diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9022 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 15012,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.998667732480682e-05,
+      "loss": 3.8577,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.997335464961364e-05,
+      "loss": 1.113,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9960031974420465e-05,
+      "loss": 1.2262,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.994670929922729e-05,
+      "loss": 1.5109,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.993338662403411e-05,
+      "loss": 1.5867,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.992006394884093e-05,
+      "loss": 1.342,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.990674127364775e-05,
+      "loss": 1.2,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.989341859845457e-05,
+      "loss": 1.0629,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9880095923261392e-05,
+      "loss": 1.515,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9866773248068215e-05,
+      "loss": 1.1302,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9853450572875035e-05,
+      "loss": 1.0941,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9840127897681855e-05,
+      "loss": 0.9305,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9826805222488675e-05,
+      "loss": 0.9049,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.98134825472955e-05,
+      "loss": 1.3378,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.980015987210232e-05,
+      "loss": 1.5296,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9786837196909142e-05,
+      "loss": 0.9969,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9773514521715962e-05,
+      "loss": 1.0424,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9760191846522782e-05,
+      "loss": 1.1835,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9746869171329606e-05,
+      "loss": 0.8778,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9733546496136426e-05,
+      "loss": 1.0943,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9720223820943246e-05,
+      "loss": 1.3601,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.970690114575007e-05,
+      "loss": 1.2044,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.969357847055689e-05,
+      "loss": 1.0574,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.968025579536371e-05,
+      "loss": 1.1926,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9666933120170533e-05,
+      "loss": 1.1607,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9653610444977353e-05,
+      "loss": 0.9255,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9640287769784173e-05,
+      "loss": 1.3423,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9626965094590996e-05,
+      "loss": 1.2185,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9613642419397816e-05,
+      "loss": 0.8564,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.960031974420464e-05,
+      "loss": 1.0111,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.958699706901146e-05,
+      "loss": 1.1537,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.957367439381828e-05,
+      "loss": 0.8285,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.95603517186251e-05,
+      "loss": 0.7506,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9547029043431923e-05,
+      "loss": 0.9619,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9533706368238743e-05,
+      "loss": 1.1323,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9520383693045567e-05,
+      "loss": 1.1748,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9507061017852387e-05,
+      "loss": 1.0394,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9493738342659207e-05,
+      "loss": 1.4735,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.948041566746603e-05,
+      "loss": 1.0793,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.946709299227285e-05,
+      "loss": 1.1222,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9453770317079673e-05,
+      "loss": 1.2979,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9440447641886493e-05,
+      "loss": 1.2262,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9427124966693313e-05,
+      "loss": 0.7893,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9413802291500133e-05,
+      "loss": 1.0507,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9400479616306957e-05,
+      "loss": 1.0554,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9387156941113777e-05,
+      "loss": 0.8673,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.93738342659206e-05,
+      "loss": 0.8038,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.936051159072742e-05,
+      "loss": 1.1252,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.934718891553424e-05,
+      "loss": 0.798,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.933386624034106e-05,
+      "loss": 1.0203,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9320543565147884e-05,
+      "loss": 1.1765,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9307220889954704e-05,
+      "loss": 1.1677,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9293898214761527e-05,
+      "loss": 0.959,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9280575539568347e-05,
+      "loss": 1.2401,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9267252864375167e-05,
+      "loss": 1.1293,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9253930189181987e-05,
+      "loss": 1.0167,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.924060751398881e-05,
+      "loss": 1.0203,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.922728483879563e-05,
+      "loss": 1.01,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9213962163602454e-05,
+      "loss": 0.8721,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9200639488409274e-05,
+      "loss": 0.9725,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9187316813216094e-05,
+      "loss": 1.1241,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9173994138022914e-05,
+      "loss": 1.0535,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9160671462829738e-05,
+      "loss": 0.9679,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.914734878763656e-05,
+      "loss": 1.1622,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.913402611244338e-05,
+      "loss": 1.1936,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.91207034372502e-05,
+      "loss": 0.771,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.910738076205702e-05,
+      "loss": 0.6295,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9094058086863845e-05,
+      "loss": 1.0405,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9080735411670665e-05,
+      "loss": 1.2107,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9067412736477488e-05,
+      "loss": 0.9424,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9054090061284308e-05,
+      "loss": 1.1574,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9040767386091128e-05,
+      "loss": 0.7862,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9027444710897948e-05,
+      "loss": 1.1039,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.901412203570477e-05,
+      "loss": 0.9475,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.900079936051159e-05,
+      "loss": 0.7481,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8987476685318415e-05,
+      "loss": 0.7635,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8974154010125235e-05,
+      "loss": 0.9573,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8960831334932055e-05,
+      "loss": 0.9435,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8947508659738878e-05,
+      "loss": 0.8267,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.89341859845457e-05,
+      "loss": 0.9648,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.892086330935252e-05,
+      "loss": 0.6978,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8907540634159342e-05,
+      "loss": 0.7828,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8894217958966162e-05,
+      "loss": 1.3017,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8880895283772985e-05,
+      "loss": 1.0878,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8867572608579805e-05,
+      "loss": 0.8126,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8854249933386625e-05,
+      "loss": 1.1181,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8840927258193445e-05,
+      "loss": 1.1542,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.882760458300027e-05,
+      "loss": 0.7235,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.881428190780709e-05,
+      "loss": 0.9951,
+      "step": 890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8800959232613912e-05,
+      "loss": 0.957,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8787636557420732e-05,
+      "loss": 1.2426,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8774313882227552e-05,
+      "loss": 0.8761,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8760991207034372e-05,
+      "loss": 1.0079,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8747668531841196e-05,
+      "loss": 1.1776,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.873434585664802e-05,
+      "loss": 1.1795,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.872102318145484e-05,
+      "loss": 1.0256,
+      "step": 960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.870770050626166e-05,
+      "loss": 0.9522,
+      "step": 970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.869437783106848e-05,
+      "loss": 0.6613,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.86810551558753e-05,
+      "loss": 1.5441,
+      "step": 990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8667732480682123e-05,
+      "loss": 1.221,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8654409805488946e-05,
+      "loss": 0.8684,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8641087130295766e-05,
+      "loss": 0.8281,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8627764455102586e-05,
+      "loss": 0.8654,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8614441779909406e-05,
+      "loss": 1.2252,
+      "step": 1040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8601119104716226e-05,
+      "loss": 0.8553,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.858779642952305e-05,
+      "loss": 0.9223,
+      "step": 1060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8574473754329873e-05,
+      "loss": 1.1264,
+      "step": 1070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8561151079136693e-05,
+      "loss": 1.1312,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8547828403943513e-05,
+      "loss": 0.9532,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8534505728750333e-05,
+      "loss": 1.1151,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8521183053557156e-05,
+      "loss": 0.8711,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8507860378363976e-05,
+      "loss": 1.0257,
+      "step": 1120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.84945377031708e-05,
+      "loss": 1.0568,
+      "step": 1130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.848121502797762e-05,
+      "loss": 1.0497,
+      "step": 1140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.846789235278444e-05,
+      "loss": 0.8945,
+      "step": 1150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.845456967759126e-05,
+      "loss": 1.0013,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8441247002398083e-05,
+      "loss": 1.3454,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8427924327204903e-05,
+      "loss": 0.9102,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8414601652011727e-05,
+      "loss": 1.1129,
+      "step": 1190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8401278976818547e-05,
+      "loss": 1.0036,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8387956301625367e-05,
+      "loss": 0.9195,
+      "step": 1210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.837463362643219e-05,
+      "loss": 0.8959,
+      "step": 1220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.836131095123901e-05,
+      "loss": 1.0366,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.834798827604583e-05,
+      "loss": 1.3046,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8334665600852654e-05,
+      "loss": 0.8456,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8321342925659474e-05,
+      "loss": 0.8431,
+      "step": 1260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8308020250466294e-05,
+      "loss": 0.7821,
+      "step": 1270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8294697575273117e-05,
+      "loss": 1.1036,
+      "step": 1280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8281374900079937e-05,
+      "loss": 0.6821,
+      "step": 1290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8268052224886757e-05,
+      "loss": 0.9491,
+      "step": 1300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.825472954969358e-05,
+      "loss": 0.9321,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.82414068745004e-05,
+      "loss": 1.08,
+      "step": 1320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8228084199307224e-05,
+      "loss": 0.9724,
+      "step": 1330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8214761524114044e-05,
+      "loss": 0.9978,
+      "step": 1340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8201438848920864e-05,
+      "loss": 1.1282,
+      "step": 1350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8188116173727687e-05,
+      "loss": 1.3154,
+      "step": 1360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8174793498534507e-05,
+      "loss": 0.9287,
+      "step": 1370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.816147082334133e-05,
+      "loss": 1.0567,
+      "step": 1380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.814814814814815e-05,
+      "loss": 1.0243,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.813482547295497e-05,
+      "loss": 0.6169,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.812150279776179e-05,
+      "loss": 0.9887,
+      "step": 1410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8108180122568614e-05,
+      "loss": 0.9247,
+      "step": 1420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8094857447375434e-05,
+      "loss": 0.7982,
+      "step": 1430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8081534772182258e-05,
+      "loss": 0.9119,
+      "step": 1440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8068212096989078e-05,
+      "loss": 0.9946,
+      "step": 1450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8054889421795898e-05,
+      "loss": 0.732,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8041566746602718e-05,
+      "loss": 1.1512,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.802824407140954e-05,
+      "loss": 1.1482,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.801492139621636e-05,
+      "loss": 1.1026,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8001598721023185e-05,
+      "loss": 1.0065,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7988276045830005e-05,
+      "loss": 0.9553,
+      "step": 1510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7974953370636825e-05,
+      "loss": 1.1751,
+      "step": 1520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7961630695443645e-05,
+      "loss": 0.954,
+      "step": 1530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7948308020250468e-05,
+      "loss": 0.9903,
+      "step": 1540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7934985345057288e-05,
+      "loss": 1.2137,
+      "step": 1550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792166266986411e-05,
+      "loss": 1.2528,
+      "step": 1560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790833999467093e-05,
+      "loss": 1.2044,
+      "step": 1570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.789501731947775e-05,
+      "loss": 1.1712,
+      "step": 1580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.788169464428457e-05,
+      "loss": 0.8754,
+      "step": 1590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7868371969091395e-05,
+      "loss": 1.2087,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7855049293898215e-05,
+      "loss": 0.8691,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.784172661870504e-05,
+      "loss": 0.7455,
+      "step": 1620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.782840394351186e-05,
+      "loss": 0.7186,
+      "step": 1630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.781508126831868e-05,
+      "loss": 0.8787,
+      "step": 1640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7801758593125502e-05,
+      "loss": 1.0877,
+      "step": 1650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7788435917932322e-05,
+      "loss": 1.0954,
+      "step": 1660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7775113242739145e-05,
+      "loss": 1.0374,
+      "step": 1670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7761790567545965e-05,
+      "loss": 0.6339,
+      "step": 1680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7748467892352785e-05,
+      "loss": 0.8952,
+      "step": 1690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7735145217159605e-05,
+      "loss": 0.8032,
+      "step": 1700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.772182254196643e-05,
+      "loss": 0.8893,
+      "step": 1710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.770849986677325e-05,
+      "loss": 0.9514,
+      "step": 1720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7695177191580072e-05,
+      "loss": 0.9611,
+      "step": 1730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7681854516386892e-05,
+      "loss": 0.8059,
+      "step": 1740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7668531841193712e-05,
+      "loss": 0.6649,
+      "step": 1750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7655209166000536e-05,
+      "loss": 1.0976,
+      "step": 1760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7641886490807356e-05,
+      "loss": 0.7004,
+      "step": 1770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7628563815614176e-05,
+      "loss": 1.4345,
+      "step": 1780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7615241140421e-05,
+      "loss": 0.8439,
+      "step": 1790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.760191846522782e-05,
+      "loss": 0.7677,
+      "step": 1800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.758859579003464e-05,
+      "loss": 0.7943,
+      "step": 1810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7575273114841463e-05,
+      "loss": 0.9815,
+      "step": 1820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7561950439648283e-05,
+      "loss": 1.0038,
+      "step": 1830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548627764455103e-05,
+      "loss": 1.4571,
+      "step": 1840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7535305089261926e-05,
+      "loss": 0.7428,
+      "step": 1850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521982414068746e-05,
+      "loss": 1.1553,
+      "step": 1860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750865973887557e-05,
