "step": 165280
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.92062394012327e-05,
+      "loss": 0.0059,
+      "step": 165290
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920619136149423e-05,
+      "loss": 0.0081,
+      "step": 165300
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206143321755758e-05,
+      "loss": 0.0107,
+      "step": 165310
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206095282017284e-05,
+      "loss": 0.0069,
+      "step": 165320
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9206047242278813e-05,
+      "loss": 0.0052,
+      "step": 165330
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9205999202540343e-05,
+      "loss": 0.0135,
+      "step": 165340
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9205951162801872e-05,
+      "loss": 0.007,
+      "step": 165350
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9205903123063398e-05,
+      "loss": 0.0085,
+      "step": 165360
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9205855083324927e-05,
+      "loss": 0.0072,
+      "step": 165370
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9205807043586456e-05,
+      "loss": 0.0077,
+      "step": 165380
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9205759003847982e-05,
+      "loss": 0.0076,
+      "step": 165390
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920571096410951e-05,
+      "loss": 0.0073,
+      "step": 165400
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9205662924371044e-05,
+      "loss": 0.0082,
+      "step": 165410
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920561488463257e-05,
+      "loss": 0.0086,
+      "step": 165420
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.92055668448941e-05,
+      "loss": 0.0096,
+      "step": 165430
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.920551880515563e-05,
+      "loss": 0.0065,
+      "step": 165440
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9205470765417154e-05,
+      "loss": 0.0046,
+      "step": 165450
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9205422725678684e-05,
+      "loss": 0.0075,
+      "step": 165460
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9205374685940213e-05,
+      "loss": 0.0116,
+      "step": 165470
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9205326646201742e-05,
+      "loss": 0.0082,
+      "step": 165480
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9205278606463268e-05,
+      "loss": 0.0049,
+      "step": 165490
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9205230566724797e-05,
+      "loss": 0.0084,
+      "step": 165500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9205182526986327e-05,
+      "loss": 0.0077,
+      "step": 165510
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9205134487247852e-05,
+      "loss": 0.0068,
+      "step": 165520
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920508644750938e-05,
+      "loss": 0.0104,
+      "step": 165530
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920503840777091e-05,
+      "loss": 0.0084,
+      "step": 165540
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920499036803244e-05,
+      "loss": 0.0067,
+      "step": 165550
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204942328293966e-05,
+      "loss": 0.0066,
+      "step": 165560
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204894288555495e-05,
+      "loss": 0.0137,
+      "step": 165570
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204846248817025e-05,
+      "loss": 0.0093,
+      "step": 165580
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204798209078554e-05,
+      "loss": 0.0081,
+      "step": 165590
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920475016934008e-05,
+      "loss": 0.0082,
+      "step": 165600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920470212960161e-05,
+      "loss": 0.0088,
+      "step": 165610
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920465408986314e-05,
+      "loss": 0.0065,
+      "step": 165620
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204606050124664e-05,
+      "loss": 0.0081,
+      "step": 165630
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204558010386193e-05,
+      "loss": 0.0079,
+      "step": 165640
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204509970647723e-05,
+      "loss": 0.0068,
+      "step": 165650
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204461930909252e-05,
+      "loss": 0.0069,
+      "step": 165660
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204413891170778e-05,
+      "loss": 0.0126,
+      "step": 165670
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204365851432307e-05,
+      "loss": 0.0056,
+      "step": 165680
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204317811693836e-05,
+      "loss": 0.0087,
+      "step": 165690
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204269771955362e-05,
+      "loss": 0.0075,
+      "step": 165700
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920422173221689e-05,
+      "loss": 0.0073,
+      "step": 165710
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920417369247842e-05,
+      "loss": 0.0069,
+      "step": 165720
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920412565273995e-05,
+      "loss": 0.0113,
+      "step": 165730
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204077613001476e-05,
+      "loss": 0.0055,
+      "step": 165740
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9204029573263005e-05,
+      "loss": 0.0049,
+      "step": 165750
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203981533524535e-05,
+      "loss": 0.0058,
+      "step": 165760
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203933493786064e-05,
+      "loss": 0.0068,
+      "step": 165770
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920388545404759e-05,
+      "loss": 0.0063,
+      "step": 165780
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920383741430912e-05,
+      "loss": 0.0086,
+      "step": 165790
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203789374570648e-05,
+      "loss": 0.006,
+      "step": 165800
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203741334832174e-05,
+      "loss": 0.0078,
+      "step": 165810
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203693295093703e-05,
+      "loss": 0.0073,
+      "step": 165820
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203645255355233e-05,
+      "loss": 0.0075,
+      "step": 165830
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203597215616762e-05,
+      "loss": 0.0052,
+      "step": 165840
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203549175878288e-05,
+      "loss": 0.0114,
+      "step": 165850
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203501136139817e-05,
+      "loss": 0.0071,
+      "step": 165860
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203453096401346e-05,
+      "loss": 0.007,
+      "step": 165870
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203405056662872e-05,
+      "loss": 0.0082,
+      "step": 165880
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.92033570169244e-05,
+      "loss": 0.0081,
+      "step": 165890
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920330897718593e-05,
+      "loss": 0.0096,
+      "step": 165900
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920326093744746e-05,
+      "loss": 0.0098,
+      "step": 165910
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203212897708986e-05,
+      "loss": 0.0075,
+      "step": 165920
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203164857970515e-05,
+      "loss": 0.0081,
+      "step": 165930
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203116818232044e-05,
+      "loss": 0.0078,
+      "step": 165940
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9203068778493574e-05,
+      "loss": 0.0078,
+      "step": 165950
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.92030207387551e-05,
+      "loss": 0.0133,
+      "step": 165960
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920297269901663e-05,
+      "loss": 0.008,
+      "step": 165970
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202924659278158e-05,
+      "loss": 0.0073,
+      "step": 165980
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202876619539684e-05,
+      "loss": 0.0062,
+      "step": 165990
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202828579801213e-05,
+      "loss": 0.0076,
+      "step": 166000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202780540062743e-05,
+      "loss": 0.0075,
+      "step": 166010
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202732500324272e-05,
+      "loss": 0.0046,
+      "step": 166020
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202684460585798e-05,
+      "loss": 0.0069,
+      "step": 166030
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202636420847327e-05,
+      "loss": 0.0065,
+      "step": 166040
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202588381108856e-05,
+      "loss": 0.0082,
+      "step": 166050
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202540341370382e-05,
+      "loss": 0.0082,
+      "step": 166060
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920249230163191e-05,
+      "loss": 0.0058,
+      "step": 166070
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920244426189344e-05,
+      "loss": 0.0059,
+      "step": 166080
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920239622215497e-05,
+      "loss": 0.0085,
+      "step": 166090
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202348182416496e-05,
+      "loss": 0.0091,
+      "step": 166100
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202300142678025e-05,
+      "loss": 0.0066,
+      "step": 166110
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202252102939554e-05,
+      "loss": 0.0081,
+      "step": 166120
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202204063201084e-05,
+      "loss": 0.0115,
+      "step": 166130
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920215602346261e-05,
+      "loss": 0.0061,
+      "step": 166140
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.920210798372414e-05,
+      "loss": 0.0062,
+      "step": 166150
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202059943985668e-05,
+      "loss": 0.0096,
+      "step": 166160
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9202011904247194e-05,
+      "loss": 0.0068,
+      "step": 166170
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9201963864508723e-05,
+      "loss": 0.0049,
+      "step": 166180
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201915824770253e-05,
+      "loss": 0.0058,
+      "step": 166190
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201867785031782e-05,
+      "loss": 0.0084,
+      "step": 166200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201819745293308e-05,
+      "loss": 0.0074,
+      "step": 166210
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201771705554837e-05,
+      "loss": 0.0094,
+      "step": 166220
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201723665816366e-05,
+      "loss": 0.0065,
+      "step": 166230
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201675626077892e-05,
+      "loss": 0.0073,
+      "step": 166240
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.920162758633942e-05,
+      "loss": 0.0066,
+      "step": 166250
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.920157954660095e-05,
+      "loss": 0.0053,
+      "step": 166260
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.920153150686248e-05,
+      "loss": 0.008,
+      "step": 166270
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201483467124006e-05,
+      "loss": 0.0076,
+      "step": 166280
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201435427385535e-05,
+      "loss": 0.0056,
+      "step": 166290
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201387387647064e-05,
+      "loss": 0.006,
+      "step": 166300
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.920133934790859e-05,
+      "loss": 0.0074,
+      "step": 166310
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.920129130817012e-05,
+      "loss": 0.0096,
+      "step": 166320
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.920124326843165e-05,
+      "loss": 0.0129,
+      "step": 166330
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201195228693178e-05,
+      "loss": 0.0086,
+      "step": 166340
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201147188954704e-05,
+      "loss": 0.0086,
+      "step": 166350
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201099149216233e-05,
+      "loss": 0.0084,
+      "step": 166360
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201051109477762e-05,
+      "loss": 0.0103,
+      "step": 166370
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9201003069739292e-05,
+      "loss": 0.0078,
+      "step": 166380
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200955030000818e-05,
+      "loss": 0.0084,
+      "step": 166390
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200906990262347e-05,
+      "loss": 0.0075,
+      "step": 166400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200858950523876e-05,
+      "loss": 0.0085,
+      "step": 166410
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200810910785402e-05,
+      "loss": 0.0067,
+      "step": 166420
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.920076287104693e-05,
+      "loss": 0.0111,
+      "step": 166430
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.920071483130846e-05,
+      "loss": 0.0075,
+      "step": 166440
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.920066679156999e-05,
+      "loss": 0.0063,
+      "step": 166450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200618751831516e-05,
+      "loss": 0.0101,
+      "step": 166460
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200570712093045e-05,
+      "loss": 0.0088,
+      "step": 166470
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200522672354574e-05,
+      "loss": 0.0075,
+      "step": 166480
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.92004746326161e-05,
+      "loss": 0.0074,
+      "step": 166490
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.920042659287763e-05,
+      "loss": 0.0081,
+      "step": 166500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.920037855313916e-05,
+      "loss": 0.0095,
+      "step": 166510
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200330513400688e-05,
+      "loss": 0.0139,
+      "step": 166520
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200282473662214e-05,
+      "loss": 0.007,
+      "step": 166530
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200234433923743e-05,
+      "loss": 0.0088,
+      "step": 166540
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200186394185272e-05,
+      "loss": 0.0076,
+      "step": 166550
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.92001383544468e-05,
+      "loss": 0.0096,
+      "step": 166560
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200090314708327e-05,
+      "loss": 0.0082,
+      "step": 166570
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9200042274969857e-05,
+      "loss": 0.0078,
+      "step": 166580
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199994235231386e-05,
+      "loss": 0.0082,
+      "step": 166590
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199946195492912e-05,
+      "loss": 0.0069,
+      "step": 166600
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.919989815575444e-05,
+      "loss": 0.0068,
+      "step": 166610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.919985011601597e-05,
+      "loss": 0.0078,
+      "step": 166620
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.91998020762775e-05,
+      "loss": 0.0053,
+      "step": 166630
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199754036539026e-05,
+      "loss": 0.0088,
+      "step": 166640
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199705996800555e-05,
+      "loss": 0.0076,
+      "step": 166650
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199657957062084e-05,
+      "loss": 0.007,
+      "step": 166660
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.919960991732361e-05,
+      "loss": 0.0064,
+      "step": 166670
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.919956187758514e-05,
+      "loss": 0.0053,
+      "step": 166680
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.919951383784667e-05,
+      "loss": 0.0051,
+      "step": 166690
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199465798108198e-05,
+      "loss": 0.0082,
+      "step": 166700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199417758369724e-05,
+      "loss": 0.0072,
+      "step": 166710
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199369718631253e-05,
+      "loss": 0.0043,
+      "step": 166720
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199321678892782e-05,
+      "loss": 0.0065,
+      "step": 166730
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.919927363915431e-05,
+      "loss": 0.007,
+      "step": 166740
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199225599415837e-05,
+      "loss": 0.0067,
+      "step": 166750
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199177559677367e-05,
+      "loss": 0.0045,
+      "step": 166760
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199129519938896e-05,
+      "loss": 0.0094,
+      "step": 166770
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9199081480200422e-05,
+      "loss": 0.0077,
+      "step": 166780
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.919903344046195e-05,
+      "loss": 0.0074,
+      "step": 166790
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.919898540072348e-05,
+      "loss": 0.0084,
+      "step": 166800
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.919893736098501e-05,
+      "loss": 0.0076,
+      "step": 166810
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9198889321246536e-05,
+      "loss": 0.0083,
+      "step": 166820
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9198841281508065e-05,
+      "loss": 0.007,
+      "step": 166830
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9198793241769594e-05,
+      "loss": 0.0102,
+      "step": 166840
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.919874520203112e-05,
+      "loss": 0.0071,
+      "step": 166850
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.919869716229265e-05,
+      "loss": 0.0066,
+      "step": 166860
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.919864912255418e-05,
+      "loss": 0.0089,
+      "step": 166870
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9198601082815708e-05,
+      "loss": 0.0089,
+      "step": 166880
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9198553043077234e-05,
+      "loss": 0.0098,
+      "step": 166890
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9198505003338763e-05,
+      "loss": 0.0073,
+      "step": 166900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9198456963600292e-05,
+      "loss": 0.0089,
+      "step": 166910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919840892386182e-05,
+      "loss": 0.0075,
+      "step": 166920
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9198360884123347e-05,
+      "loss": 0.0081,
+      "step": 166930
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9198312844384877e-05,
+      "loss": 0.0088,
+      "step": 166940
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9198264804646406e-05,
+      "loss": 0.0088,
+      "step": 166950
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9198216764907932e-05,
+      "loss": 0.0064,
+      "step": 166960
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919816872516946e-05,
+      "loss": 0.0062,
+      "step": 166970
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919812068543099e-05,
+      "loss": 0.006,
+      "step": 166980
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919807264569252e-05,
+      "loss": 0.0128,
+      "step": 166990
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9198024605954045e-05,
+      "loss": 0.0055,
+      "step": 167000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9197976566215575e-05,
+      "loss": 0.008,
+      "step": 167010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9197928526477104e-05,
+      "loss": 0.0103,
+      "step": 167020
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919788048673863e-05,
+      "loss": 0.0096,
+      "step": 167030
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919783244700016e-05,
+      "loss": 0.008,
+      "step": 167040
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919778440726169e-05,
+      "loss": 0.0082,
+      "step": 167050
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9197736367523218e-05,
+      "loss": 0.0084,
+      "step": 167060
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9197688327784744e-05,
+      "loss": 0.0078,
+      "step": 167070
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9197640288046273e-05,
+      "loss": 0.0071,
+      "step": 167080
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9197592248307802e-05,
+      "loss": 0.0047,
+      "step": 167090
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919754420856933e-05,
+      "loss": 0.0051,
+      "step": 167100
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9197496168830857e-05,
+      "loss": 0.0129,
+      "step": 167110
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9197448129092387e-05,
+      "loss": 0.0075,
+      "step": 167120
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9197400089353916e-05,
+      "loss": 0.006,
+      "step": 167130
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919735204961544e-05,
+      "loss": 0.0054,
+      "step": 167140
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919730881385082e-05,
+      "loss": 0.0072,
+      "step": 167150
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919726077411235e-05,
+      "loss": 0.0057,
+      "step": 167160
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9197212734373875e-05,
+      "loss": 0.0055,
+      "step": 167170
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9197164694635404e-05,
+      "loss": 0.0068,
+      "step": 167180
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9197116654896933e-05,
+      "loss": 0.0063,
+      "step": 167190
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919706861515846e-05,
+      "loss": 0.011,
+      "step": 167200
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919702057541999e-05,
+      "loss": 0.0064,
+      "step": 167210
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196972535681518e-05,
+      "loss": 0.0077,
+      "step": 167220
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196924495943047e-05,
+      "loss": 0.008,
+      "step": 167230
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196876456204573e-05,
+      "loss": 0.0089,
+      "step": 167240
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196828416466102e-05,
+      "loss": 0.0082,
+      "step": 167250
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919678037672763e-05,
+      "loss": 0.0061,
+      "step": 167260
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196732336989157e-05,
+      "loss": 0.0078,
+      "step": 167270
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196684297250687e-05,
+      "loss": 0.0088,
+      "step": 167280
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196636257512216e-05,
+      "loss": 0.0079,
+      "step": 167290
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196588217773745e-05,
+      "loss": 0.0068,
+      "step": 167300
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919654017803527e-05,
+      "loss": 0.0089,
+      "step": 167310
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.91964921382968e-05,
+      "loss": 0.0061,
+      "step": 167320
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919644409855833e-05,
+      "loss": 0.008,
+      "step": 167330
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919639605881986e-05,
+      "loss": 0.0089,
+      "step": 167340
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196348019081385e-05,
+      "loss": 0.0089,
+      "step": 167350
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196299979342914e-05,
+      "loss": 0.0083,
+      "step": 167360
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196251939604443e-05,
+      "loss": 0.0063,
+      "step": 167370
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919620389986597e-05,
+      "loss": 0.0074,
+      "step": 167380
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.91961558601275e-05,
+      "loss": 0.0098,
+      "step": 167390
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196107820389028e-05,
+      "loss": 0.0097,
+      "step": 167400
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196059780650557e-05,
+      "loss": 0.0108,
+      "step": 167410
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9196011740912083e-05,
+      "loss": 0.0071,
+      "step": 167420
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9195963701173612e-05,
+      "loss": 0.0096,
+      "step": 167430
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919591566143514e-05,
+      "loss": 0.0068,
+      "step": 167440
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9195867621696667e-05,
+      "loss": 0.0071,
+      "step": 167450
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9195819581958197e-05,
+      "loss": 0.0068,
+      "step": 167460
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9195771542219726e-05,
+      "loss": 0.0083,
+      "step": 167470
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9195723502481255e-05,
+      "loss": 0.0062,
+      "step": 167480
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919567546274278e-05,
+      "loss": 0.0093,
+      "step": 167490
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919562742300431e-05,
+      "loss": 0.008,
+      "step": 167500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919557938326584e-05,
+      "loss": 0.0123,
+      "step": 167510
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919553134352737e-05,
+      "loss": 0.0083,
+      "step": 167520
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9195483303788895e-05,
+      "loss": 0.0071,
+      "step": 167530
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9195435264050424e-05,
+      "loss": 0.0074,
+      "step": 167540
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9195387224311953e-05,
+      "loss": 0.0073,
+      "step": 167550
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.919533918457348e-05,
+      "loss": 0.0069,
+      "step": 167560
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919529114483501e-05,
+      "loss": 0.0068,
+      "step": 167570
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9195243105096538e-05,
+      "loss": 0.006,
+      "step": 167580
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9195195065358067e-05,
+      "loss": 0.0069,
+      "step": 167590
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9195147025619593e-05,
+      "loss": 0.0091,
+      "step": 167600
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9195098985881122e-05,
+      "loss": 0.0079,
+      "step": 167610
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919505094614265e-05,
+      "loss": 0.0067,
+      "step": 167620
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9195002906404177e-05,
+      "loss": 0.006,
+      "step": 167630
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194954866665706e-05,
+      "loss": 0.0104,
+      "step": 167640
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194906826927236e-05,
+      "loss": 0.0111,
+      "step": 167650
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194858787188765e-05,
+      "loss": 0.0063,
+      "step": 167660
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919481074745029e-05,
+      "loss": 0.0092,
+      "step": 167670
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919476270771182e-05,
+      "loss": 0.0085,
+      "step": 167680
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919471466797335e-05,
+      "loss": 0.0113,
+      "step": 167690
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919466662823488e-05,
+      "loss": 0.0056,
+      "step": 167700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194618588496405e-05,
+      "loss": 0.0068,
+      "step": 167710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194570548757934e-05,
+      "loss": 0.0073,
+      "step": 167720
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194522509019463e-05,
+      "loss": 0.0088,
+      "step": 167730
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919447446928099e-05,
+      "loss": 0.0057,
+      "step": 167740
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194426429542518e-05,
+      "loss": 0.0072,
+      "step": 167750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194378389804048e-05,
+      "loss": 0.0067,
+      "step": 167760
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194330350065577e-05,
+      "loss": 0.0055,
+      "step": 167770
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194282310327103e-05,
+      "loss": 0.011,
+      "step": 167780
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194234270588632e-05,
+      "loss": 0.0081,
+      "step": 167790
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919418623085016e-05,
+      "loss": 0.0099,
+      "step": 167800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194138191111687e-05,
+      "loss": 0.0098,
+      "step": 167810
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194090151373216e-05,
+      "loss": 0.007,
+      "step": 167820
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9194042111634746e-05,
+      "loss": 0.0064,
+      "step": 167830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193994071896275e-05,
+      "loss": 0.0081,
+      "step": 167840
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.91939460321578e-05,
+      "loss": 0.007,
+      "step": 167850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919389799241933e-05,
+      "loss": 0.0104,
+      "step": 167860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919384995268086e-05,
+      "loss": 0.0069,
+      "step": 167870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193801912942385e-05,
+      "loss": 0.0077,
+      "step": 167880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193753873203915e-05,
+      "loss": 0.0102,
+      "step": 167890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193705833465444e-05,
+      "loss": 0.0074,
+      "step": 167900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193657793726973e-05,
+      "loss": 0.0076,
+      "step": 167910
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.91936097539885e-05,
+      "loss": 0.0065,
+      "step": 167920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193561714250028e-05,
+      "loss": 0.0098,
+      "step": 167930
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193513674511557e-05,
+      "loss": 0.0037,
+      "step": 167940
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193465634773087e-05,
+      "loss": 0.0054,
+      "step": 167950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193417595034613e-05,
+      "loss": 0.0095,
+      "step": 167960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193369555296142e-05,
+      "loss": 0.0081,
+      "step": 167970
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919332151555767e-05,
+      "loss": 0.0096,
+      "step": 167980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193273475819197e-05,
+      "loss": 0.0073,
+      "step": 167990
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193225436080726e-05,
+      "loss": 0.0051,
+      "step": 168000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193177396342256e-05,
+      "loss": 0.0092,
+      "step": 168010
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9193129356603785e-05,
+      "loss": 0.0064,
+      "step": 168020
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919308131686531e-05,
+      "loss": 0.009,
+      "step": 168030
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919303327712684e-05,
+      "loss": 0.01,
+      "step": 168040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919298523738837e-05,
+      "loss": 0.0055,
+      "step": 168050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192937197649895e-05,
+      "loss": 0.0063,
+      "step": 168060
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192889157911424e-05,
+      "loss": 0.0068,
+      "step": 168070
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192841118172954e-05,
+      "loss": 0.0069,
+      "step": 168080
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192793078434483e-05,
+      "loss": 0.0072,
+      "step": 168090
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919274503869601e-05,
+      "loss": 0.0059,
+      "step": 168100
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192696998957538e-05,
+      "loss": 0.0087,
+      "step": 168110
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192648959219067e-05,
+      "loss": 0.009,
+      "step": 168120
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192600919480597e-05,
+      "loss": 0.0053,
+      "step": 168130
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919255768371597e-05,
+      "loss": 0.006,
+      "step": 168140
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.91925096439775e-05,
+      "loss": 0.01,
+      "step": 168150
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192461604239026e-05,
+      "loss": 0.0061,
+      "step": 168160
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192413564500556e-05,
+      "loss": 0.0086,
+      "step": 168170
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192365524762085e-05,
+      "loss": 0.0089,
+      "step": 168180
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192317485023614e-05,
+      "loss": 0.0056,
+      "step": 168190
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919226944528514e-05,
+      "loss": 0.009,
+      "step": 168200
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919222140554667e-05,
+      "loss": 0.0069,
+      "step": 168210
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.91921733658082e-05,
+      "loss": 0.0088,
+      "step": 168220
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192125326069725e-05,
+      "loss": 0.0065,
+      "step": 168230
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192077286331254e-05,
+      "loss": 0.0076,
+      "step": 168240
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192029246592783e-05,
+      "loss": 0.0084,
+      "step": 168250
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9191981206854312e-05,
+      "loss": 0.0084,
+      "step": 168260
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191933167115838e-05,
+      "loss": 0.0081,
+      "step": 168270
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191885127377368e-05,
+      "loss": 0.0116,
+      "step": 168280
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191837087638897e-05,
+      "loss": 0.0092,
+      "step": 168290
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191789047900423e-05,
+      "loss": 0.0063,
+      "step": 168300
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191741008161952e-05,
+      "loss": 0.0077,
+      "step": 168310
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.919169296842348e-05,
+      "loss": 0.0049,
+      "step": 168320
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.919164492868501e-05,
+      "loss": 0.0079,
+      "step": 168330
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191596888946536e-05,
+      "loss": 0.0074,
+      "step": 168340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191548849208066e-05,
+      "loss": 0.0053,
+      "step": 168350
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191500809469595e-05,
+      "loss": 0.0088,
+      "step": 168360
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191452769731124e-05,
+      "loss": 0.0072,
+      "step": 168370
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.919140472999265e-05,
+      "loss": 0.0078,
+      "step": 168380
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.919135669025418e-05,
+      "loss": 0.0042,
+      "step": 168390
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.919130865051571e-05,
+      "loss": 0.0078,
+      "step": 168400
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191260610777234e-05,
+      "loss": 0.0076,
+      "step": 168410
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191212571038764e-05,
+      "loss": 0.005,
+      "step": 168420
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191164531300293e-05,
+      "loss": 0.0078,
+      "step": 168430
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191116491561822e-05,
+      "loss": 0.0058,
+      "step": 168440
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191068451823348e-05,
+      "loss": 0.0089,
+      "step": 168450
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9191020412084877e-05,
+      "loss": 0.0055,
+      "step": 168460
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190972372346407e-05,
+      "loss": 0.0053,
+      "step": 168470
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190924332607933e-05,
+      "loss": 0.0091,
+      "step": 168480
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190876292869462e-05,
+      "loss": 0.0069,
+      "step": 168490
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.919082825313099e-05,
+      "loss": 0.0092,
+      "step": 168500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.919078021339252e-05,
+      "loss": 0.0066,
+      "step": 168510
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190732173654046e-05,
+      "loss": 0.0077,
+      "step": 168520
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190684133915576e-05,
+      "loss": 0.008,
+      "step": 168530
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190636094177105e-05,
+      "loss": 0.006,
+      "step": 168540
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190588054438634e-05,
+      "loss": 0.0074,
+      "step": 168550
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.919054001470016e-05,
+      "loss": 0.0056,
+      "step": 168560
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.919049197496169e-05,
+      "loss": 0.0069,
+      "step": 168570
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.919044393522322e-05,
+      "loss": 0.0079,
+      "step": 168580
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190395895484744e-05,
+      "loss": 0.0098,
+      "step": 168590
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190347855746274e-05,
+      "loss": 0.0102,
+      "step": 168600
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190299816007803e-05,
+      "loss": 0.0074,
+      "step": 168610
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190251776269332e-05,
+      "loss": 0.0056,
+      "step": 168620
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190203736530858e-05,
+      "loss": 0.0086,
+      "step": 168630
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190155696792387e-05,
+      "loss": 0.0085,
+      "step": 168640
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190107657053917e-05,
+      "loss": 0.0077,
+      "step": 168650
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190059617315443e-05,
+      "loss": 0.0054,
+      "step": 168660
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9190011577576972e-05,
+      "loss": 0.007,
+      "step": 168670
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.91899635378385e-05,
+      "loss": 0.0064,
+      "step": 168680
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.918991549810003e-05,
+      "loss": 0.0102,
+      "step": 168690
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189867458361556e-05,
+      "loss": 0.0084,
+      "step": 168700
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189819418623085e-05,
+      "loss": 0.0063,
+      "step": 168710
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189771378884615e-05,
+      "loss": 0.0056,
+      "step": 168720
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189723339146144e-05,
+      "loss": 0.0087,
+      "step": 168730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.918967529940767e-05,
+      "loss": 0.0071,
+      "step": 168740
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.91896272596692e-05,
+      "loss": 0.0066,
+      "step": 168750
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.918957921993073e-05,
+      "loss": 0.0195,
+      "step": 168760
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189531180192254e-05,
+      "loss": 0.0088,
+      "step": 168770
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189483140453784e-05,
+      "loss": 0.0077,
+      "step": 168780
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189435100715313e-05,
+      "loss": 0.005,
+      "step": 168790
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189387060976842e-05,
+      "loss": 0.0069,
+      "step": 168800
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189339021238368e-05,
+      "loss": 0.0067,
+      "step": 168810
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189290981499897e-05,
+      "loss": 0.0068,
+      "step": 168820
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189242941761427e-05,
+      "loss": 0.0057,
+      "step": 168830
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189194902022952e-05,
+      "loss": 0.0065,
+      "step": 168840
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.918914686228448e-05,
+      "loss": 0.0064,
+      "step": 168850
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.918909882254601e-05,
+      "loss": 0.0061,
+      "step": 168860
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.918905078280754e-05,
+      "loss": 0.0087,
+      "step": 168870
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9189002743069066e-05,
+      "loss": 0.0076,
+      "step": 168880
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9188954703330595e-05,
+      "loss": 0.01,
+      "step": 168890
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9188906663592125e-05,
+      "loss": 0.0085,
+      "step": 168900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9188858623853654e-05,
+      "loss": 0.0069,
+      "step": 168910
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.918881058411518e-05,
+      "loss": 0.0062,
+      "step": 168920
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.918876254437671e-05,
+      "loss": 0.0082,
+      "step": 168930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.918871450463824e-05,
+      "loss": 0.0145,
+      "step": 168940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9188666464899764e-05,
+      "loss": 0.0082,
+      "step": 168950
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9188618425161293e-05,
+      "loss": 0.0077,
+      "step": 168960
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9188570385422823e-05,
+      "loss": 0.0087,
+      "step": 168970
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9188522345684352e-05,
+      "loss": 0.0078,
+      "step": 168980
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9188474305945878e-05,
+      "loss": 0.0064,
+      "step": 168990
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9188426266207407e-05,
+      "loss": 0.0058,
+      "step": 169000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9188378226468936e-05,
+      "loss": 0.0069,
+      "step": 169010
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9188330186730466e-05,
+      "loss": 0.0091,
+      "step": 169020
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9188282146991995e-05,
+      "loss": 0.0132,
+      "step": 169030
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9188234107253524e-05,
+      "loss": 0.0049,
+      "step": 169040
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918818606751505e-05,
+      "loss": 0.0072,
+      "step": 169050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918813802777658e-05,
+      "loss": 0.0069,
+      "step": 169060
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918808998803811e-05,
+      "loss": 0.0074,
+      "step": 169070
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9188041948299635e-05,
+      "loss": 0.0095,
+      "step": 169080
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187993908561164e-05,
+      "loss": 0.0075,
+      "step": 169090
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187945868822693e-05,
+      "loss": 0.0072,
+      "step": 169100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187897829084222e-05,
+      "loss": 0.0076,
+      "step": 169110
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187849789345748e-05,
+      "loss": 0.0069,
+      "step": 169120
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187801749607278e-05,
+      "loss": 0.0069,
+      "step": 169130
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187753709868807e-05,
+      "loss": 0.0071,
+      "step": 169140
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187705670130333e-05,
+      "loss": 0.0099,
+      "step": 169150
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187657630391862e-05,
+      "loss": 0.0075,
+      "step": 169160
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918760959065339e-05,
+      "loss": 0.0068,
+      "step": 169170
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918756155091492e-05,
+      "loss": 0.0078,
+      "step": 169180
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187513511176446e-05,
+      "loss": 0.0051,
+      "step": 169190
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187465471437976e-05,
+      "loss": 0.0073,
+      "step": 169200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187417431699505e-05,
+      "loss": 0.0064,
+      "step": 169210
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187369391961034e-05,
+      "loss": 0.0072,
+      "step": 169220
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918732135222256e-05,
+      "loss": 0.0107,
+      "step": 169230
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918727331248409e-05,
+      "loss": 0.0078,
+      "step": 169240
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918722527274562e-05,
+      "loss": 0.0075,
+      "step": 169250
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187177233007144e-05,
+      "loss": 0.0086,
+      "step": 169260
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187129193268674e-05,
+      "loss": 0.0092,
+      "step": 169270
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187081153530203e-05,
+      "loss": 0.0084,
+      "step": 169280
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9187033113791732e-05,
+      "loss": 0.0088,
+      "step": 169290
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186985074053258e-05,
+      "loss": 0.0083,
+      "step": 169300
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186937034314787e-05,
+      "loss": 0.0055,
+      "step": 169310
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186888994576317e-05,
+      "loss": 0.0081,
+      "step": 169320
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186840954837843e-05,
+      "loss": 0.007,
+      "step": 169330
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186792915099372e-05,
+      "loss": 0.0068,
+      "step": 169340
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.91867448753609e-05,
+      "loss": 0.0092,
+      "step": 169350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918669683562243e-05,
+      "loss": 0.0042,
+      "step": 169360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186648795883956e-05,
+      "loss": 0.0091,
+      "step": 169370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186600756145486e-05,
+      "loss": 0.01,
+      "step": 169380
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186552716407015e-05,
+      "loss": 0.0099,
+      "step": 169390
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186504676668544e-05,
+      "loss": 0.0086,
+      "step": 169400
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918645663693007e-05,
+      "loss": 0.0066,
+      "step": 169410
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.91864085971916e-05,
+      "loss": 0.0089,
+      "step": 169420
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918636055745313e-05,
+      "loss": 0.006,
+      "step": 169430
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186312517714654e-05,
+      "loss": 0.0059,
+      "step": 169440
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186264477976184e-05,
+      "loss": 0.0065,
+      "step": 169450
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186216438237713e-05,
+      "loss": 0.005,
+      "step": 169460
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186168398499242e-05,
+      "loss": 0.0068,
+      "step": 169470
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186120358760768e-05,
+      "loss": 0.0087,
+      "step": 169480
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186072319022297e-05,
+      "loss": 0.0068,
+      "step": 169490
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9186024279283827e-05,
+      "loss": 0.012,
+      "step": 169500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9185976239545353e-05,
+      "loss": 0.0089,
+      "step": 169510
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9185928199806882e-05,
+      "loss": 0.0085,
+      "step": 169520
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918588016006841e-05,
+      "loss": 0.0087,
+      "step": 169530
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918583212032994e-05,
+      "loss": 0.0066,
+      "step": 169540
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9185784080591466e-05,
+      "loss": 0.0075,
+      "step": 169550
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9185736040852995e-05,
+      "loss": 0.0104,
+      "step": 169560
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9185688001114525e-05,
+      "loss": 0.0095,
+      "step": 169570
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9185639961376054e-05,
+      "loss": 0.0075,
+      "step": 169580
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918559192163758e-05,
+      "loss": 0.0092,
+      "step": 169590
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918554388189911e-05,
+      "loss": 0.0072,
+      "step": 169600
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.918549584216064e-05,
+      "loss": 0.0076,
+      "step": 169610
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9185447802422164e-05,
+      "loss": 0.0108,
+      "step": 169620