+      "loss": 1.1216,
+      "step": 1870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749533706368239e-05,
+      "loss": 0.8908,
+      "step": 1880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748201438848921e-05,
+      "loss": 0.8248,
+      "step": 1890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746869171329603e-05,
+      "loss": 0.893,
+      "step": 1900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7455369038102853e-05,
+      "loss": 1.123,
+      "step": 1910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442046362909676e-05,
+      "loss": 0.9023,
+      "step": 1920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428723687716496e-05,
+      "loss": 0.8609,
+      "step": 1930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415401012523316e-05,
+      "loss": 1.041,
+      "step": 1940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402078337330136e-05,
+      "loss": 0.9551,
+      "step": 1950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388755662136956e-05,
+      "loss": 0.9415,
+      "step": 1960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737543298694378e-05,
+      "loss": 1.0175,
+      "step": 1970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362110311750603e-05,
+      "loss": 0.9307,
+      "step": 1980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7348787636557423e-05,
+      "loss": 1.1073,
+      "step": 1990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7335464961364243e-05,
+      "loss": 0.8744,
+      "step": 2000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7322142286171063e-05,
+      "loss": 0.9754,
+      "step": 2010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7308819610977883e-05,
+      "loss": 1.2237,
+      "step": 2020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7295496935784707e-05,
+      "loss": 1.0859,
+      "step": 2030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.728217426059153e-05,
+      "loss": 0.7246,
+      "step": 2040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.726885158539835e-05,
+      "loss": 1.1179,
+      "step": 2050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.725552891020517e-05,
+      "loss": 1.0577,
+      "step": 2060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.724220623501199e-05,
+      "loss": 0.8672,
+      "step": 2070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.722888355981881e-05,
+      "loss": 1.0886,
+      "step": 2080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7215560884625634e-05,
+      "loss": 0.9956,
+      "step": 2090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7202238209432457e-05,
+      "loss": 1.1679,
+      "step": 2100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7188915534239277e-05,
+      "loss": 0.7628,
+      "step": 2110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7175592859046097e-05,
+      "loss": 1.0862,
+      "step": 2120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7162270183852917e-05,
+      "loss": 1.1052,
+      "step": 2130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.714894750865974e-05,
+      "loss": 0.977,
+      "step": 2140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.713562483346656e-05,
+      "loss": 1.0277,
+      "step": 2150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7122302158273384e-05,
+      "loss": 0.8481,
+      "step": 2160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7108979483080204e-05,
+      "loss": 0.9281,
+      "step": 2170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7095656807887024e-05,
+      "loss": 0.8152,
+      "step": 2180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7082334132693844e-05,
+      "loss": 0.8787,
+      "step": 2190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7069011457500668e-05,
+      "loss": 0.6508,
+      "step": 2200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7055688782307488e-05,
+      "loss": 1.0112,
+      "step": 2210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.704236610711431e-05,
+      "loss": 0.8638,
+      "step": 2220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.702904343192113e-05,
+      "loss": 0.7121,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.701572075672795e-05,
+      "loss": 0.9776,
+      "step": 2240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7002398081534774e-05,
+      "loss": 1.076,
+      "step": 2250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6989075406341594e-05,
+      "loss": 1.0184,
+      "step": 2260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6975752731148414e-05,
+      "loss": 1.1203,
+      "step": 2270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6962430055955238e-05,
+      "loss": 0.6124,
+      "step": 2280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6949107380762058e-05,
+      "loss": 1.0818,
+      "step": 2290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.693578470556888e-05,
+      "loss": 0.7722,
+      "step": 2300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.69224620303757e-05,
+      "loss": 0.8718,
+      "step": 2310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.690913935518252e-05,
+      "loss": 1.4208,
+      "step": 2320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.689581667998934e-05,
+      "loss": 0.641,
+      "step": 2330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6882494004796165e-05,
+      "loss": 1.188,
+      "step": 2340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6869171329602985e-05,
+      "loss": 0.6591,
+      "step": 2350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6855848654409808e-05,
+      "loss": 1.0678,
+      "step": 2360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6842525979216628e-05,
+      "loss": 0.8758,
+      "step": 2370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6829203304023448e-05,
+      "loss": 1.2125,
+      "step": 2380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.681588062883027e-05,
+      "loss": 0.9793,
+      "step": 2390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6802557953637092e-05,
+      "loss": 0.8843,
+      "step": 2400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6789235278443915e-05,
+      "loss": 0.8358,
+      "step": 2410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6775912603250735e-05,
+      "loss": 0.9347,
+      "step": 2420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6762589928057555e-05,
+      "loss": 0.9711,
+      "step": 2430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6749267252864375e-05,
+      "loss": 0.8316,
+      "step": 2440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.67359445776712e-05,
+      "loss": 0.9148,
+      "step": 2450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.672262190247802e-05,
+      "loss": 0.9615,
+      "step": 2460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6709299227284842e-05,
+      "loss": 0.902,
+      "step": 2470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6695976552091662e-05,
+      "loss": 0.8277,
+      "step": 2480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6682653876898482e-05,
+      "loss": 0.9935,
+      "step": 2490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6669331201705302e-05,
+      "loss": 1.1443,
+      "step": 2500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6656008526512126e-05,
+      "loss": 0.8983,
+      "step": 2510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6642685851318946e-05,
+      "loss": 0.8136,
+      "step": 2520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.662936317612577e-05,
+      "loss": 0.9253,
+      "step": 2530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.661604050093259e-05,
+      "loss": 1.2926,
+      "step": 2540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.660271782573941e-05,
+      "loss": 1.0296,
+      "step": 2550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.658939515054623e-05,
+      "loss": 0.9648,
+      "step": 2560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6576072475353052e-05,
+      "loss": 1.1027,
+      "step": 2570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6562749800159872e-05,
+      "loss": 1.277,
+      "step": 2580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6549427124966696e-05,
+      "loss": 0.7621,
+      "step": 2590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6536104449773516e-05,
+      "loss": 0.969,
+      "step": 2600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6522781774580336e-05,
+      "loss": 1.2513,
+      "step": 2610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6509459099387156e-05,
+      "loss": 0.9775,
+      "step": 2620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.649613642419398e-05,
+      "loss": 0.7067,
+      "step": 2630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6482813749000803e-05,
+      "loss": 1.2218,
+      "step": 2640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6469491073807623e-05,
+      "loss": 0.9326,
+      "step": 2650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6456168398614443e-05,
+      "loss": 0.8712,
+      "step": 2660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6442845723421263e-05,
+      "loss": 0.9815,
+      "step": 2670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6429523048228086e-05,
+      "loss": 0.8492,
+      "step": 2680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6416200373034906e-05,
+      "loss": 1.2447,
+      "step": 2690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.640287769784173e-05,
+      "loss": 0.954,
+      "step": 2700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.638955502264855e-05,
+      "loss": 0.9911,
+      "step": 2710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.637623234745537e-05,
+      "loss": 0.6388,
+      "step": 2720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.636290967226219e-05,
+      "loss": 0.8665,
+      "step": 2730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6349586997069013e-05,
+      "loss": 1.04,
+      "step": 2740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6336264321875833e-05,
+      "loss": 1.1303,
+      "step": 2750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6322941646682657e-05,
+      "loss": 0.7516,
+      "step": 2760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6309618971489477e-05,
+      "loss": 0.9283,
+      "step": 2770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6296296296296297e-05,
+      "loss": 0.9545,
+      "step": 2780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.628297362110312e-05,
+      "loss": 0.9962,
+      "step": 2790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.626965094590994e-05,
+      "loss": 0.6158,
+      "step": 2800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.625632827071676e-05,
+      "loss": 1.098,
+      "step": 2810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6243005595523583e-05,
+      "loss": 1.0136,
+      "step": 2820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6229682920330404e-05,
+      "loss": 1.2423,
+      "step": 2830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6216360245137227e-05,
+      "loss": 1.0788,
+      "step": 2840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6203037569944047e-05,
+      "loss": 1.0617,
+      "step": 2850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6189714894750867e-05,
+      "loss": 1.2612,
+      "step": 2860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6176392219557687e-05,
+      "loss": 0.9274,
+      "step": 2870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.616306954436451e-05,
+      "loss": 1.3712,
+      "step": 2880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.614974686917133e-05,
+      "loss": 1.1587,
+      "step": 2890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6136424193978154e-05,
+      "loss": 0.9279,
+      "step": 2900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6123101518784974e-05,
+      "loss": 0.9461,
+      "step": 2910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6109778843591794e-05,
+      "loss": 0.5225,
+      "step": 2920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6096456168398614e-05,
+      "loss": 0.7481,
+      "step": 2930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6083133493205437e-05,
+      "loss": 0.8047,
+      "step": 2940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.606981081801226e-05,
+      "loss": 1.2123,
+      "step": 2950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.605648814281908e-05,
+      "loss": 0.6625,
+      "step": 2960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.60431654676259e-05,
+      "loss": 0.8275,
+      "step": 2970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.602984279243272e-05,
+      "loss": 0.8503,
+      "step": 2980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.601652011723954e-05,
+      "loss": 1.1244,
+      "step": 2990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6003197442046364e-05,
+      "loss": 1.022,
+      "step": 3000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5989874766853188e-05,
+      "loss": 0.8131,
+      "step": 3010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5976552091660008e-05,
+      "loss": 0.7888,
+      "step": 3020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5963229416466828e-05,
+      "loss": 0.8245,
+      "step": 3030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5949906741273648e-05,
+      "loss": 0.7496,
+      "step": 3040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5936584066080468e-05,
+      "loss": 1.0171,
+      "step": 3050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.592326139088729e-05,
+      "loss": 0.8417,
+      "step": 3060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5909938715694115e-05,
+      "loss": 0.792,
+      "step": 3070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5896616040500935e-05,
+      "loss": 1.0214,
+      "step": 3080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5883293365307755e-05,
+      "loss": 0.9708,
+      "step": 3090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5869970690114575e-05,
+      "loss": 0.7127,
+      "step": 3100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5856648014921398e-05,
+      "loss": 0.7655,
+      "step": 3110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5843325339728218e-05,
+      "loss": 0.7671,
+      "step": 3120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.583000266453504e-05,
+      "loss": 0.7252,
+      "step": 3130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.581667998934186e-05,
+      "loss": 0.8863,
+      "step": 3140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.580335731414868e-05,
+      "loss": 1.1643,
+      "step": 3150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.57900346389555e-05,
+      "loss": 0.9615,
+      "step": 3160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5776711963762325e-05,
+      "loss": 1.1006,
+      "step": 3170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5763389288569145e-05,
+      "loss": 1.1006,
+      "step": 3180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.575006661337597e-05,
+      "loss": 0.7646,
+      "step": 3190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.573674393818279e-05,
+      "loss": 0.7419,
+      "step": 3200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.572342126298961e-05,
+      "loss": 0.8336,
+      "step": 3210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5710098587796432e-05,
+      "loss": 1.0134,
+      "step": 3220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5696775912603252e-05,
+      "loss": 0.9505,
+      "step": 3230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5683453237410072e-05,
+      "loss": 0.6422,
+      "step": 3240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5670130562216895e-05,
+      "loss": 0.929,
+      "step": 3250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5656807887023715e-05,
+      "loss": 1.0643,
+      "step": 3260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5643485211830535e-05,
+      "loss": 0.9868,
+      "step": 3270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.563016253663736e-05,
+      "loss": 1.0235,
+      "step": 3280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.561683986144418e-05,
+      "loss": 0.8494,
+      "step": 3290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5603517186251e-05,
+      "loss": 0.5709,
+      "step": 3300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5590194511057822e-05,
+      "loss": 0.978,
+      "step": 3310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5576871835864642e-05,
+      "loss": 0.9563,
+      "step": 3320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5563549160671466e-05,
+      "loss": 0.8088,
+      "step": 3330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5550226485478286e-05,
+      "loss": 1.116,
+      "step": 3340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5536903810285106e-05,
+      "loss": 1.0865,
+      "step": 3350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.552358113509193e-05,
+      "loss": 0.8222,
+      "step": 3360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.551025845989875e-05,
+      "loss": 0.8851,
+      "step": 3370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5496935784705573e-05,
+      "loss": 0.7073,
+      "step": 3380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5483613109512393e-05,
+      "loss": 0.8308,
+      "step": 3390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5470290434319213e-05,
+      "loss": 0.9575,
+      "step": 3400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5456967759126033e-05,
+      "loss": 1.0262,
+      "step": 3410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5443645083932856e-05,
+      "loss": 0.7715,
+      "step": 3420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5430322408739676e-05,
+      "loss": 0.6188,
+      "step": 3430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.54169997335465e-05,
+      "loss": 1.0241,
+      "step": 3440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.540367705835332e-05,
+      "loss": 1.2733,
+      "step": 3450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.539035438316014e-05,
+      "loss": 1.0629,
+      "step": 3460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.537703170796696e-05,
+      "loss": 1.0838,
+      "step": 3470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5363709032773783e-05,
+      "loss": 0.9539,
+      "step": 3480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5350386357580603e-05,
+      "loss": 0.9519,
+      "step": 3490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5337063682387426e-05,
+      "loss": 1.0351,
+      "step": 3500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5323741007194246e-05,
+      "loss": 1.3052,
+      "step": 3510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5310418332001066e-05,
+      "loss": 0.7774,
+      "step": 3520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5297095656807886e-05,
+      "loss": 0.7707,
+      "step": 3530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.528377298161471e-05,
+      "loss": 0.5777,
+      "step": 3540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.527045030642153e-05,
+      "loss": 1.0646,
+      "step": 3550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5257127631228352e-05,
+      "loss": 0.7333,
+      "step": 3560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5243804956035173e-05,
+      "loss": 0.854,
+      "step": 3570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5230482280841993e-05,
+      "loss": 1.2105,
+      "step": 3580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5217159605648815e-05,
+      "loss": 0.7533,
+      "step": 3590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5203836930455638e-05,
+      "loss": 0.8481,
+      "step": 3600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5190514255262458e-05,
+      "loss": 1.169,
+      "step": 3610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5177191580069278e-05,
+      "loss": 1.2347,
+      "step": 3620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.51638689048761e-05,
+      "loss": 0.8002,
+      "step": 3630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.515054622968292e-05,
+      "loss": 0.9941,
+      "step": 3640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5137223554489742e-05,
+      "loss": 1.1608,
+      "step": 3650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5123900879296565e-05,
+      "loss": 1.2734,
+      "step": 3660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5110578204103385e-05,
+      "loss": 1.1145,
+      "step": 3670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5097255528910207e-05,
+      "loss": 0.8585,
+      "step": 3680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5083932853717027e-05,
+      "loss": 0.7982,
+      "step": 3690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5070610178523847e-05,
+      "loss": 0.8173,
+      "step": 3700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.505728750333067e-05,
+      "loss": 0.8473,
+      "step": 3710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5043964828137492e-05,
+      "loss": 0.8075,
+      "step": 3720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5030642152944312e-05,
+      "loss": 0.7314,
+      "step": 3730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5017319477751134e-05,
+      "loss": 0.8119,
+      "step": 3740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5003996802557954e-05,
+      "loss": 0.8956,
+      "step": 3750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4990674127364777e-05,
+      "loss": 1.1716,
+      "step": 3760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4977351452171597e-05,
+      "loss": 0.8228,
+      "step": 3770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.496402877697842e-05,
+      "loss": 0.7261,
+      "step": 3780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.495070610178524e-05,
+      "loss": 0.8364,
+      "step": 3790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4937383426592061e-05,
+      "loss": 0.965,
+      "step": 3800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4924060751398881e-05,
+      "loss": 0.8599,
+      "step": 3810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4910738076205704e-05,
+      "loss": 1.0879,
+      "step": 3820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4897415401012524e-05,
+      "loss": 0.9431,
+      "step": 3830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4884092725819346e-05,
+      "loss": 1.1419,
+      "step": 3840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4870770050626166e-05,
+      "loss": 0.8545,
+      "step": 3850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4857447375432988e-05,
+      "loss": 0.8475,
+      "step": 3860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.484412470023981e-05,
+      "loss": 0.7707,
+      "step": 3870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4830802025046631e-05,
+      "loss": 0.9314,
+      "step": 3880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4817479349853451e-05,
+      "loss": 0.773,
+      "step": 3890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4804156674660273e-05,
+      "loss": 0.7724,
+      "step": 3900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4790833999467093e-05,
+      "loss": 1.022,
+      "step": 3910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4777511324273915e-05,
+      "loss": 0.5351,
+      "step": 3920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4764188649080738e-05,
+      "loss": 1.0508,
+      "step": 3930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4750865973887558e-05,
+      "loss": 0.7651,
+      "step": 3940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4737543298694378e-05,
+      "loss": 1.1721,
+      "step": 3950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.47242206235012e-05,
+      "loss": 0.9186,
+      "step": 3960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.471089794830802e-05,
+      "loss": 0.9167,
+      "step": 3970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4697575273114843e-05,
+      "loss": 0.8532,
+      "step": 3980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4684252597921665e-05,
+      "loss": 0.8338,
+      "step": 3990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4670929922728485e-05,
+      "loss": 0.6336,
+      "step": 4000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4657607247535305e-05,
+      "loss": 0.7279,
+      "step": 4010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4644284572342127e-05,
+      "loss": 0.8743,
+      "step": 4020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.463096189714895e-05,
+      "loss": 0.8711,
+      "step": 4030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.461763922195577e-05,
+      "loss": 0.8079,
+      "step": 4040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4604316546762592e-05,
+      "loss": 0.8619,
+      "step": 4050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4590993871569412e-05,
+      "loss": 0.7443,
+      "step": 4060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4577671196376234e-05,
+      "loss": 0.7558,
+      "step": 4070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4564348521183054e-05,
+      "loss": 0.9821,
+      "step": 4080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4551025845989877e-05,
+      "loss": 0.7893,
+      "step": 4090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4537703170796697e-05,
+      "loss": 0.9171,
+      "step": 4100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4524380495603519e-05,
+      "loss": 0.6518,
+      "step": 4110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4511057820410339e-05,
+      "loss": 0.9738,
+      "step": 4120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.449773514521716e-05,
+      "loss": 0.9537,
+      "step": 4130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4484412470023982e-05,
+      "loss": 0.8501,
+      "step": 4140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4471089794830804e-05,
+      "loss": 0.8564,
+      "step": 4150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4457767119637624e-05,
+      "loss": 0.7312,
+      "step": 4160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4444444444444446e-05,
+      "loss": 0.7642,
+      "step": 4170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4431121769251266e-05,
+      "loss": 1.0306,
+      "step": 4180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4417799094058088e-05,
+      "loss": 0.8579,
+      "step": 4190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.440447641886491e-05,
+      "loss": 0.8951,
+      "step": 4200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4391153743671731e-05,
+      "loss": 0.8076,
+      "step": 4210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4377831068478551e-05,
+      "loss": 1.0002,
+      "step": 4220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4364508393285373e-05,
+      "loss": 1.0057,
+      "step": 4230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4351185718092193e-05,
+      "loss": 1.1086,
+      "step": 4240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4337863042899016e-05,
+      "loss": 0.8413,
+      "step": 4250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4324540367705836e-05,
+      "loss": 0.9097,
+      "step": 4260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4311217692512658e-05,
+      "loss": 1.0422,
+      "step": 4270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4297895017319478e-05,
+      "loss": 0.91,
+      "step": 4280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.42845723421263e-05,
+      "loss": 1.0644,
+      "step": 4290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4271249666933123e-05,
+      "loss": 0.7929,
+      "step": 4300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4257926991739943e-05,
+      "loss": 0.8431,
+      "step": 4310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4244604316546765e-05,
+      "loss": 0.9138,
+      "step": 4320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4231281641353585e-05,
+      "loss": 1.1477,
+      "step": 4330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4217958966160405e-05,
+      "loss": 0.6862,
+      "step": 4340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4204636290967227e-05,
+      "loss": 1.0002,
+      "step": 4350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.419131361577405e-05,
+      "loss": 1.1307,
+      "step": 4360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.417799094058087e-05,
+      "loss": 0.804,
+      "step": 4370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4164668265387692e-05,
+      "loss": 0.8695,
+      "step": 4380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4151345590194512e-05,
+      "loss": 0.7704,
+      "step": 4390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4138022915001333e-05,
+      "loss": 0.7866,
+      "step": 4400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4124700239808155e-05,
+      "loss": 0.9987,
+      "step": 4410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4111377564614977e-05,
+      "loss": 0.6727,
+      "step": 4420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4098054889421797e-05,
+      "loss": 0.7705,
+      "step": 4430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4084732214228619e-05,
+      "loss": 0.8893,
+      "step": 4440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4071409539035439e-05,
+      "loss": 0.9634,
+      "step": 4450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.405808686384226e-05,
+      "loss": 0.8511,
+      "step": 4460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4044764188649082e-05,
+      "loss": 1.1291,
+      "step": 4470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4031441513455904e-05,
+      "loss": 0.8015,
+      "step": 4480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4018118838262724e-05,
+      "loss": 0.8958,
+      "step": 4490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4004796163069546e-05,
+      "loss": 0.8095,
+      "step": 4500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3991473487876366e-05,
+      "loss": 1.0297,
+      "step": 4510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3978150812683189e-05,
+      "loss": 0.9276,
+      "step": 4520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3964828137490009e-05,
+      "loss": 0.7591,
+      "step": 4530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.395150546229683e-05,
+      "loss": 0.9263,
+      "step": 4540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.393818278710365e-05,
+      "loss": 1.0132,
+      "step": 4550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3924860111910472e-05,
+      "loss": 0.8801,
+      "step": 4560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3911537436717296e-05,
+      "loss": 0.8833,
+      "step": 4570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3898214761524116e-05,
+      "loss": 0.9065,
+      "step": 4580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3884892086330936e-05,
+      "loss": 0.748,
+      "step": 4590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3871569411137758e-05,
+      "loss": 1.0739,
+      "step": 4600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3858246735944578e-05,
+      "loss": 1.0306,
+      "step": 4610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.38449240607514e-05,
+      "loss": 1.0399,
+      "step": 4620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3831601385558223e-05,
+      "loss": 0.8172,
+      "step": 4630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3818278710365043e-05,
+      "loss": 0.9688,
+      "step": 4640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3804956035171865e-05,
+      "loss": 1.1328,
+      "step": 4650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3791633359978685e-05,
+      "loss": 0.5959,
+      "step": 4660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3778310684785505e-05,
+      "loss": 0.8724,
+      "step": 4670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3764988009592328e-05,
+      "loss": 1.0068,
+      "step": 4680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.375166533439915e-05,
+      "loss": 1.1397,
+      "step": 4690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.373834265920597e-05,
+      "loss": 1.1376,
+      "step": 4700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3725019984012791e-05,
+      "loss": 0.617,
+      "step": 4710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3711697308819611e-05,
+      "loss": 0.8408,
+      "step": 4720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3698374633626431e-05,
+      "loss": 0.8512,
+      "step": 4730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3685051958433255e-05,
+      "loss": 0.8946,
+      "step": 4740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3671729283240077e-05,
+      "loss": 0.8904,
+      "step": 4750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3658406608046897e-05,
+      "loss": 0.992,
+      "step": 4760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3645083932853718e-05,
+      "loss": 1.024,
+      "step": 4770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3631761257660538e-05,
+      "loss": 0.9548,
+      "step": 4780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3618438582467362e-05,
+      "loss": 0.5628,
+      "step": 4790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3605115907274182e-05,
+      "loss": 0.924,
+      "step": 4800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3591793232081004e-05,
+      "loss": 0.7587,
+      "step": 4810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3578470556887824e-05,
+      "loss": 1.0084,
+      "step": 4820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3565147881694645e-05,
+      "loss": 0.8048,
+      "step": 4830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3551825206501467e-05,
+      "loss": 0.6874,
+      "step": 4840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3538502531308289e-05,
+      "loss": 0.7296,
+      "step": 4850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3525179856115109e-05,
+      "loss": 0.9631,
+      "step": 4860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.351185718092193e-05,
+      "loss": 0.9696,
+      "step": 4870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.349853450572875e-05,
+      "loss": 0.9157,
+      "step": 4880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3485211830535572e-05,
+      "loss": 0.556,
+      "step": 4890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3471889155342394e-05,
+      "loss": 0.8368,
+      "step": 4900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3458566480149216e-05,
+      "loss": 0.8747,
+      "step": 4910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3445243804956036e-05,
+      "loss": 1.0028,
+      "step": 4920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3431921129762857e-05,
+      "loss": 0.9508,
+      "step": 4930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3418598454569677e-05,
+      "loss": 0.9811,
+      "step": 4940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.34052757793765e-05,
+      "loss": 0.8648,
+      "step": 4950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3391953104183322e-05,
+      "loss": 0.4586,
+      "step": 4960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3378630428990143e-05,
+      "loss": 0.8933,
+      "step": 4970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3365307753796963e-05,
+      "loss": 0.9248,
+      "step": 4980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3351985078603784e-05,
+      "loss": 1.0249,
+      "step": 4990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3338662403410604e-05,
+      "loss": 0.8078,
+      "step": 5000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3325339728217428e-05,
+      "loss": 0.8139,
+      "step": 5010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.331201705302425e-05,
+      "loss": 0.5048,
+      "step": 5020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.329869437783107e-05,
+      "loss": 0.7647,
+      "step": 5030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3285371702637891e-05,
+      "loss": 0.7331,
+      "step": 5040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3272049027444711e-05,
+      "loss": 1.1092,
+      "step": 5050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3258726352251535e-05,
+      "loss": 0.7277,
+      "step": 5060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3245403677058355e-05,
+      "loss": 0.7868,
+      "step": 5070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3232081001865176e-05,
+      "loss": 0.8461,
+      "step": 5080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3218758326671996e-05,
+      "loss": 0.7978,
+      "step": 5090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3205435651478818e-05,
+      "loss": 0.9252,
+      "step": 5100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3192112976285638e-05,
+      "loss": 0.7401,
+      "step": 5110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3178790301092461e-05,
+      "loss": 0.7592,
+      "step": 5120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3165467625899282e-05,
+      "loss": 1.158,
+      "step": 5130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3152144950706103e-05,
+      "loss": 0.6913,
+      "step": 5140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3138822275512923e-05,
+      "loss": 0.6624,
+      "step": 5150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3125499600319745e-05,
+      "loss": 0.8496,
+      "step": 5160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3112176925126567e-05,
+      "loss": 0.7897,
+      "step": 5170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3098854249933388e-05,
+      "loss": 0.862,
+      "step": 5180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3085531574740208e-05,
+      "loss": 0.7539,
+      "step": 5190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.307220889954703e-05,
+      "loss": 0.7561,