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9185399762683694e-05,
+      "loss": 0.0091,
+      "step": 169630
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9185351722945223e-05,
+      "loss": 0.0085,
+      "step": 169640
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9185303683206752e-05,
+      "loss": 0.0059,
+      "step": 169650
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9185255643468278e-05,
+      "loss": 0.0045,
+      "step": 169660
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9185207603729807e-05,
+      "loss": 0.0063,
+      "step": 169670
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9185159563991337e-05,
+      "loss": 0.0074,
+      "step": 169680
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9185111524252862e-05,
+      "loss": 0.0143,
+      "step": 169690
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918506348451439e-05,
+      "loss": 0.0102,
+      "step": 169700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918501544477592e-05,
+      "loss": 0.0092,
+      "step": 169710
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918496740503745e-05,
+      "loss": 0.0078,
+      "step": 169720
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184919365298976e-05,
+      "loss": 0.0064,
+      "step": 169730
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184871325560505e-05,
+      "loss": 0.0095,
+      "step": 169740
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184823285822035e-05,
+      "loss": 0.0095,
+      "step": 169750
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184775246083564e-05,
+      "loss": 0.0065,
+      "step": 169760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918472720634509e-05,
+      "loss": 0.0064,
+      "step": 169770
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918467916660662e-05,
+      "loss": 0.0047,
+      "step": 169780
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918463112686815e-05,
+      "loss": 0.0087,
+      "step": 169790
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184583087129674e-05,
+      "loss": 0.0072,
+      "step": 169800
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184535047391203e-05,
+      "loss": 0.0055,
+      "step": 169810
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184487007652733e-05,
+      "loss": 0.0061,
+      "step": 169820
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184438967914262e-05,
+      "loss": 0.0075,
+      "step": 169830
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184390928175788e-05,
+      "loss": 0.0079,
+      "step": 169840
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184342888437317e-05,
+      "loss": 0.0074,
+      "step": 169850
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184294848698846e-05,
+      "loss": 0.0069,
+      "step": 169860
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184246808960372e-05,
+      "loss": 0.0068,
+      "step": 169870
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.91841987692219e-05,
+      "loss": 0.0098,
+      "step": 169880
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918415072948343e-05,
+      "loss": 0.0062,
+      "step": 169890
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918410268974496e-05,
+      "loss": 0.0065,
+      "step": 169900
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184054650006486e-05,
+      "loss": 0.0092,
+      "step": 169910
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9184006610268015e-05,
+      "loss": 0.0077,
+      "step": 169920
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183958570529545e-05,
+      "loss": 0.0071,
+      "step": 169930
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183910530791074e-05,
+      "loss": 0.0067,
+      "step": 169940
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.91838624910526e-05,
+      "loss": 0.0086,
+      "step": 169950
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918381445131413e-05,
+      "loss": 0.0091,
+      "step": 169960
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183766411575658e-05,
+      "loss": 0.0065,
+      "step": 169970
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183718371837184e-05,
+      "loss": 0.006,
+      "step": 169980
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183670332098713e-05,
+      "loss": 0.0046,
+      "step": 169990
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183622292360243e-05,
+      "loss": 0.0098,
+      "step": 170000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183574252621772e-05,
+      "loss": 0.0089,
+      "step": 170010
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183526212883298e-05,
+      "loss": 0.0103,
+      "step": 170020
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183478173144827e-05,
+      "loss": 0.0061,
+      "step": 170030
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183430133406356e-05,
+      "loss": 0.0066,
+      "step": 170040
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183382093667882e-05,
+      "loss": 0.008,
+      "step": 170050
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918333405392941e-05,
+      "loss": 0.0057,
+      "step": 170060
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918328601419094e-05,
+      "loss": 0.0058,
+      "step": 170070
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918323797445247e-05,
+      "loss": 0.0045,
+      "step": 170080
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183189934713996e-05,
+      "loss": 0.0107,
+      "step": 170090
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183141894975525e-05,
+      "loss": 0.0083,
+      "step": 170100
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183093855237054e-05,
+      "loss": 0.0091,
+      "step": 170110
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9183045815498584e-05,
+      "loss": 0.0087,
+      "step": 170120
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918299777576011e-05,
+      "loss": 0.0084,
+      "step": 170130
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918294973602164e-05,
+      "loss": 0.0049,
+      "step": 170140
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182901696283168e-05,
+      "loss": 0.0071,
+      "step": 170150
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182853656544694e-05,
+      "loss": 0.006,
+      "step": 170160
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182805616806223e-05,
+      "loss": 0.009,
+      "step": 170170
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182757577067753e-05,
+      "loss": 0.0069,
+      "step": 170180
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182709537329282e-05,
+      "loss": 0.0057,
+      "step": 170190
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182661497590808e-05,
+      "loss": 0.0124,
+      "step": 170200
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182613457852337e-05,
+      "loss": 0.0063,
+      "step": 170210
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182565418113866e-05,
+      "loss": 0.007,
+      "step": 170220
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182517378375392e-05,
+      "loss": 0.008,
+      "step": 170230
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918246933863692e-05,
+      "loss": 0.0101,
+      "step": 170240
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918242129889845e-05,
+      "loss": 0.0084,
+      "step": 170250
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918237325915998e-05,
+      "loss": 0.0079,
+      "step": 170260
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182325219421506e-05,
+      "loss": 0.0075,
+      "step": 170270
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182277179683035e-05,
+      "loss": 0.0068,
+      "step": 170280
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182229139944564e-05,
+      "loss": 0.0089,
+      "step": 170290
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182181100206094e-05,
+      "loss": 0.0089,
+      "step": 170300
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918213306046762e-05,
+      "loss": 0.0066,
+      "step": 170310
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.918208502072915e-05,
+      "loss": 0.0091,
+      "step": 170320
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9182036980990678e-05,
+      "loss": 0.0065,
+      "step": 170330
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9181988941252204e-05,
+      "loss": 0.0071,
+      "step": 170340
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181940901513733e-05,
+      "loss": 0.0072,
+      "step": 170350
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181892861775263e-05,
+      "loss": 0.008,
+      "step": 170360
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181844822036792e-05,
+      "loss": 0.0084,
+      "step": 170370
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181796782298318e-05,
+      "loss": 0.0112,
+      "step": 170380
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181748742559847e-05,
+      "loss": 0.0066,
+      "step": 170390
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181700702821376e-05,
+      "loss": 0.0072,
+      "step": 170400
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181652663082902e-05,
+      "loss": 0.0073,
+      "step": 170410
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.918160462334443e-05,
+      "loss": 0.0074,
+      "step": 170420
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.918155658360596e-05,
+      "loss": 0.0085,
+      "step": 170430
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.918150854386749e-05,
+      "loss": 0.006,
+      "step": 170440
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181460504129016e-05,
+      "loss": 0.0098,
+      "step": 170450
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181412464390545e-05,
+      "loss": 0.0055,
+      "step": 170460
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181364424652074e-05,
+      "loss": 0.0073,
+      "step": 170470
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181316384913604e-05,
+      "loss": 0.0083,
+      "step": 170480
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.918126834517513e-05,
+      "loss": 0.009,
+      "step": 170490
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.918122030543666e-05,
+      "loss": 0.0056,
+      "step": 170500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181172265698188e-05,
+      "loss": 0.0062,
+      "step": 170510
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181124225959714e-05,
+      "loss": 0.0086,
+      "step": 170520
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181076186221243e-05,
+      "loss": 0.0055,
+      "step": 170530
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9181028146482772e-05,
+      "loss": 0.0106,
+      "step": 170540
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.91809801067443e-05,
+      "loss": 0.0104,
+      "step": 170550
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180932067005828e-05,
+      "loss": 0.0062,
+      "step": 170560
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180884027267357e-05,
+      "loss": 0.0053,
+      "step": 170570
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180835987528886e-05,
+      "loss": 0.006,
+      "step": 170580
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180787947790412e-05,
+      "loss": 0.0072,
+      "step": 170590
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.918073990805194e-05,
+      "loss": 0.0054,
+      "step": 170600
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.918069186831347e-05,
+      "loss": 0.0068,
+      "step": 170610
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180643828575e-05,
+      "loss": 0.0073,
+      "step": 170620
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180595788836526e-05,
+      "loss": 0.0071,
+      "step": 170630
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180547749098055e-05,
+      "loss": 0.0067,
+      "step": 170640
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180499709359584e-05,
+      "loss": 0.0092,
+      "step": 170650
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180451669621113e-05,
+      "loss": 0.0047,
+      "step": 170660
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.918040362988264e-05,
+      "loss": 0.0065,
+      "step": 170670
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.918035559014417e-05,
+      "loss": 0.0074,
+      "step": 170680
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180307550405698e-05,
+      "loss": 0.0074,
+      "step": 170690
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180259510667224e-05,
+      "loss": 0.0051,
+      "step": 170700
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180211470928753e-05,
+      "loss": 0.0091,
+      "step": 170710
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180163431190282e-05,
+      "loss": 0.0063,
+      "step": 170720
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.918011539145181e-05,
+      "loss": 0.0071,
+      "step": 170730
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180067351713337e-05,
+      "loss": 0.009,
+      "step": 170740
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9180019311974867e-05,
+      "loss": 0.0094,
+      "step": 170750
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179971272236396e-05,
+      "loss": 0.0065,
+      "step": 170760
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179923232497922e-05,
+      "loss": 0.0063,
+      "step": 170770
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.917987519275945e-05,
+      "loss": 0.0064,
+      "step": 170780
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.917982715302098e-05,
+      "loss": 0.0059,
+      "step": 170790
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.917977911328251e-05,
+      "loss": 0.0075,
+      "step": 170800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179731073544036e-05,
+      "loss": 0.0099,
+      "step": 170810
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179683033805565e-05,
+      "loss": 0.0089,
+      "step": 170820
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179634994067094e-05,
+      "loss": 0.0106,
+      "step": 170830
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179586954328623e-05,
+      "loss": 0.007,
+      "step": 170840
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.917953891459015e-05,
+      "loss": 0.0066,
+      "step": 170850
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.917949087485168e-05,
+      "loss": 0.0056,
+      "step": 170860
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179442835113208e-05,
+      "loss": 0.0067,
+      "step": 170870
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179394795374734e-05,
+      "loss": 0.0071,
+      "step": 170880
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179346755636263e-05,
+      "loss": 0.0064,
+      "step": 170890
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179298715897792e-05,
+      "loss": 0.0089,
+      "step": 170900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.917925067615932e-05,
+      "loss": 0.0087,
+      "step": 170910
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179202636420847e-05,
+      "loss": 0.0075,
+      "step": 170920
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179154596682377e-05,
+      "loss": 0.0087,
+      "step": 170930
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179106556943906e-05,
+      "loss": 0.0047,
+      "step": 170940
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9179058517205432e-05,
+      "loss": 0.0063,
+      "step": 170950
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.917901047746696e-05,
+      "loss": 0.0053,
+      "step": 170960
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.917896243772849e-05,
+      "loss": 0.0067,
+      "step": 170970
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.917891439799002e-05,
+      "loss": 0.0056,
+      "step": 170980
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9178866358251546e-05,
+      "loss": 0.0074,
+      "step": 170990
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9178818318513075e-05,
+      "loss": 0.005,
+      "step": 171000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9178770278774604e-05,
+      "loss": 0.0101,
+      "step": 171010
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.917872223903613e-05,
+      "loss": 0.0097,
+      "step": 171020
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.917867419929766e-05,
+      "loss": 0.0064,
+      "step": 171030
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917862615955919e-05,
+      "loss": 0.0062,
+      "step": 171040
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9178578119820718e-05,
+      "loss": 0.0074,
+      "step": 171050
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9178530080082244e-05,
+      "loss": 0.0067,
+      "step": 171060
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9178482040343773e-05,
+      "loss": 0.0055,
+      "step": 171070
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9178434000605302e-05,
+      "loss": 0.0055,
+      "step": 171080
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917838596086683e-05,
+      "loss": 0.0059,
+      "step": 171090
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9178337921128357e-05,
+      "loss": 0.007,
+      "step": 171100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9178289881389887e-05,
+      "loss": 0.008,
+      "step": 171110
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9178241841651416e-05,
+      "loss": 0.009,
+      "step": 171120
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9178193801912942e-05,
+      "loss": 0.0112,
+      "step": 171130
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917814576217447e-05,
+      "loss": 0.0117,
+      "step": 171140
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9178097722436e-05,
+      "loss": 0.008,
+      "step": 171150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917804968269753e-05,
+      "loss": 0.0075,
+      "step": 171160
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9178001642959055e-05,
+      "loss": 0.0067,
+      "step": 171170
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9177953603220585e-05,
+      "loss": 0.0037,
+      "step": 171180
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9177905563482114e-05,
+      "loss": 0.0088,
+      "step": 171190
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917785752374364e-05,
+      "loss": 0.0074,
+      "step": 171200
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917780948400517e-05,
+      "loss": 0.0099,
+      "step": 171210
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.91777614442667e-05,
+      "loss": 0.0098,
+      "step": 171220
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9177713404528228e-05,
+      "loss": 0.0086,
+      "step": 171230
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9177665364789754e-05,
+      "loss": 0.0082,
+      "step": 171240
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9177617325051283e-05,
+      "loss": 0.0065,
+      "step": 171250
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9177569285312812e-05,
+      "loss": 0.0072,
+      "step": 171260
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917752124557434e-05,
+      "loss": 0.0081,
+      "step": 171270
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9177473205835867e-05,
+      "loss": 0.0076,
+      "step": 171280
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9177425166097397e-05,
+      "loss": 0.0087,
+      "step": 171290
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9177377126358926e-05,
+      "loss": 0.0056,
+      "step": 171300
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917732908662045e-05,
+      "loss": 0.0061,
+      "step": 171310
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917728104688198e-05,
+      "loss": 0.0074,
+      "step": 171320
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917723300714351e-05,
+      "loss": 0.0078,
+      "step": 171330
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917718496740504e-05,
+      "loss": 0.0086,
+      "step": 171340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9177136927666565e-05,
+      "loss": 0.009,
+      "step": 171350
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9177088887928095e-05,
+      "loss": 0.0082,
+      "step": 171360
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9177040848189624e-05,
+      "loss": 0.0081,
+      "step": 171370
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917699280845115e-05,
+      "loss": 0.0077,
+      "step": 171380
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917694476871268e-05,
+      "loss": 0.0088,
+      "step": 171390
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9176896728974208e-05,
+      "loss": 0.0068,
+      "step": 171400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9176848689235738e-05,
+      "loss": 0.0057,
+      "step": 171410
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9176800649497263e-05,
+      "loss": 0.0067,
+      "step": 171420
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9176752609758793e-05,
+      "loss": 0.0064,
+      "step": 171430
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9176704570020322e-05,
+      "loss": 0.0087,
+      "step": 171440
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917665653028185e-05,
+      "loss": 0.0061,
+      "step": 171450
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9176608490543377e-05,
+      "loss": 0.0086,
+      "step": 171460
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9176560450804906e-05,
+      "loss": 0.006,
+      "step": 171470
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9176512411066436e-05,
+      "loss": 0.0076,
+      "step": 171480
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917646437132796e-05,
+      "loss": 0.008,
+      "step": 171490
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917641633158949e-05,
+      "loss": 0.0054,
+      "step": 171500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917636829185102e-05,
+      "loss": 0.0057,
+      "step": 171510
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917632025211255e-05,
+      "loss": 0.0064,
+      "step": 171520
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9176272212374075e-05,
+      "loss": 0.0065,
+      "step": 171530
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9176224172635605e-05,
+      "loss": 0.0053,
+      "step": 171540
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9176176132897134e-05,
+      "loss": 0.0079,
+      "step": 171550
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917612809315866e-05,
+      "loss": 0.0047,
+      "step": 171560
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9176080053420192e-05,
+      "loss": 0.0077,
+      "step": 171570
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917603201368172e-05,
+      "loss": 0.008,
+      "step": 171580
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9175983973943247e-05,
+      "loss": 0.0048,
+      "step": 171590
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9175935934204777e-05,
+      "loss": 0.008,
+      "step": 171600
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9175887894466306e-05,
+      "loss": 0.0072,
+      "step": 171610
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9175839854727832e-05,
+      "loss": 0.0077,
+      "step": 171620
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917579181498936e-05,
+      "loss": 0.0082,
+      "step": 171630
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917574377525089e-05,
+      "loss": 0.0073,
+      "step": 171640
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917569573551242e-05,
+      "loss": 0.0082,
+      "step": 171650
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9175647695773946e-05,
+      "loss": 0.0066,
+      "step": 171660
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9175599656035475e-05,
+      "loss": 0.0097,
+      "step": 171670
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9175551616297004e-05,
+      "loss": 0.0092,
+      "step": 171680
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9175503576558533e-05,
+      "loss": 0.014,
+      "step": 171690
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917545553682006e-05,
+      "loss": 0.0056,
+      "step": 171700
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.917540749708159e-05,
+      "loss": 0.0088,
+      "step": 171710
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9175359457343118e-05,
+      "loss": 0.0076,
+      "step": 171720
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9175311417604644e-05,
+      "loss": 0.0081,
+      "step": 171730
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9175263377866173e-05,
+      "loss": 0.0065,
+      "step": 171740
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9175215338127702e-05,
+      "loss": 0.0116,
+      "step": 171750
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917516729838923e-05,
+      "loss": 0.0093,
+      "step": 171760
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9175119258650757e-05,
+      "loss": 0.0064,
+      "step": 171770
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9175071218912287e-05,
+      "loss": 0.0067,
+      "step": 171780
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9175023179173816e-05,
+      "loss": 0.0058,
+      "step": 171790
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174975139435342e-05,
+      "loss": 0.0057,
+      "step": 171800
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917492709969687e-05,
+      "loss": 0.0077,
+      "step": 171810
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.91748790599584e-05,
+      "loss": 0.0059,
+      "step": 171820
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917483102021993e-05,
+      "loss": 0.0058,
+      "step": 171830
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174782980481456e-05,
+      "loss": 0.006,
+      "step": 171840
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174734940742985e-05,
+      "loss": 0.0115,
+      "step": 171850
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174686901004514e-05,
+      "loss": 0.0053,
+      "step": 171860
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174638861266043e-05,
+      "loss": 0.0082,
+      "step": 171870
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917459082152757e-05,
+      "loss": 0.006,
+      "step": 171880
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.91745427817891e-05,
+      "loss": 0.0083,
+      "step": 171890
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174494742050628e-05,
+      "loss": 0.013,
+      "step": 171900
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174446702312154e-05,
+      "loss": 0.0098,
+      "step": 171910
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174398662573683e-05,
+      "loss": 0.0078,
+      "step": 171920
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174350622835212e-05,
+      "loss": 0.0096,
+      "step": 171930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917430258309674e-05,
+      "loss": 0.0063,
+      "step": 171940
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174254543358267e-05,
+      "loss": 0.0095,
+      "step": 171950
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174206503619797e-05,
+      "loss": 0.0088,
+      "step": 171960
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174158463881326e-05,
+      "loss": 0.0082,
+      "step": 171970
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9174110424142852e-05,
+      "loss": 0.006,
+      "step": 171980
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917406238440438e-05,
+      "loss": 0.0063,
+      "step": 171990
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917401434466591e-05,
+      "loss": 0.0072,
+      "step": 172000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917396630492744e-05,
+      "loss": 0.0076,
+      "step": 172010
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173918265188965e-05,
+      "loss": 0.0115,
+      "step": 172020
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173870225450495e-05,
+      "loss": 0.0063,
+      "step": 172030
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173822185712024e-05,
+      "loss": 0.0072,
+      "step": 172040
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173774145973553e-05,
+      "loss": 0.0083,
+      "step": 172050
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917372610623508e-05,
+      "loss": 0.0066,
+      "step": 172060
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917367806649661e-05,
+      "loss": 0.0064,
+      "step": 172070
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173630026758138e-05,
+      "loss": 0.0081,
+      "step": 172080
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173581987019664e-05,
+      "loss": 0.0089,
+      "step": 172090
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173533947281193e-05,
+      "loss": 0.0071,
+      "step": 172100
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173485907542722e-05,
+      "loss": 0.0063,
+      "step": 172110
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917343786780425e-05,
+      "loss": 0.0085,
+      "step": 172120
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173389828065777e-05,
+      "loss": 0.0077,
+      "step": 172130
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173341788327307e-05,
+      "loss": 0.0054,
+      "step": 172140
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173293748588836e-05,
+      "loss": 0.0067,
+      "step": 172150
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917324570885036e-05,
+      "loss": 0.0067,
+      "step": 172160
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917319766911189e-05,
+      "loss": 0.0075,
+      "step": 172170
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917314962937342e-05,
+      "loss": 0.0076,
+      "step": 172180
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917310158963495e-05,
+      "loss": 0.006,
+      "step": 172190
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173053549896475e-05,
+      "loss": 0.0096,
+      "step": 172200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9173005510158005e-05,
+      "loss": 0.0096,
+      "step": 172210
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9172957470419534e-05,
+      "loss": 0.0081,
+      "step": 172220
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9172909430681063e-05,
+      "loss": 0.0079,
+      "step": 172230
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917286139094259e-05,
+      "loss": 0.0068,
+      "step": 172240
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917281335120412e-05,
+      "loss": 0.0052,
+      "step": 172250
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9172765311465648e-05,
+      "loss": 0.0066,
+      "step": 172260
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9172717271727173e-05,
+      "loss": 0.0118,
+      "step": 172270
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9172669231988703e-05,
+      "loss": 0.0056,
+      "step": 172280
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9172621192250232e-05,
+      "loss": 0.007,
+      "step": 172290
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917257315251176e-05,
+      "loss": 0.0063,
+      "step": 172300
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9172525112773287e-05,
+      "loss": 0.0069,
+      "step": 172310
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9172477073034816e-05,
+      "loss": 0.0113,
+      "step": 172320
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9172429033296346e-05,
+      "loss": 0.005,
+      "step": 172330
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917238099355787e-05,
+      "loss": 0.0064,
+      "step": 172340
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.91723329538194e-05,
+      "loss": 0.0087,
+      "step": 172350
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917228491408093e-05,
+      "loss": 0.0076,
+      "step": 172360
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917223687434246e-05,
+      "loss": 0.0075,
+      "step": 172370
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9172188834603985e-05,
+      "loss": 0.0085,
+      "step": 172380
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9172140794865515e-05,
+      "loss": 0.0075,
+      "step": 172390
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9172092755127044e-05,
+      "loss": 0.0089,
+      "step": 172400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.917204471538857e-05,
+      "loss": 0.007,
+      "step": 172410
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.91719966756501e-05,
+      "loss": 0.0056,
+      "step": 172420
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171948635911628e-05,
+      "loss": 0.0061,
+      "step": 172430
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171900596173157e-05,
+      "loss": 0.0059,
+      "step": 172440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171852556434683e-05,
+      "loss": 0.0074,
+      "step": 172450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171804516696213e-05,
+      "loss": 0.0098,
+      "step": 172460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171756476957742e-05,
+      "loss": 0.0081,
+      "step": 172470
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917170843721927e-05,
+      "loss": 0.004,
+      "step": 172480
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171660397480797e-05,
+      "loss": 0.0068,
+      "step": 172490
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171612357742326e-05,
+      "loss": 0.0089,
+      "step": 172500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171564318003856e-05,
+      "loss": 0.0089,
+      "step": 172510
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917151627826538e-05,
+      "loss": 0.0052,
+      "step": 172520
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917146823852691e-05,
+      "loss": 0.0079,
+      "step": 172530
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917142019878844e-05,
+      "loss": 0.0064,
+      "step": 172540
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917137215904997e-05,
+      "loss": 0.0082,
+      "step": 172550
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171324119311495e-05,
+      "loss": 0.0095,
+      "step": 172560
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171276079573024e-05,
+      "loss": 0.0083,
+      "step": 172570
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171228039834554e-05,
+      "loss": 0.0094,
+      "step": 172580
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917118000009608e-05,
+      "loss": 0.0065,
+      "step": 172590
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917113196035761e-05,
+      "loss": 0.0075,
+      "step": 172600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171083920619138e-05,
+      "loss": 0.0084,
+      "step": 172610
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9171035880880667e-05,
+      "loss": 0.0061,
+      "step": 172620
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170987841142193e-05,
+      "loss": 0.0103,
+      "step": 172630
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170939801403723e-05,
+      "loss": 0.0094,
+      "step": 172640
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170891761665252e-05,
+      "loss": 0.007,
+      "step": 172650
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917084372192678e-05,
+      "loss": 0.006,
+      "step": 172660
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170795682188307e-05,
+      "loss": 0.0085,
+      "step": 172670
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170747642449836e-05,
+      "loss": 0.0094,
+      "step": 172680
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170699602711366e-05,
+      "loss": 0.0082,
+      "step": 172690
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917065156297289e-05,
+      "loss": 0.007,
+      "step": 172700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917060352323442e-05,
+      "loss": 0.0083,
+      "step": 172710
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917055548349595e-05,
+      "loss": 0.0071,
+      "step": 172720
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917050744375748e-05,
+      "loss": 0.0074,
+      "step": 172730
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170459404019005e-05,
+      "loss": 0.0107,
+      "step": 172740
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170411364280534e-05,
+      "loss": 0.007,
+      "step": 172750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170363324542064e-05,
+      "loss": 0.0065,
+      "step": 172760
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917031528480359e-05,
+      "loss": 0.0048,
+      "step": 172770
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.917026724506512e-05,
+      "loss": 0.0045,
+      "step": 172780
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170219205326648e-05,
+      "loss": 0.0056,
+      "step": 172790
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170171165588177e-05,
+      "loss": 0.0066,
+      "step": 172800
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170123125849703e-05,
+      "loss": 0.0041,
+      "step": 172810
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170075086111232e-05,
+      "loss": 0.0073,
+      "step": 172820
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9170027046372762e-05,
+      "loss": 0.0083,
+      "step": 172830
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.916997900663429e-05,
+      "loss": 0.0068,
+      "step": 172840
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9169930966895817e-05,
+      "loss": 0.0092,
+      "step": 172850
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9169882927157346e-05,
+      "loss": 0.0084,
+      "step": 172860
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9169834887418875e-05,
+      "loss": 0.0079,
+      "step": 172870
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.91697868476804e-05,
+      "loss": 0.0073,
+      "step": 172880
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.916973880794193e-05,
+      "loss": 0.0089,
+      "step": 172890
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.916969076820346e-05,
+      "loss": 0.0068,
+      "step": 172900
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.916964272846499e-05,
+      "loss": 0.009,
+      "step": 172910
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9169594688726515e-05,
+      "loss": 0.006,
+      "step": 172920
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9169546648988044e-05,
+      "loss": 0.0086,
+      "step": 172930
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9169498609249574e-05,
+      "loss": 0.0056,
+      "step": 172940
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.91694505695111e-05,
+      "loss": 0.008,
+      "step": 172950
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.916940252977263e-05,
+      "loss": 0.005,
+      "step": 172960
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9169354490034158e-05,
+      "loss": 0.0074,
+      "step": 172970
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9169306450295687e-05,
+      "loss": 0.007,
+      "step": 172980
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9169258410557213e-05,
+      "loss": 0.0077,
+      "step": 172990
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9169210370818742e-05,
+      "loss": 0.0066,
+      "step": 173000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.916916233108027e-05,
+      "loss": 0.0081,
+      "step": 173010
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.91691142913418e-05,
+      "loss": 0.0065,
+      "step": 173020
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9169066251603327e-05,
+      "loss": 0.005,
+      "step": 173030
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9169018211864856e-05,
+      "loss": 0.0064,
+      "step": 173040
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9168970172126385e-05,
+      "loss": 0.0058,
+      "step": 173050
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.916892213238791e-05,
+      "loss": 0.0091,
+      "step": 173060
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.916887409264944e-05,
+      "loss": 0.0086,
+      "step": 173070
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.916882605291097e-05,
+      "loss": 0.0072,
+      "step": 173080
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.91687780131725e-05,
+      "loss": 0.0068,
+      "step": 173090
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9168729973434025e-05,
+      "loss": 0.0076,
+      "step": 173100
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9168681933695554e-05,
+      "loss": 0.0072,
+      "step": 173110
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9168633893957083e-05,
+      "loss": 0.0091,
+      "step": 173120
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916858585421861e-05,
+      "loss": 0.0056,
+      "step": 173130
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916853781448014e-05,
+      "loss": 0.008,
+      "step": 173140
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9168489774741668e-05,
+      "loss": 0.0076,
+      "step": 173150
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9168441735003197e-05,
+      "loss": 0.0076,
+      "step": 173160
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9168393695264723e-05,
+      "loss": 0.0086,
+      "step": 173170
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9168345655526252e-05,
+      "loss": 0.0059,
+      "step": 173180
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916829761578778e-05,
+      "loss": 0.0081,
+      "step": 173190
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916824957604931e-05,
+      "loss": 0.0092,
+      "step": 173200
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9168201536310837e-05,
+      "loss": 0.0076,
+      "step": 173210
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9168153496572366e-05,
+      "loss": 0.005,
+      "step": 173220
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9168105456833895e-05,
+      "loss": 0.0075,
+      "step": 173230
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916805741709542e-05,
+      "loss": 0.0104,
+      "step": 173240
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916800937735695e-05,
+      "loss": 0.0098,
+      "step": 173250
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916796133761848e-05,
+      "loss": 0.0094,
+      "step": 173260
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916791329788001e-05,
+      "loss": 0.0059,
+      "step": 173270
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9167865258141535e-05,
+      "loss": 0.0073,
+      "step": 173280
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9167817218403064e-05,
+      "loss": 0.005,
+      "step": 173290
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9167769178664593e-05,
+      "loss": 0.0058,
+      "step": 173300
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916772113892612e-05,
+      "loss": 0.0042,
+      "step": 173310
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916767309918765e-05,
+      "loss": 0.0086,
+      "step": 173320
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9167625059449178e-05,
+      "loss": 0.0075,
+      "step": 173330
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9167577019710707e-05,
+      "loss": 0.0089,
+      "step": 173340
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9167528979972233e-05,
+      "loss": 0.0078,
+      "step": 173350
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9167480940233762e-05,
+      "loss": 0.0082,
+      "step": 173360
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916743290049529e-05,
+      "loss": 0.0063,
+      "step": 173370
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916738486075682e-05,
+      "loss": 0.0071,
+      "step": 173380
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9167336821018347e-05,
+      "loss": 0.0064,
+      "step": 173390
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9167288781279876e-05,
+      "loss": 0.008,
+      "step": 173400
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9167240741541405e-05,
+      "loss": 0.0109,
+      "step": 173410
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916719270180293e-05,
+      "loss": 0.0077,
+      "step": 173420
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916714466206446e-05,
+      "loss": 0.0072,
+      "step": 173430
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916709662232599e-05,
+      "loss": 0.0065,
+      "step": 173440
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916704858258752e-05,
+      "loss": 0.0109,
+      "step": 173450
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9167000542849045e-05,
+      "loss": 0.0069,
+      "step": 173460
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9166952503110574e-05,
+      "loss": 0.0076,
+      "step": 173470
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9166904463372103e-05,
+      "loss": 0.0061,
+      "step": 173480
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916685642363363e-05,
+      "loss": 0.0076,
+      "step": 173490
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916680838389516e-05,
+      "loss": 0.0105,
+      "step": 173500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9166760344156688e-05,
+      "loss": 0.0087,
+      "step": 173510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9166712304418217e-05,
+      "loss": 0.0079,
+      "step": 173520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9166664264679743e-05,
+      "loss": 0.0068,
+      "step": 173530
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9166616224941272e-05,
+      "loss": 0.0115,
+      "step": 173540
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.91665681852028e-05,
+      "loss": 0.0092,
+      "step": 173550
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916652014546433e-05,
+      "loss": 0.007,
+      "step": 173560
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9166472105725857e-05,
+      "loss": 0.0076,
+      "step": 173570
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9166424065987386e-05,
+      "loss": 0.0063,
+      "step": 173580
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9166376026248915e-05,
+      "loss": 0.0077,
+      "step": 173590
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916632798651044e-05,
+      "loss": 0.0052,
+      "step": 173600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916627994677197e-05,
+      "loss": 0.0095,
+      "step": 173610
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.91662319070335e-05,
+      "loss": 0.0064,
+      "step": 173620
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916618386729503e-05,
+      "loss": 0.0082,
+      "step": 173630
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9166135827556555e-05,
+      "loss": 0.0067,
+      "step": 173640
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9166087787818084e-05,
+      "loss": 0.0074,
+      "step": 173650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9166039748079613e-05,
+      "loss": 0.0055,
+      "step": 173660
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916599170834114e-05,
+      "loss": 0.0066,
+      "step": 173670
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916594366860267e-05,
+      "loss": 0.0081,
+      "step": 173680
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9165895628864198e-05,
+      "loss": 0.0064,
+      "step": 173690
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9165847589125727e-05,
+      "loss": 0.0067,
+      "step": 173700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9165799549387253e-05,
+      "loss": 0.0078,
+      "step": 173710
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9165751509648782e-05,
+      "loss": 0.0042,
+      "step": 173720
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916570346991031e-05,
+      "loss": 0.009,
+      "step": 173730
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916565543017184e-05,
+      "loss": 0.0084,
+      "step": 173740
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9165607390433366e-05,
+      "loss": 0.0074,
+      "step": 173750
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9165559350694896e-05,
+      "loss": 0.0072,
+      "step": 173760
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9165511310956425e-05,
+      "loss": 0.0061,
+      "step": 173770
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916546327121795e-05,
+      "loss": 0.0074,
+      "step": 173780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916541523147948e-05,
+      "loss": 0.008,
+      "step": 173790
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916536719174101e-05,
+      "loss": 0.0045,
+      "step": 173800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.916531915200254e-05,
+      "loss": 0.0071,
+      "step": 173810
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9165271112264065e-05,
+      "loss": 0.0072,
+      "step": 173820
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9165223072525594e-05,
+      "loss": 0.011,
+      "step": 173830
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9165175032787123e-05,
+      "loss": 0.0061,
+      "step": 173840
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916512699304865e-05,
+      "loss": 0.0095,
+      "step": 173850
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9165078953310178e-05,
+      "loss": 0.0061,
+      "step": 173860
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9165030913571708e-05,
+      "loss": 0.0076,
+      "step": 173870
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9164982873833237e-05,
+      "loss": 0.0065,
+      "step": 173880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9164934834094763e-05,
+      "loss": 0.0072,
+      "step": 173890
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9164886794356292e-05,
+      "loss": 0.0085,
+      "step": 173900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916483875461782e-05,
+      "loss": 0.0064,
+      "step": 173910
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916479071487935e-05,
+      "loss": 0.0069,
+      "step": 173920
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9164742675140876e-05,
+      "loss": 0.0057,
+      "step": 173930
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9164694635402406e-05,
+      "loss": 0.0071,
+      "step": 173940
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9164646595663935e-05,
+      "loss": 0.009,
+      "step": 173950
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916459855592546e-05,
+      "loss": 0.006,
+      "step": 173960
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916455051618699e-05,
+      "loss": 0.0088,
+      "step": 173970
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916450247644852e-05,
+      "loss": 0.0089,
+      "step": 173980
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916445443671005e-05,
+      "loss": 0.0077,
+      "step": 173990
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9164406396971574e-05,
+      "loss": 0.0055,
+      "step": 174000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9164358357233104e-05,
+      "loss": 0.0061,
+      "step": 174010
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9164310317494633e-05,
+      "loss": 0.0085,
+      "step": 174020
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916426227775616e-05,
+      "loss": 0.0052,
+      "step": 174030
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9164214238017688e-05,
+      "loss": 0.0067,
+      "step": 174040
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9164166198279217e-05,
+      "loss": 0.0049,
+      "step": 174050
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9164122962514592e-05,
+      "loss": 0.0067,
+      "step": 174060
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916407492277612e-05,
+      "loss": 0.0069,
+      "step": 174070
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916402688303765e-05,
+      "loss": 0.0073,
+      "step": 174080
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163978843299177e-05,
+      "loss": 0.0068,
+      "step": 174090
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163930803560706e-05,
+      "loss": 0.006,
+      "step": 174100
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163882763822235e-05,
+      "loss": 0.0053,
+      "step": 174110
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163834724083764e-05,
+      "loss": 0.0073,
+      "step": 174120
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916378668434529e-05,
+      "loss": 0.0072,
+      "step": 174130
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916373864460682e-05,
+      "loss": 0.0075,
+      "step": 174140
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916369060486835e-05,
+      "loss": 0.0098,
+      "step": 174150
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163642565129875e-05,
+      "loss": 0.0078,
+      "step": 174160
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163594525391404e-05,
+      "loss": 0.0081,
+      "step": 174170
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163546485652933e-05,
+      "loss": 0.0052,
+      "step": 174180
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163498445914462e-05,
+      "loss": 0.005,
+      "step": 174190
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916345040617599e-05,
+      "loss": 0.0081,
+      "step": 174200
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163402366437518e-05,
+      "loss": 0.0061,
+      "step": 174210
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163354326699047e-05,
+      "loss": 0.0069,
+      "step": 174220
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163306286960576e-05,
+      "loss": 0.0046,
+      "step": 174230
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163258247222102e-05,
+      "loss": 0.0108,
+      "step": 174240
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916321020748363e-05,
+      "loss": 0.0095,
+      "step": 174250
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916316216774516e-05,
+      "loss": 0.0063,
+      "step": 174260
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163114128006686e-05,
+      "loss": 0.0083,
+      "step": 174270
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163066088268216e-05,
+      "loss": 0.0073,
+      "step": 174280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9163018048529745e-05,
+      "loss": 0.0069,
+      "step": 174290
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162970008791274e-05,
+      "loss": 0.0065,
+      "step": 174300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.91629219690528e-05,
+      "loss": 0.0058,
+      "step": 174310
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916287392931433e-05,
+      "loss": 0.0105,
+      "step": 174320
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916282588957586e-05,
+      "loss": 0.0071,
+      "step": 174330
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162777849837385e-05,
+      "loss": 0.005,
+      "step": 174340
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162729810098914e-05,
+      "loss": 0.0065,
+      "step": 174350
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162681770360443e-05,
+      "loss": 0.0075,
+      "step": 174360
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162633730621972e-05,
+      "loss": 0.006,
+      "step": 174370
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162585690883498e-05,
+      "loss": 0.0073,
+      "step": 174380
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162537651145028e-05,
+      "loss": 0.0091,
+      "step": 174390
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162489611406557e-05,
+      "loss": 0.0111,
+      "step": 174400
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162441571668086e-05,
+      "loss": 0.0085,
+      "step": 174410
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162393531929612e-05,
+      "loss": 0.0049,
+      "step": 174420
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916234549219114e-05,
+      "loss": 0.0072,
+      "step": 174430
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916229745245267e-05,
+      "loss": 0.0055,
+      "step": 174440
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162249412714196e-05,
+      "loss": 0.0072,
+      "step": 174450
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162201372975726e-05,
+      "loss": 0.0095,
+      "step": 174460
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162153333237255e-05,
+      "loss": 0.0088,
+      "step": 174470
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9162105293498784e-05,
+      "loss": 0.0075,
+      "step": 174480
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916205725376031e-05,
+      "loss": 0.0067,
+      "step": 174490
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.916200921402184e-05,
+      "loss": 0.0065,
+      "step": 174500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.916196117428337e-05,
+      "loss": 0.0083,
+      "step": 174510
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161913134544894e-05,
+      "loss": 0.0096,
+      "step": 174520
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161865094806424e-05,
+      "loss": 0.014,
+      "step": 174530
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161817055067953e-05,
+      "loss": 0.0077,
+      "step": 174540
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161769015329482e-05,
+      "loss": 0.0052,
+      "step": 174550
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161720975591008e-05,
+      "loss": 0.0068,
+      "step": 174560
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161672935852537e-05,
+      "loss": 0.007,
+      "step": 174570
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161624896114067e-05,
+      "loss": 0.0082,
+      "step": 174580
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161576856375596e-05,
+      "loss": 0.0142,
+      "step": 174590
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161528816637122e-05,
+      "loss": 0.0075,
+      "step": 174600
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.916148077689865e-05,
+      "loss": 0.0066,
+      "step": 174610
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.916143273716018e-05,
+      "loss": 0.0088,
+      "step": 174620
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161384697421706e-05,
+      "loss": 0.0049,
+      "step": 174630
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161336657683236e-05,
+      "loss": 0.0091,
+      "step": 174640
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161288617944768e-05,
+      "loss": 0.0062,
+      "step": 174650
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161240578206294e-05,
+      "loss": 0.0089,
+      "step": 174660
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161192538467823e-05,
+      "loss": 0.0085,
+      "step": 174670
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161144498729353e-05,
+      "loss": 0.007,
+      "step": 174680
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.916109645899088e-05,
+      "loss": 0.0058,
+      "step": 174690
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161048419252408e-05,
+      "loss": 0.0095,
+      "step": 174700
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161000379513937e-05,
+      "loss": 0.0075,
+      "step": 174710
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160952339775466e-05,
+      "loss": 0.0083,
+      "step": 174720
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160904300036992e-05,
+      "loss": 0.0103,
+      "step": 174730
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.916085626029852e-05,
+      "loss": 0.0053,
+      "step": 174740
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.916080822056005e-05,
+      "loss": 0.0087,
+      "step": 174750
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160760180821577e-05,
+      "loss": 0.0077,
+      "step": 174760
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160712141083106e-05,
+      "loss": 0.0057,
+      "step": 174770
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160664101344635e-05,
+      "loss": 0.0074,
+      "step": 174780
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160616061606164e-05,
+      "loss": 0.011,
+      "step": 174790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.916056802186769e-05,
+      "loss": 0.007,
+      "step": 174800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.916051998212922e-05,
+      "loss": 0.0084,
+      "step": 174810
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.916047194239075e-05,
+      "loss": 0.0049,
+      "step": 174820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160423902652278e-05,
+      "loss": 0.0059,
+      "step": 174830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160375862913804e-05,
+      "loss": 0.0066,
+      "step": 174840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160327823175333e-05,
+      "loss": 0.0108,
+      "step": 174850
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160279783436863e-05,
+      "loss": 0.0095,
+      "step": 174860
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.916023174369839e-05,
+      "loss": 0.0097,
+      "step": 174870
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160183703959918e-05,
+      "loss": 0.0047,
+      "step": 174880
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160135664221447e-05,
+      "loss": 0.0062,
+      "step": 174890
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160087624482976e-05,
+      "loss": 0.0097,
+      "step": 174900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9160039584744502e-05,
+      "loss": 0.0073,
+      "step": 174910
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.915999154500603e-05,
+      "loss": 0.0074,
+      "step": 174920
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.915994350526756e-05,
+      "loss": 0.0085,
+      "step": 174930
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159895465529087e-05,
+      "loss": 0.0067,
+      "step": 174940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159847425790616e-05,
+      "loss": 0.0049,
+      "step": 174950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159799386052145e-05,
+      "loss": 0.0074,
+      "step": 174960
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159751346313674e-05,
+      "loss": 0.0051,
+      "step": 174970
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.91597033065752e-05,
+      "loss": 0.0076,
+      "step": 174980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.915965526683673e-05,
+      "loss": 0.0106,
+      "step": 174990
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.915960722709826e-05,
+      "loss": 0.0075,
+      "step": 175000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159559187359788e-05,
+      "loss": 0.0071,
+      "step": 175010
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159511147621314e-05,
+      "loss": 0.0075,
+      "step": 175020
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159463107882843e-05,
+      "loss": 0.0081,
+      "step": 175030
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159415068144372e-05,
+      "loss": 0.0057,
+      "step": 175040
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.91593670284059e-05,
+      "loss": 0.0083,
+      "step": 175050
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159318988667428e-05,
+      "loss": 0.0086,
+      "step": 175060
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159270948928957e-05,
+      "loss": 0.0069,
+      "step": 175070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159222909190486e-05,
+      "loss": 0.0087,
+      "step": 175080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159174869452012e-05,
+      "loss": 0.0066,
+      "step": 175090
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.915912682971354e-05,
+      "loss": 0.0139,
+      "step": 175100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.915907878997507e-05,
+      "loss": 0.0084,
+      "step": 175110
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9159030750236596e-05,
+      "loss": 0.0054,
+      "step": 175120
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9158982710498126e-05,
+      "loss": 0.0069,
+      "step": 175130
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9158934670759655e-05,
+      "loss": 0.0059,
+      "step": 175140
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9158886631021184e-05,
+      "loss": 0.0094,
+      "step": 175150
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.915883859128271e-05,
+      "loss": 0.0064,
+      "step": 175160
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.915879055154424e-05,
+      "loss": 0.0065,
+      "step": 175170
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.915874251180577e-05,
+      "loss": 0.0083,
+      "step": 175180
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9158694472067298e-05,
+      "loss": 0.0083,
+      "step": 175190
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9158646432328824e-05,
+      "loss": 0.0062,
+      "step": 175200
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9158598392590353e-05,
+      "loss": 0.0069,
+      "step": 175210
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9158550352851882e-05,
+      "loss": 0.0071,
+      "step": 175220
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9158502313113408e-05,
+      "loss": 0.0089,
+      "step": 175230
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9158454273374938e-05,
+      "loss": 0.0068,
+      "step": 175240
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9158406233636467e-05,
+      "loss": 0.0104,
+      "step": 175250
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9158358193897996e-05,
+      "loss": 0.0092,
+      "step": 175260
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9158310154159522e-05,
+      "loss": 0.0095,
+      "step": 175270
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915826211442105e-05,
+      "loss": 0.0057,
+      "step": 175280
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915821407468258e-05,
+      "loss": 0.0036,
+      "step": 175290
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9158166034944106e-05,
+      "loss": 0.0083,
+      "step": 175300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9158117995205636e-05,
+      "loss": 0.008,
+      "step": 175310
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9158069955467165e-05,
+      "loss": 0.006,
+      "step": 175320
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9158021915728694e-05,
+      "loss": 0.0051,
+      "step": 175330
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915797387599022e-05,
+      "loss": 0.0073,
+      "step": 175340
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915792583625175e-05,
+      "loss": 0.0074,
+      "step": 175350
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915787779651328e-05,
+      "loss": 0.0061,
+      "step": 175360
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157829756774808e-05,
+      "loss": 0.0081,
+      "step": 175370
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157781717036334e-05,
+      "loss": 0.0048,
+      "step": 175380
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157733677297863e-05,
+      "loss": 0.0063,
+      "step": 175390
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157685637559392e-05,
+      "loss": 0.0086,
+      "step": 175400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157637597820918e-05,
+      "loss": 0.0058,
+      "step": 175410
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157589558082447e-05,
+      "loss": 0.0081,
+      "step": 175420
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157541518343977e-05,
+      "loss": 0.0079,
+      "step": 175430
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157493478605506e-05,
+      "loss": 0.0059,
+      "step": 175440
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157445438867032e-05,
+      "loss": 0.0068,
+      "step": 175450
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915739739912856e-05,
+      "loss": 0.0066,
+      "step": 175460
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915734935939009e-05,
+      "loss": 0.0093,
+      "step": 175470
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157301319651616e-05,
+      "loss": 0.0057,
+      "step": 175480
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157253279913146e-05,
+      "loss": 0.0088,
+      "step": 175490
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157205240174675e-05,
+      "loss": 0.0068,
+      "step": 175500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9157157200436204e-05,
+      "loss": 0.0071,
+      "step": 175510
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915710916069773e-05,
+      "loss": 0.0081,
+      "step": 175520
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915706112095926e-05,
+      "loss": 0.0062,
+      "step": 175530
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915701308122079e-05,
+      "loss": 0.0077,
+      "step": 175540
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156965041482314e-05,
+      "loss": 0.0075,
+      "step": 175550
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156917001743844e-05,
+      "loss": 0.0067,
+      "step": 175560
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156868962005373e-05,
+      "loss": 0.0042,
+      "step": 175570
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156820922266902e-05,
+      "loss": 0.0056,
+      "step": 175580
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156772882528428e-05,
+      "loss": 0.0098,
+      "step": 175590
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156724842789957e-05,
+      "loss": 0.0085,
+      "step": 175600
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156676803051487e-05,
+      "loss": 0.0063,
+      "step": 175610
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156628763313016e-05,
+      "loss": 0.0094,
+      "step": 175620
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156580723574542e-05,
+      "loss": 0.0087,
+      "step": 175630
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915653268383607e-05,
+      "loss": 0.0072,
+      "step": 175640
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.91564846440976e-05,
+      "loss": 0.0054,
+      "step": 175650
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156436604359126e-05,
+      "loss": 0.0061,
+      "step": 175660
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156388564620655e-05,
+      "loss": 0.0092,
+      "step": 175670
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156340524882185e-05,
+      "loss": 0.0047,
+      "step": 175680
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156292485143714e-05,
+      "loss": 0.0097,
+      "step": 175690
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915624444540524e-05,
+      "loss": 0.0067,
+      "step": 175700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915619640566677e-05,
+      "loss": 0.0071,
+      "step": 175710
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.91561483659283e-05,
+      "loss": 0.0059,
+      "step": 175720
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156100326189824e-05,
+      "loss": 0.0056,
+      "step": 175730
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156052286451354e-05,
+      "loss": 0.01,
+      "step": 175740
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9156004246712883e-05,
+      "loss": 0.0087,
+      "step": 175750
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9155956206974412e-05,
+      "loss": 0.0052,
+      "step": 175760
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9155908167235938e-05,
+      "loss": 0.0064,
+      "step": 175770
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9155860127497467e-05,
+      "loss": 0.0092,
+      "step": 175780
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9155812087758997e-05,
+      "loss": 0.0064,
+      "step": 175790
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9155764048020526e-05,
+      "loss": 0.0076,
+      "step": 175800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915571600828205e-05,
+      "loss": 0.006,
+      "step": 175810
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915566796854358e-05,
+      "loss": 0.0048,
+      "step": 175820
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915561992880511e-05,
+      "loss": 0.0056,
+      "step": 175830
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9155571889066636e-05,
+      "loss": 0.008,
+      "step": 175840
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9155523849328165e-05,
+      "loss": 0.0055,
+      "step": 175850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9155475809589695e-05,
+      "loss": 0.0083,
+      "step": 175860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9155427769851224e-05,
+      "loss": 0.0083,
+      "step": 175870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915537973011275e-05,
+      "loss": 0.0078,
+      "step": 175880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.915533169037428e-05,
+      "loss": 0.0053,
+      "step": 175890
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915528365063581e-05,
+      "loss": 0.0066,
+      "step": 175900
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9155235610897334e-05,
+      "loss": 0.0075,
+      "step": 175910
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9155187571158863e-05,
+      "loss": 0.0059,
+      "step": 175920
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9155139531420393e-05,
+      "loss": 0.0065,
+      "step": 175930
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9155091491681922e-05,
+      "loss": 0.0084,
+      "step": 175940
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9155043451943448e-05,
+      "loss": 0.0051,
+      "step": 175950
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154995412204977e-05,
+      "loss": 0.0099,
+      "step": 175960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154947372466506e-05,
+      "loss": 0.0078,
+      "step": 175970
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154899332728036e-05,
+      "loss": 0.0068,
+      "step": 175980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915485129298956e-05,
+      "loss": 0.0078,
+      "step": 175990
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915480325325109e-05,
+      "loss": 0.0059,
+      "step": 176000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915475521351262e-05,
+      "loss": 0.005,
+      "step": 176010
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154707173774146e-05,
+      "loss": 0.0076,
+      "step": 176020
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154659134035675e-05,
+      "loss": 0.0095,
+      "step": 176030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154611094297205e-05,
+      "loss": 0.0068,
+      "step": 176040
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154563054558734e-05,
+      "loss": 0.0088,
+      "step": 176050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915451501482026e-05,
+      "loss": 0.0084,
+      "step": 176060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915446697508179e-05,
+      "loss": 0.007,
+      "step": 176070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154418935343318e-05,
+      "loss": 0.006,
+      "step": 176080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154370895604844e-05,
+      "loss": 0.0088,
+      "step": 176090
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154322855866373e-05,
+      "loss": 0.0099,
+      "step": 176100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154274816127903e-05,
+      "loss": 0.006,
+      "step": 176110
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154226776389432e-05,
+      "loss": 0.0089,
+      "step": 176120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154178736650958e-05,
+      "loss": 0.0089,
+      "step": 176130
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154130696912487e-05,
+      "loss": 0.0048,
+      "step": 176140
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154082657174016e-05,
+      "loss": 0.009,
+      "step": 176150
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9154034617435546e-05,
+      "loss": 0.0048,
+      "step": 176160
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915398657769707e-05,
+      "loss": 0.0104,
+      "step": 176170
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.91539385379586e-05,
+      "loss": 0.0106,
+      "step": 176180
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915389049822013e-05,
+      "loss": 0.0095,
+      "step": 176190
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153842458481656e-05,
+      "loss": 0.0086,
+      "step": 176200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153794418743185e-05,
+      "loss": 0.0066,
+      "step": 176210
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153746379004714e-05,
+      "loss": 0.0071,
+      "step": 176220
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153698339266244e-05,
+      "loss": 0.0051,
+      "step": 176230
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915365029952777e-05,
+      "loss": 0.0111,
+      "step": 176240
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.91536022597893e-05,
+      "loss": 0.0077,
+      "step": 176250
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153554220050828e-05,
+      "loss": 0.0068,
+      "step": 176260
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153506180312354e-05,
+      "loss": 0.0065,
+      "step": 176270
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153458140573883e-05,
+      "loss": 0.0056,
+      "step": 176280
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153410100835413e-05,
+      "loss": 0.0062,
+      "step": 176290
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153362061096942e-05,
+      "loss": 0.0063,
+      "step": 176300
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153314021358468e-05,
+      "loss": 0.0086,
+      "step": 176310
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153265981619997e-05,
+      "loss": 0.0058,
+      "step": 176320
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153217941881526e-05,
+      "loss": 0.0068,
+      "step": 176330
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9153169902143056e-05,
+      "loss": 0.0124,
+      "step": 176340
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915312186240458e-05,
+      "loss": 0.0091,
+      "step": 176350
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915307382266611e-05,
+      "loss": 0.0072,
+      "step": 176360
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915302578292764e-05,
+      "loss": 0.011,
+      "step": 176370
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152977743189166e-05,
+      "loss": 0.0111,
+      "step": 176380
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152929703450695e-05,
+      "loss": 0.011,
+      "step": 176390
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152881663712224e-05,
+      "loss": 0.006,
+      "step": 176400
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152833623973754e-05,
+      "loss": 0.0085,
+      "step": 176410
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915278558423528e-05,
+      "loss": 0.009,
+      "step": 176420
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915273754449681e-05,
+      "loss": 0.0078,
+      "step": 176430
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152689504758338e-05,
+      "loss": 0.0056,
+      "step": 176440
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152641465019864e-05,
+      "loss": 0.007,
+      "step": 176450
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152593425281393e-05,
+      "loss": 0.0081,
+      "step": 176460
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152545385542922e-05,
+      "loss": 0.0047,
+      "step": 176470
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152497345804452e-05,
+      "loss": 0.0094,
+      "step": 176480
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152449306065978e-05,
+      "loss": 0.0059,
+      "step": 176490
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152401266327507e-05,
+      "loss": 0.0068,
+      "step": 176500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152353226589036e-05,
+      "loss": 0.0056,
+      "step": 176510
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152305186850565e-05,
+      "loss": 0.0068,
+      "step": 176520
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915225714711209e-05,
+      "loss": 0.0044,
+      "step": 176530
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915220910737362e-05,
+      "loss": 0.0065,
+      "step": 176540
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.915216106763515e-05,
+      "loss": 0.01,
+      "step": 176550
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152113027896676e-05,
+      "loss": 0.0071,
+      "step": 176560
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152064988158205e-05,
+      "loss": 0.0065,
+      "step": 176570
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9152016948419734e-05,
+      "loss": 0.011,
+      "step": 176580
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151968908681264e-05,
+      "loss": 0.0062,
+      "step": 176590
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.915192086894279e-05,
+      "loss": 0.0062,
+      "step": 176600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.915187282920432e-05,
+      "loss": 0.0048,
+      "step": 176610
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151824789465848e-05,
+      "loss": 0.0073,
+      "step": 176620
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151776749727374e-05,
+      "loss": 0.0074,
+      "step": 176630
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151728709988903e-05,
+      "loss": 0.0068,
+      "step": 176640
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151680670250432e-05,
+      "loss": 0.0118,
+      "step": 176650
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.915163263051196e-05,
+      "loss": 0.0053,
+      "step": 176660
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151584590773488e-05,
+      "loss": 0.0047,
+      "step": 176670
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151536551035017e-05,
+      "loss": 0.0094,
+      "step": 176680
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151488511296546e-05,
+      "loss": 0.0095,
+      "step": 176690
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151440471558075e-05,
+      "loss": 0.0113,
+      "step": 176700
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.915139723579345e-05,
+      "loss": 0.0055,
+      "step": 176710
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.915134919605498e-05,
+      "loss": 0.0064,
+      "step": 176720
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151301156316505e-05,
+      "loss": 0.011,
+      "step": 176730
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151253116578034e-05,
+      "loss": 0.0052,
+      "step": 176740
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151205076839564e-05,
+      "loss": 0.0077,
+      "step": 176750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151157037101093e-05,
+      "loss": 0.0092,
+      "step": 176760
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.915110899736262e-05,
+      "loss": 0.0079,
+      "step": 176770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151060957624148e-05,
+      "loss": 0.0079,
+      "step": 176780
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9151012917885677e-05,
+      "loss": 0.0082,
+      "step": 176790
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150964878147203e-05,
+      "loss": 0.0061,
+      "step": 176800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150916838408733e-05,
+      "loss": 0.0086,
+      "step": 176810
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150868798670262e-05,
+      "loss": 0.0076,
+      "step": 176820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.915082075893179e-05,
+      "loss": 0.007,
+      "step": 176830
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150772719193317e-05,
+      "loss": 0.0059,
+      "step": 176840
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150724679454846e-05,
+      "loss": 0.0084,
+      "step": 176850
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150676639716376e-05,
+      "loss": 0.0057,
+      "step": 176860
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.91506285999779e-05,
+      "loss": 0.0075,
+      "step": 176870
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.915058056023943e-05,
+      "loss": 0.0069,
+      "step": 176880
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.915053252050096e-05,
+      "loss": 0.0086,
+      "step": 176890
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.915048448076249e-05,
+      "loss": 0.0047,
+      "step": 176900
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150436441024015e-05,
+      "loss": 0.0063,
+      "step": 176910
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150388401285544e-05,
+      "loss": 0.0094,
+      "step": 176920
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150340361547074e-05,
+      "loss": 0.0062,
+      "step": 176930
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150292321808603e-05,
+      "loss": 0.0084,
+      "step": 176940
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.915024428207013e-05,
+      "loss": 0.0052,
+      "step": 176950
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150196242331658e-05,
+      "loss": 0.0042,
+      "step": 176960
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150148202593187e-05,
+      "loss": 0.0051,
+      "step": 176970
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150100162854713e-05,
+      "loss": 0.0042,
+      "step": 176980
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150052123116242e-05,
+      "loss": 0.0063,
+      "step": 176990
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9150004083377772e-05,
+      "loss": 0.0085,
+      "step": 177000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.91499560436393e-05,
+      "loss": 0.0057,
+      "step": 177010
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9149908003900827e-05,
+      "loss": 0.0081,
+      "step": 177020
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9149859964162356e-05,
+      "loss": 0.0074,
+      "step": 177030
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9149811924423885e-05,
+      "loss": 0.0076,
+      "step": 177040
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.914976388468541e-05,
+      "loss": 0.0067,
+      "step": 177050
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.914971584494694e-05,
+      "loss": 0.0061,
+      "step": 177060
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.914966780520847e-05,
+      "loss": 0.0084,
+      "step": 177070
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.914961976547e-05,
+      "loss": 0.0055,
+      "step": 177080
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9149571725731525e-05,
+      "loss": 0.0088,
+      "step": 177090
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9149523685993054e-05,
+      "loss": 0.0061,
+      "step": 177100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9149475646254584e-05,
+      "loss": 0.0085,
+      "step": 177110
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.914942760651611e-05,
+      "loss": 0.0066,
+      "step": 177120
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.914937956677764e-05,
+      "loss": 0.0064,
+      "step": 177130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9149331527039168e-05,
+      "loss": 0.0073,
+      "step": 177140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9149283487300697e-05,
+      "loss": 0.0074,
+      "step": 177150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9149235447562223e-05,
+      "loss": 0.0067,
+      "step": 177160
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9149187407823752e-05,
+      "loss": 0.0076,
+      "step": 177170
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.914913936808528e-05,
+      "loss": 0.007,
+      "step": 177180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.914909132834681e-05,
+      "loss": 0.0072,
+      "step": 177190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9149043288608337e-05,
+      "loss": 0.0056,
+      "step": 177200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9148995248869866e-05,
+      "loss": 0.006,
+      "step": 177210
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9148947209131395e-05,
+      "loss": 0.0072,
+      "step": 177220
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.914889916939292e-05,
+      "loss": 0.0093,
+      "step": 177230
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.914885112965445e-05,
+      "loss": 0.0076,
+      "step": 177240
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.914880308991598e-05,
+      "loss": 0.0068,
+      "step": 177250
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.914875505017751e-05,
+      "loss": 0.0073,
+      "step": 177260
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9148707010439035e-05,
+      "loss": 0.005,
+      "step": 177270
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9148658970700564e-05,
+      "loss": 0.0096,
+      "step": 177280
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9148610930962093e-05,
+      "loss": 0.007,
+      "step": 177290
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914856289122362e-05,
+      "loss": 0.0061,
+      "step": 177300
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914851485148515e-05,
+      "loss": 0.0086,
+      "step": 177310
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9148466811746678e-05,
+      "loss": 0.0093,
+      "step": 177320
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9148418772008207e-05,
+      "loss": 0.0052,
+      "step": 177330
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9148370732269733e-05,
+      "loss": 0.0085,
+      "step": 177340
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9148322692531262e-05,
+      "loss": 0.0063,
+      "step": 177350
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914827465279279e-05,
+      "loss": 0.0071,
+      "step": 177360
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914822661305432e-05,
+      "loss": 0.0056,
+      "step": 177370
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9148178573315847e-05,
+      "loss": 0.0083,
+      "step": 177380
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9148130533577376e-05,
+      "loss": 0.0073,
+      "step": 177390
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9148082493838905e-05,
+      "loss": 0.0058,
+      "step": 177400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914803445410043e-05,
+      "loss": 0.0088,
+      "step": 177410
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914798641436196e-05,
+      "loss": 0.0062,
+      "step": 177420
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914793837462349e-05,
+      "loss": 0.0062,
+      "step": 177430
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914789033488502e-05,
+      "loss": 0.006,
+      "step": 177440
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9147842295146545e-05,
+      "loss": 0.0049,
+      "step": 177450
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9147794255408074e-05,
+      "loss": 0.0064,
+      "step": 177460
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9147746215669603e-05,
+      "loss": 0.0068,
+      "step": 177470
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914769817593113e-05,
+      "loss": 0.0086,
+      "step": 177480
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914765013619266e-05,
+      "loss": 0.0076,
+      "step": 177490
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9147602096454188e-05,
+      "loss": 0.0049,
+      "step": 177500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9147554056715717e-05,
+      "loss": 0.0098,
+      "step": 177510
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9147506016977243e-05,
+      "loss": 0.0075,
+      "step": 177520
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9147457977238772e-05,
+      "loss": 0.0052,
+      "step": 177530
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.91474099375003e-05,
+      "loss": 0.01,
+      "step": 177540
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914736189776183e-05,
+      "loss": 0.009,
+      "step": 177550
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9147313858023357e-05,
+      "loss": 0.0095,
+      "step": 177560
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9147265818284886e-05,
+      "loss": 0.0079,
+      "step": 177570
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9147217778546415e-05,
+      "loss": 0.0064,
+      "step": 177580
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914716973880794e-05,
+      "loss": 0.0093,
+      "step": 177590
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914712169906947e-05,
+      "loss": 0.0063,
+      "step": 177600
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9147073659331e-05,
+      "loss": 0.007,
+      "step": 177610
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914702561959253e-05,
+      "loss": 0.0081,
+      "step": 177620
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146977579854055e-05,
+      "loss": 0.0089,
+      "step": 177630
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146929540115584e-05,
+      "loss": 0.0078,
+      "step": 177640
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146881500377113e-05,
+      "loss": 0.0071,
+      "step": 177650
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914683346063864e-05,
+      "loss": 0.0051,
+      "step": 177660
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914678542090017e-05,
+      "loss": 0.0091,
+      "step": 177670
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146737381161698e-05,
+      "loss": 0.0092,
+      "step": 177680
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146689341423227e-05,
+      "loss": 0.0081,
+      "step": 177690
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146641301684753e-05,
+      "loss": 0.0074,
+      "step": 177700
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146593261946282e-05,
+      "loss": 0.0065,
+      "step": 177710
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914654522220781e-05,
+      "loss": 0.0065,
+      "step": 177720
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914649718246934e-05,
+      "loss": 0.0068,
+      "step": 177730
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914644914273087e-05,
+      "loss": 0.0048,
+      "step": 177740
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.91464011029924e-05,
+      "loss": 0.0082,
+      "step": 177750
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146353063253925e-05,
+      "loss": 0.0096,
+      "step": 177760
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146305023515454e-05,
+      "loss": 0.0054,
+      "step": 177770
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146256983776984e-05,
+      "loss": 0.007,
+      "step": 177780
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146208944038513e-05,
+      "loss": 0.0068,
+      "step": 177790
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914616090430004e-05,
+      "loss": 0.009,
+      "step": 177800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146112864561568e-05,
+      "loss": 0.0047,
+      "step": 177810
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146064824823097e-05,
+      "loss": 0.0059,
+      "step": 177820
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9146016785084623e-05,
+      "loss": 0.0042,
+      "step": 177830
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9145968745346152e-05,
+      "loss": 0.0051,
+      "step": 177840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9145920705607682e-05,
+      "loss": 0.0055,
+      "step": 177850
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914587266586921e-05,
+      "loss": 0.0078,
+      "step": 177860
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9145824626130737e-05,
+      "loss": 0.0071,
+      "step": 177870
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9145776586392266e-05,
+      "loss": 0.0061,
+      "step": 177880
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9145728546653795e-05,
+      "loss": 0.0074,
+      "step": 177890
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914568050691532e-05,
+      "loss": 0.0069,
+      "step": 177900
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914563246717685e-05,
+      "loss": 0.008,
+      "step": 177910
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914558442743838e-05,
+      "loss": 0.0073,
+      "step": 177920
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.914553638769991e-05,
+      "loss": 0.0077,
+      "step": 177930
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9145488347961435e-05,
+      "loss": 0.0078,
+      "step": 177940
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9145440308222964e-05,
+      "loss": 0.0073,
+      "step": 177950
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9145392268484494e-05,
+      "loss": 0.0063,
+      "step": 177960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9145344228746023e-05,
+      "loss": 0.0084,
+      "step": 177970
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914529618900755e-05,
+      "loss": 0.0066,
+      "step": 177980
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9145248149269078e-05,
+      "loss": 0.0057,
+      "step": 177990
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9145200109530607e-05,
+      "loss": 0.0091,
+      "step": 178000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9145152069792133e-05,
+      "loss": 0.0058,
+      "step": 178010
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9145104030053662e-05,
+      "loss": 0.0077,
+      "step": 178020
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914505599031519e-05,
+      "loss": 0.0101,
+      "step": 178030
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914500795057672e-05,
+      "loss": 0.0064,
+      "step": 178040
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144959910838247e-05,
+      "loss": 0.0079,
+      "step": 178050
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144911871099776e-05,
+      "loss": 0.0078,
+      "step": 178060
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144863831361305e-05,
+      "loss": 0.0095,
+      "step": 178070
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914481579162283e-05,
+      "loss": 0.0059,
+      "step": 178080
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914476775188436e-05,
+      "loss": 0.006,
+      "step": 178090
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914471971214589e-05,
+      "loss": 0.0064,
+      "step": 178100
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914467167240742e-05,
+      "loss": 0.007,
+      "step": 178110
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144623632668945e-05,
+      "loss": 0.0061,
+      "step": 178120
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144575592930474e-05,
+      "loss": 0.0071,
+      "step": 178130
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144527553192003e-05,
+      "loss": 0.0048,
+      "step": 178140
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144479513453533e-05,
+      "loss": 0.0055,
+      "step": 178150
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914443147371506e-05,
+      "loss": 0.0074,
+      "step": 178160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144383433976588e-05,
+      "loss": 0.0067,
+      "step": 178170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144335394238117e-05,
+      "loss": 0.0048,
+      "step": 178180
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144287354499643e-05,
+      "loss": 0.0068,
+      "step": 178190
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144239314761172e-05,
+      "loss": 0.0083,
+      "step": 178200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.91441912750227e-05,
+      "loss": 0.0082,
+      "step": 178210
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914414323528423e-05,
+      "loss": 0.0077,
+      "step": 178220
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144095195545757e-05,
+      "loss": 0.007,
+      "step": 178230
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9144047155807286e-05,
+      "loss": 0.0099,
+      "step": 178240
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9143999116068815e-05,
+      "loss": 0.008,
+      "step": 178250
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914395107633034e-05,
+      "loss": 0.008,
+      "step": 178260
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914390303659187e-05,
+      "loss": 0.0087,
+      "step": 178270
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.91438549968534e-05,
+      "loss": 0.0079,
+      "step": 178280
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914380695711493e-05,
+      "loss": 0.0068,
+      "step": 178290
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9143758917376455e-05,
+      "loss": 0.0071,
+      "step": 178300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9143710877637984e-05,