+      "step": 5200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.305888622435385e-05,
+      "loss": 0.7537,
+      "step": 5210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3045563549160674e-05,
+      "loss": 1.1637,
+      "step": 5220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3032240873967494e-05,
+      "loss": 0.9744,
+      "step": 5230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3018918198774315e-05,
+      "loss": 1.0448,
+      "step": 5240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3005595523581135e-05,
+      "loss": 0.8154,
+      "step": 5250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2992272848387957e-05,
+      "loss": 0.9379,
+      "step": 5260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2978950173194777e-05,
+      "loss": 1.0857,
+      "step": 5270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.29656274980016e-05,
+      "loss": 0.7569,
+      "step": 5280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2952304822808422e-05,
+      "loss": 0.8747,
+      "step": 5290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2938982147615242e-05,
+      "loss": 0.8236,
+      "step": 5300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2925659472422062e-05,
+      "loss": 0.9229,
+      "step": 5310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2912336797228884e-05,
+      "loss": 0.7699,
+      "step": 5320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2899014122035707e-05,
+      "loss": 1.0339,
+      "step": 5330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2885691446842527e-05,
+      "loss": 0.5565,
+      "step": 5340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2872368771649349e-05,
+      "loss": 1.1413,
+      "step": 5350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2859046096456169e-05,
+      "loss": 0.7633,
+      "step": 5360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.284572342126299e-05,
+      "loss": 0.9841,
+      "step": 5370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2832400746069811e-05,
+      "loss": 0.8538,
+      "step": 5380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2819078070876634e-05,
+      "loss": 0.8028,
+      "step": 5390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2805755395683454e-05,
+      "loss": 1.1235,
+      "step": 5400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2792432720490276e-05,
+      "loss": 1.1321,
+      "step": 5410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2779110045297096e-05,
+      "loss": 0.9859,
+      "step": 5420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2765787370103918e-05,
+      "loss": 0.987,
+      "step": 5430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.275246469491074e-05,
+      "loss": 0.9717,
+      "step": 5440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2739142019717561e-05,
+      "loss": 1.0352,
+      "step": 5450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2725819344524381e-05,
+      "loss": 0.8472,
+      "step": 5460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2712496669331203e-05,
+      "loss": 0.6774,
+      "step": 5470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2699173994138023e-05,
+      "loss": 0.9617,
+      "step": 5480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2685851318944846e-05,
+      "loss": 0.7842,
+      "step": 5490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2672528643751666e-05,
+      "loss": 0.9939,
+      "step": 5500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2659205968558488e-05,
+      "loss": 0.9137,
+      "step": 5510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2645883293365308e-05,
+      "loss": 0.7807,
+      "step": 5520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.263256061817213e-05,
+      "loss": 0.7988,
+      "step": 5530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.261923794297895e-05,
+      "loss": 0.9518,
+      "step": 5540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2605915267785773e-05,
+      "loss": 0.6778,
+      "step": 5550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2592592592592593e-05,
+      "loss": 1.0503,
+      "step": 5560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2579269917399415e-05,
+      "loss": 0.7174,
+      "step": 5570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2565947242206235e-05,
+      "loss": 0.7978,
+      "step": 5580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2552624567013057e-05,
+      "loss": 0.839,
+      "step": 5590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.253930189181988e-05,
+      "loss": 1.3887,
+      "step": 5600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.25259792166267e-05,
+      "loss": 0.708,
+      "step": 5610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.251265654143352e-05,
+      "loss": 0.8197,
+      "step": 5620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2499333866240342e-05,
+      "loss": 0.5705,
+      "step": 5630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2486011191047162e-05,
+      "loss": 0.8979,
+      "step": 5640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2472688515853984e-05,
+      "loss": 0.7944,
+      "step": 5650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2459365840660807e-05,
+      "loss": 0.8836,
+      "step": 5660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2446043165467627e-05,
+      "loss": 0.8557,
+      "step": 5670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2432720490274449e-05,
+      "loss": 0.8535,
+      "step": 5680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2419397815081269e-05,
+      "loss": 0.5644,
+      "step": 5690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2406075139888089e-05,
+      "loss": 0.9361,
+      "step": 5700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2392752464694912e-05,
+      "loss": 0.9234,
+      "step": 5710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2379429789501734e-05,
+      "loss": 0.9151,
+      "step": 5720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2366107114308554e-05,
+      "loss": 1.1469,
+      "step": 5730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2352784439115376e-05,
+      "loss": 0.7642,
+      "step": 5740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2339461763922196e-05,
+      "loss": 0.8959,
+      "step": 5750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.232613908872902e-05,
+      "loss": 0.9791,
+      "step": 5760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.231281641353584e-05,
+      "loss": 0.7513,
+      "step": 5770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2299493738342661e-05,
+      "loss": 0.8127,
+      "step": 5780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2286171063149481e-05,
+      "loss": 0.6569,
+      "step": 5790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2272848387956303e-05,
+      "loss": 0.9072,
+      "step": 5800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2259525712763123e-05,
+      "loss": 0.8966,
+      "step": 5810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2246203037569946e-05,
+      "loss": 0.7972,
+      "step": 5820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2232880362376766e-05,
+      "loss": 0.7422,
+      "step": 5830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2219557687183588e-05,
+      "loss": 0.9659,
+      "step": 5840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2206235011990408e-05,
+      "loss": 1.1856,
+      "step": 5850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.219291233679723e-05,
+      "loss": 0.788,
+      "step": 5860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2179589661604051e-05,
+      "loss": 0.7689,
+      "step": 5870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2166266986410873e-05,
+      "loss": 1.1042,
+      "step": 5880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2152944311217693e-05,
+      "loss": 0.7513,
+      "step": 5890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2139621636024515e-05,
+      "loss": 0.6981,
+      "step": 5900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2126298960831335e-05,
+      "loss": 0.6604,
+      "step": 5910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2112976285638156e-05,
+      "loss": 0.8738,
+      "step": 5920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.209965361044498e-05,
+      "loss": 0.7364,
+      "step": 5930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.20863309352518e-05,
+      "loss": 0.9592,
+      "step": 5940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.207300826005862e-05,
+      "loss": 1.0524,
+      "step": 5950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2059685584865442e-05,
+      "loss": 0.5619,
+      "step": 5960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2046362909672262e-05,
+      "loss": 0.9171,
+      "step": 5970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2033040234479085e-05,
+      "loss": 0.8864,
+      "step": 5980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2019717559285907e-05,
+      "loss": 0.8777,
+      "step": 5990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2006394884092727e-05,
+      "loss": 1.2906,
+      "step": 6000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1993072208899549e-05,
+      "loss": 0.865,
+      "step": 6010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1979749533706369e-05,
+      "loss": 0.615,
+      "step": 6020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1966426858513192e-05,
+      "loss": 0.7379,
+      "step": 6030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1953104183320012e-05,
+      "loss": 0.6808,
+      "step": 6040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1939781508126834e-05,
+      "loss": 0.8351,
+      "step": 6050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1926458832933654e-05,
+      "loss": 0.7282,
+      "step": 6060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1913136157740475e-05,
+      "loss": 0.934,
+      "step": 6070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1899813482547295e-05,
+      "loss": 0.8292,
+      "step": 6080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1886490807354119e-05,
+      "loss": 0.8739,
+      "step": 6090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1873168132160939e-05,
+      "loss": 0.7058,
+      "step": 6100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.185984545696776e-05,
+      "loss": 0.7018,
+      "step": 6110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.184652278177458e-05,
+      "loss": 1.0218,
+      "step": 6120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1833200106581402e-05,
+      "loss": 1.1008,
+      "step": 6130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1819877431388224e-05,
+      "loss": 0.8833,
+      "step": 6140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1806554756195046e-05,
+      "loss": 0.9845,
+      "step": 6150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1793232081001866e-05,
+      "loss": 1.2978,
+      "step": 6160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1779909405808688e-05,
+      "loss": 1.0189,
+      "step": 6170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1766586730615508e-05,
+      "loss": 1.2776,
+      "step": 6180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.175326405542233e-05,
+      "loss": 0.7908,
+      "step": 6190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1739941380229151e-05,
+      "loss": 0.8986,
+      "step": 6200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1726618705035973e-05,
+      "loss": 0.8974,
+      "step": 6210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1713296029842793e-05,
+      "loss": 0.4973,
+      "step": 6220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1699973354649614e-05,
+      "loss": 0.9393,
+      "step": 6230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1686650679456434e-05,
+      "loss": 1.0344,
+      "step": 6240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1673328004263258e-05,
+      "loss": 0.8439,
+      "step": 6250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1660005329070078e-05,
+      "loss": 0.9039,
+      "step": 6260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.16466826538769e-05,
+      "loss": 0.862,
+      "step": 6270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.163335997868372e-05,
+      "loss": 0.8327,
+      "step": 6280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1620037303490541e-05,
+      "loss": 0.8622,
+      "step": 6290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1606714628297361e-05,
+      "loss": 0.8518,
+      "step": 6300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1593391953104185e-05,
+      "loss": 0.8339,
+      "step": 6310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1580069277911007e-05,
+      "loss": 1.0355,
+      "step": 6320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1566746602717827e-05,
+      "loss": 0.6166,
+      "step": 6330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1553423927524647e-05,
+      "loss": 0.7943,
+      "step": 6340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1540101252331468e-05,
+      "loss": 0.5717,
+      "step": 6350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1526778577138292e-05,
+      "loss": 1.0663,
+      "step": 6360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1513455901945112e-05,
+      "loss": 1.1115,
+      "step": 6370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1500133226751933e-05,
+      "loss": 0.9026,
+      "step": 6380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1486810551558753e-05,
+      "loss": 1.0676,
+      "step": 6390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1473487876365575e-05,
+      "loss": 0.9403,
+      "step": 6400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1460165201172397e-05,
+      "loss": 0.9409,
+      "step": 6410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1446842525979219e-05,
+      "loss": 0.7303,
+      "step": 6420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1433519850786039e-05,
+      "loss": 0.7126,
+      "step": 6430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.142019717559286e-05,
+      "loss": 0.7029,
+      "step": 6440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.140687450039968e-05,
+      "loss": 0.8466,
+      "step": 6450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1393551825206502e-05,
+      "loss": 0.827,
+      "step": 6460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1380229150013324e-05,
+      "loss": 1.2105,
+      "step": 6470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1366906474820146e-05,
+      "loss": 1.0531,
+      "step": 6480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1353583799626966e-05,
+      "loss": 0.7658,
+      "step": 6490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1340261124433787e-05,
+      "loss": 0.7935,
+      "step": 6500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1326938449240607e-05,
+      "loss": 1.148,
+      "step": 6510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.131361577404743e-05,
+      "loss": 1.3408,
+      "step": 6520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.130029309885425e-05,
+      "loss": 0.9053,
+      "step": 6530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1286970423661072e-05,
+      "loss": 0.4991,
+      "step": 6540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1273647748467892e-05,
+      "loss": 1.0191,
+      "step": 6550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1260325073274714e-05,
+      "loss": 0.8286,
+      "step": 6560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1247002398081534e-05,
+      "loss": 0.8756,
+      "step": 6570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1233679722888358e-05,
+      "loss": 0.8976,
+      "step": 6580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1220357047695178e-05,
+      "loss": 0.7993,
+      "step": 6590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1207034372502e-05,
+      "loss": 0.8111,
+      "step": 6600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.119371169730882e-05,
+      "loss": 0.8845,
+      "step": 6610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1180389022115641e-05,
+      "loss": 1.0341,
+      "step": 6620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1167066346922464e-05,
+      "loss": 0.8975,
+      "step": 6630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1153743671729285e-05,
+      "loss": 0.7685,
+      "step": 6640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1140420996536106e-05,
+      "loss": 1.2886,
+      "step": 6650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1127098321342926e-05,
+      "loss": 0.646,
+      "step": 6660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1113775646149746e-05,
+      "loss": 1.0076,
+      "step": 6670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.110045297095657e-05,
+      "loss": 0.7607,
+      "step": 6680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1087130295763391e-05,
+      "loss": 0.7927,
+      "step": 6690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1073807620570211e-05,
+      "loss": 0.7231,
+      "step": 6700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1060484945377033e-05,
+      "loss": 0.7715,
+      "step": 6710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1047162270183853e-05,
+      "loss": 0.8849,
+      "step": 6720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1033839594990673e-05,
+      "loss": 1.0985,
+      "step": 6730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1020516919797497e-05,
+      "loss": 0.8239,
+      "step": 6740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1007194244604318e-05,
+      "loss": 0.8532,
+      "step": 6750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0993871569411138e-05,
+      "loss": 0.9088,
+      "step": 6760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.098054889421796e-05,
+      "loss": 0.7076,
+      "step": 6770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.096722621902478e-05,
+      "loss": 0.8952,
+      "step": 6780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0953903543831603e-05,
+      "loss": 0.9333,
+      "step": 6790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0940580868638424e-05,
+      "loss": 1.0662,
+      "step": 6800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0927258193445245e-05,
+      "loss": 0.9924,
+      "step": 6810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0913935518252065e-05,
+      "loss": 0.6257,
+      "step": 6820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0900612843058887e-05,
+      "loss": 0.9909,
+      "step": 6830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0887290167865707e-05,
+      "loss": 0.6456,
+      "step": 6840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.087396749267253e-05,
+      "loss": 0.955,
+      "step": 6850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.086064481747935e-05,
+      "loss": 0.9396,
+      "step": 6860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0847322142286172e-05,