+      "loss": 0.0089,
+      "step": 178310
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9143662837899513e-05,
+      "loss": 0.0075,
+      "step": 178320
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9143614798161043e-05,
+      "loss": 0.0053,
+      "step": 178330
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914356675842257e-05,
+      "loss": 0.0072,
+      "step": 178340
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9143518718684098e-05,
+      "loss": 0.0084,
+      "step": 178350
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9143470678945627e-05,
+      "loss": 0.0061,
+      "step": 178360
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9143422639207153e-05,
+      "loss": 0.0093,
+      "step": 178370
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9143374599468682e-05,
+      "loss": 0.009,
+      "step": 178380
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914332655973021e-05,
+      "loss": 0.0066,
+      "step": 178390
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914327851999174e-05,
+      "loss": 0.0077,
+      "step": 178400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9143230480253267e-05,
+      "loss": 0.0082,
+      "step": 178410
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9143182440514796e-05,
+      "loss": 0.0047,
+      "step": 178420
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9143134400776325e-05,
+      "loss": 0.0088,
+      "step": 178430
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914308636103785e-05,
+      "loss": 0.0095,
+      "step": 178440
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914303832129938e-05,
+      "loss": 0.0067,
+      "step": 178450
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914299028156091e-05,
+      "loss": 0.0057,
+      "step": 178460
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914294224182244e-05,
+      "loss": 0.0059,
+      "step": 178470
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9142894202083965e-05,
+      "loss": 0.0066,
+      "step": 178480
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9142846162345494e-05,
+      "loss": 0.0059,
+      "step": 178490
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9142798122607023e-05,
+      "loss": 0.004,
+      "step": 178500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914275008286855e-05,
+      "loss": 0.0055,
+      "step": 178510
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914270204313008e-05,
+      "loss": 0.0063,
+      "step": 178520
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9142654003391608e-05,
+      "loss": 0.008,
+      "step": 178530
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9142605963653137e-05,
+      "loss": 0.0064,
+      "step": 178540
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9142557923914663e-05,
+      "loss": 0.0062,
+      "step": 178550
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9142509884176192e-05,
+      "loss": 0.008,
+      "step": 178560
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914246184443772e-05,
+      "loss": 0.006,
+      "step": 178570
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914241380469925e-05,
+      "loss": 0.0058,
+      "step": 178580
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9142365764960777e-05,
+      "loss": 0.0096,
+      "step": 178590
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9142317725222306e-05,
+      "loss": 0.0085,
+      "step": 178600
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9142269685483835e-05,
+      "loss": 0.0092,
+      "step": 178610
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914222164574536e-05,
+      "loss": 0.0078,
+      "step": 178620
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914217360600689e-05,
+      "loss": 0.0079,
+      "step": 178630
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914212556626842e-05,
+      "loss": 0.0057,
+      "step": 178640
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.914207752652995e-05,
+      "loss": 0.0098,
+      "step": 178650
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9142029486791475e-05,
+      "loss": 0.0053,
+      "step": 178660
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9141981447053004e-05,
+      "loss": 0.0073,
+      "step": 178670
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9141933407314533e-05,
+      "loss": 0.0053,
+      "step": 178680
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.914188536757606e-05,
+      "loss": 0.0067,
+      "step": 178690
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.914183732783759e-05,
+      "loss": 0.0057,
+      "step": 178700
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9141789288099118e-05,
+      "loss": 0.0069,
+      "step": 178710
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9141741248360647e-05,
+      "loss": 0.0063,
+      "step": 178720
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9141693208622173e-05,
+      "loss": 0.0056,
+      "step": 178730
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9141645168883702e-05,
+      "loss": 0.0066,
+      "step": 178740
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.914159712914523e-05,
+      "loss": 0.0068,
+      "step": 178750
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.914154908940676e-05,
+      "loss": 0.0057,
+      "step": 178760
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9141501049668286e-05,
+      "loss": 0.0074,
+      "step": 178770
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9141453009929816e-05,
+      "loss": 0.0102,
+      "step": 178780
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9141404970191345e-05,
+      "loss": 0.0097,
+      "step": 178790
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.914135693045287e-05,
+      "loss": 0.0075,
+      "step": 178800
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.91413088907144e-05,
+      "loss": 0.0052,
+      "step": 178810
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.914126085097593e-05,
+      "loss": 0.0082,
+      "step": 178820
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.914121281123746e-05,
+      "loss": 0.0053,
+      "step": 178830
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9141164771498985e-05,
+      "loss": 0.0104,
+      "step": 178840
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9141116731760514e-05,
+      "loss": 0.0107,
+      "step": 178850
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9141068692022043e-05,
+      "loss": 0.0064,
+      "step": 178860
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.914102065228357e-05,
+      "loss": 0.0092,
+      "step": 178870
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140972612545098e-05,
+      "loss": 0.01,
+      "step": 178880
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140929376780476e-05,
+      "loss": 0.0077,
+      "step": 178890
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140881337042002e-05,
+      "loss": 0.0072,
+      "step": 178900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.914083329730353e-05,
+      "loss": 0.0101,
+      "step": 178910
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.914078525756506e-05,
+      "loss": 0.0066,
+      "step": 178920
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140737217826587e-05,
+      "loss": 0.008,
+      "step": 178930
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140689178088116e-05,
+      "loss": 0.0056,
+      "step": 178940
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140641138349645e-05,
+      "loss": 0.0071,
+      "step": 178950
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140593098611174e-05,
+      "loss": 0.0095,
+      "step": 178960
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.91405450588727e-05,
+      "loss": 0.0074,
+      "step": 178970
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.914049701913423e-05,
+      "loss": 0.0058,
+      "step": 178980
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.914044897939576e-05,
+      "loss": 0.0062,
+      "step": 178990
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140400939657288e-05,
+      "loss": 0.01,
+      "step": 179000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140352899918814e-05,
+      "loss": 0.0049,
+      "step": 179010
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140304860180343e-05,
+      "loss": 0.0077,
+      "step": 179020
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140256820441873e-05,
+      "loss": 0.0056,
+      "step": 179030
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.91402087807034e-05,
+      "loss": 0.0079,
+      "step": 179040
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140160740964928e-05,
+      "loss": 0.0051,
+      "step": 179050
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140112701226457e-05,
+      "loss": 0.007,
+      "step": 179060
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140064661487986e-05,
+      "loss": 0.0116,
+      "step": 179070
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9140016621749512e-05,
+      "loss": 0.0083,
+      "step": 179080
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.913996858201104e-05,
+      "loss": 0.0054,
+      "step": 179090
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.913992054227257e-05,
+      "loss": 0.0071,
+      "step": 179100
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139872502534097e-05,
+      "loss": 0.0091,
+      "step": 179110
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139824462795626e-05,
+      "loss": 0.0072,
+      "step": 179120
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139776423057155e-05,
+      "loss": 0.0061,
+      "step": 179130
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139728383318684e-05,
+      "loss": 0.0086,
+      "step": 179140
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.913968034358021e-05,
+      "loss": 0.0081,
+      "step": 179150
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.913963230384174e-05,
+      "loss": 0.0076,
+      "step": 179160
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.913958426410327e-05,
+      "loss": 0.0129,
+      "step": 179170
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139536224364798e-05,
+      "loss": 0.0052,
+      "step": 179180
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139488184626324e-05,
+      "loss": 0.0067,
+      "step": 179190
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139440144887853e-05,
+      "loss": 0.01,
+      "step": 179200
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139392105149382e-05,
+      "loss": 0.0079,
+      "step": 179210
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.913934406541091e-05,
+      "loss": 0.0091,
+      "step": 179220
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139296025672438e-05,
+      "loss": 0.0076,
+      "step": 179230
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139247985933967e-05,
+      "loss": 0.0102,
+      "step": 179240
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139199946195496e-05,
+      "loss": 0.0102,
+      "step": 179250
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139151906457022e-05,
+      "loss": 0.0063,
+      "step": 179260
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.913910386671855e-05,
+      "loss": 0.0049,
+      "step": 179270
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.913905582698008e-05,
+      "loss": 0.0093,
+      "step": 179280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9139007787241606e-05,
+      "loss": 0.0072,
+      "step": 179290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9138959747503136e-05,
+      "loss": 0.0132,
+      "step": 179300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9138911707764665e-05,
+      "loss": 0.0061,
+      "step": 179310
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9138863668026194e-05,
+      "loss": 0.0063,
+      "step": 179320
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.913881562828772e-05,
+      "loss": 0.0092,
+      "step": 179330
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.913876758854925e-05,
+      "loss": 0.0092,
+      "step": 179340
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.913871954881078e-05,
+      "loss": 0.0077,
+      "step": 179350
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9138671509072308e-05,
+      "loss": 0.009,
+      "step": 179360
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9138623469333834e-05,
+      "loss": 0.0065,
+      "step": 179370
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9138575429595363e-05,
+      "loss": 0.01,
+      "step": 179380
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9138527389856892e-05,
+      "loss": 0.0082,
+      "step": 179390
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9138479350118418e-05,
+      "loss": 0.0038,
+      "step": 179400
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9138431310379948e-05,
+      "loss": 0.0076,
+      "step": 179410
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9138383270641477e-05,
+      "loss": 0.0087,
+      "step": 179420
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9138335230903006e-05,
+      "loss": 0.0074,
+      "step": 179430
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9138287191164532e-05,
+      "loss": 0.0052,
+      "step": 179440
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913823915142606e-05,
+      "loss": 0.0062,
+      "step": 179450
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913819111168759e-05,
+      "loss": 0.0081,
+      "step": 179460
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9138143071949116e-05,
+      "loss": 0.0078,
+      "step": 179470
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9138095032210646e-05,
+      "loss": 0.0064,
+      "step": 179480
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9138046992472175e-05,
+      "loss": 0.0047,
+      "step": 179490
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137998952733704e-05,
+      "loss": 0.0083,
+      "step": 179500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913795091299523e-05,
+      "loss": 0.0083,
+      "step": 179510
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913790287325676e-05,
+      "loss": 0.0085,
+      "step": 179520
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913785483351829e-05,
+      "loss": 0.0065,
+      "step": 179530
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137806793779818e-05,
+      "loss": 0.0073,
+      "step": 179540
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137758754041344e-05,
+      "loss": 0.0048,
+      "step": 179550
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137710714302873e-05,
+      "loss": 0.0115,
+      "step": 179560
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137662674564402e-05,
+      "loss": 0.0075,
+      "step": 179570
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137614634825928e-05,
+      "loss": 0.0078,
+      "step": 179580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137566595087457e-05,
+      "loss": 0.0091,
+      "step": 179590
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137518555348987e-05,
+      "loss": 0.0096,
+      "step": 179600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137470515610516e-05,
+      "loss": 0.0094,
+      "step": 179610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137422475872042e-05,
+      "loss": 0.0083,
+      "step": 179620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913737443613357e-05,
+      "loss": 0.0056,
+      "step": 179630
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.91373263963951e-05,
+      "loss": 0.0077,
+      "step": 179640
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137278356656626e-05,
+      "loss": 0.0065,
+      "step": 179650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137230316918156e-05,
+      "loss": 0.0069,
+      "step": 179660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137182277179685e-05,
+      "loss": 0.0094,
+      "step": 179670
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9137134237441214e-05,
+      "loss": 0.0088,
+      "step": 179680
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913708619770274e-05,
+      "loss": 0.0048,
+      "step": 179690
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913703815796427e-05,
+      "loss": 0.0064,
+      "step": 179700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.91369901182258e-05,
+      "loss": 0.0083,
+      "step": 179710
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136942078487328e-05,
+      "loss": 0.007,
+      "step": 179720
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136894038748854e-05,
+      "loss": 0.0083,
+      "step": 179730
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136845999010383e-05,
+      "loss": 0.0084,
+      "step": 179740
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136797959271912e-05,
+      "loss": 0.0058,
+      "step": 179750
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136749919533438e-05,
+      "loss": 0.0135,
+      "step": 179760
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136701879794967e-05,
+      "loss": 0.0056,
+      "step": 179770
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136653840056497e-05,
+      "loss": 0.0082,
+      "step": 179780
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136605800318026e-05,
+      "loss": 0.0058,
+      "step": 179790
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136557760579552e-05,
+      "loss": 0.0053,
+      "step": 179800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913650972084108e-05,
+      "loss": 0.0063,
+      "step": 179810
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913646168110261e-05,
+      "loss": 0.0062,
+      "step": 179820
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136413641364136e-05,
+      "loss": 0.0074,
+      "step": 179830
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136365601625665e-05,
+      "loss": 0.0066,
+      "step": 179840
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136317561887195e-05,
+      "loss": 0.0081,
+      "step": 179850
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136269522148724e-05,
+      "loss": 0.0069,
+      "step": 179860
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913622148241025e-05,
+      "loss": 0.0071,
+      "step": 179870
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913617344267178e-05,
+      "loss": 0.0075,
+      "step": 179880
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913612540293331e-05,
+      "loss": 0.0064,
+      "step": 179890
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136077363194838e-05,
+      "loss": 0.0086,
+      "step": 179900
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9136029323456364e-05,
+      "loss": 0.0062,
+      "step": 179910
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9135981283717893e-05,
+      "loss": 0.006,
+      "step": 179920
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9135933243979422e-05,
+      "loss": 0.0092,
+      "step": 179930
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9135885204240948e-05,
+      "loss": 0.0087,
+      "step": 179940
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9135837164502477e-05,
+      "loss": 0.0078,
+      "step": 179950
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9135789124764007e-05,
+      "loss": 0.0048,
+      "step": 179960
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9135741085025536e-05,
+      "loss": 0.0051,
+      "step": 179970
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913569304528706e-05,
+      "loss": 0.0073,
+      "step": 179980
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913564500554859e-05,
+      "loss": 0.0049,
+      "step": 179990
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913559696581012e-05,
+      "loss": 0.008,
+      "step": 180000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9135548926071646e-05,
+      "loss": 0.0068,
+      "step": 180010
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9135500886333175e-05,
+      "loss": 0.0101,
+      "step": 180020
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9135452846594705e-05,
+      "loss": 0.0071,
+      "step": 180030
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9135404806856234e-05,
+      "loss": 0.0048,
+      "step": 180040
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.913535676711776e-05,
+      "loss": 0.01,
+      "step": 180050
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913530872737929e-05,
+      "loss": 0.0073,
+      "step": 180060
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913526068764082e-05,
+      "loss": 0.0074,
+      "step": 180070
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9135212647902344e-05,
+      "loss": 0.0074,
+      "step": 180080
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9135164608163873e-05,
+      "loss": 0.0083,
+      "step": 180090
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9135116568425403e-05,
+      "loss": 0.0068,
+      "step": 180100
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9135068528686932e-05,
+      "loss": 0.006,
+      "step": 180110
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9135020488948458e-05,
+      "loss": 0.0061,
+      "step": 180120
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134972449209987e-05,
+      "loss": 0.0075,
+      "step": 180130
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134924409471516e-05,
+      "loss": 0.0067,
+      "step": 180140
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134876369733046e-05,
+      "loss": 0.0064,
+      "step": 180150
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913482832999457e-05,
+      "loss": 0.0065,
+      "step": 180160
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.91347802902561e-05,
+      "loss": 0.0043,
+      "step": 180170
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913473225051763e-05,
+      "loss": 0.0073,
+      "step": 180180
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134684210779156e-05,
+      "loss": 0.0099,
+      "step": 180190
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134636171040685e-05,
+      "loss": 0.0061,
+      "step": 180200
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134588131302215e-05,
+      "loss": 0.0085,
+      "step": 180210
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134540091563744e-05,
+      "loss": 0.0063,
+      "step": 180220
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913449205182527e-05,
+      "loss": 0.0053,
+      "step": 180230
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.91344440120868e-05,
+      "loss": 0.0087,
+      "step": 180240
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134395972348328e-05,
+      "loss": 0.0083,
+      "step": 180250
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134347932609854e-05,
+      "loss": 0.0093,
+      "step": 180260
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134299892871383e-05,
+      "loss": 0.0052,
+      "step": 180270
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134251853132913e-05,
+      "loss": 0.0048,
+      "step": 180280
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134203813394442e-05,
+      "loss": 0.009,
+      "step": 180290
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134155773655968e-05,
+      "loss": 0.0046,
+      "step": 180300
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134107733917497e-05,
+      "loss": 0.0056,
+      "step": 180310
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134059694179026e-05,
+      "loss": 0.005,
+      "step": 180320
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9134011654440556e-05,
+      "loss": 0.0049,
+      "step": 180330
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913396361470208e-05,
+      "loss": 0.0045,
+      "step": 180340
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913391557496361e-05,
+      "loss": 0.0073,
+      "step": 180350
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913386753522514e-05,
+      "loss": 0.0097,
+      "step": 180360
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133819495486666e-05,
+      "loss": 0.0096,
+      "step": 180370
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133771455748195e-05,
+      "loss": 0.0063,
+      "step": 180380
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133723416009724e-05,
+      "loss": 0.0062,
+      "step": 180390
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133675376271254e-05,
+      "loss": 0.009,
+      "step": 180400
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913362733653278e-05,
+      "loss": 0.0043,
+      "step": 180410
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913357929679431e-05,
+      "loss": 0.0087,
+      "step": 180420
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133531257055838e-05,
+      "loss": 0.0056,
+      "step": 180430
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133483217317364e-05,
+      "loss": 0.0064,
+      "step": 180440
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133435177578893e-05,
+      "loss": 0.0056,
+      "step": 180450
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133387137840423e-05,
+      "loss": 0.005,
+      "step": 180460
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133339098101952e-05,
+      "loss": 0.0122,
+      "step": 180470
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133291058363478e-05,
+      "loss": 0.0068,
+      "step": 180480
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133243018625007e-05,
+      "loss": 0.0085,
+      "step": 180490
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133194978886536e-05,
+      "loss": 0.0086,
+      "step": 180500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9133146939148066e-05,
+      "loss": 0.009,
+      "step": 180510
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913309889940959e-05,
+      "loss": 0.0069,
+      "step": 180520
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913305085967112e-05,
+      "loss": 0.0101,
+      "step": 180530
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913300281993265e-05,
+      "loss": 0.007,
+      "step": 180540
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132954780194176e-05,
+      "loss": 0.0066,
+      "step": 180550
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132906740455705e-05,
+      "loss": 0.0056,
+      "step": 180560
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132858700717234e-05,
+      "loss": 0.0103,
+      "step": 180570
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132810660978764e-05,
+      "loss": 0.007,
+      "step": 180580
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913276262124029e-05,
+      "loss": 0.0079,
+      "step": 180590
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913271458150182e-05,
+      "loss": 0.006,
+      "step": 180600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132666541763348e-05,
+      "loss": 0.007,
+      "step": 180610
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132618502024874e-05,
+      "loss": 0.0163,
+      "step": 180620
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132570462286403e-05,
+      "loss": 0.0066,
+      "step": 180630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132522422547932e-05,
+      "loss": 0.0069,
+      "step": 180640
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132474382809462e-05,
+      "loss": 0.0065,
+      "step": 180650
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132426343070988e-05,
+      "loss": 0.0065,
+      "step": 180660
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132378303332517e-05,
+      "loss": 0.0062,
+      "step": 180670
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132330263594046e-05,
+      "loss": 0.0083,
+      "step": 180680
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132282223855575e-05,
+      "loss": 0.0064,
+      "step": 180690
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.91322341841171e-05,
+      "loss": 0.0059,
+      "step": 180700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913218614437863e-05,
+      "loss": 0.0068,
+      "step": 180710
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.913213810464016e-05,
+      "loss": 0.0082,
+      "step": 180720
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132090064901686e-05,
+      "loss": 0.0055,
+      "step": 180730
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9132042025163215e-05,
+      "loss": 0.008,
+      "step": 180740
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9131993985424744e-05,
+      "loss": 0.0063,
+      "step": 180750
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9131945945686274e-05,
+      "loss": 0.0071,
+      "step": 180760
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.91318979059478e-05,
+      "loss": 0.0065,
+      "step": 180770
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.913184986620933e-05,
+      "loss": 0.003,
+      "step": 180780
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9131801826470858e-05,
+      "loss": 0.0085,
+      "step": 180790
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9131753786732387e-05,
+      "loss": 0.0073,
+      "step": 180800
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9131705746993917e-05,
+      "loss": 0.0075,
+      "step": 180810
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9131657707255446e-05,
+      "loss": 0.0043,
+      "step": 180820
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.913160966751697e-05,
+      "loss": 0.0062,
+      "step": 180830
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.91315616277785e-05,
+      "loss": 0.0053,
+      "step": 180840
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.913151358804003e-05,
+      "loss": 0.0073,
+      "step": 180850
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9131465548301556e-05,
+      "loss": 0.005,
+      "step": 180860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9131417508563085e-05,
+      "loss": 0.0095,
+      "step": 180870
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9131369468824615e-05,
+      "loss": 0.0076,
+      "step": 180880
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9131321429086144e-05,
+      "loss": 0.0059,
+      "step": 180890
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.913127338934767e-05,
+      "loss": 0.0087,
+      "step": 180900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.91312253496092e-05,
+      "loss": 0.0057,
+      "step": 180910
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.913117730987073e-05,
+      "loss": 0.009,
+      "step": 180920
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9131129270132258e-05,
+      "loss": 0.0071,
+      "step": 180930
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9131081230393783e-05,
+      "loss": 0.0062,
+      "step": 180940
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9131033190655313e-05,
+      "loss": 0.0073,
+      "step": 180950
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130985150916842e-05,
+      "loss": 0.0075,
+      "step": 180960
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130937111178368e-05,
+      "loss": 0.006,
+      "step": 180970
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130889071439897e-05,
+      "loss": 0.0074,
+      "step": 180980
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130841031701426e-05,
+      "loss": 0.0058,
+      "step": 180990
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130792991962956e-05,
+      "loss": 0.0073,
+      "step": 181000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.913074495222448e-05,
+      "loss": 0.0084,
+      "step": 181010
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.913069691248601e-05,
+      "loss": 0.0054,
+      "step": 181020
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.913064887274754e-05,
+      "loss": 0.0086,
+      "step": 181030
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130600833009066e-05,
+      "loss": 0.0088,
+      "step": 181040
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130552793270595e-05,
+      "loss": 0.0064,
+      "step": 181050
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130504753532125e-05,
+      "loss": 0.0089,
+      "step": 181060
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130456713793654e-05,
+      "loss": 0.0059,
+      "step": 181070
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.913040867405518e-05,
+      "loss": 0.0068,
+      "step": 181080
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.913036063431671e-05,
+      "loss": 0.0058,
+      "step": 181090
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130312594578238e-05,
+      "loss": 0.0105,
+      "step": 181100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130264554839768e-05,
+      "loss": 0.0078,
+      "step": 181110
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130216515101293e-05,
+      "loss": 0.0089,
+      "step": 181120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130168475362823e-05,
+      "loss": 0.0065,
+      "step": 181130
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130120435624352e-05,
+      "loss": 0.0066,
+      "step": 181140
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130072395885878e-05,
+      "loss": 0.0103,
+      "step": 181150
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9130024356147407e-05,
+      "loss": 0.007,
+      "step": 181160
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129976316408936e-05,
+      "loss": 0.0045,
+      "step": 181170
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129928276670466e-05,
+      "loss": 0.0071,
+      "step": 181180
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.912988023693199e-05,
+      "loss": 0.0082,
+      "step": 181190
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.912983219719352e-05,
+      "loss": 0.008,
+      "step": 181200
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.912978415745505e-05,
+      "loss": 0.0127,
+      "step": 181210
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129736117716576e-05,
+      "loss": 0.0045,
+      "step": 181220
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129688077978105e-05,
+      "loss": 0.0089,
+      "step": 181230
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129640038239634e-05,
+      "loss": 0.0059,
+      "step": 181240
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129591998501164e-05,
+      "loss": 0.0069,
+      "step": 181250
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.912954395876269e-05,
+      "loss": 0.0065,
+      "step": 181260
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.912949591902422e-05,
+      "loss": 0.0057,
+      "step": 181270
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129447879285748e-05,
+      "loss": 0.006,
+      "step": 181280
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129399839547277e-05,
+      "loss": 0.0057,
+      "step": 181290
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129351799808803e-05,
+      "loss": 0.0072,
+      "step": 181300
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129303760070333e-05,
+      "loss": 0.0077,
+      "step": 181310
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129255720331862e-05,
+      "loss": 0.0065,
+      "step": 181320
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129207680593388e-05,
+      "loss": 0.0049,
+      "step": 181330
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129159640854917e-05,
+      "loss": 0.0065,
+      "step": 181340
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129111601116446e-05,
+      "loss": 0.0099,
+      "step": 181350
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9129063561377976e-05,
+      "loss": 0.0077,
+      "step": 181360
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.91290155216395e-05,
+      "loss": 0.0107,
+      "step": 181370
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.912896748190103e-05,
+      "loss": 0.0058,
+      "step": 181380
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.912891944216256e-05,
+      "loss": 0.0069,
+      "step": 181390
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9128871402424086e-05,
+      "loss": 0.0056,
+      "step": 181400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9128823362685615e-05,
+      "loss": 0.0081,
+      "step": 181410
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9128775322947144e-05,
+      "loss": 0.0058,
+      "step": 181420
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9128727283208674e-05,
+      "loss": 0.0078,
+      "step": 181430
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.91286792434702e-05,
+      "loss": 0.0071,
+      "step": 181440
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912863120373173e-05,
+      "loss": 0.0071,
+      "step": 181450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9128583163993258e-05,
+      "loss": 0.0081,
+      "step": 181460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9128535124254784e-05,
+      "loss": 0.0062,
+      "step": 181470
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9128487084516313e-05,
+      "loss": 0.008,
+      "step": 181480
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9128439044777842e-05,
+      "loss": 0.0074,
+      "step": 181490
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9128391005039372e-05,
+      "loss": 0.0109,
+      "step": 181500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9128342965300898e-05,
+      "loss": 0.0077,
+      "step": 181510
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9128294925562427e-05,
+      "loss": 0.0063,
+      "step": 181520
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9128246885823956e-05,
+      "loss": 0.0055,
+      "step": 181530
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9128198846085485e-05,
+      "loss": 0.0069,
+      "step": 181540
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912815080634701e-05,
+      "loss": 0.0058,
+      "step": 181550
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912810276660854e-05,
+      "loss": 0.0063,
+      "step": 181560
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912805472687007e-05,
+      "loss": 0.0066,
+      "step": 181570
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9128006687131596e-05,
+      "loss": 0.0065,
+      "step": 181580
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127958647393125e-05,
+      "loss": 0.0059,
+      "step": 181590
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127910607654654e-05,
+      "loss": 0.0082,
+      "step": 181600
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127862567916184e-05,
+      "loss": 0.007,
+      "step": 181610
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912781452817771e-05,
+      "loss": 0.0061,
+      "step": 181620
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912776648843924e-05,
+      "loss": 0.0073,
+      "step": 181630
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127718448700768e-05,
+      "loss": 0.0062,
+      "step": 181640
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127670408962294e-05,
+      "loss": 0.007,
+      "step": 181650
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127622369223823e-05,
+      "loss": 0.0086,
+      "step": 181660
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127574329485352e-05,
+      "loss": 0.0075,
+      "step": 181670
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912752628974688e-05,
+      "loss": 0.0056,
+      "step": 181680
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127478250008408e-05,
+      "loss": 0.0091,
+      "step": 181690
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127430210269937e-05,
+      "loss": 0.009,
+      "step": 181700
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127382170531466e-05,
+      "loss": 0.0075,
+      "step": 181710
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127334130792995e-05,
+      "loss": 0.0066,
+      "step": 181720
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912728609105452e-05,
+      "loss": 0.0061,
+      "step": 181730
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912723805131605e-05,
+      "loss": 0.0065,
+      "step": 181740
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912719001157758e-05,
+      "loss": 0.0061,
+      "step": 181750
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127141971839106e-05,
+      "loss": 0.0087,
+      "step": 181760
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127093932100635e-05,
+      "loss": 0.0065,
+      "step": 181770
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127045892362164e-05,
+      "loss": 0.0081,
+      "step": 181780
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126997852623693e-05,
+      "loss": 0.0067,
+      "step": 181790
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912694981288522e-05,
+      "loss": 0.0062,
+      "step": 181800
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912690177314675e-05,
+      "loss": 0.0059,
+      "step": 181810
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126853733408278e-05,
+      "loss": 0.0064,
+      "step": 181820
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126805693669804e-05,
+      "loss": 0.007,
+      "step": 181830
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126757653931333e-05,
+      "loss": 0.0068,
+      "step": 181840
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126709614192862e-05,
+      "loss": 0.0068,
+      "step": 181850
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912666157445439e-05,
+      "loss": 0.01,
+      "step": 181860
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126613534715917e-05,
+      "loss": 0.0081,
+      "step": 181870
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126565494977447e-05,
+      "loss": 0.0078,
+      "step": 181880
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126517455238976e-05,
+      "loss": 0.0054,
+      "step": 181890
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126469415500505e-05,
+      "loss": 0.0073,
+      "step": 181900
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912642137576203e-05,
+      "loss": 0.0055,
+      "step": 181910
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912637333602356e-05,
+      "loss": 0.007,
+      "step": 181920
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912632529628509e-05,
+      "loss": 0.0087,
+      "step": 181930
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126277256546616e-05,
+      "loss": 0.0072,
+      "step": 181940
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126229216808145e-05,
+      "loss": 0.0081,
+      "step": 181950
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126181177069674e-05,
+      "loss": 0.0051,
+      "step": 181960
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9126133137331203e-05,
+      "loss": 0.0119,
+      "step": 181970
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912608509759273e-05,
+      "loss": 0.0077,
+      "step": 181980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912603705785426e-05,
+      "loss": 0.0072,
+      "step": 181990
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9125989018115788e-05,
+      "loss": 0.0066,
+      "step": 182000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9125940978377314e-05,
+      "loss": 0.0076,
+      "step": 182010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9125892938638843e-05,
+      "loss": 0.0067,
+      "step": 182020
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9125844898900372e-05,
+      "loss": 0.0089,
+      "step": 182030
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.91257968591619e-05,
+      "loss": 0.0104,
+      "step": 182040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9125748819423427e-05,
+      "loss": 0.0078,
+      "step": 182050
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9125700779684957e-05,
+      "loss": 0.0049,
+      "step": 182060
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9125652739946486e-05,
+      "loss": 0.0058,
+      "step": 182070
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9125604700208015e-05,
+      "loss": 0.0093,
+      "step": 182080
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912555666046954e-05,
+      "loss": 0.0079,
+      "step": 182090
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912550862073107e-05,
+      "loss": 0.0068,
+      "step": 182100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.91254605809926e-05,
+      "loss": 0.0084,
+      "step": 182110
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9125412541254126e-05,
+      "loss": 0.0081,
+      "step": 182120
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9125364501515655e-05,
+      "loss": 0.0049,
+      "step": 182130
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9125316461777184e-05,
+      "loss": 0.0103,
+      "step": 182140
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9125268422038713e-05,
+      "loss": 0.0063,
+      "step": 182150
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912522038230024e-05,
+      "loss": 0.0082,
+      "step": 182160
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912517234256177e-05,
+      "loss": 0.0085,
+      "step": 182170
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9125124302823298e-05,
+      "loss": 0.0111,
+      "step": 182180
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9125076263084824e-05,
+      "loss": 0.0069,
+      "step": 182190
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9125028223346353e-05,
+      "loss": 0.0092,
+      "step": 182200
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124980183607882e-05,
+      "loss": 0.0074,
+      "step": 182210
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912493214386941e-05,
+      "loss": 0.0082,
+      "step": 182220
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124884104130937e-05,
+      "loss": 0.004,
+      "step": 182230
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124836064392467e-05,
+      "loss": 0.0106,
+      "step": 182240
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124788024653996e-05,
+      "loss": 0.0092,
+      "step": 182250
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124739984915525e-05,
+      "loss": 0.0075,
+      "step": 182260
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912469194517705e-05,
+      "loss": 0.0117,
+      "step": 182270
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912464390543858e-05,
+      "loss": 0.0069,
+      "step": 182280
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912459586570011e-05,
+      "loss": 0.0073,
+      "step": 182290
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124547825961635e-05,
+      "loss": 0.0088,
+      "step": 182300
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124499786223165e-05,
+      "loss": 0.0075,
+      "step": 182310
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124451746484694e-05,
+      "loss": 0.0064,
+      "step": 182320
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124403706746223e-05,
+      "loss": 0.0063,
+      "step": 182330
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912435566700775e-05,
+      "loss": 0.0071,
+      "step": 182340
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912430762726928e-05,
+      "loss": 0.0052,
+      "step": 182350
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124259587530808e-05,
+      "loss": 0.0062,
+      "step": 182360
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124211547792334e-05,
+      "loss": 0.0054,
+      "step": 182370
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124163508053863e-05,
+      "loss": 0.0069,
+      "step": 182380
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124115468315392e-05,
+      "loss": 0.0085,
+      "step": 182390
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912406742857692e-05,
+      "loss": 0.0073,
+      "step": 182400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9124019388838447e-05,
+      "loss": 0.0046,
+      "step": 182410
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123971349099976e-05,
+      "loss": 0.0062,
+      "step": 182420
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123923309361506e-05,
+      "loss": 0.0053,
+      "step": 182430
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123875269623035e-05,
+      "loss": 0.0064,
+      "step": 182440
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912382722988456e-05,
+      "loss": 0.0058,
+      "step": 182450
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912377919014609e-05,
+      "loss": 0.0111,
+      "step": 182460
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912373115040762e-05,
+      "loss": 0.0089,
+      "step": 182470
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123683110669145e-05,
+      "loss": 0.007,
+      "step": 182480
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123635070930675e-05,
+      "loss": 0.009,
+      "step": 182490
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123587031192204e-05,
+      "loss": 0.0057,
+      "step": 182500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123538991453733e-05,
+      "loss": 0.0079,
+      "step": 182510
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912349095171526e-05,
+      "loss": 0.0084,
+      "step": 182520
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123442911976788e-05,
+      "loss": 0.0047,
+      "step": 182530
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123394872238318e-05,
+      "loss": 0.008,
+      "step": 182540
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123346832499843e-05,
+      "loss": 0.0064,
+      "step": 182550
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123298792761373e-05,
+      "loss": 0.0069,
+      "step": 182560
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123250753022902e-05,
+      "loss": 0.0082,
+      "step": 182570
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912320271328443e-05,
+      "loss": 0.0041,
+      "step": 182580
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123154673545957e-05,
+      "loss": 0.0084,
+      "step": 182590
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123106633807486e-05,
+      "loss": 0.0058,
+      "step": 182600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123058594069016e-05,
+      "loss": 0.009,
+      "step": 182610
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9123010554330545e-05,
+      "loss": 0.006,
+      "step": 182620
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912296251459207e-05,
+      "loss": 0.0063,
+      "step": 182630
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.91229144748536e-05,
+      "loss": 0.006,
+      "step": 182640
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912286643511513e-05,
+      "loss": 0.0036,
+      "step": 182650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122818395376655e-05,
+      "loss": 0.0088,
+      "step": 182660
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122770355638185e-05,
+      "loss": 0.0076,
+      "step": 182670
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122722315899714e-05,
+      "loss": 0.0074,
+      "step": 182680
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122674276161243e-05,
+      "loss": 0.0082,
+      "step": 182690
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912262623642277e-05,
+      "loss": 0.0074,
+      "step": 182700
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122578196684298e-05,
+      "loss": 0.0109,
+      "step": 182710
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122530156945827e-05,
+      "loss": 0.0062,
+      "step": 182720
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122482117207353e-05,
+      "loss": 0.0059,
+      "step": 182730
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122434077468883e-05,
+      "loss": 0.0091,
+      "step": 182740
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122386037730412e-05,
+      "loss": 0.009,
+      "step": 182750
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912233799799194e-05,
+      "loss": 0.009,
+      "step": 182760
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122289958253467e-05,
+      "loss": 0.0093,
+      "step": 182770
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122241918514996e-05,
+      "loss": 0.0073,
+      "step": 182780
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122193878776526e-05,
+      "loss": 0.0065,
+      "step": 182790
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9122145839038055e-05,
+      "loss": 0.0083,
+      "step": 182800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912209779929958e-05,
+      "loss": 0.0074,
+      "step": 182810
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912204975956111e-05,
+      "loss": 0.0073,
+      "step": 182820
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.912200171982264e-05,
+      "loss": 0.0063,
+      "step": 182830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121953680084165e-05,
+      "loss": 0.0072,
+      "step": 182840
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121905640345694e-05,
+      "loss": 0.008,
+      "step": 182850
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121857600607224e-05,
+      "loss": 0.0074,
+      "step": 182860
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121809560868753e-05,
+      "loss": 0.007,
+      "step": 182870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.912176152113028e-05,
+      "loss": 0.0096,
+      "step": 182880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121713481391808e-05,
+      "loss": 0.0088,
+      "step": 182890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121665441653337e-05,
+      "loss": 0.0067,
+      "step": 182900
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121617401914863e-05,
+      "loss": 0.0106,
+      "step": 182910
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121569362176393e-05,
+      "loss": 0.0047,
+      "step": 182920
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121521322437922e-05,
+      "loss": 0.009,
+      "step": 182930
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.912147328269945e-05,
+      "loss": 0.0067,
+      "step": 182940
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121425242960977e-05,
+      "loss": 0.0097,
+      "step": 182950
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121377203222506e-05,
+      "loss": 0.0088,
+      "step": 182960
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121329163484036e-05,
+      "loss": 0.0079,
+      "step": 182970
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121281123745565e-05,
+      "loss": 0.0064,
+      "step": 182980
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.912123308400709e-05,
+      "loss": 0.0087,
+      "step": 182990
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.912118504426862e-05,
+      "loss": 0.0064,
+      "step": 183000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.912113700453015e-05,
+      "loss": 0.009,
+      "step": 183010
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121088964791675e-05,
+      "loss": 0.0081,
+      "step": 183020
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9121040925053204e-05,
+      "loss": 0.0077,
+      "step": 183030
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120992885314734e-05,
+      "loss": 0.0075,
+      "step": 183040
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120944845576263e-05,
+      "loss": 0.0075,
+      "step": 183050
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.912089680583779e-05,
+      "loss": 0.005,
+      "step": 183060
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120848766099318e-05,
+      "loss": 0.0075,
+      "step": 183070
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120800726360847e-05,
+      "loss": 0.0065,
+      "step": 183080
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120752686622373e-05,
+      "loss": 0.0081,
+      "step": 183090
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120704646883902e-05,
+      "loss": 0.0073,
+      "step": 183100
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120656607145432e-05,
+      "loss": 0.0075,
+      "step": 183110
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.912060856740696e-05,
+      "loss": 0.0073,
+      "step": 183120
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120560527668487e-05,
+      "loss": 0.0079,
+      "step": 183130
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120512487930016e-05,
+      "loss": 0.0058,
+      "step": 183140
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120464448191545e-05,
+      "loss": 0.008,
+      "step": 183150
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120416408453075e-05,
+      "loss": 0.0074,
+      "step": 183160
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.91203683687146e-05,
+      "loss": 0.0085,
+      "step": 183170
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.912032032897613e-05,
+      "loss": 0.0091,
+      "step": 183180
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.912027228923766e-05,
+      "loss": 0.0054,
+      "step": 183190
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120224249499185e-05,
+      "loss": 0.0069,
+      "step": 183200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120176209760714e-05,
+      "loss": 0.0057,
+      "step": 183210
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120128170022244e-05,
+      "loss": 0.0078,
+      "step": 183220
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9120080130283773e-05,
+      "loss": 0.0076,
+      "step": 183230
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.91200320905453e-05,
+      "loss": 0.0092,
+      "step": 183240
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119984050806828e-05,
+      "loss": 0.0075,
+      "step": 183250
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119936011068357e-05,
+      "loss": 0.0076,
+      "step": 183260
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119887971329883e-05,
+      "loss": 0.0057,
+      "step": 183270
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119839931591412e-05,
+      "loss": 0.0089,
+      "step": 183280
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.911979189185294e-05,
+      "loss": 0.0074,
+      "step": 183290
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.911974385211447e-05,
+      "loss": 0.0045,
+      "step": 183300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119695812375997e-05,
+      "loss": 0.0074,
+      "step": 183310
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119647772637526e-05,
+      "loss": 0.008,
+      "step": 183320
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119599732899055e-05,
+      "loss": 0.0053,
+      "step": 183330
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.911955169316058e-05,
+      "loss": 0.0071,
+      "step": 183340
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.911950365342211e-05,
+      "loss": 0.0083,
+      "step": 183350
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119455613683643e-05,
+      "loss": 0.0068,
+      "step": 183360
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.911940757394517e-05,
+      "loss": 0.0067,
+      "step": 183370
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119359534206698e-05,
+      "loss": 0.0066,
+      "step": 183380
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119311494468228e-05,
+      "loss": 0.0063,
+      "step": 183390
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119263454729753e-05,
+      "loss": 0.0058,
+      "step": 183400
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119215414991283e-05,
+      "loss": 0.0076,
+      "step": 183410
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119167375252812e-05,
+      "loss": 0.0057,
+      "step": 183420
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.911911933551434e-05,
+      "loss": 0.006,
+      "step": 183430
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119071295775867e-05,
+      "loss": 0.0061,
+      "step": 183440
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9119023256037396e-05,
+      "loss": 0.0047,
+      "step": 183450
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9118975216298926e-05,
+      "loss": 0.0065,
+      "step": 183460
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9118927176560455e-05,
+      "loss": 0.0085,
+      "step": 183470
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.911887913682198e-05,
+      "loss": 0.0082,
+      "step": 183480
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.911883109708351e-05,
+      "loss": 0.0087,
+      "step": 183490
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.911878305734504e-05,
+      "loss": 0.0067,
+      "step": 183500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9118735017606565e-05,
+      "loss": 0.0069,
+      "step": 183510
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9118686977868095e-05,
+      "loss": 0.0077,
+      "step": 183520
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9118638938129624e-05,
+      "loss": 0.0075,
+      "step": 183530
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9118590898391153e-05,
+      "loss": 0.0074,
+      "step": 183540
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911854285865268e-05,
+      "loss": 0.0081,
+      "step": 183550
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9118494818914208e-05,
+      "loss": 0.011,
+      "step": 183560
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9118446779175737e-05,
+      "loss": 0.0076,
+      "step": 183570
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9118398739437263e-05,
+      "loss": 0.0079,
+      "step": 183580
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9118350699698793e-05,
+      "loss": 0.0078,
+      "step": 183590
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9118302659960322e-05,
+      "loss": 0.0062,
+      "step": 183600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911825462022185e-05,
+      "loss": 0.0074,
+      "step": 183610
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9118206580483377e-05,
+      "loss": 0.0078,
+      "step": 183620
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9118158540744906e-05,
+      "loss": 0.0061,
+      "step": 183630
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9118110501006436e-05,
+      "loss": 0.0079,
+      "step": 183640
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9118062461267965e-05,
+      "loss": 0.0074,
+      "step": 183650
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911801442152949e-05,
+      "loss": 0.0073,
+      "step": 183660
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911796638179102e-05,
+      "loss": 0.0075,
+      "step": 183670
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911791834205255e-05,
+      "loss": 0.0059,
+      "step": 183680
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117870302314075e-05,
+      "loss": 0.007,
+      "step": 183690
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117822262575604e-05,
+      "loss": 0.0092,
+      "step": 183700
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117774222837134e-05,
+      "loss": 0.0079,
+      "step": 183710
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117726183098663e-05,
+      "loss": 0.0059,
+      "step": 183720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911767814336019e-05,
+      "loss": 0.0076,
+      "step": 183730
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117630103621718e-05,
+      "loss": 0.0036,
+      "step": 183740
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117582063883247e-05,
+      "loss": 0.0084,
+      "step": 183750
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117534024144773e-05,
+      "loss": 0.0071,
+      "step": 183760
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117485984406303e-05,
+      "loss": 0.0069,
+      "step": 183770
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117437944667832e-05,
+      "loss": 0.0043,
+      "step": 183780
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911738990492936e-05,
+      "loss": 0.0057,
+      "step": 183790
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117341865190887e-05,
+      "loss": 0.0112,
+      "step": 183800
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117293825452416e-05,
+      "loss": 0.0066,
+      "step": 183810
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117245785713946e-05,
+      "loss": 0.0067,
+      "step": 183820
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117197745975475e-05,
+      "loss": 0.0098,
+      "step": 183830
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117149706237e-05,
+      "loss": 0.0073,
+      "step": 183840
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911710166649853e-05,
+      "loss": 0.008,
+      "step": 183850
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911705362676006e-05,
+      "loss": 0.0055,
+      "step": 183860
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9117005587021585e-05,
+      "loss": 0.0059,
+      "step": 183870
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116957547283114e-05,
+      "loss": 0.0075,
+      "step": 183880
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116909507544644e-05,
+      "loss": 0.0076,
+      "step": 183890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116861467806173e-05,
+      "loss": 0.0057,
+      "step": 183900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.91168134280677e-05,
+      "loss": 0.0069,
+      "step": 183910
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116765388329228e-05,
+      "loss": 0.0068,
+      "step": 183920
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116717348590757e-05,
+      "loss": 0.0079,
+      "step": 183930
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116669308852283e-05,
+      "loss": 0.0045,
+      "step": 183940
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116621269113812e-05,
+      "loss": 0.0052,
+      "step": 183950
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116573229375342e-05,
+      "loss": 0.0091,
+      "step": 183960
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911652518963687e-05,
+      "loss": 0.0081,
+      "step": 183970
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116477149898397e-05,
+      "loss": 0.0052,
+      "step": 183980
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116429110159926e-05,
+      "loss": 0.0077,
+      "step": 183990
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116381070421455e-05,
+      "loss": 0.0089,
+      "step": 184000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116333030682985e-05,
+      "loss": 0.0077,
+      "step": 184010
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911628499094451e-05,
+      "loss": 0.0051,
+      "step": 184020
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911623695120604e-05,
+      "loss": 0.011,
+      "step": 184030
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911618891146757e-05,
+      "loss": 0.0111,
+      "step": 184040
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116140871729095e-05,
+      "loss": 0.0074,
+      "step": 184050
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116092831990624e-05,
+      "loss": 0.0087,
+      "step": 184060
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9116044792252154e-05,
+      "loss": 0.0043,
+      "step": 184070
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9115996752513683e-05,
+      "loss": 0.0152,
+      "step": 184080
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911594871277521e-05,
+      "loss": 0.0096,
+      "step": 184090
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9115900673036738e-05,
+      "loss": 0.0052,
+      "step": 184100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9115852633298267e-05,
+      "loss": 0.0066,
+      "step": 184110
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9115804593559793e-05,
+      "loss": 0.0073,
+      "step": 184120
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9115756553821322e-05,
+      "loss": 0.0059,
+      "step": 184130
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911570851408285e-05,
+      "loss": 0.0057,
+      "step": 184140
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911566047434438e-05,
+      "loss": 0.0063,
+      "step": 184150
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9115612434605907e-05,
+      "loss": 0.0044,
+      "step": 184160
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9115564394867436e-05,
+      "loss": 0.0095,
+      "step": 184170
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9115516355128965e-05,
+      "loss": 0.0074,
+      "step": 184180
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9115468315390495e-05,
+      "loss": 0.0069,
+      "step": 184190
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911542027565202e-05,
+      "loss": 0.0068,
+      "step": 184200
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911537223591355e-05,
+      "loss": 0.0037,
+      "step": 184210
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.911532419617508e-05,
+      "loss": 0.0045,
+      "step": 184220
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9115276156436605e-05,
+      "loss": 0.0054,
+      "step": 184230
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9115228116698134e-05,
+      "loss": 0.0058,
+      "step": 184240
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9115180076959663e-05,
+      "loss": 0.0084,
+      "step": 184250
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9115132037221193e-05,
+      "loss": 0.0074,
+      "step": 184260
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911508399748272e-05,
+      "loss": 0.004,
+      "step": 184270
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9115035957744248e-05,
+      "loss": 0.0066,
+      "step": 184280
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114987918005777e-05,
+      "loss": 0.006,
+      "step": 184290
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114939878267303e-05,
+      "loss": 0.0072,
+      "step": 184300
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114891838528832e-05,
+      "loss": 0.0069,
+      "step": 184310
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911484379879036e-05,
+      "loss": 0.0093,
+      "step": 184320
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911479575905189e-05,
+      "loss": 0.0065,
+      "step": 184330
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114747719313417e-05,
+      "loss": 0.0075,
+      "step": 184340
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114699679574946e-05,
+      "loss": 0.0076,
+      "step": 184350
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114651639836475e-05,
+      "loss": 0.0053,
+      "step": 184360
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114603600098005e-05,
+      "loss": 0.0081,
+      "step": 184370
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911455556035953e-05,
+      "loss": 0.0057,
+      "step": 184380
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911450752062106e-05,
+      "loss": 0.0086,
+      "step": 184390
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911445948088259e-05,
+      "loss": 0.0095,
+      "step": 184400
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114411441144115e-05,
+      "loss": 0.006,
+      "step": 184410
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114363401405644e-05,
+      "loss": 0.0089,
+      "step": 184420
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114315361667173e-05,
+      "loss": 0.0043,
+      "step": 184430
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114267321928703e-05,
+      "loss": 0.0055,
+      "step": 184440
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911421928219023e-05,
+      "loss": 0.0079,
+      "step": 184450
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114171242451758e-05,
+      "loss": 0.0053,
+      "step": 184460
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114123202713287e-05,
+      "loss": 0.0066,
+      "step": 184470
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114075162974813e-05,
+      "loss": 0.0067,
+      "step": 184480
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9114027123236342e-05,
+      "loss": 0.0075,
+      "step": 184490
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911397908349787e-05,
+      "loss": 0.0065,
+      "step": 184500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.91139310437594e-05,
+      "loss": 0.0089,
+      "step": 184510
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113883004020927e-05,
+      "loss": 0.0057,
+      "step": 184520
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113834964282456e-05,
+      "loss": 0.007,
+      "step": 184530
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113786924543985e-05,
+      "loss": 0.0056,
+      "step": 184540
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113738884805514e-05,
+      "loss": 0.0096,
+      "step": 184550
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911369084506704e-05,
+      "loss": 0.0065,
+      "step": 184560
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911364280532857e-05,
+      "loss": 0.0065,
+      "step": 184570
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.91135947655901e-05,
+      "loss": 0.0069,
+      "step": 184580
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113546725851625e-05,
+      "loss": 0.0041,
+      "step": 184590
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113498686113154e-05,
+      "loss": 0.0084,
+      "step": 184600
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113450646374683e-05,
+      "loss": 0.0074,
+      "step": 184610
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113402606636213e-05,
+      "loss": 0.0043,
+      "step": 184620
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911335456689774e-05,
+      "loss": 0.0061,
+      "step": 184630
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113306527159268e-05,
+      "loss": 0.0075,
+      "step": 184640
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113258487420797e-05,
+      "loss": 0.0088,
+      "step": 184650
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113210447682323e-05,
+      "loss": 0.0057,
+      "step": 184660
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113162407943852e-05,
+      "loss": 0.0091,
+      "step": 184670
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911311436820538e-05,
+      "loss": 0.006,
+      "step": 184680
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911306632846691e-05,
+      "loss": 0.0062,
+      "step": 184690
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9113018288728437e-05,
+      "loss": 0.0069,
+      "step": 184700
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112970248989966e-05,
+      "loss": 0.0079,
+      "step": 184710
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112922209251495e-05,
+      "loss": 0.0068,
+      "step": 184720
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112874169513024e-05,
+      "loss": 0.0073,
+      "step": 184730
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911282612977455e-05,
+      "loss": 0.0072,
+      "step": 184740
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911277809003608e-05,
+      "loss": 0.0092,
+      "step": 184750
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911273005029761e-05,
+      "loss": 0.0068,
+      "step": 184760
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112682010559135e-05,
+      "loss": 0.0065,
+      "step": 184770
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112633970820664e-05,
+      "loss": 0.0053,
+      "step": 184780
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112585931082193e-05,
+      "loss": 0.0104,
+      "step": 184790
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112537891343722e-05,
+      "loss": 0.0071,
+      "step": 184800
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911248985160525e-05,
+      "loss": 0.0095,
+      "step": 184810
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112441811866778e-05,
+      "loss": 0.0075,
+      "step": 184820
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112393772128307e-05,
+      "loss": 0.0073,
+      "step": 184830
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112345732389833e-05,
+      "loss": 0.008,
+      "step": 184840
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112297692651362e-05,
+      "loss": 0.007,
+      "step": 184850
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911224965291289e-05,
+      "loss": 0.0045,
+      "step": 184860
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911220161317442e-05,
+      "loss": 0.0091,
+      "step": 184870
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112153573435946e-05,
+      "loss": 0.0087,
+      "step": 184880
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112105533697476e-05,
+      "loss": 0.0067,
+      "step": 184890
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9112057493959005e-05,
+      "loss": 0.0059,
+      "step": 184900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.911200945422053e-05,
+      "loss": 0.0081,
+      "step": 184910
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.911196141448206e-05,
+      "loss": 0.0081,
+      "step": 184920
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.911191337474359e-05,
+      "loss": 0.0066,
+      "step": 184930
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.911186533500512e-05,
+      "loss": 0.0066,
+      "step": 184940
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9111817295266645e-05,
+      "loss": 0.0141,
+      "step": 184950
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9111769255528174e-05,
+      "loss": 0.0072,
+      "step": 184960
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9111721215789703e-05,
+      "loss": 0.006,
+      "step": 184970
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9111673176051232e-05,
+      "loss": 0.0062,
+      "step": 184980
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9111625136312758e-05,
+      "loss": 0.0071,
+      "step": 184990
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9111577096574288e-05,
+      "loss": 0.0089,
+      "step": 185000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9111529056835817e-05,
+      "loss": 0.0055,
+      "step": 185010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9111481017097343e-05,
+      "loss": 0.0071,
+      "step": 185020
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9111432977358872e-05,
+      "loss": 0.0076,
+      "step": 185030
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.91113849376204e-05,
+      "loss": 0.0054,
+      "step": 185040
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.911133689788193e-05,
+      "loss": 0.0059,
+      "step": 185050
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9111288858143456e-05,
+      "loss": 0.0066,
+      "step": 185060
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9111240818404986e-05,
+      "loss": 0.0091,
+      "step": 185070
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9111192778666515e-05,
+      "loss": 0.0059,
+      "step": 185080
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.911114473892804e-05,
+      "loss": 0.0085,
+      "step": 185090
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.911109669918957e-05,
+      "loss": 0.009,
+      "step": 185100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.91110486594511e-05,
+      "loss": 0.0079,
+      "step": 185110
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.911100061971263e-05,
+      "loss": 0.0055,
+      "step": 185120
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110952579974154e-05,
+      "loss": 0.006,
+      "step": 185130
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110904540235684e-05,
+      "loss": 0.006,
+      "step": 185140
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110856500497213e-05,
+      "loss": 0.0054,
+      "step": 185150
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110808460758742e-05,
+      "loss": 0.0071,
+      "step": 185160
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110760421020268e-05,
+      "loss": 0.0061,
+      "step": 185170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110712381281797e-05,
+      "loss": 0.0057,
+      "step": 185180
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110664341543327e-05,
+      "loss": 0.0052,
+      "step": 185190
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110616301804853e-05,
+      "loss": 0.0053,
+      "step": 185200
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110568262066382e-05,
+      "loss": 0.0076,
+      "step": 185210
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.911052022232791e-05,
+      "loss": 0.0054,
+      "step": 185220
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.911047218258944e-05,
+      "loss": 0.0063,
+      "step": 185230
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110424142850966e-05,
+      "loss": 0.0101,
+      "step": 185240
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110376103112496e-05,
+      "loss": 0.0065,
+      "step": 185250
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110328063374025e-05,
+      "loss": 0.0079,
+      "step": 185260
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.911028002363555e-05,
+      "loss": 0.0056,
+      "step": 185270
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.911023198389708e-05,
+      "loss": 0.0077,
+      "step": 185280
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.911018394415861e-05,
+      "loss": 0.0075,
+      "step": 185290
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110140708393984e-05,
+      "loss": 0.005,
+      "step": 185300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110092668655513e-05,
+      "loss": 0.0047,
+      "step": 185310
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9110044628917042e-05,
+      "loss": 0.0081,
+      "step": 185320
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.910999658917857e-05,
+      "loss": 0.0115,
+      "step": 185330
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109948549440098e-05,
+      "loss": 0.0067,
+      "step": 185340
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109900509701627e-05,
+      "loss": 0.0111,
+      "step": 185350
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109852469963156e-05,
+      "loss": 0.006,
+      "step": 185360
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109804430224682e-05,
+      "loss": 0.0049,
+      "step": 185370
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.910975639048621e-05,
+      "loss": 0.0052,
+      "step": 185380
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.910970835074774e-05,
+      "loss": 0.0065,
+      "step": 185390
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.910966031100927e-05,
+      "loss": 0.0073,
+      "step": 185400
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109612271270796e-05,
+      "loss": 0.0052,
+      "step": 185410
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109564231532325e-05,
+      "loss": 0.0052,
+      "step": 185420
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109516191793854e-05,
+      "loss": 0.0071,
+      "step": 185430
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.910946815205538e-05,
+      "loss": 0.0062,
+      "step": 185440
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.910942011231691e-05,
+      "loss": 0.0062,
+      "step": 185450
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.910937207257844e-05,
+      "loss": 0.0084,
+      "step": 185460
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109324032839968e-05,
+      "loss": 0.0074,
+      "step": 185470
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109275993101494e-05,
+      "loss": 0.0063,
+      "step": 185480
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109227953363023e-05,
+      "loss": 0.0047,
+      "step": 185490
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109179913624552e-05,
+      "loss": 0.0081,
+      "step": 185500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109131873886078e-05,
+      "loss": 0.0106,
+      "step": 185510
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109083834147608e-05,
+      "loss": 0.0102,
+      "step": 185520
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9109035794409137e-05,
+      "loss": 0.0071,
+      "step": 185530
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9108987754670666e-05,
+      "loss": 0.0075,
+      "step": 185540
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9108939714932192e-05,
+      "loss": 0.0065,
+      "step": 185550
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.910889167519372e-05,
+      "loss": 0.0097,
+      "step": 185560
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.910884363545525e-05,
+      "loss": 0.0074,
+      "step": 185570
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.910879559571678e-05,
+      "loss": 0.0082,
+      "step": 185580
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9108747555978306e-05,
+      "loss": 0.0062,
+      "step": 185590
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9108699516239835e-05,
+      "loss": 0.0052,
+      "step": 185600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9108651476501364e-05,
+      "loss": 0.0057,
+      "step": 185610
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910860343676289e-05,
+      "loss": 0.0049,
+      "step": 185620
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910855539702442e-05,
+      "loss": 0.0049,
+      "step": 185630
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910850735728595e-05,
+      "loss": 0.0062,
+      "step": 185640
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9108459317547478e-05,
+      "loss": 0.0061,
+      "step": 185650
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9108411277809004e-05,
+      "loss": 0.0086,
+      "step": 185660
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9108363238070533e-05,
+      "loss": 0.0082,
+      "step": 185670
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9108315198332062e-05,
+      "loss": 0.0096,
+      "step": 185680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9108267158593588e-05,
+      "loss": 0.0068,
+      "step": 185690
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9108219118855117e-05,
+      "loss": 0.0048,
+      "step": 185700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9108171079116647e-05,
+      "loss": 0.0082,
+      "step": 185710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9108123039378176e-05,
+      "loss": 0.0036,
+      "step": 185720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9108074999639702e-05,
+      "loss": 0.0068,
+      "step": 185730
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910802695990123e-05,
+      "loss": 0.0039,
+      "step": 185740
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910797892016276e-05,
+      "loss": 0.0064,
+      "step": 185750
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910793088042429e-05,
+      "loss": 0.0063,
+      "step": 185760
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107882840685816e-05,
+      "loss": 0.0073,
+      "step": 185770
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107834800947345e-05,
+      "loss": 0.0059,
+      "step": 185780
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107786761208874e-05,
+      "loss": 0.0066,
+      "step": 185790
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.91077387214704e-05,
+      "loss": 0.0079,
+      "step": 185800
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910769068173193e-05,
+      "loss": 0.0075,
+      "step": 185810
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910764264199346e-05,
+      "loss": 0.0048,
+      "step": 185820
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107594602254988e-05,
+      "loss": 0.0074,
+      "step": 185830
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107546562516514e-05,
+      "loss": 0.0078,
+      "step": 185840
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107498522778043e-05,
+      "loss": 0.008,
+      "step": 185850
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107450483039572e-05,
+      "loss": 0.0059,
+      "step": 185860
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107402443301098e-05,
+      "loss": 0.0055,
+      "step": 185870
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107354403562627e-05,
+      "loss": 0.0074,
+      "step": 185880
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107306363824157e-05,
+      "loss": 0.0063,
+      "step": 185890
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107258324085686e-05,
+      "loss": 0.0091,
+      "step": 185900
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107210284347212e-05,
+      "loss": 0.0084,
+      "step": 185910
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910716224460874e-05,
+      "loss": 0.0047,
+      "step": 185920
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910711420487027e-05,
+      "loss": 0.0081,
+      "step": 185930
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.91070661651318e-05,
+      "loss": 0.0052,
+      "step": 185940
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9107018125393325e-05,
+      "loss": 0.0084,
+      "step": 185950
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106970085654855e-05,
+      "loss": 0.0056,
+      "step": 185960
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106922045916384e-05,
+      "loss": 0.0052,
+      "step": 185970
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910687400617791e-05,
+      "loss": 0.0036,
+      "step": 185980
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910682596643944e-05,
+      "loss": 0.0075,
+      "step": 185990
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910677792670097e-05,
+      "loss": 0.0051,
+      "step": 186000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106729886962498e-05,
+      "loss": 0.0065,
+      "step": 186010
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106681847224024e-05,
+      "loss": 0.0037,
+      "step": 186020
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106633807485553e-05,
+      "loss": 0.0077,
+      "step": 186030
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106585767747082e-05,
+      "loss": 0.0098,
+      "step": 186040
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106537728008608e-05,
+      "loss": 0.0085,
+      "step": 186050
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106489688270137e-05,
+      "loss": 0.0072,
+      "step": 186060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106441648531667e-05,
+      "loss": 0.0059,
+      "step": 186070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106393608793196e-05,
+      "loss": 0.0078,
+      "step": 186080
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910634556905472e-05,
+      "loss": 0.0058,
+      "step": 186090
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910629752931625e-05,
+      "loss": 0.0072,
+      "step": 186100
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910624948957778e-05,
+      "loss": 0.0067,
+      "step": 186110
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910620144983931e-05,
+      "loss": 0.006,
+      "step": 186120
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106153410100835e-05,
+      "loss": 0.0105,
+      "step": 186130
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106105370362365e-05,
+      "loss": 0.0048,
+      "step": 186140
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9106057330623894e-05,
+      "loss": 0.0085,
+      "step": 186150
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910600929088542e-05,
+      "loss": 0.0075,
+      "step": 186160
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910596125114695e-05,
+      "loss": 0.0069,
+      "step": 186170
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910591321140848e-05,
+      "loss": 0.0071,
+      "step": 186180
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9105865171670008e-05,
+      "loss": 0.0069,
+      "step": 186190
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9105817131931533e-05,
+      "loss": 0.0091,
+      "step": 186200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9105769092193063e-05,
+      "loss": 0.0056,
+      "step": 186210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9105721052454592e-05,
+      "loss": 0.0085,
+      "step": 186220
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9105673012716118e-05,
+      "loss": 0.0059,
+      "step": 186230
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9105624972977647e-05,
+      "loss": 0.0054,
+      "step": 186240
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9105576933239176e-05,
+      "loss": 0.0063,
+      "step": 186250
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9105528893500706e-05,
+      "loss": 0.0038,
+      "step": 186260
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910548085376223e-05,
+      "loss": 0.006,
+      "step": 186270
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910543281402376e-05,
+      "loss": 0.0099,
+      "step": 186280
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910538477428529e-05,
+      "loss": 0.0054,
+      "step": 186290
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.910533673454682e-05,
+      "loss": 0.0091,
+      "step": 186300
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9105288694808345e-05,
+      "loss": 0.0071,
+      "step": 186310
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9105240655069875e-05,
+      "loss": 0.0083,
+      "step": 186320
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9105192615331404e-05,
+      "loss": 0.0047,
+      "step": 186330
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910514457559293e-05,
+      "loss": 0.006,
+      "step": 186340
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910509653585446e-05,
+      "loss": 0.0054,
+      "step": 186350
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9105048496115988e-05,
+      "loss": 0.0064,
+      "step": 186360
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9105000456377518e-05,
+      "loss": 0.0061,
+      "step": 186370
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104952416639043e-05,
+      "loss": 0.0075,
+      "step": 186380
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104904376900573e-05,
+      "loss": 0.0062,
+      "step": 186390
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104856337162102e-05,
+      "loss": 0.004,
+      "step": 186400
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104808297423628e-05,
+      "loss": 0.0052,
+      "step": 186410
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104760257685157e-05,
+      "loss": 0.0064,
+      "step": 186420
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104712217946686e-05,
+      "loss": 0.008,
+      "step": 186430
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104664178208216e-05,
+      "loss": 0.0065,
+      "step": 186440
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104616138469745e-05,
+      "loss": 0.0072,
+      "step": 186450
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104568098731274e-05,
+      "loss": 0.0085,
+      "step": 186460
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.91045200589928e-05,
+      "loss": 0.0058,
+      "step": 186470
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910447201925433e-05,
+      "loss": 0.0071,
+      "step": 186480
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910442397951586e-05,
+      "loss": 0.005,
+      "step": 186490
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104375939777388e-05,
+      "loss": 0.0095,
+      "step": 186500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104327900038914e-05,
+      "loss": 0.0052,
+      "step": 186510
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104279860300443e-05,
+      "loss": 0.0075,
+      "step": 186520
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104231820561972e-05,
+      "loss": 0.0075,
+      "step": 186530
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104183780823498e-05,
+      "loss": 0.0071,
+      "step": 186540
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104135741085027e-05,
+      "loss": 0.0079,
+      "step": 186550
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104087701346557e-05,
+      "loss": 0.0057,
+      "step": 186560
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9104039661608086e-05,
+      "loss": 0.0078,
+      "step": 186570
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103991621869612e-05,
+      "loss": 0.0076,
+      "step": 186580
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910394358213114e-05,
+      "loss": 0.0069,
+      "step": 186590
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910389554239267e-05,
+      "loss": 0.0052,
+      "step": 186600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.91038475026542e-05,
+      "loss": 0.0064,
+      "step": 186610
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103799462915726e-05,
+      "loss": 0.0068,
+      "step": 186620
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103751423177255e-05,
+      "loss": 0.0083,
+      "step": 186630
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103703383438784e-05,
+      "loss": 0.0051,
+      "step": 186640
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910365534370031e-05,
+      "loss": 0.0073,
+      "step": 186650
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910360730396184e-05,
+      "loss": 0.0054,
+      "step": 186660
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910355926422337e-05,
+      "loss": 0.0076,
+      "step": 186670
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103511224484898e-05,
+      "loss": 0.0049,
+      "step": 186680
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103463184746424e-05,
+      "loss": 0.0048,
+      "step": 186690
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103415145007953e-05,
+      "loss": 0.0075,
+      "step": 186700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103367105269482e-05,
+      "loss": 0.0093,
+      "step": 186710
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103319065531008e-05,
+      "loss": 0.0068,
+      "step": 186720
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103271025792537e-05,
+      "loss": 0.0066,
+      "step": 186730
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103222986054067e-05,
+      "loss": 0.0059,
+      "step": 186740
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103174946315596e-05,
+      "loss": 0.0062,
+      "step": 186750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9103126906577122e-05,
+      "loss": 0.0055,
+      "step": 186760
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910307886683865e-05,
+      "loss": 0.0086,
+      "step": 186770
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910303082710018e-05,
+      "loss": 0.0065,
+      "step": 186780
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910298278736171e-05,
+      "loss": 0.0049,
+      "step": 186790
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102934747623235e-05,
+      "loss": 0.0057,
+      "step": 186800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102886707884765e-05,
+      "loss": 0.006,
+      "step": 186810
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102838668146294e-05,
+      "loss": 0.0113,
+      "step": 186820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910279062840782e-05,
+      "loss": 0.0075,
+      "step": 186830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910274258866935e-05,
+      "loss": 0.006,
+      "step": 186840
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910269454893088e-05,
+      "loss": 0.0066,
+      "step": 186850
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102646509192408e-05,
+      "loss": 0.0061,
+      "step": 186860
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102598469453934e-05,
+      "loss": 0.0067,
+      "step": 186870
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102550429715463e-05,
+      "loss": 0.0064,
+      "step": 186880
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102502389976992e-05,
+      "loss": 0.006,
+      "step": 186890
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102454350238518e-05,
+      "loss": 0.007,
+      "step": 186900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102406310500047e-05,
+      "loss": 0.009,
+      "step": 186910
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102358270761577e-05,
+      "loss": 0.0068,
+      "step": 186920
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102310231023106e-05,
+      "loss": 0.006,
+      "step": 186930
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910226219128463e-05,
+      "loss": 0.0088,
+      "step": 186940
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910221415154616e-05,
+      "loss": 0.0082,
+      "step": 186950
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910216611180769e-05,
+      "loss": 0.0047,
+      "step": 186960
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.910211807206922e-05,
+      "loss": 0.0086,
+      "step": 186970
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102070032330745e-05,
+      "loss": 0.0068,
+      "step": 186980
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9102021992592275e-05,