+      "loss": 1.0635,
+      "step": 6870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0833999467092992e-05,
+      "loss": 0.8497,
+      "step": 6880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0820676791899814e-05,
+      "loss": 0.6953,
+      "step": 6890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0807354116706637e-05,
+      "loss": 0.8164,
+      "step": 6900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0794031441513457e-05,
+      "loss": 0.8728,
+      "step": 6910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0780708766320277e-05,
+      "loss": 0.8216,
+      "step": 6920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0767386091127099e-05,
+      "loss": 0.8249,
+      "step": 6930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0754063415933919e-05,
+      "loss": 0.9111,
+      "step": 6940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0740740740740742e-05,
+      "loss": 0.8234,
+      "step": 6950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0727418065547564e-05,
+      "loss": 0.7209,
+      "step": 6960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0714095390354384e-05,
+      "loss": 0.6214,
+      "step": 6970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0700772715161204e-05,
+      "loss": 0.8723,
+      "step": 6980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0687450039968026e-05,
+      "loss": 0.8791,
+      "step": 6990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0674127364774846e-05,
+      "loss": 1.0159,
+      "step": 7000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.066080468958167e-05,
+      "loss": 1.1373,
+      "step": 7010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0647482014388491e-05,
+      "loss": 0.8018,
+      "step": 7020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0634159339195311e-05,
+      "loss": 0.7598,
+      "step": 7030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0620836664002133e-05,
+      "loss": 0.8811,
+      "step": 7040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0607513988808953e-05,
+      "loss": 0.5787,
+      "step": 7050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0594191313615776e-05,
+      "loss": 0.6979,
+      "step": 7060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0580868638422596e-05,
+      "loss": 0.8698,
+      "step": 7070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0567545963229418e-05,
+      "loss": 0.7918,
+      "step": 7080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0554223288036238e-05,
+      "loss": 0.791,
+      "step": 7090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.054090061284306e-05,
+      "loss": 0.9026,
+      "step": 7100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.052757793764988e-05,
+      "loss": 0.7427,
+      "step": 7110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0514255262456703e-05,
+      "loss": 0.8489,
+      "step": 7120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0500932587263523e-05,
+      "loss": 0.5992,
+      "step": 7130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0487609912070345e-05,
+      "loss": 0.948,
+      "step": 7140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0474287236877165e-05,
+      "loss": 0.7375,
+      "step": 7150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0460964561683987e-05,
+      "loss": 0.8703,
+      "step": 7160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0447641886490808e-05,
+      "loss": 0.588,
+      "step": 7170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.043431921129763e-05,
+      "loss": 0.9549,
+      "step": 7180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.042099653610445e-05,
+      "loss": 0.7655,
+      "step": 7190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0407673860911272e-05,
+      "loss": 0.9914,
+      "step": 7200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0394351185718092e-05,
+      "loss": 0.716,
+      "step": 7210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0381028510524915e-05,
+      "loss": 0.8625,
+      "step": 7220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0367705835331735e-05,
+      "loss": 0.546,
+      "step": 7230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0354383160138557e-05,
+      "loss": 0.9452,
+      "step": 7240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0341060484945377e-05,
+      "loss": 0.7429,
+      "step": 7250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0327737809752199e-05,
+      "loss": 0.8323,
+      "step": 7260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0314415134559019e-05,
+      "loss": 0.8861,
+      "step": 7270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0301092459365842e-05,
+      "loss": 0.5798,
+      "step": 7280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0287769784172664e-05,
+      "loss": 0.6907,
+      "step": 7290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0274447108979484e-05,
+      "loss": 1.1387,
+      "step": 7300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0261124433786304e-05,
+      "loss": 0.851,
+      "step": 7310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0247801758593126e-05,
+      "loss": 0.7026,
+      "step": 7320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0234479083399949e-05,
+      "loss": 1.0397,
+      "step": 7330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0221156408206769e-05,
+      "loss": 0.8539,
+      "step": 7340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.020783373301359e-05,
+      "loss": 0.7249,
+      "step": 7350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0194511057820411e-05,
+      "loss": 0.6053,
+      "step": 7360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0181188382627233e-05,
+      "loss": 0.9206,
+      "step": 7370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0167865707434053e-05,
+      "loss": 0.72,
+      "step": 7380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0154543032240876e-05,
+      "loss": 0.7217,
+      "step": 7390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0141220357047696e-05,
+      "loss": 0.9969,
+      "step": 7400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0127897681854518e-05,
+      "loss": 0.9146,
+      "step": 7410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0114575006661338e-05,
+      "loss": 0.8879,
+      "step": 7420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.010125233146816e-05,
+      "loss": 0.6825,
+      "step": 7430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0087929656274981e-05,
+      "loss": 1.2015,
+      "step": 7440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0074606981081803e-05,
+      "loss": 0.8784,
+      "step": 7450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0061284305888623e-05,
+      "loss": 0.7296,
+      "step": 7460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0047961630695445e-05,
+      "loss": 0.5832,
+      "step": 7470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0034638955502265e-05,
+      "loss": 0.9754,
+      "step": 7480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0021316280309088e-05,
+      "loss": 0.8457,
+      "step": 7490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0007993605115908e-05,
+      "loss": 0.8754,
+      "step": 7500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.99467092992273e-06,
+      "loss": 0.6651,
+      "step": 7510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.98134825472955e-06,
+      "loss": 0.5887,
+      "step": 7520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.968025579536372e-06,
+      "loss": 0.5617,
+      "step": 7530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.954702904343193e-06,
+      "loss": 0.6945,
+      "step": 7540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.941380229150013e-06,
+      "loss": 0.6885,
+      "step": 7550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.928057553956835e-06,
+      "loss": 0.6411,
+      "step": 7560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.914734878763657e-06,
+      "loss": 0.6029,
+      "step": 7570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.901412203570477e-06,
+      "loss": 0.479,
+      "step": 7580
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.888089528377298e-06,
+      "loss": 0.4233,
+      "step": 7590
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.87476685318412e-06,
+      "loss": 0.7149,
+      "step": 7600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.86144417799094e-06,
+      "loss": 0.6039,
+      "step": 7610
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.848121502797764e-06,
+      "loss": 0.3904,
+      "step": 7620
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.834798827604584e-06,
+      "loss": 0.554,
+      "step": 7630
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.821476152411405e-06,
+      "loss": 0.6223,
+      "step": 7640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.808153477218227e-06,
+      "loss": 0.6661,
+      "step": 7650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.794830802025047e-06,
+      "loss": 0.6223,
+      "step": 7660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.781508126831869e-06,
+      "loss": 0.4074,
+      "step": 7670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.76818545163869e-06,
+      "loss": 0.5239,
+      "step": 7680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.75486277644551e-06,
+      "loss": 0.6673,
+      "step": 7690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.741540101252332e-06,
+      "loss": 0.4243,
+      "step": 7700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.728217426059154e-06,
+      "loss": 0.5662,
+      "step": 7710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.714894750865976e-06,
+      "loss": 0.5088,
+      "step": 7720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.701572075672796e-06,
+      "loss": 0.6867,
+      "step": 7730
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.688249400479617e-06,
+      "loss": 0.6084,
+      "step": 7740
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.67492672528644e-06,
+      "loss": 0.5469,
+      "step": 7750
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.66160405009326e-06,
+      "loss": 0.4736,
+      "step": 7760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.648281374900081e-06,
+      "loss": 0.5609,
+      "step": 7770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.634958699706903e-06,
+      "loss": 0.4694,
+      "step": 7780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.621636024513723e-06,
+      "loss": 0.6748,
+      "step": 7790
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.608313349320544e-06,
+      "loss": 0.6605,
+      "step": 7800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.594990674127366e-06,
+      "loss": 0.5726,
+      "step": 7810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.581667998934186e-06,
+      "loss": 0.8413,
+      "step": 7820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.568345323741008e-06,
+      "loss": 0.4299,
+      "step": 7830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.55502264854783e-06,
+      "loss": 0.6246,
+      "step": 7840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.54169997335465e-06,
+      "loss": 0.7487,
+      "step": 7850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.528377298161471e-06,
+      "loss": 0.7082,
+      "step": 7860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.515054622968293e-06,
+      "loss": 0.5832,
+      "step": 7870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.501731947775113e-06,
+      "loss": 0.5197,
+      "step": 7880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.488409272581935e-06,
+      "loss": 0.4889,
+      "step": 7890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.475086597388756e-06,
+      "loss": 0.3017,
+      "step": 7900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.461763922195578e-06,
+      "loss": 0.5872,
+      "step": 7910
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.448441247002398e-06,
+      "loss": 0.6167,
+      "step": 7920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.43511857180922e-06,
+      "loss": 0.5968,
+      "step": 7930
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.421795896616042e-06,
+      "loss": 0.7126,
+      "step": 7940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.408473221422862e-06,
+      "loss": 0.6875,
+      "step": 7950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.395150546229683e-06,
+      "loss": 0.4871,
+      "step": 7960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.381827871036505e-06,
+      "loss": 0.6521,
+      "step": 7970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.368505195843325e-06,
+      "loss": 0.5513,
+      "step": 7980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.355182520650149e-06,
+      "loss": 0.438,
+      "step": 7990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.341859845456969e-06,
+      "loss": 0.6675,
+      "step": 8000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.32853717026379e-06,
+      "loss": 0.7628,
+      "step": 8010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.315214495070612e-06,
+      "loss": 0.48,
+      "step": 8020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.301891819877432e-06,
+      "loss": 0.5852,
+      "step": 8030
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.288569144684254e-06,
+      "loss": 0.5102,
+      "step": 8040
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.275246469491075e-06,
+      "loss": 0.3851,
+      "step": 8050
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.261923794297895e-06,
+      "loss": 0.7704,
+      "step": 8060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.248601119104717e-06,
+      "loss": 0.7697,
+      "step": 8070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.235278443911539e-06,
+      "loss": 0.4341,
+      "step": 8080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.221955768718359e-06,
+      "loss": 0.4089,
+      "step": 8090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.20863309352518e-06,
+      "loss": 0.825,
+      "step": 8100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.195310418332002e-06,
+      "loss": 0.667,
+      "step": 8110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.181987743138822e-06,
+      "loss": 0.6688,
+      "step": 8120
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.168665067945644e-06,
+      "loss": 0.426,
+      "step": 8130
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.155342392752466e-06,
+      "loss": 0.6672,
+      "step": 8140
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.142019717559286e-06,
+      "loss": 0.4767,
+      "step": 8150
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.128697042366108e-06,
+      "loss": 0.6029,
+      "step": 8160
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.11537436717293e-06,
+      "loss": 0.6964,
+      "step": 8170
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.102051691979751e-06,
+      "loss": 0.7647,
+      "step": 8180
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.088729016786571e-06,
+      "loss": 0.5858,
+      "step": 8190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.075406341593393e-06,
+      "loss": 0.6002,
+      "step": 8200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.062083666400214e-06,
+      "loss": 0.6143,
+      "step": 8210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.048760991207034e-06,
+      "loss": 0.4929,
+      "step": 8220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.035438316013856e-06,
+      "loss": 0.6795,
+      "step": 8230
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.022115640820678e-06,
+      "loss": 0.5844,
+      "step": 8240
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.008792965627498e-06,
+      "loss": 0.597,
+      "step": 8250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.99547029043432e-06,
+      "loss": 0.5097,
+      "step": 8260
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.982147615241141e-06,
+      "loss": 0.8274,
+      "step": 8270
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.968824940047961e-06,
+      "loss": 0.4721,
+      "step": 8280
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.955502264854785e-06,
+      "loss": 0.7477,
+      "step": 8290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.942179589661605e-06,
+      "loss": 0.515,
+      "step": 8300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.928856914468425e-06,
+      "loss": 0.6419,
+      "step": 8310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.915534239275248e-06,
+      "loss": 0.548,
+      "step": 8320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.902211564082068e-06,
+      "loss": 0.6217,
+      "step": 8330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.515,
+      "step": 8340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.875566213695712e-06,
+      "loss": 0.5288,
+      "step": 8350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.862243538502532e-06,
+      "loss": 0.6349,
+      "step": 8360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.848920863309353e-06,
+      "loss": 0.6642,
+      "step": 8370
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.835598188116175e-06,
+      "loss": 0.6257,
+      "step": 8380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.822275512922995e-06,
+      "loss": 0.5945,
+      "step": 8390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.808952837729817e-06,
+      "loss": 0.4554,
+      "step": 8400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.795630162536639e-06,
+      "loss": 0.5942,
+      "step": 8410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.782307487343459e-06,
+      "loss": 0.5292,
+      "step": 8420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.76898481215028e-06,
+      "loss": 0.517,
+      "step": 8430
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.755662136957102e-06,
+      "loss": 0.4412,
+      "step": 8440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.742339461763924e-06,
+      "loss": 0.6529,
+      "step": 8450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.729016786570744e-06,
+      "loss": 0.3173,
+      "step": 8460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.715694111377566e-06,
+      "loss": 0.6011,
+      "step": 8470
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.702371436184387e-06,
+      "loss": 0.511,
+      "step": 8480
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.689048760991207e-06,
+      "loss": 0.5281,
+      "step": 8490
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.675726085798029e-06,
+      "loss": 0.5546,
+      "step": 8500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.66240341060485e-06,
+      "loss": 0.4424,
+      "step": 8510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.64908073541167e-06,
+      "loss": 0.5005,
+      "step": 8520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.635758060218492e-06,
+      "loss": 0.5843,
+      "step": 8530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.622435385025314e-06,
+      "loss": 0.5101,
+      "step": 8540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.609112709832134e-06,
+      "loss": 0.5558,
+      "step": 8550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.595790034638956e-06,
+      "loss": 0.5574,
+      "step": 8560
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.582467359445778e-06,
+      "loss": 0.5099,
+      "step": 8570
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.569144684252598e-06,
+      "loss": 0.5025,
+      "step": 8580
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.55582200905942e-06,
+      "loss": 0.646,
+      "step": 8590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.542499333866241e-06,
+      "loss": 0.511,
+      "step": 8600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.529176658673061e-06,
+      "loss": 0.5669,
+      "step": 8610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.515853983479885e-06,
+      "loss": 0.8414,
+      "step": 8620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.502531308286705e-06,
+      "loss": 0.5391,
+      "step": 8630
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.489208633093526e-06,
+      "loss": 0.841,
+      "step": 8640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.475885957900348e-06,
+      "loss": 0.8282,
+      "step": 8650
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.462563282707168e-06,
+      "loss": 0.4807,
+      "step": 8660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.44924060751399e-06,
+      "loss": 0.5238,
+      "step": 8670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.435917932320811e-06,
+      "loss": 0.7307,
+      "step": 8680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.422595257127631e-06,
+      "loss": 0.6926,
+      "step": 8690
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.409272581934453e-06,
+      "loss": 0.6897,
+      "step": 8700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.395949906741275e-06,
+      "loss": 0.6377,
+      "step": 8710
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.382627231548095e-06,
+      "loss": 0.6711,
+      "step": 8720
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.369304556354917e-06,
+      "loss": 0.4469,
+      "step": 8730
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.355981881161738e-06,
+      "loss": 0.4468,
+      "step": 8740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.34265920596856e-06,
+      "loss": 0.5768,
+      "step": 8750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.32933653077538e-06,
+      "loss": 0.5792,
+      "step": 8760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.316013855582202e-06,
+      "loss": 0.6167,
+      "step": 8770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.302691180389024e-06,
+      "loss": 0.5216,
+      "step": 8780
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.289368505195844e-06,
+      "loss": 0.706,
+      "step": 8790
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.276045830002665e-06,
+      "loss": 0.5537,
+      "step": 8800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.262723154809487e-06,
+      "loss": 0.6349,
+      "step": 8810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.249400479616307e-06,
+      "loss": 0.4659,
+      "step": 8820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.236077804423129e-06,
+      "loss": 0.4473,
+      "step": 8830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.22275512922995e-06,
+      "loss": 0.5256,
+      "step": 8840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.20943245403677e-06,
+      "loss": 0.5484,
+      "step": 8850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.196109778843592e-06,
+      "loss": 0.6666,
+      "step": 8860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.182787103650414e-06,
+      "loss": 0.9443,
+      "step": 8870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.169464428457234e-06,
+      "loss": 0.4171,
+      "step": 8880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.156141753264056e-06,
+      "loss": 0.5668,
+      "step": 8890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.142819078070877e-06,
+      "loss": 0.7022,
+      "step": 8900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.129496402877699e-06,
+      "loss": 0.5253,
+      "step": 8910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.116173727684519e-06,
+      "loss": 0.8242,
+      "step": 8920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.10285105249134e-06,
+      "loss": 0.6424,
+      "step": 8930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.089528377298163e-06,
+      "loss": 0.4108,
+      "step": 8940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.076205702104983e-06,
+      "loss": 0.7285,
+      "step": 8950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.062883026911804e-06,
+      "loss": 0.6103,
+      "step": 8960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.049560351718626e-06,
+      "loss": 0.4268,
+      "step": 8970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.036237676525448e-06,
+      "loss": 0.6346,
+      "step": 8980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.022915001332268e-06,
+      "loss": 0.5144,
+      "step": 8990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.00959232613909e-06,
+      "loss": 0.622,
+      "step": 9000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.996269650945911e-06,
+      "loss": 0.6551,
+      "step": 9010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.982946975752733e-06,
+      "loss": 0.5867,
+      "step": 9020
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.969624300559553e-06,
+      "loss": 0.5306,
+      "step": 9030
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.956301625366375e-06,
+      "loss": 0.7457,
+      "step": 9040
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.942978950173196e-06,
+      "loss": 0.4717,
+      "step": 9050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.929656274980016e-06,
+      "loss": 0.7841,
+      "step": 9060
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.916333599786838e-06,
+      "loss": 0.4112,
+      "step": 9070
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.90301092459366e-06,
+      "loss": 0.6972,
+      "step": 9080
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.88968824940048e-06,
+      "loss": 0.7216,
+      "step": 9090
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.876365574207302e-06,
+      "loss": 0.5126,
+      "step": 9100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.863042899014123e-06,
+      "loss": 0.7825,
+      "step": 9110
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.849720223820943e-06,
+      "loss": 0.5064,
+      "step": 9120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.836397548627765e-06,
+      "loss": 0.5202,
+      "step": 9130
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.823074873434587e-06,
+      "loss": 0.7074,
+      "step": 9140
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.809752198241407e-06,
+      "loss": 0.6743,
+      "step": 9150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.796429523048228e-06,
+      "loss": 0.5713,
+      "step": 9160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.78310684785505e-06,
+      "loss": 0.5938,
+      "step": 9170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.769784172661872e-06,
+      "loss": 0.5986,
+      "step": 9180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.756461497468692e-06,
+      "loss": 0.6494,
+      "step": 9190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.743138822275514e-06,
+      "loss": 0.7668,
+      "step": 9200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.729816147082335e-06,
+      "loss": 0.5961,
+      "step": 9210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.716493471889155e-06,
+      "loss": 0.5373,
+      "step": 9220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.703170796695977e-06,
+      "loss": 0.5084,
+      "step": 9230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.689848121502799e-06,
+      "loss": 0.4927,
+      "step": 9240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.676525446309619e-06,
+      "loss": 0.5338,
+      "step": 9250
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.66320277111644e-06,
+      "loss": 0.5202,
+      "step": 9260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.649880095923262e-06,
+      "loss": 0.7087,
+      "step": 9270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.636557420730082e-06,
+      "loss": 0.4755,
+      "step": 9280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.623234745536905e-06,
+      "loss": 0.6859,
+      "step": 9290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.609912070343726e-06,
+      "loss": 0.5637,
+      "step": 9300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.5965893951505466e-06,
+      "loss": 0.4301,
+      "step": 9310
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.583266719957368e-06,
+      "loss": 0.555,
+      "step": 9320
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.569944044764189e-06,
+      "loss": 0.5708,
+      "step": 9330
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.55662136957101e-06,
+      "loss": 0.3748,
+      "step": 9340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.543298694377832e-06,
+      "loss": 0.3104,
+      "step": 9350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.529976019184653e-06,
+      "loss": 0.4811,
+      "step": 9360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.516653343991474e-06,
+      "loss": 0.7626,
+      "step": 9370
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.503330668798295e-06,
+      "loss": 0.7432,
+      "step": 9380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.490007993605116e-06,
+      "loss": 0.4564,
+      "step": 9390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.476685318411938e-06,
+      "loss": 0.4931,
+      "step": 9400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.463362643218759e-06,
+      "loss": 0.5394,
+      "step": 9410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.4500399680255795e-06,
+      "loss": 0.6123,
+      "step": 9420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.436717292832401e-06,
+      "loss": 0.4299,
+      "step": 9430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.423394617639222e-06,
+      "loss": 0.5161,
+      "step": 9440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.410071942446043e-06,
+      "loss": 0.5089,
+      "step": 9450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.3967492672528655e-06,
+      "loss": 0.6182,
+      "step": 9460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.3834265920596856e-06,
+      "loss": 0.5111,
+      "step": 9470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.370103916866508e-06,
+      "loss": 0.6544,
+      "step": 9480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.356781241673329e-06,
+      "loss": 0.5086,
+      "step": 9490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.343458566480149e-06,
+      "loss": 0.5772,
+      "step": 9500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.3301358912869716e-06,
+      "loss": 0.7216,
+      "step": 9510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.3168132160937924e-06,
+      "loss": 0.3456,
+      "step": 9520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.303490540900613e-06,
+      "loss": 0.5831,
+      "step": 9530
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.290167865707435e-06,
+      "loss": 0.4155,
+      "step": 9540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.276845190514256e-06,
+      "loss": 0.4813,
+      "step": 9550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.263522515321078e-06,
+      "loss": 0.4247,
+      "step": 9560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.2501998401278985e-06,
+      "loss": 0.3583,
+      "step": 9570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.236877164934719e-06,
+      "loss": 0.5876,
+      "step": 9580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.223554489741541e-06,
+      "loss": 0.4301,
+      "step": 9590
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.210231814548362e-06,
+      "loss": 0.6511,
+      "step": 9600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.196909139355183e-06,
+      "loss": 0.3719,
+      "step": 9610
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.1835864641620045e-06,
+      "loss": 0.8085,
+      "step": 9620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.170263788968825e-06,
+      "loss": 0.495,
+      "step": 9630
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.156941113775647e-06,
+      "loss": 0.4479,
+      "step": 9640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.143618438582468e-06,
+      "loss": 0.5258,
+      "step": 9650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.130295763389289e-06,
+      "loss": 0.5833,
+      "step": 9660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.1169730881961106e-06,
+      "loss": 0.6848,
+      "step": 9670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.1036504130029314e-06,
+      "loss": 0.681,
+      "step": 9680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.090327737809752e-06,
+      "loss": 0.7938,
+      "step": 9690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.077005062616574e-06,
+      "loss": 0.7305,
+      "step": 9700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.063682387423395e-06,
+      "loss": 0.6472,
+      "step": 9710
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.050359712230216e-06,
+      "loss": 0.6139,
+      "step": 9720
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.0370370370370375e-06,
+      "loss": 0.6646,
+      "step": 9730
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.023714361843858e-06,
+      "loss": 0.6314,
+      "step": 9740
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.01039168665068e-06,
+      "loss": 0.6679,
+      "step": 9750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.997069011457501e-06,
+      "loss": 0.3637,
+      "step": 9760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.983746336264322e-06,
+      "loss": 0.5498,
+      "step": 9770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.970423661071144e-06,
+      "loss": 0.5565,
+      "step": 9780
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.957100985877964e-06,
+      "loss": 0.7444,
+      "step": 9790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.943778310684785e-06,
+      "loss": 0.4358,
+      "step": 9800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.930455635491608e-06,
+      "loss": 0.3845,
+      "step": 9810
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.917132960298429e-06,
+      "loss": 0.7262,
+      "step": 9820
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.90381028510525e-06,
+      "loss": 0.5637,
+      "step": 9830
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.890487609912071e-06,
+      "loss": 0.5843,
+      "step": 9840
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.877164934718892e-06,
+      "loss": 0.6286,
+      "step": 9850
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.863842259525714e-06,
+      "loss": 0.4262,
+      "step": 9860
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.850519584332535e-06,
+      "loss": 0.6395,
+      "step": 9870
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.837196909139356e-06,
+      "loss": 0.618,
+      "step": 9880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.823874233946177e-06,
+      "loss": 0.4814,
+      "step": 9890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.810551558752998e-06,
+      "loss": 0.474,
+      "step": 9900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.79722888355982e-06,
+      "loss": 0.5267,
+      "step": 9910
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.783906208366641e-06,
+      "loss": 0.6874,
+      "step": 9920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.770583533173462e-06,
+      "loss": 0.5276,
+      "step": 9930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.757260857980283e-06,
+      "loss": 0.5303,
+      "step": 9940
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.743938182787104e-06,
+      "loss": 0.4643,
+      "step": 9950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.730615507593925e-06,
+      "loss": 0.4328,
+      "step": 9960
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.717292832400747e-06,
+      "loss": 0.542,
+      "step": 9970
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.703970157207568e-06,
+      "loss": 0.502,
+      "step": 9980
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.6906474820143886e-06,
+      "loss": 0.603,
+      "step": 9990
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.67732480682121e-06,
+      "loss": 0.7646,
+      "step": 10000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.664002131628031e-06,
+      "loss": 0.6183,
+      "step": 10010
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.650679456434853e-06,
+      "loss": 0.462,
+      "step": 10020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.637356781241674e-06,
+      "loss": 0.4754,
+      "step": 10030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.624034106048495e-06,
+      "loss": 0.632,
+      "step": 10040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.610711430855316e-06,
+      "loss": 0.5148,
+      "step": 10050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.597388755662137e-06,
+      "loss": 0.4957,
+      "step": 10060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.584066080468958e-06,
+      "loss": 0.5495,
+      "step": 10070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.57074340527578e-06,
+      "loss": 0.5744,
+      "step": 10080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.557420730082601e-06,
+      "loss": 0.8109,
+      "step": 10090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.544098054889423e-06,
+      "loss": 0.4862,
+      "step": 10100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.530775379696243e-06,
+      "loss": 0.4743,
+      "step": 10110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.517452704503064e-06,
+      "loss": 0.6421,
+      "step": 10120
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.504130029309887e-06,
+      "loss": 0.376,
+      "step": 10130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.4908073541167076e-06,
+      "loss": 0.5927,
+      "step": 10140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.4774846789235276e-06,
+      "loss": 0.805,
+      "step": 10150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.46416200373035e-06,
+      "loss": 0.6305,
+      "step": 10160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.450839328537171e-06,
+      "loss": 0.5092,
+      "step": 10170
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.437516653343991e-06,
+      "loss": 0.7112,
+      "step": 10180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.424193978150814e-06,
+      "loss": 0.7156,
+      "step": 10190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.4108713029576345e-06,