+      "loss": 0.0097,
+      "step": 186990
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9101973952853804e-05,
+      "loss": 0.006,
+      "step": 187000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910192591311533e-05,
+      "loss": 0.0086,
+      "step": 187010
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910187787337686e-05,
+      "loss": 0.0076,
+      "step": 187020
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910182983363839e-05,
+      "loss": 0.0099,
+      "step": 187030
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9101781793899918e-05,
+      "loss": 0.0111,
+      "step": 187040
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9101733754161443e-05,
+      "loss": 0.0076,
+      "step": 187050
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9101685714422973e-05,
+      "loss": 0.0086,
+      "step": 187060
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9101637674684502e-05,
+      "loss": 0.008,
+      "step": 187070
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9101589634946028e-05,
+      "loss": 0.0099,
+      "step": 187080
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9101541595207557e-05,
+      "loss": 0.0099,
+      "step": 187090
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9101493555469086e-05,
+      "loss": 0.0066,
+      "step": 187100
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9101445515730616e-05,
+      "loss": 0.0077,
+      "step": 187110
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910139747599214e-05,
+      "loss": 0.0074,
+      "step": 187120
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910134943625367e-05,
+      "loss": 0.0077,
+      "step": 187130
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.91013013965152e-05,
+      "loss": 0.0077,
+      "step": 187140
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910125335677673e-05,
+      "loss": 0.005,
+      "step": 187150
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9101205317038255e-05,
+      "loss": 0.0049,
+      "step": 187160
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9101157277299785e-05,
+      "loss": 0.0077,
+      "step": 187170
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9101109237561314e-05,
+      "loss": 0.0053,
+      "step": 187180
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910106119782284e-05,
+      "loss": 0.0082,
+      "step": 187190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910101315808437e-05,
+      "loss": 0.0039,
+      "step": 187200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100965118345898e-05,
+      "loss": 0.0069,
+      "step": 187210
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100917078607428e-05,
+      "loss": 0.0063,
+      "step": 187220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100869038868953e-05,
+      "loss": 0.0038,
+      "step": 187230
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100820999130483e-05,
+      "loss": 0.0062,
+      "step": 187240
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100772959392012e-05,
+      "loss": 0.0065,
+      "step": 187250
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100724919653538e-05,
+      "loss": 0.0107,
+      "step": 187260
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100676879915067e-05,
+      "loss": 0.0069,
+      "step": 187270
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100628840176596e-05,
+      "loss": 0.007,
+      "step": 187280
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100580800438126e-05,
+      "loss": 0.0082,
+      "step": 187290
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910053276069965e-05,
+      "loss": 0.0106,
+      "step": 187300
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910048472096118e-05,
+      "loss": 0.0066,
+      "step": 187310
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910043668122271e-05,
+      "loss": 0.0082,
+      "step": 187320
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910038864148424e-05,
+      "loss": 0.0053,
+      "step": 187330
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100340601745765e-05,
+      "loss": 0.0065,
+      "step": 187340
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100292562007294e-05,
+      "loss": 0.0062,
+      "step": 187350
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100244522268824e-05,
+      "loss": 0.0068,
+      "step": 187360
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910019648253035e-05,
+      "loss": 0.0068,
+      "step": 187370
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.910014844279188e-05,
+      "loss": 0.0062,
+      "step": 187380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100100403053408e-05,
+      "loss": 0.0059,
+      "step": 187390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100052363314937e-05,
+      "loss": 0.009,
+      "step": 187400
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9100004323576463e-05,
+      "loss": 0.0082,
+      "step": 187410
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099956283837993e-05,
+      "loss": 0.01,
+      "step": 187420
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099908244099522e-05,
+      "loss": 0.0076,
+      "step": 187430
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099860204361048e-05,
+      "loss": 0.0082,
+      "step": 187440
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099816968596426e-05,
+      "loss": 0.0058,
+      "step": 187450
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099768928857955e-05,
+      "loss": 0.0048,
+      "step": 187460
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.909972088911948e-05,
+      "loss": 0.0045,
+      "step": 187470
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.909967284938101e-05,
+      "loss": 0.0069,
+      "step": 187480
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.909962480964254e-05,
+      "loss": 0.007,
+      "step": 187490
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099576769904065e-05,
+      "loss": 0.0101,
+      "step": 187500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099528730165595e-05,
+      "loss": 0.0083,
+      "step": 187510
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099480690427124e-05,
+      "loss": 0.0055,
+      "step": 187520
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099432650688653e-05,
+      "loss": 0.0079,
+      "step": 187530
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.909938461095018e-05,
+      "loss": 0.0067,
+      "step": 187540
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099336571211708e-05,
+      "loss": 0.0073,
+      "step": 187550
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099288531473238e-05,
+      "loss": 0.0056,
+      "step": 187560
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099240491734763e-05,
+      "loss": 0.0054,
+      "step": 187570
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099192451996293e-05,
+      "loss": 0.0078,
+      "step": 187580
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099144412257822e-05,
+      "loss": 0.0101,
+      "step": 187590
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.909909637251935e-05,
+      "loss": 0.0076,
+      "step": 187600
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099048332780877e-05,
+      "loss": 0.0074,
+      "step": 187610
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9099000293042406e-05,
+      "loss": 0.0057,
+      "step": 187620
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9098952253303936e-05,
+      "loss": 0.0067,
+      "step": 187630
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9098904213565465e-05,
+      "loss": 0.0076,
+      "step": 187640
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.909885617382699e-05,
+      "loss": 0.0075,
+      "step": 187650
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.909880813408852e-05,
+      "loss": 0.007,
+      "step": 187660
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.909876009435005e-05,
+      "loss": 0.0065,
+      "step": 187670
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9098712054611575e-05,
+      "loss": 0.0063,
+      "step": 187680
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9098664014873105e-05,
+      "loss": 0.0052,
+      "step": 187690
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9098615975134634e-05,
+      "loss": 0.0052,
+      "step": 187700
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9098567935396163e-05,
+      "loss": 0.0058,
+      "step": 187710
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909851989565769e-05,
+      "loss": 0.0077,
+      "step": 187720
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9098471855919218e-05,
+      "loss": 0.0068,
+      "step": 187730
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9098423816180747e-05,
+      "loss": 0.0084,
+      "step": 187740
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9098375776442273e-05,
+      "loss": 0.0066,
+      "step": 187750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9098327736703803e-05,
+      "loss": 0.0049,
+      "step": 187760
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9098279696965332e-05,
+      "loss": 0.0065,
+      "step": 187770
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909823165722686e-05,
+      "loss": 0.0063,
+      "step": 187780
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9098183617488387e-05,
+      "loss": 0.0069,
+      "step": 187790
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9098135577749916e-05,
+      "loss": 0.0084,
+      "step": 187800
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9098087538011446e-05,
+      "loss": 0.0082,
+      "step": 187810
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9098039498272975e-05,
+      "loss": 0.0055,
+      "step": 187820
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.90979914585345e-05,
+      "loss": 0.0076,
+      "step": 187830
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909794341879603e-05,
+      "loss": 0.0055,
+      "step": 187840
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909789537905756e-05,
+      "loss": 0.0074,
+      "step": 187850
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097847339319085e-05,
+      "loss": 0.0083,
+      "step": 187860
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097799299580614e-05,
+      "loss": 0.0062,
+      "step": 187870
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097751259842144e-05,
+      "loss": 0.0051,
+      "step": 187880
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097703220103673e-05,
+      "loss": 0.0047,
+      "step": 187890
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.90976551803652e-05,
+      "loss": 0.0072,
+      "step": 187900
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097607140626728e-05,
+      "loss": 0.0058,
+      "step": 187910
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097559100888257e-05,
+      "loss": 0.0073,
+      "step": 187920
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097511061149783e-05,
+      "loss": 0.0064,
+      "step": 187930
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097463021411313e-05,
+      "loss": 0.0085,
+      "step": 187940
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097414981672842e-05,
+      "loss": 0.0086,
+      "step": 187950
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909736694193437e-05,
+      "loss": 0.0087,
+      "step": 187960
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097318902195897e-05,
+      "loss": 0.0065,
+      "step": 187970
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097270862457426e-05,
+      "loss": 0.0098,
+      "step": 187980
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097222822718956e-05,
+      "loss": 0.008,
+      "step": 187990
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9097174782980485e-05,
+      "loss": 0.007,
+      "step": 188000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909712674324201e-05,
+      "loss": 0.0081,
+      "step": 188010
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909707870350354e-05,
+      "loss": 0.0065,
+      "step": 188020
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909703066376507e-05,
+      "loss": 0.0083,
+      "step": 188030
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096982624026595e-05,
+      "loss": 0.0085,
+      "step": 188040
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096934584288124e-05,
+      "loss": 0.0083,
+      "step": 188050
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096886544549654e-05,
+      "loss": 0.007,
+      "step": 188060
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096838504811183e-05,
+      "loss": 0.0055,
+      "step": 188070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909679046507271e-05,
+      "loss": 0.008,
+      "step": 188080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096742425334238e-05,
+      "loss": 0.0083,
+      "step": 188090
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096694385595767e-05,
+      "loss": 0.0065,
+      "step": 188100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096646345857293e-05,
+      "loss": 0.0063,
+      "step": 188110
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096598306118822e-05,
+      "loss": 0.0045,
+      "step": 188120
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096550266380352e-05,
+      "loss": 0.0053,
+      "step": 188130
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909650222664188e-05,
+      "loss": 0.0115,
+      "step": 188140
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096454186903407e-05,
+      "loss": 0.0063,
+      "step": 188150
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096406147164936e-05,
+      "loss": 0.0068,
+      "step": 188160
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096358107426465e-05,
+      "loss": 0.0052,
+      "step": 188170
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096310067687995e-05,
+      "loss": 0.0053,
+      "step": 188180
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909626202794952e-05,
+      "loss": 0.0088,
+      "step": 188190
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909621398821105e-05,
+      "loss": 0.0078,
+      "step": 188200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909616594847258e-05,
+      "loss": 0.0069,
+      "step": 188210
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096117908734105e-05,
+      "loss": 0.006,
+      "step": 188220
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096069868995634e-05,
+      "loss": 0.007,
+      "step": 188230
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9096021829257164e-05,
+      "loss": 0.0067,
+      "step": 188240
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9095973789518693e-05,
+      "loss": 0.0051,
+      "step": 188250
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909592574978022e-05,
+      "loss": 0.0071,
+      "step": 188260
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9095877710041748e-05,
+      "loss": 0.0072,
+      "step": 188270
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9095829670303277e-05,
+      "loss": 0.006,
+      "step": 188280
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9095781630564803e-05,
+      "loss": 0.0098,
+      "step": 188290
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9095733590826332e-05,
+      "loss": 0.0068,
+      "step": 188300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909568555108786e-05,
+      "loss": 0.0073,
+      "step": 188310
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909563751134939e-05,
+      "loss": 0.0085,
+      "step": 188320
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9095589471610917e-05,
+      "loss": 0.0097,
+      "step": 188330
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9095541431872446e-05,
+      "loss": 0.0072,
+      "step": 188340
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9095493392133975e-05,
+      "loss": 0.0098,
+      "step": 188350
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9095445352395505e-05,
+      "loss": 0.0046,
+      "step": 188360
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909539731265703e-05,
+      "loss": 0.0066,
+      "step": 188370
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909534927291856e-05,
+      "loss": 0.0082,
+      "step": 188380
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909530123318009e-05,
+      "loss": 0.0078,
+      "step": 188390
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9095253193441615e-05,
+      "loss": 0.0089,
+      "step": 188400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9095205153703144e-05,
+      "loss": 0.007,
+      "step": 188410
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9095157113964673e-05,
+      "loss": 0.0164,
+      "step": 188420
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9095109074226203e-05,
+      "loss": 0.0072,
+      "step": 188430
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909506103448773e-05,
+      "loss": 0.0095,
+      "step": 188440
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9095012994749258e-05,
+      "loss": 0.0063,
+      "step": 188450
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094964955010787e-05,
+      "loss": 0.0077,
+      "step": 188460
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094916915272313e-05,
+      "loss": 0.0066,
+      "step": 188470
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094868875533842e-05,
+      "loss": 0.0054,
+      "step": 188480
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909482083579537e-05,
+      "loss": 0.0102,
+      "step": 188490
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.90947727960569e-05,
+      "loss": 0.0082,
+      "step": 188500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094724756318427e-05,
+      "loss": 0.0057,
+      "step": 188510
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094676716579956e-05,
+      "loss": 0.0119,
+      "step": 188520
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094628676841485e-05,
+      "loss": 0.0085,
+      "step": 188530
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094580637103015e-05,
+      "loss": 0.0048,
+      "step": 188540
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909453259736454e-05,
+      "loss": 0.0057,
+      "step": 188550
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909448455762607e-05,
+      "loss": 0.0047,
+      "step": 188560
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.90944365178876e-05,
+      "loss": 0.0105,
+      "step": 188570
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094388478149125e-05,
+      "loss": 0.0101,
+      "step": 188580
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094340438410654e-05,
+      "loss": 0.0062,
+      "step": 188590
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094292398672183e-05,
+      "loss": 0.0057,
+      "step": 188600
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094244358933713e-05,
+      "loss": 0.008,
+      "step": 188610
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909419631919524e-05,
+      "loss": 0.0103,
+      "step": 188620
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094148279456768e-05,
+      "loss": 0.0105,
+      "step": 188630
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094100239718297e-05,
+      "loss": 0.0104,
+      "step": 188640
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094052199979823e-05,
+      "loss": 0.0081,
+      "step": 188650
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9094004160241352e-05,
+      "loss": 0.0062,
+      "step": 188660
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909395612050288e-05,
+      "loss": 0.0052,
+      "step": 188670
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909390808076441e-05,
+      "loss": 0.0067,
+      "step": 188680
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9093860041025937e-05,
+      "loss": 0.0066,
+      "step": 188690
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9093812001287466e-05,
+      "loss": 0.006,
+      "step": 188700
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9093763961548995e-05,
+      "loss": 0.006,
+      "step": 188710
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9093715921810524e-05,
+      "loss": 0.0049,
+      "step": 188720
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909366788207205e-05,
+      "loss": 0.0084,
+      "step": 188730
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909361984233358e-05,
+      "loss": 0.0057,
+      "step": 188740
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909357180259511e-05,
+      "loss": 0.0049,
+      "step": 188750
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9093523762856635e-05,
+      "loss": 0.0071,
+      "step": 188760
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9093475723118164e-05,
+      "loss": 0.0044,
+      "step": 188770
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9093427683379693e-05,
+      "loss": 0.0075,
+      "step": 188780
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9093379643641223e-05,
+      "loss": 0.0051,
+      "step": 188790
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909333160390275e-05,
+      "loss": 0.0057,
+      "step": 188800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9093283564164278e-05,
+      "loss": 0.0092,
+      "step": 188810
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9093235524425807e-05,
+      "loss": 0.0069,
+      "step": 188820
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9093187484687333e-05,
+      "loss": 0.0074,
+      "step": 188830
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9093139444948862e-05,
+      "loss": 0.0053,
+      "step": 188840
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909309140521039e-05,
+      "loss": 0.0063,
+      "step": 188850
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909304336547192e-05,
+      "loss": 0.0078,
+      "step": 188860
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092995325733447e-05,
+      "loss": 0.0095,
+      "step": 188870
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092947285994976e-05,
+      "loss": 0.0055,
+      "step": 188880
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092899246256505e-05,
+      "loss": 0.0084,
+      "step": 188890
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092851206518034e-05,
+      "loss": 0.0067,
+      "step": 188900
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909280316677956e-05,
+      "loss": 0.0072,
+      "step": 188910
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909275512704109e-05,
+      "loss": 0.0068,
+      "step": 188920
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909270708730262e-05,
+      "loss": 0.0064,
+      "step": 188930
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092659047564145e-05,
+      "loss": 0.0068,
+      "step": 188940
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092611007825674e-05,
+      "loss": 0.0059,
+      "step": 188950
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092562968087203e-05,
+      "loss": 0.0076,
+      "step": 188960
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092514928348732e-05,
+      "loss": 0.0079,
+      "step": 188970
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909246688861026e-05,
+      "loss": 0.0049,
+      "step": 188980
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092418848871788e-05,
+      "loss": 0.0062,
+      "step": 188990
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092370809133317e-05,
+      "loss": 0.0067,
+      "step": 189000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092322769394843e-05,
+      "loss": 0.007,
+      "step": 189010
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092274729656372e-05,
+      "loss": 0.0058,
+      "step": 189020
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.90922266899179e-05,
+      "loss": 0.0048,
+      "step": 189030
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.909217865017943e-05,
+      "loss": 0.0076,
+      "step": 189040
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092130610440956e-05,
+      "loss": 0.0058,
+      "step": 189050
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092082570702486e-05,
+      "loss": 0.0085,
+      "step": 189060
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9092034530964015e-05,
+      "loss": 0.0083,
+      "step": 189070
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091986491225544e-05,
+      "loss": 0.0063,
+      "step": 189080
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909193845148707e-05,
+      "loss": 0.0052,
+      "step": 189090
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.90918904117486e-05,
+      "loss": 0.0052,
+      "step": 189100
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909184237201013e-05,
+      "loss": 0.0105,
+      "step": 189110
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091794332271655e-05,
+      "loss": 0.0092,
+      "step": 189120
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091746292533184e-05,
+      "loss": 0.0056,
+      "step": 189130
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091698252794713e-05,
+      "loss": 0.0108,
+      "step": 189140
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091650213056242e-05,
+      "loss": 0.0083,
+      "step": 189150
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091602173317768e-05,
+      "loss": 0.0067,
+      "step": 189160
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091554133579298e-05,
+      "loss": 0.0079,
+      "step": 189170
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091506093840827e-05,
+      "loss": 0.0065,
+      "step": 189180
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091458054102353e-05,
+      "loss": 0.0062,
+      "step": 189190
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091410014363882e-05,
+      "loss": 0.0097,
+      "step": 189200
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909136197462541e-05,
+      "loss": 0.009,
+      "step": 189210
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909131393488694e-05,
+      "loss": 0.0096,
+      "step": 189220
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091265895148466e-05,
+      "loss": 0.0064,
+      "step": 189230
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091217855409996e-05,
+      "loss": 0.008,
+      "step": 189240
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091169815671525e-05,
+      "loss": 0.0083,
+      "step": 189250
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9091121775933054e-05,
+      "loss": 0.0074,
+      "step": 189260
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909107373619458e-05,
+      "loss": 0.007,
+      "step": 189270
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909102569645611e-05,
+      "loss": 0.0076,
+      "step": 189280
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909097765671764e-05,
+      "loss": 0.0064,
+      "step": 189290
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090929616979164e-05,
+      "loss": 0.0094,
+      "step": 189300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090881577240694e-05,
+      "loss": 0.0053,
+      "step": 189310
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090833537502223e-05,
+      "loss": 0.0104,
+      "step": 189320
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090785497763752e-05,
+      "loss": 0.0056,
+      "step": 189330
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090737458025278e-05,
+      "loss": 0.0057,
+      "step": 189340
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090689418286807e-05,
+      "loss": 0.0097,
+      "step": 189350
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090641378548337e-05,
+      "loss": 0.0068,
+      "step": 189360
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090593338809863e-05,
+      "loss": 0.0057,
+      "step": 189370
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090545299071392e-05,
+      "loss": 0.009,
+      "step": 189380
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909049725933292e-05,
+      "loss": 0.0047,
+      "step": 189390
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909044921959445e-05,
+      "loss": 0.006,
+      "step": 189400
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090401179855976e-05,
+      "loss": 0.0078,
+      "step": 189410
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090353140117506e-05,
+      "loss": 0.0074,
+      "step": 189420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090305100379035e-05,
+      "loss": 0.0073,
+      "step": 189430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909025706064056e-05,
+      "loss": 0.0076,
+      "step": 189440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909020902090209e-05,
+      "loss": 0.008,
+      "step": 189450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909016098116362e-05,
+      "loss": 0.0079,
+      "step": 189460
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.909011294142515e-05,
+      "loss": 0.0065,
+      "step": 189470
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090064901686674e-05,
+      "loss": 0.0086,
+      "step": 189480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9090016861948204e-05,
+      "loss": 0.0058,
+      "step": 189490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089968822209733e-05,
+      "loss": 0.0078,
+      "step": 189500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089920782471262e-05,
+      "loss": 0.0078,
+      "step": 189510
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.908987274273279e-05,
+      "loss": 0.0072,
+      "step": 189520
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.908982470299432e-05,
+      "loss": 0.0051,
+      "step": 189530
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089776663255847e-05,
+      "loss": 0.0069,
+      "step": 189540
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089728623517376e-05,
+      "loss": 0.0061,
+      "step": 189550
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089680583778905e-05,
+      "loss": 0.0082,
+      "step": 189560
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089632544040434e-05,
+      "loss": 0.0087,
+      "step": 189570
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.908958450430196e-05,
+      "loss": 0.0061,
+      "step": 189580
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.908953646456349e-05,
+      "loss": 0.0084,
+      "step": 189590
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.908948842482502e-05,
+      "loss": 0.0112,
+      "step": 189600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089440385086545e-05,
+      "loss": 0.0069,
+      "step": 189610
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089392345348074e-05,
+      "loss": 0.0106,
+      "step": 189620
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089344305609603e-05,
+      "loss": 0.0075,
+      "step": 189630
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089296265871133e-05,
+      "loss": 0.0066,
+      "step": 189640
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.908924822613266e-05,
+      "loss": 0.0051,
+      "step": 189650
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089200186394188e-05,
+      "loss": 0.0057,
+      "step": 189660
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089152146655717e-05,
+      "loss": 0.0088,
+      "step": 189670
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089104106917243e-05,
+      "loss": 0.0083,
+      "step": 189680
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9089056067178772e-05,
+      "loss": 0.0055,
+      "step": 189690
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.90890080274403e-05,
+      "loss": 0.0059,
+      "step": 189700
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.908895998770183e-05,
+      "loss": 0.0105,
+      "step": 189710
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9088911947963357e-05,
+      "loss": 0.0091,
+      "step": 189720
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9088863908224886e-05,
+      "loss": 0.0073,
+      "step": 189730
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9088815868486415e-05,
+      "loss": 0.0059,
+      "step": 189740
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9088767828747944e-05,
+      "loss": 0.0081,
+      "step": 189750
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.908871978900947e-05,
+      "loss": 0.0088,
+      "step": 189760
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9088671749271e-05,
+      "loss": 0.01,
+      "step": 189770
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908862370953253e-05,
+      "loss": 0.009,
+      "step": 189780
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9088575669794055e-05,
+      "loss": 0.0061,
+      "step": 189790
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9088527630055584e-05,
+      "loss": 0.0077,
+      "step": 189800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9088479590317113e-05,
+      "loss": 0.0061,
+      "step": 189810
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9088431550578642e-05,
+      "loss": 0.0102,
+      "step": 189820
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908838351084017e-05,
+      "loss": 0.0093,
+      "step": 189830
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9088335471101698e-05,
+      "loss": 0.0058,
+      "step": 189840
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9088287431363227e-05,
+      "loss": 0.0059,
+      "step": 189850
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9088239391624753e-05,
+      "loss": 0.0068,
+      "step": 189860
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9088191351886282e-05,
+      "loss": 0.0048,
+      "step": 189870
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908814331214781e-05,
+      "loss": 0.0071,
+      "step": 189880
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908809527240934e-05,
+      "loss": 0.0066,
+      "step": 189890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9088047232670866e-05,
+      "loss": 0.0122,
+      "step": 189900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9087999192932396e-05,
+      "loss": 0.0079,
+      "step": 189910
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908795595716777e-05,
+      "loss": 0.0112,
+      "step": 189920
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.90879079174293e-05,
+      "loss": 0.0063,
+      "step": 189930
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9087859877690826e-05,
+      "loss": 0.0129,
+      "step": 189940
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9087811837952355e-05,
+      "loss": 0.0057,
+      "step": 189950
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9087763798213884e-05,
+      "loss": 0.0073,
+      "step": 189960
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908771575847541e-05,
+      "loss": 0.0039,
+      "step": 189970
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908766771873694e-05,
+      "loss": 0.0063,
+      "step": 189980
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908761967899847e-05,
+      "loss": 0.0052,
+      "step": 189990
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9087571639259998e-05,
+      "loss": 0.008,
+      "step": 190000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9087523599521524e-05,
+      "loss": 0.007,
+      "step": 190010
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9087475559783053e-05,
+      "loss": 0.0079,
+      "step": 190020
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9087427520044582e-05,
+      "loss": 0.0053,
+      "step": 190030
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908737948030611e-05,
+      "loss": 0.0047,
+      "step": 190040
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908733144056764e-05,
+      "loss": 0.0083,
+      "step": 190050
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908728340082917e-05,
+      "loss": 0.0075,
+      "step": 190060
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9087235361090696e-05,
+      "loss": 0.0087,
+      "step": 190070
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9087187321352225e-05,
+      "loss": 0.0103,
+      "step": 190080
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9087139281613754e-05,
+      "loss": 0.0065,
+      "step": 190090
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908709124187528e-05,
+      "loss": 0.0094,
+      "step": 190100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908704320213681e-05,
+      "loss": 0.0079,
+      "step": 190110
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908699516239834e-05,
+      "loss": 0.01,
+      "step": 190120
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086947122659868e-05,
+      "loss": 0.0057,
+      "step": 190130
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086899082921394e-05,
+      "loss": 0.0054,
+      "step": 190140
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086851043182923e-05,
+      "loss": 0.007,
+      "step": 190150
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086803003444453e-05,
+      "loss": 0.0097,
+      "step": 190160
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086754963705982e-05,
+      "loss": 0.0048,
+      "step": 190170
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086706923967508e-05,
+      "loss": 0.0066,
+      "step": 190180
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086658884229037e-05,
+      "loss": 0.0085,
+      "step": 190190
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086610844490566e-05,
+      "loss": 0.0063,
+      "step": 190200
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086562804752092e-05,
+      "loss": 0.0079,
+      "step": 190210
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908651476501362e-05,
+      "loss": 0.0048,
+      "step": 190220
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908646672527515e-05,
+      "loss": 0.0062,
+      "step": 190230
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908641868553668e-05,
+      "loss": 0.0078,
+      "step": 190240
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086370645798206e-05,
+      "loss": 0.0048,
+      "step": 190250
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086322606059735e-05,
+      "loss": 0.0086,
+      "step": 190260
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086274566321264e-05,
+      "loss": 0.0079,
+      "step": 190270
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908622652658279e-05,
+      "loss": 0.0065,
+      "step": 190280
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908617848684432e-05,
+      "loss": 0.0086,
+      "step": 190290
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908613044710585e-05,
+      "loss": 0.007,
+      "step": 190300
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086082407367378e-05,
+      "loss": 0.0057,
+      "step": 190310
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9086034367628904e-05,
+      "loss": 0.0054,
+      "step": 190320
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9085986327890433e-05,
+      "loss": 0.0068,
+      "step": 190330
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9085938288151962e-05,
+      "loss": 0.0114,
+      "step": 190340
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9085890248413492e-05,
+      "loss": 0.0078,
+      "step": 190350
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9085842208675018e-05,
+      "loss": 0.0079,
+      "step": 190360
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9085794168936547e-05,
+      "loss": 0.0046,
+      "step": 190370
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9085746129198076e-05,
+      "loss": 0.0094,
+      "step": 190380
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9085698089459602e-05,
+      "loss": 0.0064,
+      "step": 190390
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908565004972113e-05,
+      "loss": 0.0058,
+      "step": 190400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908560200998266e-05,
+      "loss": 0.0067,
+      "step": 190410
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.908555397024419e-05,
+      "loss": 0.0105,
+      "step": 190420
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9085505930505716e-05,
+      "loss": 0.0066,
+      "step": 190430
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9085457890767245e-05,
+      "loss": 0.0051,
+      "step": 190440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9085409851028774e-05,
+      "loss": 0.0062,
+      "step": 190450
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9085366615264145e-05,
+      "loss": 0.014,
+      "step": 190460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9085318575525675e-05,
+      "loss": 0.0058,
+      "step": 190470
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9085270535787204e-05,
+      "loss": 0.0082,
+      "step": 190480
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9085222496048733e-05,
+      "loss": 0.0064,
+      "step": 190490
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908517445631026e-05,
+      "loss": 0.0076,
+      "step": 190500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908512641657179e-05,
+      "loss": 0.006,
+      "step": 190510
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9085078376833318e-05,
+      "loss": 0.0059,
+      "step": 190520
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9085030337094847e-05,
+      "loss": 0.007,
+      "step": 190530
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084982297356373e-05,
+      "loss": 0.0054,
+      "step": 190540
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084934257617906e-05,
+      "loss": 0.0084,
+      "step": 190550
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908488621787943e-05,
+      "loss": 0.0049,
+      "step": 190560
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908483817814096e-05,
+      "loss": 0.0074,
+      "step": 190570
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908479013840249e-05,
+      "loss": 0.0041,
+      "step": 190580
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084742098664016e-05,
+      "loss": 0.0061,
+      "step": 190590
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084694058925545e-05,
+      "loss": 0.0087,
+      "step": 190600
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084646019187074e-05,
+      "loss": 0.0056,
+      "step": 190610
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084597979448604e-05,
+      "loss": 0.0063,
+      "step": 190620
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908454993971013e-05,
+      "loss": 0.0078,
+      "step": 190630
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908450189997166e-05,
+      "loss": 0.0063,
+      "step": 190640
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084453860233188e-05,
+      "loss": 0.0102,
+      "step": 190650
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084405820494717e-05,
+      "loss": 0.009,
+      "step": 190660
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084357780756243e-05,
+      "loss": 0.0067,
+      "step": 190670
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084309741017772e-05,
+      "loss": 0.0064,
+      "step": 190680
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084261701279302e-05,
+      "loss": 0.0056,
+      "step": 190690
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084213661540828e-05,
+      "loss": 0.0056,
+      "step": 190700
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084165621802357e-05,
+      "loss": 0.0073,
+      "step": 190710
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084117582063886e-05,
+      "loss": 0.0062,
+      "step": 190720
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9084069542325415e-05,
+      "loss": 0.0053,
+      "step": 190730
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908402150258694e-05,
+      "loss": 0.0043,
+      "step": 190740
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908397346284847e-05,
+      "loss": 0.0065,
+      "step": 190750
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908392542311e-05,
+      "loss": 0.0062,
+      "step": 190760
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083877383371526e-05,
+      "loss": 0.0105,
+      "step": 190770
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083829343633055e-05,
+      "loss": 0.0065,
+      "step": 190780
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083781303894584e-05,
+      "loss": 0.0066,
+      "step": 190790
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083733264156114e-05,
+      "loss": 0.0042,
+      "step": 190800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908368522441764e-05,
+      "loss": 0.0045,
+      "step": 190810
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908363718467917e-05,
+      "loss": 0.0062,
+      "step": 190820
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083589144940698e-05,
+      "loss": 0.0075,
+      "step": 190830
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083541105202227e-05,
+      "loss": 0.0079,
+      "step": 190840
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083493065463753e-05,
+      "loss": 0.0068,
+      "step": 190850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083445025725282e-05,
+      "loss": 0.008,
+      "step": 190860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908339698598681e-05,
+      "loss": 0.0083,
+      "step": 190870
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083348946248338e-05,
+      "loss": 0.0076,
+      "step": 190880
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083300906509867e-05,
+      "loss": 0.0097,
+      "step": 190890
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083252866771396e-05,
+      "loss": 0.0056,
+      "step": 190900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083204827032925e-05,
+      "loss": 0.0071,
+      "step": 190910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908315678729445e-05,
+      "loss": 0.0048,
+      "step": 190920
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908310874755598e-05,
+      "loss": 0.008,
+      "step": 190930
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908306070781751e-05,
+      "loss": 0.0076,
+      "step": 190940
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9083012668079036e-05,
+      "loss": 0.0082,
+      "step": 190950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082964628340565e-05,
+      "loss": 0.0076,
+      "step": 190960
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082916588602094e-05,
+      "loss": 0.0089,
+      "step": 190970
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082868548863623e-05,
+      "loss": 0.011,
+      "step": 190980
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908282050912515e-05,
+      "loss": 0.0088,
+      "step": 190990
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908277246938668e-05,
+      "loss": 0.0064,
+      "step": 191000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082724429648208e-05,
+      "loss": 0.0053,
+      "step": 191010
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082676389909737e-05,
+      "loss": 0.0041,
+      "step": 191020
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082628350171263e-05,
+      "loss": 0.0092,
+      "step": 191030
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082580310432792e-05,
+      "loss": 0.007,
+      "step": 191040
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908253227069432e-05,
+      "loss": 0.0069,
+      "step": 191050
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082484230955847e-05,
+      "loss": 0.0052,
+      "step": 191060
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082436191217377e-05,
+      "loss": 0.0072,
+      "step": 191070
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082388151478906e-05,
+      "loss": 0.0074,
+      "step": 191080
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082340111740435e-05,
+      "loss": 0.0083,
+      "step": 191090
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908229207200196e-05,
+      "loss": 0.005,
+      "step": 191100
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908224403226349e-05,
+      "loss": 0.0062,
+      "step": 191110
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.908219599252502e-05,
+      "loss": 0.0078,
+      "step": 191120
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082147952786546e-05,
+      "loss": 0.0061,
+      "step": 191130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082099913048075e-05,
+      "loss": 0.0065,
+      "step": 191140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082051873309604e-05,
+      "loss": 0.0055,
+      "step": 191150
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9082003833571133e-05,
+      "loss": 0.0054,
+      "step": 191160
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.908195579383266e-05,
+      "loss": 0.0084,
+      "step": 191170
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.908190775409419e-05,
+      "loss": 0.0044,
+      "step": 191180
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081859714355718e-05,
+      "loss": 0.0057,
+      "step": 191190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081811674617247e-05,
+      "loss": 0.0043,
+      "step": 191200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081763634878773e-05,
+      "loss": 0.0069,
+      "step": 191210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081715595140302e-05,
+      "loss": 0.0082,
+      "step": 191220
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.908166755540183e-05,
+      "loss": 0.009,
+      "step": 191230
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081619515663357e-05,
+      "loss": 0.0065,
+      "step": 191240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081571475924887e-05,
+      "loss": 0.0074,
+      "step": 191250
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081523436186416e-05,
+      "loss": 0.0089,
+      "step": 191260
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081475396447945e-05,
+      "loss": 0.0082,
+      "step": 191270
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.908142735670947e-05,
+      "loss": 0.0065,
+      "step": 191280
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081379316971e-05,
+      "loss": 0.0079,
+      "step": 191290
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.908133127723253e-05,
+      "loss": 0.0073,
+      "step": 191300
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081283237494055e-05,
+      "loss": 0.008,
+      "step": 191310
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081235197755585e-05,
+      "loss": 0.0055,
+      "step": 191320
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081187158017114e-05,
+      "loss": 0.0061,
+      "step": 191330
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9081139118278643e-05,
+      "loss": 0.0071,
+      "step": 191340
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.908109107854017e-05,
+      "loss": 0.0071,
+      "step": 191350
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.90810430388017e-05,
+      "loss": 0.0058,
+      "step": 191360
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080994999063228e-05,
+      "loss": 0.0063,
+      "step": 191370
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080946959324757e-05,
+      "loss": 0.006,
+      "step": 191380
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080898919586283e-05,
+      "loss": 0.011,
+      "step": 191390
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080850879847812e-05,
+      "loss": 0.0065,
+      "step": 191400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.908080284010934e-05,
+      "loss": 0.0127,
+      "step": 191410
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080754800370867e-05,
+      "loss": 0.0078,
+      "step": 191420
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080706760632397e-05,
+      "loss": 0.0065,
+      "step": 191430
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080658720893926e-05,
+      "loss": 0.0112,
+      "step": 191440
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080610681155455e-05,
+      "loss": 0.0112,
+      "step": 191450
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.908056264141698e-05,
+      "loss": 0.0048,
+      "step": 191460
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.908051460167851e-05,
+      "loss": 0.01,
+      "step": 191470
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.908046656194004e-05,
+      "loss": 0.0081,
+      "step": 191480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080418522201565e-05,
+      "loss": 0.006,
+      "step": 191490
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080370482463095e-05,
+      "loss": 0.006,
+      "step": 191500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080322442724624e-05,
+      "loss": 0.0094,
+      "step": 191510
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080274402986153e-05,
+      "loss": 0.0067,
+      "step": 191520
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.908022636324768e-05,
+      "loss": 0.0079,
+      "step": 191530
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.908017832350921e-05,
+      "loss": 0.0038,
+      "step": 191540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080130283770738e-05,
+      "loss": 0.0051,
+      "step": 191550
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080082244032267e-05,
+      "loss": 0.0083,
+      "step": 191560
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9080034204293793e-05,
+      "loss": 0.0054,
+      "step": 191570
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079986164555322e-05,
+      "loss": 0.0052,
+      "step": 191580
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.907993812481685e-05,
+      "loss": 0.0066,
+      "step": 191590
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079890085078377e-05,
+      "loss": 0.007,
+      "step": 191600
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079842045339906e-05,
+      "loss": 0.0065,
+      "step": 191610
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079794005601436e-05,
+      "loss": 0.0064,
+      "step": 191620
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079745965862965e-05,
+      "loss": 0.0058,
+      "step": 191630
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.907969792612449e-05,
+      "loss": 0.0061,
+      "step": 191640
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.907964988638602e-05,
+      "loss": 0.0077,
+      "step": 191650
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.907960184664755e-05,
+      "loss": 0.0076,
+      "step": 191660
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079553806909075e-05,
+      "loss": 0.0081,
+      "step": 191670
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079505767170605e-05,
+      "loss": 0.008,
+      "step": 191680
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079457727432134e-05,
+      "loss": 0.0088,
+      "step": 191690
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079409687693663e-05,
+      "loss": 0.0073,
+      "step": 191700
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.907936164795519e-05,
+      "loss": 0.006,
+      "step": 191710
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079313608216718e-05,
+      "loss": 0.004,
+      "step": 191720
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079265568478248e-05,
+      "loss": 0.0062,
+      "step": 191730
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079217528739777e-05,
+      "loss": 0.0085,
+      "step": 191740
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079169489001303e-05,
+      "loss": 0.0084,
+      "step": 191750
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079121449262832e-05,
+      "loss": 0.0076,
+      "step": 191760
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.907907340952436e-05,
+      "loss": 0.0048,
+      "step": 191770
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9079025369785887e-05,
+      "loss": 0.0092,
+      "step": 191780
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9078977330047416e-05,
+      "loss": 0.0077,
+      "step": 191790
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9078929290308946e-05,
+      "loss": 0.0061,
+      "step": 191800
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9078881250570475e-05,
+      "loss": 0.01,
+      "step": 191810
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9078833210832e-05,
+      "loss": 0.005,
+      "step": 191820
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.907878517109353e-05,
+      "loss": 0.0063,
+      "step": 191830
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.907873713135506e-05,
+      "loss": 0.0078,
+      "step": 191840
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9078689091616585e-05,
+      "loss": 0.0056,
+      "step": 191850
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9078641051878115e-05,
+      "loss": 0.0068,
+      "step": 191860
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9078593012139644e-05,
+      "loss": 0.0045,
+      "step": 191870
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9078544972401173e-05,
+      "loss": 0.0053,
+      "step": 191880
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.90784969326627e-05,
+      "loss": 0.0067,
+      "step": 191890
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9078448892924228e-05,
+      "loss": 0.0056,
+      "step": 191900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9078400853185757e-05,
+      "loss": 0.0042,
+      "step": 191910
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9078352813447287e-05,
+      "loss": 0.0057,
+      "step": 191920
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9078304773708813e-05,
+      "loss": 0.0062,
+      "step": 191930
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9078256733970342e-05,
+      "loss": 0.0055,
+      "step": 191940
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907820869423187e-05,
+      "loss": 0.007,
+      "step": 191950
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9078160654493397e-05,
+      "loss": 0.0061,
+      "step": 191960
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9078112614754926e-05,
+      "loss": 0.0068,
+      "step": 191970
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9078064575016456e-05,
+      "loss": 0.0078,
+      "step": 191980
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9078016535277985e-05,
+      "loss": 0.0084,
+      "step": 191990
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907796849553951e-05,
+      "loss": 0.0056,
+      "step": 192000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907792045580104e-05,
+      "loss": 0.0062,
+      "step": 192010
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907787241606257e-05,
+      "loss": 0.0069,
+      "step": 192020
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077824376324095e-05,
+      "loss": 0.0049,
+      "step": 192030
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077776336585624e-05,
+      "loss": 0.0056,
+      "step": 192040
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077728296847154e-05,
+      "loss": 0.0049,
+      "step": 192050
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077680257108683e-05,
+      "loss": 0.006,
+      "step": 192060
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907763221737021e-05,
+      "loss": 0.0069,
+      "step": 192070
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077584177631738e-05,
+      "loss": 0.0073,
+      "step": 192080
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077536137893267e-05,
+      "loss": 0.0062,
+      "step": 192090
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077488098154797e-05,
+      "loss": 0.0064,
+      "step": 192100
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077440058416323e-05,
+      "loss": 0.0078,
+      "step": 192110
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077392018677852e-05,
+      "loss": 0.0058,
+      "step": 192120
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907734397893938e-05,
+      "loss": 0.0042,
+      "step": 192130
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077295939200907e-05,
+      "loss": 0.0065,
+      "step": 192140
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077247899462436e-05,
+      "loss": 0.0081,
+      "step": 192150
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077199859723966e-05,
+      "loss": 0.0056,
+      "step": 192160
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9077151819985495e-05,
+      "loss": 0.0063,
+      "step": 192170
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907710378024702e-05,
+      "loss": 0.0091,
+      "step": 192180
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907705574050855e-05,
+      "loss": 0.005,
+      "step": 192190
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907700770077008e-05,
+      "loss": 0.0091,
+      "step": 192200
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076959661031605e-05,
+      "loss": 0.0054,
+      "step": 192210
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076911621293134e-05,
+      "loss": 0.0061,
+      "step": 192220
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076863581554664e-05,
+      "loss": 0.0064,
+      "step": 192230
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076815541816193e-05,
+      "loss": 0.0075,
+      "step": 192240
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907676750207772e-05,
+      "loss": 0.0062,
+      "step": 192250
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076719462339248e-05,
+      "loss": 0.0066,
+      "step": 192260
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076671422600777e-05,
+      "loss": 0.0102,
+      "step": 192270
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076623382862303e-05,
+      "loss": 0.0051,
+      "step": 192280
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076575343123832e-05,
+      "loss": 0.008,
+      "step": 192290
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076527303385362e-05,
+      "loss": 0.0069,
+      "step": 192300
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907647926364689e-05,
+      "loss": 0.0057,
+      "step": 192310
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076431223908417e-05,
+      "loss": 0.0074,
+      "step": 192320
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076383184169946e-05,
+      "loss": 0.0082,
+      "step": 192330
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076335144431475e-05,
+      "loss": 0.0098,
+      "step": 192340
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076287104693005e-05,
+      "loss": 0.0088,
+      "step": 192350
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907623906495453e-05,
+      "loss": 0.0066,
+      "step": 192360
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907619102521606e-05,
+      "loss": 0.008,
+      "step": 192370
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907614298547759e-05,
+      "loss": 0.0074,
+      "step": 192380
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076094945739115e-05,
+      "loss": 0.0068,
+      "step": 192390
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9076046906000644e-05,
+      "loss": 0.0099,
+      "step": 192400
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9075998866262174e-05,
+      "loss": 0.0079,
+      "step": 192410
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9075950826523703e-05,
+      "loss": 0.0082,
+      "step": 192420
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907590278678523e-05,
+      "loss": 0.0052,
+      "step": 192430
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9075854747046758e-05,
+      "loss": 0.0064,
+      "step": 192440
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9075806707308287e-05,
+      "loss": 0.0071,
+      "step": 192450
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9075758667569813e-05,
+      "loss": 0.0054,
+      "step": 192460
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9075710627831342e-05,
+      "loss": 0.0065,
+      "step": 192470
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907566258809287e-05,
+      "loss": 0.0064,
+      "step": 192480
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.90756145483544e-05,
+      "loss": 0.0083,
+      "step": 192490
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9075566508615927e-05,
+      "loss": 0.0063,
+      "step": 192500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9075518468877456e-05,
+      "loss": 0.008,
+      "step": 192510
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9075470429138985e-05,
+      "loss": 0.0074,
+      "step": 192520
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9075422389400515e-05,
+      "loss": 0.0076,
+      "step": 192530
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.907537434966204e-05,
+      "loss": 0.0035,
+      "step": 192540
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907532630992357e-05,
+      "loss": 0.0085,
+      "step": 192550
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.90752782701851e-05,
+      "loss": 0.0078,
+      "step": 192560
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9075230230446625e-05,
+      "loss": 0.008,
+      "step": 192570
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9075182190708154e-05,
+      "loss": 0.0053,
+      "step": 192580
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9075134150969683e-05,
+      "loss": 0.0049,
+      "step": 192590
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9075086111231213e-05,
+      "loss": 0.0055,
+      "step": 192600
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907503807149274e-05,
+      "loss": 0.0073,
+      "step": 192610
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074990031754268e-05,
+      "loss": 0.0067,
+      "step": 192620
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074941992015797e-05,
+      "loss": 0.0052,
+      "step": 192630
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074893952277323e-05,
+      "loss": 0.0088,
+      "step": 192640
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074845912538852e-05,
+      "loss": 0.006,
+      "step": 192650
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907479787280038e-05,
+      "loss": 0.0061,
+      "step": 192660
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907474983306191e-05,
+      "loss": 0.0051,
+      "step": 192670
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074701793323437e-05,
+      "loss": 0.0056,
+      "step": 192680
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074653753584966e-05,
+      "loss": 0.0072,
+      "step": 192690
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074605713846495e-05,
+      "loss": 0.0066,
+      "step": 192700
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074557674108025e-05,
+      "loss": 0.0078,
+      "step": 192710
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907450963436955e-05,
+      "loss": 0.0055,
+      "step": 192720
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907446159463108e-05,
+      "loss": 0.006,
+      "step": 192730
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907441355489261e-05,
+      "loss": 0.0115,
+      "step": 192740
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074365515154135e-05,
+      "loss": 0.0052,
+      "step": 192750
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074317475415664e-05,
+      "loss": 0.0069,
+      "step": 192760
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074269435677193e-05,
+      "loss": 0.0083,
+      "step": 192770
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074221395938723e-05,
+      "loss": 0.0071,
+      "step": 192780
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907417335620025e-05,
+      "loss": 0.0063,
+      "step": 192790
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074125316461778e-05,
+      "loss": 0.0077,
+      "step": 192800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074077276723307e-05,
+      "loss": 0.0051,
+      "step": 192810
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9074029236984833e-05,
+      "loss": 0.0053,
+      "step": 192820
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073981197246362e-05,
+      "loss": 0.0135,
+      "step": 192830
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907393315750789e-05,
+      "loss": 0.0074,
+      "step": 192840
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907388511776942e-05,
+      "loss": 0.0091,
+      "step": 192850
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073837078030947e-05,
+      "loss": 0.0055,
+      "step": 192860
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073789038292476e-05,
+      "loss": 0.0081,
+      "step": 192870
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073740998554005e-05,
+      "loss": 0.0078,
+      "step": 192880
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073692958815534e-05,
+      "loss": 0.0068,
+      "step": 192890
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907364491907706e-05,
+      "loss": 0.0053,
+      "step": 192900
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907359687933859e-05,
+      "loss": 0.0052,
+      "step": 192910
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907354883960012e-05,
+      "loss": 0.0074,
+      "step": 192920
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073500799861645e-05,
+      "loss": 0.0061,
+      "step": 192930
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073452760123174e-05,
+      "loss": 0.0062,
+      "step": 192940
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073404720384703e-05,
+      "loss": 0.0048,
+      "step": 192950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073356680646233e-05,
+      "loss": 0.0087,
+      "step": 192960
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907330864090776e-05,
+      "loss": 0.0067,
+      "step": 192970
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073260601169288e-05,
+      "loss": 0.0081,
+      "step": 192980
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073212561430817e-05,
+      "loss": 0.0073,
+      "step": 192990
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073164521692343e-05,
+      "loss": 0.0065,
+      "step": 193000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9073116481953872e-05,
+      "loss": 0.0098,
+      "step": 193010
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.90730684422154e-05,
+      "loss": 0.0068,
+      "step": 193020
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907302040247693e-05,
+      "loss": 0.0053,
+      "step": 193030
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072972362738457e-05,
+      "loss": 0.0061,
+      "step": 193040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072924322999986e-05,
+      "loss": 0.004,
+      "step": 193050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072876283261515e-05,
+      "loss": 0.0054,
+      "step": 193060
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072828243523044e-05,
+      "loss": 0.0053,
+      "step": 193070
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907278020378457e-05,
+      "loss": 0.0052,
+      "step": 193080
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.90727321640461e-05,
+      "loss": 0.0061,
+      "step": 193090
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907268412430763e-05,
+      "loss": 0.0056,
+      "step": 193100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072636084569158e-05,
+      "loss": 0.0069,
+      "step": 193110
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072588044830687e-05,
+      "loss": 0.0062,
+      "step": 193120
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072540005092217e-05,
+      "loss": 0.0067,
+      "step": 193130
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072491965353742e-05,
+      "loss": 0.0069,
+      "step": 193140
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072443925615272e-05,
+      "loss": 0.0102,
+      "step": 193150
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.90723958858768e-05,
+      "loss": 0.0049,
+      "step": 193160
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072347846138327e-05,
+      "loss": 0.0061,
+      "step": 193170
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072299806399856e-05,
+      "loss": 0.0053,
+      "step": 193180
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072251766661385e-05,
+      "loss": 0.0053,
+      "step": 193190
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072203726922915e-05,
+      "loss": 0.0048,
+      "step": 193200
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907215568718444e-05,
+      "loss": 0.0041,
+      "step": 193210
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.907210764744597e-05,
+      "loss": 0.0076,
+      "step": 193220
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.90720596077075e-05,
+      "loss": 0.0057,
+      "step": 193230
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9072011567969025e-05,
+      "loss": 0.0044,
+      "step": 193240
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071963528230554e-05,
+      "loss": 0.0066,
+      "step": 193250
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071915488492084e-05,
+      "loss": 0.0065,
+      "step": 193260
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071867448753613e-05,
+      "loss": 0.0074,
+      "step": 193270
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907181940901514e-05,
+      "loss": 0.0092,
+      "step": 193280
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071771369276668e-05,
+      "loss": 0.0056,
+      "step": 193290
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071723329538197e-05,
+      "loss": 0.0063,
+      "step": 193300
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071675289799726e-05,
+      "loss": 0.0086,
+      "step": 193310
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071627250061252e-05,
+      "loss": 0.0058,
+      "step": 193320
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907157921032278e-05,
+      "loss": 0.0065,
+      "step": 193330
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907153117058431e-05,
+      "loss": 0.0059,
+      "step": 193340
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071483130845837e-05,
+      "loss": 0.0061,
+      "step": 193350
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071435091107366e-05,
+      "loss": 0.0055,
+      "step": 193360
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071387051368895e-05,
+      "loss": 0.0059,
+      "step": 193370
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071339011630425e-05,
+      "loss": 0.007,
+      "step": 193380
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907129097189195e-05,
+      "loss": 0.0103,
+      "step": 193390
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907124293215348e-05,
+      "loss": 0.0085,
+      "step": 193400
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907119489241501e-05,
+      "loss": 0.0059,
+      "step": 193410
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071146852676535e-05,
+      "loss": 0.0096,
+      "step": 193420
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071098812938064e-05,
+      "loss": 0.0091,
+      "step": 193430
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071050773199593e-05,
+      "loss": 0.0071,
+      "step": 193440
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9071002733461123e-05,
+      "loss": 0.0049,
+      "step": 193450
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907095469372265e-05,
+      "loss": 0.0063,
+      "step": 193460
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070906653984178e-05,
+      "loss": 0.0077,
+      "step": 193470
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070858614245707e-05,
+      "loss": 0.0072,
+      "step": 193480
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070810574507236e-05,
+      "loss": 0.0069,
+      "step": 193490
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070762534768762e-05,
+      "loss": 0.0077,
+      "step": 193500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907071449503029e-05,
+      "loss": 0.0072,
+      "step": 193510
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907066645529182e-05,
+      "loss": 0.008,
+      "step": 193520
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070618415553347e-05,
+      "loss": 0.0062,
+      "step": 193530
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070570375814876e-05,
+      "loss": 0.0056,
+      "step": 193540
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070522336076405e-05,
+      "loss": 0.0056,
+      "step": 193550
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070474296337935e-05,
+      "loss": 0.0091,
+      "step": 193560
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907042625659946e-05,
+      "loss": 0.0059,
+      "step": 193570
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907037821686099e-05,
+      "loss": 0.0095,
+      "step": 193580
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907033017712252e-05,
+      "loss": 0.0057,
+      "step": 193590
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070282137384045e-05,
+      "loss": 0.0104,
+      "step": 193600
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070234097645574e-05,
+      "loss": 0.0058,
+      "step": 193610
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070186057907103e-05,
+      "loss": 0.0072,
+      "step": 193620
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070138018168633e-05,
+      "loss": 0.0068,
+      "step": 193630
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.907008997843016e-05,
+      "loss": 0.0057,
+      "step": 193640
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9070041938691688e-05,
+      "loss": 0.0079,
+      "step": 193650
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069993898953217e-05,
+      "loss": 0.0062,
+      "step": 193660
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069945859214743e-05,
+      "loss": 0.0082,
+      "step": 193670
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069897819476272e-05,
+      "loss": 0.007,
+      "step": 193680
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.90698497797378e-05,
+      "loss": 0.0079,
+      "step": 193690
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.906980173999933e-05,
+      "loss": 0.004,
+      "step": 193700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069753700260857e-05,
+      "loss": 0.0057,
+      "step": 193710
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069705660522386e-05,
+      "loss": 0.0059,
+      "step": 193720
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069657620783915e-05,
+      "loss": 0.0054,
+      "step": 193730
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069609581045444e-05,
+      "loss": 0.0086,
+      "step": 193740
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.906956154130697e-05,
+      "loss": 0.0071,
+      "step": 193750
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.90695135015685e-05,
+      "loss": 0.0074,
+      "step": 193760
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.906946546183003e-05,
+      "loss": 0.0062,
+      "step": 193770
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069417422091555e-05,
+      "loss": 0.0073,
+      "step": 193780
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069369382353084e-05,
+      "loss": 0.0063,
+      "step": 193790
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069321342614613e-05,
+      "loss": 0.0108,
+      "step": 193800
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069273302876143e-05,
+      "loss": 0.0051,
+      "step": 193810
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.906922526313767e-05,
+      "loss": 0.0078,
+      "step": 193820
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069177223399198e-05,
+      "loss": 0.0094,
+      "step": 193830
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069129183660727e-05,
+      "loss": 0.0055,
+      "step": 193840
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069081143922253e-05,
+      "loss": 0.0064,
+      "step": 193850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9069033104183782e-05,
+      "loss": 0.0059,
+      "step": 193860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.906898506444531e-05,
+      "loss": 0.0074,
+      "step": 193870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.906893702470684e-05,
+      "loss": 0.0078,
+      "step": 193880
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9068888984968367e-05,
+      "loss": 0.0061,
+      "step": 193890
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9068840945229896e-05,
+      "loss": 0.0078,
+      "step": 193900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9068792905491425e-05,
+      "loss": 0.0092,
+      "step": 193910
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9068744865752954e-05,
+      "loss": 0.0054,
+      "step": 193920
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.906869682601448e-05,
+      "loss": 0.0093,
+      "step": 193930
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906864878627601e-05,
+      "loss": 0.0067,
+      "step": 193940
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906860074653754e-05,
+      "loss": 0.0053,
+      "step": 193950
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9068552706799065e-05,
+      "loss": 0.0116,
+      "step": 193960
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9068504667060594e-05,
+      "loss": 0.0047,
+      "step": 193970
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9068456627322123e-05,
+      "loss": 0.0062,
+      "step": 193980
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9068408587583652e-05,
+      "loss": 0.0084,
+      "step": 193990
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906836054784518e-05,
+      "loss": 0.0074,
+      "step": 194000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9068312508106708e-05,
+      "loss": 0.0069,
+      "step": 194010
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9068264468368237e-05,
+      "loss": 0.0058,
+      "step": 194020
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9068216428629763e-05,
+      "loss": 0.0057,
+      "step": 194030
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9068168388891292e-05,
+      "loss": 0.0075,
+      "step": 194040
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906812034915282e-05,
+      "loss": 0.0064,
+      "step": 194050
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906807230941435e-05,
+      "loss": 0.0068,
+      "step": 194060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9068024269675876e-05,
+      "loss": 0.0084,
+      "step": 194070
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067976229937406e-05,
+      "loss": 0.008,
+      "step": 194080
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067928190198935e-05,
+      "loss": 0.0059,
+      "step": 194090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067880150460464e-05,
+      "loss": 0.0049,
+      "step": 194100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906783211072199e-05,
+      "loss": 0.0066,
+      "step": 194110
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906778407098352e-05,
+      "loss": 0.0048,
+      "step": 194120
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906773603124505e-05,
+      "loss": 0.0073,
+      "step": 194130
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067687991506575e-05,
+      "loss": 0.0097,
+      "step": 194140
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067639951768104e-05,
+      "loss": 0.0039,
+      "step": 194150
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067591912029633e-05,
+      "loss": 0.0051,
+      "step": 194160
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067543872291162e-05,
+      "loss": 0.009,
+      "step": 194170
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067495832552688e-05,
+      "loss": 0.0042,
+      "step": 194180
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067447792814218e-05,
+      "loss": 0.0076,
+      "step": 194190
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067399753075747e-05,
+      "loss": 0.0103,
+      "step": 194200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067351713337273e-05,
+      "loss": 0.0041,
+      "step": 194210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067303673598802e-05,
+      "loss": 0.0048,
+      "step": 194220
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906725563386033e-05,
+      "loss": 0.0072,
+      "step": 194230
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906720759412186e-05,
+      "loss": 0.0075,
+      "step": 194240
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067159554383386e-05,
+      "loss": 0.01,
+      "step": 194250
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067111514644916e-05,
+      "loss": 0.0109,
+      "step": 194260
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067063474906445e-05,
+      "loss": 0.0077,
+      "step": 194270
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9067015435167974e-05,
+      "loss": 0.0044,
+      "step": 194280
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.90669673954295e-05,
+      "loss": 0.007,
+      "step": 194290
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906691935569103e-05,
+      "loss": 0.0063,
+      "step": 194300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906687131595256e-05,
+      "loss": 0.0064,
+      "step": 194310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066823276214084e-05,
+      "loss": 0.0059,
+      "step": 194320
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066775236475614e-05,
+      "loss": 0.0134,
+      "step": 194330
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066727196737143e-05,
+      "loss": 0.007,
+      "step": 194340
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066679156998672e-05,
+      "loss": 0.006,
+      "step": 194350
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066631117260198e-05,
+      "loss": 0.006,
+      "step": 194360
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066583077521727e-05,
+      "loss": 0.0083,
+      "step": 194370
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066535037783257e-05,
+      "loss": 0.0064,
+      "step": 194380
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066486998044783e-05,
+      "loss": 0.0066,
+      "step": 194390
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066438958306312e-05,
+      "loss": 0.0086,
+      "step": 194400
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906639091856784e-05,
+      "loss": 0.0107,
+      "step": 194410
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906634287882937e-05,
+      "loss": 0.0099,
+      "step": 194420
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066294839090896e-05,
+      "loss": 0.0042,
+      "step": 194430
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066246799352426e-05,
+      "loss": 0.0088,
+      "step": 194440
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066198759613955e-05,
+      "loss": 0.0056,
+      "step": 194450
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9066150719875484e-05,
+      "loss": 0.006,
+      "step": 194460
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906610268013701e-05,
+      "loss": 0.0067,
+      "step": 194470
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906605464039854e-05,
+      "loss": 0.0106,
+      "step": 194480
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906600660066007e-05,
+      "loss": 0.0081,
+      "step": 194490
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9065958560921594e-05,
+      "loss": 0.0048,
+      "step": 194500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9065910521183124e-05,
+      "loss": 0.0085,
+      "step": 194510
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9065867285418502e-05,
+      "loss": 0.0083,
+      "step": 194520
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9065819245680028e-05,
+      "loss": 0.0064,
+      "step": 194530
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9065771205941557e-05,
+      "loss": 0.0091,
+      "step": 194540
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9065723166203086e-05,
+      "loss": 0.007,
+      "step": 194550
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9065675126464612e-05,
+      "loss": 0.0057,
+      "step": 194560
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906562708672614e-05,
+      "loss": 0.0054,
+      "step": 194570
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906557904698767e-05,
+      "loss": 0.0064,
+      "step": 194580
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.90655310072492e-05,
+      "loss": 0.0045,
+      "step": 194590
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9065482967510726e-05,
+      "loss": 0.0069,
+      "step": 194600
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9065434927772255e-05,
+      "loss": 0.0054,
+      "step": 194610
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9065386888033784e-05,
+      "loss": 0.0109,
+      "step": 194620
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.906533884829531e-05,
+      "loss": 0.0036,
+      "step": 194630
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906529080855684e-05,
+      "loss": 0.0068,
+      "step": 194640
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906524276881837e-05,
+      "loss": 0.0084,
+      "step": 194650
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9065194729079898e-05,
+      "loss": 0.0084,
+      "step": 194660
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9065146689341424e-05,
+      "loss": 0.0056,
+      "step": 194670
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9065098649602953e-05,
+      "loss": 0.0059,
+      "step": 194680
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9065050609864482e-05,
+      "loss": 0.0058,
+      "step": 194690
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906500257012601e-05,
+      "loss": 0.0075,
+      "step": 194700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064954530387537e-05,
+      "loss": 0.0085,
+      "step": 194710
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064906490649067e-05,
+      "loss": 0.0046,
+      "step": 194720
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064858450910596e-05,
+      "loss": 0.0072,
+      "step": 194730
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064810411172122e-05,
+      "loss": 0.006,
+      "step": 194740
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906476237143365e-05,
+      "loss": 0.009,
+      "step": 194750
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906471433169518e-05,
+      "loss": 0.0059,
+      "step": 194760
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906466629195671e-05,
+      "loss": 0.0067,
+      "step": 194770
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064618252218236e-05,
+      "loss": 0.0072,
+      "step": 194780
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064570212479765e-05,
+      "loss": 0.0048,
+      "step": 194790
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064522172741294e-05,
+      "loss": 0.0093,
+      "step": 194800
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906447413300282e-05,
+      "loss": 0.0068,
+      "step": 194810
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906442609326435e-05,
+      "loss": 0.0064,
+      "step": 194820
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906437805352588e-05,
+      "loss": 0.0067,
+      "step": 194830
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064330013787408e-05,
+      "loss": 0.0048,
+      "step": 194840
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064281974048934e-05,
+      "loss": 0.0057,
+      "step": 194850
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064233934310463e-05,
+      "loss": 0.0065,
+      "step": 194860
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064185894571992e-05,
+      "loss": 0.0079,
+      "step": 194870
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906413785483352e-05,
+      "loss": 0.0069,
+      "step": 194880
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064089815095047e-05,
+      "loss": 0.0065,
+      "step": 194890
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9064041775356577e-05,
+      "loss": 0.0053,
+      "step": 194900
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063993735618106e-05,
+      "loss": 0.0075,
+      "step": 194910
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063945695879632e-05,
+      "loss": 0.0063,
+      "step": 194920
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906389765614116e-05,
+      "loss": 0.0057,
+      "step": 194930
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906384961640269e-05,
+      "loss": 0.0056,
+      "step": 194940
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906380157666422e-05,
+      "loss": 0.0042,
+      "step": 194950
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063753536925746e-05,
+      "loss": 0.0079,
+      "step": 194960
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063705497187275e-05,
+      "loss": 0.005,
+      "step": 194970
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063657457448804e-05,
+      "loss": 0.0072,
+      "step": 194980
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906360941771033e-05,
+      "loss": 0.0061,
+      "step": 194990
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906356137797186e-05,
+      "loss": 0.0038,
+      "step": 195000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906351333823339e-05,
+      "loss": 0.006,
+      "step": 195010
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063465298494918e-05,
+      "loss": 0.0044,
+      "step": 195020
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063417258756444e-05,
+      "loss": 0.0071,
+      "step": 195030
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063369219017973e-05,
+      "loss": 0.0092,
+      "step": 195040
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063321179279502e-05,
+      "loss": 0.0061,
+      "step": 195050
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906327313954103e-05,
+      "loss": 0.0073,
+      "step": 195060
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063225099802557e-05,
+      "loss": 0.0056,
+      "step": 195070
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063177060064087e-05,
+      "loss": 0.0065,
+      "step": 195080
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063129020325616e-05,
+      "loss": 0.0057,
+      "step": 195090
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9063080980587142e-05,
+      "loss": 0.0067,
+      "step": 195100
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906303294084867e-05,
+      "loss": 0.0079,
+      "step": 195110
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.90629849011102e-05,
+      "loss": 0.007,
+      "step": 195120
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906293686137173e-05,
+      "loss": 0.0074,
+      "step": 195130
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9062888821633255e-05,
+      "loss": 0.0074,
+      "step": 195140
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9062840781894785e-05,
+      "loss": 0.0077,
+      "step": 195150
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9062792742156314e-05,
+      "loss": 0.0078,
+      "step": 195160
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906274470241784e-05,
+      "loss": 0.0085,
+      "step": 195170
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906269666267937e-05,
+      "loss": 0.0089,
+      "step": 195180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.90626486229409e-05,
+      "loss": 0.0056,
+      "step": 195190
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9062600583202428e-05,
+      "loss": 0.006,
+      "step": 195200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9062552543463954e-05,
+      "loss": 0.0057,
+      "step": 195210
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9062504503725483e-05,
+      "loss": 0.0057,
+      "step": 195220
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9062456463987012e-05,
+      "loss": 0.0096,
+      "step": 195230
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9062408424248538e-05,
+      "loss": 0.007,
+      "step": 195240
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9062360384510067e-05,
+      "loss": 0.005,
+      "step": 195250
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9062312344771597e-05,
+      "loss": 0.0069,
+      "step": 195260
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9062264305033126e-05,
+      "loss": 0.0066,
+      "step": 195270
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906221626529465e-05,
+      "loss": 0.0067,
+      "step": 195280
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906216822555618e-05,
+      "loss": 0.0091,
+      "step": 195290
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906212018581771e-05,
+      "loss": 0.0052,
+      "step": 195300
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906207214607924e-05,
+      "loss": 0.0058,
+      "step": 195310
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9062024106340765e-05,
+      "loss": 0.0068,
+      "step": 195320
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061976066602295e-05,
+      "loss": 0.0064,
+      "step": 195330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061928026863824e-05,
+      "loss": 0.0081,
+      "step": 195340
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906187998712535e-05,
+      "loss": 0.0051,
+      "step": 195350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906183194738688e-05,
+      "loss": 0.0093,
+      "step": 195360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906178390764841e-05,
+      "loss": 0.0078,
+      "step": 195370
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061735867909938e-05,
+      "loss": 0.0149,
+      "step": 195380
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061687828171463e-05,
+      "loss": 0.0069,
+      "step": 195390
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061639788432993e-05,
+      "loss": 0.0069,
+      "step": 195400
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061591748694522e-05,
+      "loss": 0.0083,
+      "step": 195410
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061543708956048e-05,
+      "loss": 0.0096,
+      "step": 195420
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061495669217577e-05,
+      "loss": 0.0081,
+      "step": 195430
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061447629479106e-05,
+      "loss": 0.0045,
+      "step": 195440
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061399589740636e-05,
+      "loss": 0.0062,
+      "step": 195450
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906135155000216e-05,
+      "loss": 0.0041,
+      "step": 195460
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906130351026369e-05,
+      "loss": 0.0042,
+      "step": 195470
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906125547052522e-05,
+      "loss": 0.01,
+      "step": 195480
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906120743078675e-05,
+      "loss": 0.005,
+      "step": 195490
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061159391048275e-05,
+      "loss": 0.0054,
+      "step": 195500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061111351309805e-05,
+      "loss": 0.0045,
+      "step": 195510
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061063311571334e-05,
+      "loss": 0.0054,
+      "step": 195520
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906101527183286e-05,
+      "loss": 0.0074,
+      "step": 195530
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906096723209439e-05,
+      "loss": 0.0061,
+      "step": 195540
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060919192355918e-05,
+      "loss": 0.0071,
+      "step": 195550
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060871152617447e-05,
+      "loss": 0.0081,
+      "step": 195560
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060823112878973e-05,
+      "loss": 0.0063,
+      "step": 195570
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060775073140503e-05,
+      "loss": 0.0051,
+      "step": 195580
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060727033402032e-05,
+      "loss": 0.0065,
+      "step": 195590
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060678993663558e-05,
+      "loss": 0.0076,
+      "step": 195600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060630953925087e-05,
+      "loss": 0.0078,
+      "step": 195610
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060582914186616e-05,
+      "loss": 0.0078,
+      "step": 195620
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060534874448146e-05,
+      "loss": 0.0045,
+      "step": 195630
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906048683470967e-05,
+      "loss": 0.008,
+      "step": 195640
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.90604387949712e-05,
+      "loss": 0.0075,
+      "step": 195650
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906039075523273e-05,
+      "loss": 0.0082,
+      "step": 195660
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906034271549426e-05,
+      "loss": 0.0079,
+      "step": 195670