+      "loss": 0.4235,
+      "step": 10200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.397548627764456e-06,
+      "loss": 0.3841,
+      "step": 10210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.384225952571277e-06,
+      "loss": 0.4779,
+      "step": 10220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.370903277378098e-06,
+      "loss": 0.5718,
+      "step": 10230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.35758060218492e-06,
+      "loss": 0.5171,
+      "step": 10240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.3442579269917405e-06,
+      "loss": 0.5425,
+      "step": 10250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.330935251798561e-06,
+      "loss": 0.6426,
+      "step": 10260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.317612576605383e-06,
+      "loss": 0.5412,
+      "step": 10270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.304289901412204e-06,
+      "loss": 0.6446,
+      "step": 10280
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.290967226219026e-06,
+      "loss": 0.4798,
+      "step": 10290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.2776445510258465e-06,
+      "loss": 0.601,
+      "step": 10300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.264321875832667e-06,
+      "loss": 0.481,
+      "step": 10310
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.250999200639489e-06,
+      "loss": 0.3424,
+      "step": 10320
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.23767652544631e-06,
+      "loss": 0.6193,
+      "step": 10330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.224353850253131e-06,
+      "loss": 0.462,
+      "step": 10340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.211031175059953e-06,
+      "loss": 0.518,
+      "step": 10350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.1977084998667735e-06,
+      "loss": 0.6449,
+      "step": 10360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.184385824673595e-06,
+      "loss": 0.6115,
+      "step": 10370
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.171063149480416e-06,
+      "loss": 0.5942,
+      "step": 10380
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.157740474287237e-06,
+      "loss": 0.7223,
+      "step": 10390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.144417799094059e-06,
+      "loss": 0.5872,
+      "step": 10400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.1310951239008795e-06,
+      "loss": 0.8311,
+      "step": 10410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.1177724487077e-06,
+      "loss": 0.4722,
+      "step": 10420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.104449773514522e-06,
+      "loss": 0.6166,
+      "step": 10430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.091127098321343e-06,
+      "loss": 0.5281,
+      "step": 10440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.077804423128164e-06,
+      "loss": 0.5282,
+      "step": 10450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.064481747934986e-06,
+      "loss": 0.4552,
+      "step": 10460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.051159072741806e-06,
+      "loss": 0.486,
+      "step": 10470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.037836397548629e-06,
+      "loss": 0.7482,
+      "step": 10480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.02451372235545e-06,
+      "loss": 0.6316,
+      "step": 10490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.011191047162271e-06,
+      "loss": 0.6591,
+      "step": 10500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.9978683719690924e-06,
+      "loss": 0.5151,
+      "step": 10510
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.984545696775913e-06,
+      "loss": 0.5123,
+      "step": 10520
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.971223021582734e-06,
+      "loss": 0.3546,
+      "step": 10530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.957900346389556e-06,
+      "loss": 0.5524,
+      "step": 10540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.944577671196377e-06,
+      "loss": 0.4409,
+      "step": 10550
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.9312549960031985e-06,
+      "loss": 0.5293,
+      "step": 10560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.917932320810019e-06,
+      "loss": 0.437,
+      "step": 10570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.90460964561684e-06,
+      "loss": 0.5775,
+      "step": 10580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.891286970423662e-06,
+      "loss": 0.5641,
+      "step": 10590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.877964295230483e-06,
+      "loss": 0.4839,
+      "step": 10600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.864641620037304e-06,
+      "loss": 0.4325,
+      "step": 10610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.851318944844125e-06,
+      "loss": 0.5035,
+      "step": 10620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.837996269650946e-06,
+      "loss": 0.6819,
+      "step": 10630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.824673594457768e-06,
+      "loss": 0.6849,
+      "step": 10640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.811350919264589e-06,
+      "loss": 0.4236,
+      "step": 10650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.79802824407141e-06,
+      "loss": 0.3773,
+      "step": 10660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.7847055688782314e-06,
+      "loss": 0.6302,
+      "step": 10670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.771382893685052e-06,
+      "loss": 0.4993,
+      "step": 10680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.758060218491873e-06,
+      "loss": 0.5893,
+      "step": 10690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.744737543298695e-06,
+      "loss": 0.4807,
+      "step": 10700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.731414868105516e-06,
+      "loss": 0.5702,
+      "step": 10710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.718092192912337e-06,
+      "loss": 0.512,
+      "step": 10720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.704769517719158e-06,
+      "loss": 0.5136,
+      "step": 10730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.691446842525979e-06,
+      "loss": 0.2975,
+      "step": 10740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.678124167332801e-06,
+      "loss": 0.4943,
+      "step": 10750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.664801492139622e-06,
+      "loss": 0.4108,
+      "step": 10760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.651478816946443e-06,
+      "loss": 0.647,
+      "step": 10770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.638156141753265e-06,
+      "loss": 0.4033,
+      "step": 10780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.624833466560085e-06,
+      "loss": 0.5327,
+      "step": 10790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.611510791366906e-06,
+      "loss": 0.659,
+      "step": 10800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.598188116173729e-06,
+      "loss": 0.582,
+      "step": 10810
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.5848654409805496e-06,
+      "loss": 0.5686,
+      "step": 10820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.571542765787371e-06,
+      "loss": 0.774,
+      "step": 10830
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.558220090594192e-06,
+      "loss": 0.4268,
+      "step": 10840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.544897415401013e-06,
+      "loss": 0.6283,
+      "step": 10850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.531574740207835e-06,
+      "loss": 0.7076,
+      "step": 10860
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.518252065014656e-06,
+      "loss": 0.4266,
+      "step": 10870
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5049293898214765e-06,
+      "loss": 0.615,
+      "step": 10880
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.491606714628298e-06,
+      "loss": 0.5533,
+      "step": 10890
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.478284039435119e-06,
+      "loss": 0.4839,
+      "step": 10900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.46496136424194e-06,
+      "loss": 0.5785,
+      "step": 10910
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.451638689048762e-06,
+      "loss": 0.52,
+      "step": 10920
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4383160138555825e-06,
+      "loss": 0.4498,
+      "step": 10930
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.424993338662404e-06,
+      "loss": 0.5643,
+      "step": 10940
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.411670663469225e-06,
+      "loss": 0.6726,
+      "step": 10950
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.398347988276046e-06,
+      "loss": 0.6215,
+      "step": 10960
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.385025313082868e-06,
+      "loss": 0.4321,
+      "step": 10970
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.3717026378896886e-06,
+      "loss": 0.5597,
+      "step": 10980
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.3583799626965094e-06,
+      "loss": 0.4928,
+      "step": 10990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.345057287503331e-06,
+      "loss": 0.5622,
+      "step": 11000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.331734612310152e-06,
+      "loss": 0.5592,
+      "step": 11010
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.318411937116974e-06,
+      "loss": 0.4732,
+      "step": 11020
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.305089261923795e-06,
+      "loss": 0.5939,
+      "step": 11030
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.2917665867306155e-06,
+      "loss": 0.5467,
+      "step": 11040
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.278443911537437e-06,
+      "loss": 0.6886,
+      "step": 11050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.265121236344258e-06,
+      "loss": 0.492,
+      "step": 11060
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.251798561151079e-06,
+      "loss": 0.3911,
+      "step": 11070
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.238475885957901e-06,
+      "loss": 0.7715,
+      "step": 11080
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.2251532107647215e-06,
+      "loss": 0.5695,
+      "step": 11090
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.211830535571544e-06,
+      "loss": 0.5881,
+      "step": 11100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.198507860378364e-06,
+      "loss": 0.5352,
+      "step": 11110
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 0.6824,
+      "step": 11120
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.1718625099920075e-06,
+      "loss": 0.6598,
+      "step": 11130
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.158539834798828e-06,
+      "loss": 0.5027,
+      "step": 11140
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.1452171596056484e-06,
+      "loss": 0.4774,
+      "step": 11150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.131894484412471e-06,
+      "loss": 0.309,
+      "step": 11160
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.118571809219292e-06,
+      "loss": 0.5187,
+      "step": 11170
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.105249134026113e-06,
+      "loss": 0.5329,
+      "step": 11180
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.0919264588329345e-06,
+      "loss": 0.4912,
+      "step": 11190
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.078603783639755e-06,
+      "loss": 0.5734,
+      "step": 11200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.065281108446577e-06,
+      "loss": 0.695,
+      "step": 11210
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.051958433253398e-06,
+      "loss": 0.4631,
+      "step": 11220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.038635758060219e-06,
+      "loss": 0.9621,
+      "step": 11230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.0253130828670405e-06,
+      "loss": 0.6088,
+      "step": 11240
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.011990407673861e-06,
+      "loss": 0.4807,
+      "step": 11250
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.998667732480682e-06,
+      "loss": 0.4774,
+      "step": 11260
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.985345057287504e-06,
+      "loss": 0.3755,
+      "step": 11270
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.972022382094325e-06,
+      "loss": 0.4934,
+      "step": 11280
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.958699706901146e-06,
+      "loss": 0.3653,
+      "step": 11290
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.945377031707967e-06,
+      "loss": 0.5374,
+      "step": 11300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.932054356514789e-06,
+      "loss": 0.6684,
+      "step": 11310
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.91873168132161e-06,
+      "loss": 0.3613,
+      "step": 11320
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.905409006128431e-06,
+      "loss": 0.5459,
+      "step": 11330
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.892086330935253e-06,
+      "loss": 0.4648,
+      "step": 11340
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.8787636557420735e-06,
+      "loss": 0.6087,
+      "step": 11350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.865440980548894e-06,
+      "loss": 0.4454,
+      "step": 11360
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.852118305355716e-06,
+      "loss": 0.6797,
+      "step": 11370
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.838795630162537e-06,
+      "loss": 0.444,
+      "step": 11380
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.825472954969358e-06,
+      "loss": 0.4996,
+      "step": 11390
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.8121502797761795e-06,
+      "loss": 0.5194,
+      "step": 11400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.798827604583e-06,
+      "loss": 0.7734,
+      "step": 11410
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.785504929389822e-06,
+      "loss": 0.406,
+      "step": 11420
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.772182254196643e-06,
+      "loss": 0.5761,
+      "step": 11430
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.758859579003464e-06,
+      "loss": 0.4261,
+      "step": 11440
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7455369038102855e-06,
+      "loss": 0.5295,
+      "step": 11450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.732214228617107e-06,
+      "loss": 0.5268,
+      "step": 11460
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.718891553423927e-06,
+      "loss": 0.4441,
+      "step": 11470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.705568878230749e-06,
+      "loss": 0.5511,
+      "step": 11480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.692246203037571e-06,
+      "loss": 0.4482,
+      "step": 11490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.678923527844392e-06,
+      "loss": 0.3605,
+      "step": 11500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.6656008526512125e-06,
+      "loss": 0.828,
+      "step": 11510
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.652278177458034e-06,
+      "loss": 0.4273,
+      "step": 11520
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.638955502264855e-06,
+      "loss": 0.6486,
+      "step": 11530
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.625632827071677e-06,
+      "loss": 0.5631,
+      "step": 11540
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.612310151878498e-06,
+      "loss": 0.5487,
+      "step": 11550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5989874766853185e-06,
+      "loss": 0.4099,
+      "step": 11560
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.58566480149214e-06,
+      "loss": 0.6225,
+      "step": 11570
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.572342126298961e-06,
+      "loss": 0.5682,
+      "step": 11580
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.559019451105782e-06,
+      "loss": 0.3695,
+      "step": 11590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.545696775912604e-06,
+      "loss": 0.5973,
+      "step": 11600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5323741007194245e-06,
+      "loss": 0.5033,
+      "step": 11610
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.519051425526246e-06,
+      "loss": 0.6484,
+      "step": 11620
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.505728750333067e-06,
+      "loss": 0.516,
+      "step": 11630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.492406075139889e-06,
+      "loss": 0.6044,
+      "step": 11640
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.47908339994671e-06,
+      "loss": 0.6888,
+      "step": 11650
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.465760724753531e-06,
+      "loss": 0.6246,
+      "step": 11660
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.452438049560352e-06,
+      "loss": 0.4872,
+      "step": 11670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.439115374367173e-06,
+      "loss": 0.4675,
+      "step": 11680
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.425792699173995e-06,
+      "loss": 0.594,
+      "step": 11690
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.412470023980816e-06,
+      "loss": 0.7868,
+      "step": 11700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.399147348787637e-06,
+      "loss": 0.5357,
+      "step": 11710
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.385824673594458e-06,
+      "loss": 0.4265,
+      "step": 11720
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.372501998401279e-06,
+      "loss": 0.4934,
+      "step": 11730
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.3591793232081e-06,
+      "loss": 0.5334,
+      "step": 11740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.345856648014922e-06,
+      "loss": 0.5469,
+      "step": 11750
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.332533972821743e-06,
+      "loss": 0.5974,
+      "step": 11760
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.319211297628564e-06,
+      "loss": 0.6087,
+      "step": 11770
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.305888622435385e-06,
+      "loss": 0.4159,
+      "step": 11780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.292565947242206e-06,
+      "loss": 0.4755,
+      "step": 11790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.279243272049028e-06,
+      "loss": 0.5726,
+      "step": 11800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2659205968558496e-06,
+      "loss": 0.4904,
+      "step": 11810
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2525979216626704e-06,
+      "loss": 0.6391,
+      "step": 11820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.239275246469491e-06,
+      "loss": 0.5228,
+      "step": 11830
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.225952571276313e-06,
+      "loss": 0.7083,
+      "step": 11840
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.212629896083134e-06,