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060294675755785e-05,
+      "loss": 0.0081,
+      "step": 195680
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060246636017314e-05,
+      "loss": 0.007,
+      "step": 195690
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060198596278844e-05,
+      "loss": 0.006,
+      "step": 195700
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906015055654037e-05,
+      "loss": 0.0072,
+      "step": 195710
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.90601025168019e-05,
+      "loss": 0.0086,
+      "step": 195720
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060054477063428e-05,
+      "loss": 0.0095,
+      "step": 195730
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9060006437324957e-05,
+      "loss": 0.0072,
+      "step": 195740
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059958397586483e-05,
+      "loss": 0.0054,
+      "step": 195750
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059910357848013e-05,
+      "loss": 0.0056,
+      "step": 195760
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059862318109542e-05,
+      "loss": 0.0065,
+      "step": 195770
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059814278371068e-05,
+      "loss": 0.0041,
+      "step": 195780
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059766238632597e-05,
+      "loss": 0.0076,
+      "step": 195790
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059718198894126e-05,
+      "loss": 0.0075,
+      "step": 195800
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059670159155656e-05,
+      "loss": 0.0081,
+      "step": 195810
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.905962211941718e-05,
+      "loss": 0.007,
+      "step": 195820
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.905957407967871e-05,
+      "loss": 0.0091,
+      "step": 195830
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.905952603994024e-05,
+      "loss": 0.0045,
+      "step": 195840
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.905947800020177e-05,
+      "loss": 0.0079,
+      "step": 195850
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059429960463295e-05,
+      "loss": 0.0101,
+      "step": 195860
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059381920724824e-05,
+      "loss": 0.005,
+      "step": 195870
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059333880986354e-05,
+      "loss": 0.0074,
+      "step": 195880
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.905928584124788e-05,
+      "loss": 0.0061,
+      "step": 195890
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.905923780150941e-05,
+      "loss": 0.0077,
+      "step": 195900
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059189761770938e-05,
+      "loss": 0.0063,
+      "step": 195910
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059141722032467e-05,
+      "loss": 0.0055,
+      "step": 195920
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059093682293993e-05,
+      "loss": 0.007,
+      "step": 195930
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059045642555522e-05,
+      "loss": 0.0067,
+      "step": 195940
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9058997602817052e-05,
+      "loss": 0.0089,
+      "step": 195950
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9058949563078578e-05,
+      "loss": 0.0079,
+      "step": 195960
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9058901523340107e-05,
+      "loss": 0.0171,
+      "step": 195970
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9058853483601636e-05,
+      "loss": 0.0079,
+      "step": 195980
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9058805443863165e-05,
+      "loss": 0.0047,
+      "step": 195990
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.905875740412469e-05,
+      "loss": 0.0074,
+      "step": 196000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.905870936438622e-05,
+      "loss": 0.0059,
+      "step": 196010
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905866132464775e-05,
+      "loss": 0.0069,
+      "step": 196020
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905861328490928e-05,
+      "loss": 0.004,
+      "step": 196030
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9058565245170805e-05,
+      "loss": 0.0066,
+      "step": 196040
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9058517205432334e-05,
+      "loss": 0.0053,
+      "step": 196050
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9058469165693864e-05,
+      "loss": 0.0033,
+      "step": 196060
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905842112595539e-05,
+      "loss": 0.0065,
+      "step": 196070
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905837308621692e-05,
+      "loss": 0.0088,
+      "step": 196080
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9058325046478448e-05,
+      "loss": 0.0052,
+      "step": 196090
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9058277006739977e-05,
+      "loss": 0.0075,
+      "step": 196100
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9058228967001503e-05,
+      "loss": 0.0078,
+      "step": 196110
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9058180927263032e-05,
+      "loss": 0.0037,
+      "step": 196120
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905813288752456e-05,
+      "loss": 0.0065,
+      "step": 196130
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9058084847786088e-05,
+      "loss": 0.0045,
+      "step": 196140
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9058036808047617e-05,
+      "loss": 0.0078,
+      "step": 196150
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9057988768309146e-05,
+      "loss": 0.0119,
+      "step": 196160
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9057940728570675e-05,
+      "loss": 0.0202,
+      "step": 196170
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.90578926888322e-05,
+      "loss": 0.0076,
+      "step": 196180
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9057844649093734e-05,
+      "loss": 0.0067,
+      "step": 196190
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905779660935526e-05,
+      "loss": 0.0061,
+      "step": 196200
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905774856961679e-05,
+      "loss": 0.0085,
+      "step": 196210
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905770052987832e-05,
+      "loss": 0.008,
+      "step": 196220
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9057652490139848e-05,
+      "loss": 0.0064,
+      "step": 196230
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9057604450401373e-05,
+      "loss": 0.0076,
+      "step": 196240
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9057556410662903e-05,
+      "loss": 0.0056,
+      "step": 196250
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9057508370924432e-05,
+      "loss": 0.008,
+      "step": 196260
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905746033118596e-05,
+      "loss": 0.0066,
+      "step": 196270
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9057412291447487e-05,
+      "loss": 0.0073,
+      "step": 196280
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9057364251709016e-05,
+      "loss": 0.0079,
+      "step": 196290
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9057316211970546e-05,
+      "loss": 0.0084,
+      "step": 196300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905726817223207e-05,
+      "loss": 0.0054,
+      "step": 196310
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.90572201324936e-05,
+      "loss": 0.0058,
+      "step": 196320
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905717209275513e-05,
+      "loss": 0.0059,
+      "step": 196330
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905712405301666e-05,
+      "loss": 0.0042,
+      "step": 196340
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9057076013278185e-05,
+      "loss": 0.0075,
+      "step": 196350
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9057027973539715e-05,
+      "loss": 0.0061,
+      "step": 196360
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9056979933801244e-05,
+      "loss": 0.0069,
+      "step": 196370
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905693189406277e-05,
+      "loss": 0.007,
+      "step": 196380
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.90568838543243e-05,
+      "loss": 0.0063,
+      "step": 196390
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9056835814585828e-05,
+      "loss": 0.0049,
+      "step": 196400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9056787774847357e-05,
+      "loss": 0.0047,
+      "step": 196410
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9056739735108883e-05,
+      "loss": 0.0056,
+      "step": 196420
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9056691695370413e-05,
+      "loss": 0.0047,
+      "step": 196430
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9056643655631942e-05,
+      "loss": 0.0089,
+      "step": 196440
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905659561589347e-05,
+      "loss": 0.0074,
+      "step": 196450
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9056547576154997e-05,
+      "loss": 0.0092,
+      "step": 196460
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9056499536416526e-05,
+      "loss": 0.0051,
+      "step": 196470
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9056451496678056e-05,
+      "loss": 0.005,
+      "step": 196480
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905640345693958e-05,
+      "loss": 0.0082,
+      "step": 196490
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905635541720111e-05,
+      "loss": 0.0079,
+      "step": 196500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905630737746264e-05,
+      "loss": 0.0076,
+      "step": 196510
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905625933772417e-05,
+      "loss": 0.0049,
+      "step": 196520
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9056211297985695e-05,
+      "loss": 0.0044,
+      "step": 196530
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9056163258247224e-05,
+      "loss": 0.0063,
+      "step": 196540
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9056115218508754e-05,
+      "loss": 0.0052,
+      "step": 196550
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905606717877028e-05,
+      "loss": 0.0098,
+      "step": 196560
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905601913903181e-05,
+      "loss": 0.0068,
+      "step": 196570
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9055971099293338e-05,
+      "loss": 0.0066,
+      "step": 196580
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9055923059554867e-05,
+      "loss": 0.008,
+      "step": 196590
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9055875019816393e-05,
+      "loss": 0.0086,
+      "step": 196600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9055826980077923e-05,
+      "loss": 0.0082,
+      "step": 196610
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9055778940339452e-05,
+      "loss": 0.0057,
+      "step": 196620
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9055730900600978e-05,
+      "loss": 0.007,
+      "step": 196630
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9055682860862507e-05,
+      "loss": 0.0101,
+      "step": 196640
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9055634821124036e-05,
+      "loss": 0.0077,
+      "step": 196650
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9055586781385566e-05,
+      "loss": 0.0063,
+      "step": 196660
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905553874164709e-05,
+      "loss": 0.0087,
+      "step": 196670
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905549070190862e-05,
+      "loss": 0.0058,
+      "step": 196680
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905544266217015e-05,
+      "loss": 0.0058,
+      "step": 196690
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905539462243168e-05,
+      "loss": 0.0054,
+      "step": 196700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9055346582693205e-05,
+      "loss": 0.0053,
+      "step": 196710
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9055298542954734e-05,
+      "loss": 0.0076,
+      "step": 196720
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9055250503216264e-05,
+      "loss": 0.0062,
+      "step": 196730
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905520246347779e-05,
+      "loss": 0.0071,
+      "step": 196740
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905515442373932e-05,
+      "loss": 0.0032,
+      "step": 196750
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9055106384000848e-05,
+      "loss": 0.0071,
+      "step": 196760
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9055058344262377e-05,
+      "loss": 0.0085,
+      "step": 196770
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9055010304523903e-05,
+      "loss": 0.0066,
+      "step": 196780
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054962264785432e-05,
+      "loss": 0.0055,
+      "step": 196790
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054914225046962e-05,
+      "loss": 0.0046,
+      "step": 196800
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054866185308488e-05,
+      "loss": 0.0059,
+      "step": 196810
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054818145570017e-05,
+      "loss": 0.0048,
+      "step": 196820
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054770105831546e-05,
+      "loss": 0.0063,
+      "step": 196830
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054722066093075e-05,
+      "loss": 0.0071,
+      "step": 196840
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.90546740263546e-05,
+      "loss": 0.0083,
+      "step": 196850
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905462598661613e-05,
+      "loss": 0.0059,
+      "step": 196860
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905457794687766e-05,
+      "loss": 0.009,
+      "step": 196870
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905452990713919e-05,
+      "loss": 0.0066,
+      "step": 196880
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054481867400715e-05,
+      "loss": 0.0085,
+      "step": 196890
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054433827662244e-05,
+      "loss": 0.0057,
+      "step": 196900
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054385787923774e-05,
+      "loss": 0.0062,
+      "step": 196910
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.90543377481853e-05,
+      "loss": 0.0118,
+      "step": 196920
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905428970844683e-05,
+      "loss": 0.0059,
+      "step": 196930
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054241668708358e-05,
+      "loss": 0.0079,
+      "step": 196940
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054193628969887e-05,
+      "loss": 0.0057,
+      "step": 196950
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054145589231413e-05,
+      "loss": 0.0065,
+      "step": 196960
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054097549492942e-05,
+      "loss": 0.005,
+      "step": 196970
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905404950975447e-05,
+      "loss": 0.007,
+      "step": 196980
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9054001470015998e-05,
+      "loss": 0.0067,
+      "step": 196990
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053953430277527e-05,
+      "loss": 0.0064,
+      "step": 197000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053905390539056e-05,
+      "loss": 0.0065,
+      "step": 197010
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053857350800585e-05,
+      "loss": 0.0062,
+      "step": 197020
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905380931106211e-05,
+      "loss": 0.0047,
+      "step": 197030
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905376127132364e-05,
+      "loss": 0.0079,
+      "step": 197040
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905371323158517e-05,
+      "loss": 0.006,
+      "step": 197050
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.90536651918467e-05,
+      "loss": 0.0077,
+      "step": 197060
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053617152108225e-05,
+      "loss": 0.0071,
+      "step": 197070
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053573916343603e-05,
+      "loss": 0.0045,
+      "step": 197080
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905352587660513e-05,
+      "loss": 0.0059,
+      "step": 197090
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053477836866658e-05,
+      "loss": 0.0057,
+      "step": 197100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053429797128187e-05,
+      "loss": 0.0051,
+      "step": 197110
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053381757389717e-05,
+      "loss": 0.0045,
+      "step": 197120
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053333717651243e-05,
+      "loss": 0.0074,
+      "step": 197130
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053285677912772e-05,
+      "loss": 0.0075,
+      "step": 197140
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.90532376381743e-05,
+      "loss": 0.0104,
+      "step": 197150
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053189598435827e-05,
+      "loss": 0.0062,
+      "step": 197160
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053141558697356e-05,
+      "loss": 0.0057,
+      "step": 197170
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053093518958885e-05,
+      "loss": 0.0065,
+      "step": 197180
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9053045479220415e-05,
+      "loss": 0.0049,
+      "step": 197190
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905299743948194e-05,
+      "loss": 0.0079,
+      "step": 197200
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905294939974347e-05,
+      "loss": 0.0072,
+      "step": 197210
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052901360005e-05,
+      "loss": 0.0107,
+      "step": 197220
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052853320266525e-05,
+      "loss": 0.008,
+      "step": 197230
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052805280528054e-05,
+      "loss": 0.0058,
+      "step": 197240
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052757240789584e-05,
+      "loss": 0.0057,
+      "step": 197250
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052709201051113e-05,
+      "loss": 0.0045,
+      "step": 197260
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905266116131264e-05,
+      "loss": 0.0072,
+      "step": 197270
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052613121574168e-05,
+      "loss": 0.0094,
+      "step": 197280
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052569885809543e-05,
+      "loss": 0.0058,
+      "step": 197290
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052521846071072e-05,
+      "loss": 0.0068,
+      "step": 197300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.90524738063326e-05,
+      "loss": 0.0066,
+      "step": 197310
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905242576659413e-05,
+      "loss": 0.0058,
+      "step": 197320
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052377726855656e-05,
+      "loss": 0.0057,
+      "step": 197330
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052329687117186e-05,
+      "loss": 0.0084,
+      "step": 197340
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052281647378715e-05,
+      "loss": 0.0085,
+      "step": 197350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052233607640244e-05,
+      "loss": 0.007,
+      "step": 197360
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905218556790177e-05,
+      "loss": 0.0047,
+      "step": 197370
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.90521375281633e-05,
+      "loss": 0.0071,
+      "step": 197380
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.905208948842483e-05,
+      "loss": 0.0066,
+      "step": 197390
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9052041448686354e-05,
+      "loss": 0.0089,
+      "step": 197400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051993408947884e-05,
+      "loss": 0.0052,
+      "step": 197410
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051945369209413e-05,
+      "loss": 0.006,
+      "step": 197420
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051897329470942e-05,
+      "loss": 0.0061,
+      "step": 197430
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051849289732468e-05,
+      "loss": 0.005,
+      "step": 197440
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051801249993997e-05,
+      "loss": 0.0046,
+      "step": 197450
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051753210255527e-05,
+      "loss": 0.0096,
+      "step": 197460
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051705170517053e-05,
+      "loss": 0.0065,
+      "step": 197470
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051657130778582e-05,
+      "loss": 0.0054,
+      "step": 197480
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.905160909104011e-05,
+      "loss": 0.0066,
+      "step": 197490
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.905156105130164e-05,
+      "loss": 0.0061,
+      "step": 197500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051513011563166e-05,
+      "loss": 0.0053,
+      "step": 197510
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051464971824696e-05,
+      "loss": 0.0051,
+      "step": 197520
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051416932086225e-05,
+      "loss": 0.0049,
+      "step": 197530
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051368892347754e-05,
+      "loss": 0.0065,
+      "step": 197540
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.905132085260928e-05,
+      "loss": 0.0068,
+      "step": 197550
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.905127281287081e-05,
+      "loss": 0.0063,
+      "step": 197560
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.905122477313234e-05,
+      "loss": 0.0089,
+      "step": 197570
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051176733393864e-05,
+      "loss": 0.008,
+      "step": 197580
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051128693655394e-05,
+      "loss": 0.0053,
+      "step": 197590
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051080653916923e-05,
+      "loss": 0.0057,
+      "step": 197600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051032614178452e-05,
+      "loss": 0.0082,
+      "step": 197610
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050984574439978e-05,
+      "loss": 0.004,
+      "step": 197620
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050936534701507e-05,
+      "loss": 0.0052,
+      "step": 197630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050888494963037e-05,
+      "loss": 0.0054,
+      "step": 197640
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050840455224563e-05,
+      "loss": 0.0073,
+      "step": 197650
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050792415486092e-05,
+      "loss": 0.0091,
+      "step": 197660
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.905074437574762e-05,
+      "loss": 0.005,
+      "step": 197670
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.905069633600915e-05,
+      "loss": 0.0064,
+      "step": 197680
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050648296270676e-05,
+      "loss": 0.008,
+      "step": 197690
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050600256532205e-05,
+      "loss": 0.0096,
+      "step": 197700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050552216793735e-05,
+      "loss": 0.0053,
+      "step": 197710
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050504177055264e-05,
+      "loss": 0.006,
+      "step": 197720
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.905045613731679e-05,
+      "loss": 0.0053,
+      "step": 197730
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.905040809757832e-05,
+      "loss": 0.0062,
+      "step": 197740
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.905036005783985e-05,
+      "loss": 0.0064,
+      "step": 197750
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050312018101374e-05,
+      "loss": 0.0078,
+      "step": 197760
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050263978362904e-05,
+      "loss": 0.0065,
+      "step": 197770
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050215938624433e-05,
+      "loss": 0.0072,
+      "step": 197780
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050167898885962e-05,
+      "loss": 0.0055,
+      "step": 197790
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050119859147488e-05,
+      "loss": 0.0066,
+      "step": 197800
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050071819409017e-05,
+      "loss": 0.0084,
+      "step": 197810
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9050023779670547e-05,
+      "loss": 0.0072,
+      "step": 197820
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049975739932072e-05,
+      "loss": 0.004,
+      "step": 197830
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049927700193602e-05,
+      "loss": 0.0089,
+      "step": 197840
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.904987966045513e-05,
+      "loss": 0.0077,
+      "step": 197850
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.904983162071666e-05,
+      "loss": 0.0051,
+      "step": 197860
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049783580978186e-05,
+      "loss": 0.007,
+      "step": 197870
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049735541239715e-05,
+      "loss": 0.0066,
+      "step": 197880
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049687501501245e-05,
+      "loss": 0.0069,
+      "step": 197890
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049639461762774e-05,
+      "loss": 0.0081,
+      "step": 197900
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.90495914220243e-05,
+      "loss": 0.0055,
+      "step": 197910
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.904954338228583e-05,
+      "loss": 0.0058,
+      "step": 197920
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.904949534254736e-05,
+      "loss": 0.0079,
+      "step": 197930
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049447302808884e-05,
+      "loss": 0.0047,
+      "step": 197940
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049399263070413e-05,
+      "loss": 0.0092,
+      "step": 197950
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049351223331943e-05,
+      "loss": 0.0084,
+      "step": 197960
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049303183593472e-05,
+      "loss": 0.008,
+      "step": 197970
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049255143854998e-05,
+      "loss": 0.0055,
+      "step": 197980
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049207104116527e-05,
+      "loss": 0.0064,
+      "step": 197990
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049159064378056e-05,
+      "loss": 0.0064,
+      "step": 198000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9049111024639582e-05,
+      "loss": 0.0079,
+      "step": 198010
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.904906298490111e-05,
+      "loss": 0.0081,
+      "step": 198020
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.904901494516264e-05,
+      "loss": 0.009,
+      "step": 198030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.904896690542417e-05,
+      "loss": 0.0074,
+      "step": 198040
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9048918865685696e-05,
+      "loss": 0.0051,
+      "step": 198050
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9048870825947225e-05,
+      "loss": 0.0053,
+      "step": 198060
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9048822786208755e-05,
+      "loss": 0.0069,
+      "step": 198070
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.904877474647028e-05,
+      "loss": 0.0076,
+      "step": 198080
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.904872670673181e-05,
+      "loss": 0.0065,
+      "step": 198090
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904867866699334e-05,
+      "loss": 0.0054,
+      "step": 198100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9048630627254868e-05,
+      "loss": 0.007,
+      "step": 198110
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9048582587516394e-05,
+      "loss": 0.0066,
+      "step": 198120
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9048534547777923e-05,
+      "loss": 0.0113,
+      "step": 198130
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9048486508039453e-05,
+      "loss": 0.004,
+      "step": 198140
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9048438468300982e-05,
+      "loss": 0.0073,
+      "step": 198150
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9048390428562508e-05,
+      "loss": 0.0074,
+      "step": 198160
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9048342388824037e-05,
+      "loss": 0.0057,
+      "step": 198170
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9048294349085566e-05,
+      "loss": 0.0065,
+      "step": 198180
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9048246309347092e-05,
+      "loss": 0.0053,
+      "step": 198190
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904819826960862e-05,
+      "loss": 0.0051,
+      "step": 198200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904815022987015e-05,
+      "loss": 0.0101,
+      "step": 198210
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904810219013168e-05,
+      "loss": 0.0093,
+      "step": 198220
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9048054150393206e-05,
+      "loss": 0.0081,
+      "step": 198230
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9048006110654735e-05,
+      "loss": 0.0062,
+      "step": 198240
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047958070916264e-05,
+      "loss": 0.0074,
+      "step": 198250
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904791003117779e-05,
+      "loss": 0.0059,
+      "step": 198260
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904786199143932e-05,
+      "loss": 0.0058,
+      "step": 198270
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904781395170085e-05,
+      "loss": 0.0072,
+      "step": 198280
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047765911962378e-05,
+      "loss": 0.0074,
+      "step": 198290
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047717872223904e-05,
+      "loss": 0.0052,
+      "step": 198300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047669832485433e-05,
+      "loss": 0.0086,
+      "step": 198310
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047621792746963e-05,
+      "loss": 0.007,
+      "step": 198320
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047573753008492e-05,
+      "loss": 0.0067,
+      "step": 198330
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047525713270018e-05,
+      "loss": 0.0079,
+      "step": 198340
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047477673531547e-05,
+      "loss": 0.0049,
+      "step": 198350
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047429633793076e-05,
+      "loss": 0.0074,
+      "step": 198360
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047381594054602e-05,
+      "loss": 0.0046,
+      "step": 198370
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904733355431613e-05,
+      "loss": 0.0041,
+      "step": 198380
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904729031855151e-05,
+      "loss": 0.0071,
+      "step": 198390
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047242278813035e-05,
+      "loss": 0.0062,
+      "step": 198400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047194239074565e-05,
+      "loss": 0.0081,
+      "step": 198410
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9047146199336094e-05,
+      "loss": 0.0067,
+      "step": 198420
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904709815959762e-05,
+      "loss": 0.0049,
+      "step": 198430
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904705011985915e-05,
+      "loss": 0.0086,
+      "step": 198440
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904700208012068e-05,
+      "loss": 0.0058,
+      "step": 198450
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046954040382208e-05,
+      "loss": 0.0063,
+      "step": 198460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046906000643733e-05,
+      "loss": 0.0068,
+      "step": 198470
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046857960905263e-05,
+      "loss": 0.0085,
+      "step": 198480
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046809921166792e-05,
+      "loss": 0.0058,
+      "step": 198490
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046761881428318e-05,
+      "loss": 0.0056,
+      "step": 198500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046713841689847e-05,
+      "loss": 0.0059,
+      "step": 198510
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046665801951376e-05,
+      "loss": 0.009,
+      "step": 198520
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046617762212906e-05,
+      "loss": 0.0078,
+      "step": 198530
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904656972247443e-05,
+      "loss": 0.0045,
+      "step": 198540
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904652168273596e-05,
+      "loss": 0.0063,
+      "step": 198550
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904647364299749e-05,
+      "loss": 0.008,
+      "step": 198560
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904642560325902e-05,
+      "loss": 0.0043,
+      "step": 198570
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046377563520545e-05,
+      "loss": 0.0069,
+      "step": 198580
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046329523782075e-05,
+      "loss": 0.0121,
+      "step": 198590
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046281484043604e-05,
+      "loss": 0.0056,
+      "step": 198600
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904623344430513e-05,
+      "loss": 0.0064,
+      "step": 198610
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904618540456666e-05,
+      "loss": 0.0045,
+      "step": 198620
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046137364828188e-05,
+      "loss": 0.0081,
+      "step": 198630
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046089325089718e-05,
+      "loss": 0.0061,
+      "step": 198640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046041285351243e-05,
+      "loss": 0.0042,
+      "step": 198650
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9045993245612773e-05,
+      "loss": 0.0066,
+      "step": 198660
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9045945205874302e-05,
+      "loss": 0.0057,
+      "step": 198670
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9045897166135828e-05,
+      "loss": 0.0038,
+      "step": 198680
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9045849126397357e-05,
+      "loss": 0.0059,
+      "step": 198690
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9045801086658886e-05,
+      "loss": 0.0062,
+      "step": 198700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9045753046920416e-05,
+      "loss": 0.0069,
+      "step": 198710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904570500718194e-05,
+      "loss": 0.0063,
+      "step": 198720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904565696744347e-05,
+      "loss": 0.0051,
+      "step": 198730
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9045608927705e-05,
+      "loss": 0.0048,
+      "step": 198740
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904556088796653e-05,
+      "loss": 0.0068,
+      "step": 198750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9045512848228055e-05,
+      "loss": 0.0059,
+      "step": 198760
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9045464808489584e-05,
+      "loss": 0.0051,
+      "step": 198770
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9045416768751114e-05,
+      "loss": 0.0068,
+      "step": 198780
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904536872901264e-05,
+      "loss": 0.0075,
+      "step": 198790
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904532068927417e-05,
+      "loss": 0.0065,
+      "step": 198800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9045272649535698e-05,
+      "loss": 0.008,
+      "step": 198810
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9045224609797227e-05,
+      "loss": 0.0064,
+      "step": 198820
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9045176570058753e-05,
+      "loss": 0.0048,
+      "step": 198830
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9045128530320283e-05,
+      "loss": 0.0108,
+      "step": 198840
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9045080490581812e-05,
+      "loss": 0.0071,
+      "step": 198850
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9045032450843338e-05,
+      "loss": 0.009,
+      "step": 198860
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044984411104867e-05,
+      "loss": 0.0106,
+      "step": 198870
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044936371366396e-05,
+      "loss": 0.0073,
+      "step": 198880
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044888331627926e-05,
+      "loss": 0.0064,
+      "step": 198890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904484029188945e-05,
+      "loss": 0.0082,
+      "step": 198900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904479225215098e-05,
+      "loss": 0.0068,
+      "step": 198910
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904474421241251e-05,
+      "loss": 0.008,
+      "step": 198920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904469617267404e-05,
+      "loss": 0.0085,
+      "step": 198930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044648132935565e-05,
+      "loss": 0.0065,
+      "step": 198940
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044600093197094e-05,
+      "loss": 0.0062,
+      "step": 198950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044552053458624e-05,
+      "loss": 0.0041,
+      "step": 198960
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904450401372015e-05,
+      "loss": 0.0041,
+      "step": 198970
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904445597398168e-05,
+      "loss": 0.0081,
+      "step": 198980
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044407934243208e-05,
+      "loss": 0.0092,
+      "step": 198990
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044359894504737e-05,
+      "loss": 0.0059,
+      "step": 199000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044311854766263e-05,
+      "loss": 0.0046,
+      "step": 199010
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044263815027792e-05,
+      "loss": 0.0064,
+      "step": 199020
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044215775289322e-05,
+      "loss": 0.0067,
+      "step": 199030
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044167735550848e-05,
+      "loss": 0.0054,
+      "step": 199040
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044119695812377e-05,
+      "loss": 0.0057,
+      "step": 199050
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044071656073906e-05,
+      "loss": 0.0067,
+      "step": 199060
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9044023616335435e-05,
+      "loss": 0.0055,
+      "step": 199070
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904397557659696e-05,
+      "loss": 0.0043,
+      "step": 199080
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904392753685849e-05,
+      "loss": 0.0047,
+      "step": 199090
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904387949712002e-05,
+      "loss": 0.0043,
+      "step": 199100
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904383145738155e-05,
+      "loss": 0.0043,
+      "step": 199110
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9043783417643075e-05,
+      "loss": 0.0088,
+      "step": 199120
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9043735377904604e-05,
+      "loss": 0.0058,
+      "step": 199130
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9043687338166134e-05,
+      "loss": 0.0068,
+      "step": 199140
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904363929842766e-05,
+      "loss": 0.0059,
+      "step": 199150
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904359125868919e-05,
+      "loss": 0.0073,
+      "step": 199160
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9043543218950718e-05,
+      "loss": 0.0052,
+      "step": 199170
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9043495179212247e-05,
+      "loss": 0.0053,
+      "step": 199180
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9043447139473773e-05,
+      "loss": 0.0052,
+      "step": 199190
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9043399099735302e-05,
+      "loss": 0.0101,
+      "step": 199200
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904335105999683e-05,
+      "loss": 0.0065,
+      "step": 199210
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9043303020258358e-05,
+      "loss": 0.0065,
+      "step": 199220
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9043254980519887e-05,
+      "loss": 0.0055,
+      "step": 199230
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9043206940781416e-05,
+      "loss": 0.0071,
+      "step": 199240
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9043158901042945e-05,
+      "loss": 0.0108,
+      "step": 199250
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904311086130447e-05,
+      "loss": 0.0053,
+      "step": 199260
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9043062821566e-05,
+      "loss": 0.0077,
+      "step": 199270
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904301478182753e-05,
+      "loss": 0.0075,
+      "step": 199280
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904296674208906e-05,
+      "loss": 0.0047,
+      "step": 199290
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9042918702350585e-05,
+      "loss": 0.0053,
+      "step": 199300
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9042870662612114e-05,
+      "loss": 0.005,
+      "step": 199310
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9042822622873643e-05,
+      "loss": 0.005,
+      "step": 199320
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904277458313517e-05,
+      "loss": 0.0035,
+      "step": 199330
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.90427265433967e-05,
+      "loss": 0.0055,
+      "step": 199340
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9042678503658228e-05,
+      "loss": 0.0051,
+      "step": 199350
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9042630463919757e-05,
+      "loss": 0.0049,
+      "step": 199360
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9042582424181283e-05,
+      "loss": 0.0073,
+      "step": 199370
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9042534384442812e-05,
+      "loss": 0.0093,
+      "step": 199380
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904248634470434e-05,
+      "loss": 0.0164,
+      "step": 199390
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9042438304965867e-05,
+      "loss": 0.0091,
+      "step": 199400
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9042390265227397e-05,
+      "loss": 0.0082,
+      "step": 199410
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9042342225488926e-05,
+      "loss": 0.009,
+      "step": 199420
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9042294185750455e-05,
+      "loss": 0.0075,
+      "step": 199430
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904224614601198e-05,
+      "loss": 0.0048,
+      "step": 199440
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904219810627351e-05,
+      "loss": 0.0077,
+      "step": 199450
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904215006653504e-05,
+      "loss": 0.007,
+      "step": 199460
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.904210202679657e-05,
+      "loss": 0.0064,
+      "step": 199470
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9042053987058095e-05,
+      "loss": 0.0055,
+      "step": 199480
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9042005947319624e-05,
+      "loss": 0.0082,
+      "step": 199490
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041957907581153e-05,
+      "loss": 0.0072,
+      "step": 199500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.904190986784268e-05,
+      "loss": 0.0092,
+      "step": 199510
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.904186182810421e-05,
+      "loss": 0.0057,
+      "step": 199520
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041813788365738e-05,
+      "loss": 0.0034,
+      "step": 199530
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041765748627267e-05,
+      "loss": 0.0071,
+      "step": 199540
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041717708888793e-05,
+      "loss": 0.0089,
+      "step": 199550
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041669669150322e-05,
+      "loss": 0.0063,
+      "step": 199560
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.904162162941185e-05,
+      "loss": 0.0085,
+      "step": 199570
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041573589673377e-05,
+      "loss": 0.0063,
+      "step": 199580
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041525549934907e-05,
+      "loss": 0.0074,
+      "step": 199590
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041477510196436e-05,
+      "loss": 0.005,
+      "step": 199600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041429470457965e-05,
+      "loss": 0.008,
+      "step": 199610
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.904138143071949e-05,
+      "loss": 0.0046,
+      "step": 199620
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.904133339098102e-05,
+      "loss": 0.0056,
+      "step": 199630
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.904128535124255e-05,
+      "loss": 0.007,
+      "step": 199640
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041237311504075e-05,
+      "loss": 0.0057,
+      "step": 199650
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041189271765605e-05,
+      "loss": 0.0117,
+      "step": 199660
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041141232027134e-05,
+      "loss": 0.0097,
+      "step": 199670
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9041093192288663e-05,
+      "loss": 0.0058,
+      "step": 199680
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.904104515255019e-05,
+      "loss": 0.0053,
+      "step": 199690
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.904099711281172e-05,
+      "loss": 0.0084,
+      "step": 199700
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040949073073248e-05,
+      "loss": 0.0069,
+      "step": 199710
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040901033334777e-05,
+      "loss": 0.0066,
+      "step": 199720
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040852993596303e-05,
+      "loss": 0.0063,
+      "step": 199730
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040804953857832e-05,
+      "loss": 0.0087,
+      "step": 199740
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.904075691411936e-05,
+      "loss": 0.0073,
+      "step": 199750
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040708874380887e-05,
+      "loss": 0.0079,
+      "step": 199760
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040660834642417e-05,
+      "loss": 0.0077,
+      "step": 199770
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040612794903946e-05,
+      "loss": 0.0076,
+      "step": 199780
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040564755165475e-05,
+      "loss": 0.0073,
+      "step": 199790
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040516715427e-05,
+      "loss": 0.0081,
+      "step": 199800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.904046867568853e-05,
+      "loss": 0.0072,
+      "step": 199810
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.904042063595006e-05,
+      "loss": 0.0069,
+      "step": 199820
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040372596211585e-05,
+      "loss": 0.0064,
+      "step": 199830
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040324556473115e-05,
+      "loss": 0.0038,
+      "step": 199840
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040276516734644e-05,
+      "loss": 0.0037,
+      "step": 199850
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040228476996173e-05,
+      "loss": 0.0043,
+      "step": 199860
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.90401804372577e-05,
+      "loss": 0.0068,
+      "step": 199870
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.904013239751923e-05,
+      "loss": 0.0093,
+      "step": 199880
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040084357780758e-05,
+      "loss": 0.0056,
+      "step": 199890
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9040036318042287e-05,
+      "loss": 0.0046,
+      "step": 199900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9039988278303813e-05,
+      "loss": 0.0049,
+      "step": 199910
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9039940238565342e-05,
+      "loss": 0.0081,
+      "step": 199920
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.903989219882687e-05,
+      "loss": 0.0056,
+      "step": 199930
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9039844159088397e-05,
+      "loss": 0.0056,
+      "step": 199940
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9039796119349926e-05,
+      "loss": 0.0105,
+      "step": 199950
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9039748079611456e-05,
+      "loss": 0.0093,
+      "step": 199960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9039700039872985e-05,
+      "loss": 0.0053,
+      "step": 199970
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.903965200013451e-05,
+      "loss": 0.0071,
+      "step": 199980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.903960396039604e-05,
+      "loss": 0.0036,
+      "step": 199990
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.903955592065757e-05,
+      "loss": 0.004,
+      "step": 200000
+    }
+  ],
+  "max_steps": 4163220,
+  "num_train_epochs": 60,
+  "total_flos": 1.2542511339243725e+18,
+  "trial_name": null,
+  "trial_params": null
+}