+      "loss": 0.5254,
+      "step": 11850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.199307220889955e-06,
+      "loss": 0.416,
+      "step": 11860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.1859845456967765e-06,
+      "loss": 0.6754,
+      "step": 11870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.172661870503597e-06,
+      "loss": 0.3771,
+      "step": 11880
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.159339195310418e-06,
+      "loss": 0.6722,
+      "step": 11890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.14601652011724e-06,
+      "loss": 0.5206,
+      "step": 11900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.132693844924061e-06,
+      "loss": 0.6298,
+      "step": 11910
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1193711697308825e-06,
+      "loss": 0.5899,
+      "step": 11920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.106048494537703e-06,
+      "loss": 0.4625,
+      "step": 11930
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.092725819344524e-06,
+      "loss": 0.5672,
+      "step": 11940
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.079403144151346e-06,
+      "loss": 0.5623,
+      "step": 11950
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.066080468958168e-06,
+      "loss": 0.3423,
+      "step": 11960
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.052757793764988e-06,
+      "loss": 0.4891,
+      "step": 11970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0394351185718094e-06,
+      "loss": 0.3861,
+      "step": 11980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.026112443378631e-06,
+      "loss": 0.4937,
+      "step": 11990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.012789768185452e-06,
+      "loss": 0.5285,
+      "step": 12000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.999467092992273e-06,
+      "loss": 0.3314,
+      "step": 12010
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.986144417799095e-06,
+      "loss": 0.5348,
+      "step": 12020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9728217426059155e-06,
+      "loss": 0.3142,
+      "step": 12030
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.959499067412737e-06,
+      "loss": 0.644,
+      "step": 12040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.946176392219558e-06,
+      "loss": 0.6136,
+      "step": 12050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.932853717026379e-06,
+      "loss": 0.6213,
+      "step": 12060
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.919531041833201e-06,
+      "loss": 0.5401,
+      "step": 12070
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9062083666400215e-06,
+      "loss": 0.4765,
+      "step": 12080
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.892885691446842e-06,
+      "loss": 0.6232,
+      "step": 12090
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.879563016253664e-06,
+      "loss": 0.5262,
+      "step": 12100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.866240341060486e-06,
+      "loss": 0.5294,
+      "step": 12110
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.852917665867306e-06,
+      "loss": 0.6858,
+      "step": 12120
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8395949906741276e-06,
+      "loss": 0.5539,
+      "step": 12130
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.826272315480949e-06,
+      "loss": 0.5337,
+      "step": 12140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.81294964028777e-06,
+      "loss": 0.4814,
+      "step": 12150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.799626965094591e-06,
+      "loss": 0.5019,
+      "step": 12160
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7863042899014123e-06,
+      "loss": 0.4682,
+      "step": 12170
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7729816147082336e-06,
+      "loss": 0.6597,
+      "step": 12180
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7596589395150553e-06,
+      "loss": 0.5827,
+      "step": 12190
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7463362643218758e-06,
+      "loss": 0.4335,
+      "step": 12200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.733013589128697e-06,
+      "loss": 0.4961,
+      "step": 12210
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7196909139355188e-06,
+      "loss": 0.5998,
+      "step": 12220
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.70636823874234e-06,
+      "loss": 0.5215,
+      "step": 12230
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.693045563549161e-06,
+      "loss": 0.6032,
+      "step": 12240
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.6797228883559822e-06,
+      "loss": 0.6204,
+      "step": 12250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.6664002131628035e-06,
+      "loss": 0.5401,
+      "step": 12260
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.653077537969625e-06,
+      "loss": 0.7566,
+      "step": 12270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6397548627764457e-06,
+      "loss": 0.59,
+      "step": 12280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.626432187583267e-06,
+      "loss": 0.5552,
+      "step": 12290
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6131095123900883e-06,
+      "loss": 0.4806,
+      "step": 12300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5997868371969096e-06,
+      "loss": 0.4438,
+      "step": 12310
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5864641620037304e-06,
+      "loss": 0.3571,
+      "step": 12320
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5731414868105517e-06,
+      "loss": 0.5311,
+      "step": 12330
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.559818811617373e-06,
+      "loss": 0.7225,
+      "step": 12340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.546496136424194e-06,
+      "loss": 0.5775,
+      "step": 12350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.533173461231015e-06,
+      "loss": 0.6592,
+      "step": 12360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5198507860378365e-06,
+      "loss": 0.5143,
+      "step": 12370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.506528110844658e-06,
+      "loss": 0.6855,
+      "step": 12380
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4932054356514787e-06,
+      "loss": 0.3721,
+      "step": 12390
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4798827604583004e-06,
+      "loss": 0.5709,
+      "step": 12400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4665600852651217e-06,
+      "loss": 0.5464,
+      "step": 12410
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.453237410071943e-06,
+      "loss": 0.5,
+      "step": 12420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.439914734878764e-06,
+      "loss": 0.3392,
+      "step": 12430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.426592059685585e-06,
+      "loss": 0.7597,
+      "step": 12440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4132693844924064e-06,
+      "loss": 0.6143,
+      "step": 12450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3999467092992277e-06,
+      "loss": 0.5592,
+      "step": 12460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3866240341060486e-06,
+      "loss": 0.6935,
+      "step": 12470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.37330135891287e-06,
+      "loss": 0.6525,
+      "step": 12480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.359978683719691e-06,
+      "loss": 0.4908,
+      "step": 12490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3466560085265125e-06,
+      "loss": 0.5525,
+      "step": 12500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.7124,
+      "step": 12510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3200106581401546e-06,
+      "loss": 0.7283,
+      "step": 12520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3066879829469763e-06,
+      "loss": 0.4084,
+      "step": 12530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2933653077537976e-06,
+      "loss": 0.5301,
+      "step": 12540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.280042632560618e-06,
+      "loss": 0.6944,
+      "step": 12550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.26671995736744e-06,
+      "loss": 0.545,
+      "step": 12560
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.253397282174261e-06,
+      "loss": 0.5842,
+      "step": 12570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2400746069810824e-06,
+      "loss": 0.3925,
+      "step": 12580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2267519317879032e-06,
+      "loss": 0.4888,
+      "step": 12590
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2134292565947245e-06,
+      "loss": 0.777,
+      "step": 12600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.200106581401546e-06,
+      "loss": 0.7409,
+      "step": 12610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.1867839062083667e-06,
+      "loss": 0.4189,
+      "step": 12620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.173461231015188e-06,
+      "loss": 0.4592,
+      "step": 12630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.1601385558220093e-06,
+      "loss": 0.5899,
+      "step": 12640
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1468158806288306e-06,
+      "loss": 0.5961,
+      "step": 12650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1334932054356515e-06,
+      "loss": 0.4069,
+      "step": 12660
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1201705302424727e-06,
+      "loss": 0.4548,
+      "step": 12670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.106847855049294e-06,
+      "loss": 0.4679,
+      "step": 12680
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0935251798561158e-06,
+      "loss": 0.3883,
+      "step": 12690
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.080202504662936e-06,
+      "loss": 0.3436,
+      "step": 12700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0668798294697575e-06,
+      "loss": 0.6312,
+      "step": 12710
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0535571542765792e-06,
+      "loss": 0.6167,
+      "step": 12720
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0402344790834005e-06,
+      "loss": 0.4497,
+      "step": 12730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0269118038902214e-06,
+      "loss": 0.4056,
+      "step": 12740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0135891286970427e-06,
+      "loss": 0.6267,
+      "step": 12750
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.000266453503864e-06,
+      "loss": 0.5406,
+      "step": 12760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9869437783106853e-06,
+      "loss": 0.5243,
+      "step": 12770
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.973621103117506e-06,
+      "loss": 0.5,
+      "step": 12780
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9602984279243274e-06,
+      "loss": 0.4584,
+      "step": 12790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9469757527311487e-06,
+      "loss": 0.758,
+      "step": 12800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.93365307753797e-06,
+      "loss": 0.3937,
+      "step": 12810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.920330402344791e-06,
+      "loss": 0.4183,
+      "step": 12820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.907007727151612e-06,
+      "loss": 0.518,
+      "step": 12830
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8936850519584335e-06,
+      "loss": 0.4435,
+      "step": 12840
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8803623767652543e-06,
+      "loss": 0.6761,
+      "step": 12850
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8670397015720756e-06,
+      "loss": 0.5532,
+      "step": 12860
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8537170263788973e-06,
+      "loss": 0.3782,
+      "step": 12870
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8403943511857186e-06,
+      "loss": 0.7742,
+      "step": 12880
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.827071675992539e-06,
+      "loss": 0.4839,
+      "step": 12890
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.813749000799361e-06,
+      "loss": 0.4243,
+      "step": 12900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.800426325606182e-06,
+      "loss": 0.5923,
+      "step": 12910
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7871036504130034e-06,
+      "loss": 0.5949,
+      "step": 12920
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7737809752198243e-06,
+      "loss": 0.5117,
+      "step": 12930
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7604583000266455e-06,
+      "loss": 0.7791,
+      "step": 12940
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.747135624833467e-06,
+      "loss": 0.5184,
+      "step": 12950
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.733812949640288e-06,
+      "loss": 0.5903,
+      "step": 12960
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.720490274447109e-06,
+      "loss": 0.6456,
+      "step": 12970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7071675992539303e-06,
+      "loss": 0.4857,
+      "step": 12980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6938449240607516e-06,
+      "loss": 0.5727,
+      "step": 12990
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.680522248867573e-06,
+      "loss": 0.4204,
+      "step": 13000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6671995736743938e-06,
+      "loss": 0.3454,
+      "step": 13010
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.653876898481215e-06,
+      "loss": 0.6435,
+      "step": 13020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6405542232880368e-06,
+      "loss": 0.6825,
+      "step": 13030
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.627231548094858e-06,
+      "loss": 0.7152,
+      "step": 13040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6139088729016785e-06,
+      "loss": 0.448,
+      "step": 13050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6005861977085002e-06,
+      "loss": 0.4351,
+      "step": 13060
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5872635225153215e-06,
+      "loss": 0.7678,
+      "step": 13070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5739408473221424e-06,
+      "loss": 0.3533,
+      "step": 13080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5606181721289637e-06,
+      "loss": 0.6413,
+      "step": 13090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.547295496935785e-06,
+      "loss": 0.5917,
+      "step": 13100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5339728217426063e-06,
+      "loss": 0.5904,
+      "step": 13110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.520650146549427e-06,
+      "loss": 0.5905,
+      "step": 13120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5073274713562484e-06,
+      "loss": 0.3289,
+      "step": 13130
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4940047961630697e-06,
+      "loss": 0.4646,
+      "step": 13140
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.480682120969891e-06,
+      "loss": 0.5391,
+      "step": 13150
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4673594457767123e-06,
+      "loss": 0.4321,
+      "step": 13160
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.454036770583533e-06,
+      "loss": 0.7442,
+      "step": 13170
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4407140953903545e-06,
+      "loss": 0.5778,
+      "step": 13180
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4273914201971758e-06,
+      "loss": 0.4422,
+      "step": 13190
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.414068745003997e-06,
+      "loss": 0.6558,
+      "step": 13200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4007460698108184e-06,
+      "loss": 0.7679,
+      "step": 13210
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3874233946176396e-06,
+      "loss": 0.4981,
+      "step": 13220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3741007194244605e-06,
+      "loss": 0.6829,
+      "step": 13230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.360778044231282e-06,
+      "loss": 0.3649,
+      "step": 13240
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.347455369038103e-06,
+      "loss": 0.5827,
+      "step": 13250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3341326938449244e-06,
+      "loss": 0.4577,
+      "step": 13260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3208100186517453e-06,
+      "loss": 0.3576,
+      "step": 13270
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.307487343458567e-06,
+      "loss": 0.566,
+      "step": 13280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.294164668265388e-06,
+      "loss": 0.4685,
+      "step": 13290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.280841993072209e-06,
+      "loss": 0.5552,
+      "step": 13300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2675193178790304e-06,
+      "loss": 0.4454,
+      "step": 13310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2541966426858513e-06,
+      "loss": 0.6684,
+      "step": 13320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2408739674926726e-06,
+      "loss": 0.5333,
+      "step": 13330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.227551292299494e-06,
+      "loss": 0.5584,
+      "step": 13340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.214228617106315e-06,
+      "loss": 0.592,
+      "step": 13350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.200905941913136e-06,
+      "loss": 0.5376,
+      "step": 13360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1875832667199578e-06,
+      "loss": 0.3944,
+      "step": 13370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1742605915267786e-06,
+      "loss": 0.5578,
+      "step": 13380
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1609379163336e-06,
+      "loss": 0.3291,
+      "step": 13390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1476152411404212e-06,
+      "loss": 0.6107,
+      "step": 13400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1342925659472425e-06,
+      "loss": 0.5605,
+      "step": 13410
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1209698907540634e-06,
+      "loss": 0.7272,
+      "step": 13420
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1076472155608847e-06,
+      "loss": 0.6814,
+      "step": 13430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.094324540367706e-06,
+      "loss": 0.5591,
+      "step": 13440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0810018651745273e-06,
+      "loss": 0.4281,
+      "step": 13450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0676791899813486e-06,
+      "loss": 0.4085,
+      "step": 13460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.05435651478817e-06,
+      "loss": 0.4977,
+      "step": 13470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0410338395949907e-06,
+      "loss": 0.403,
+      "step": 13480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.027711164401812e-06,
+      "loss": 0.5554,
+      "step": 13490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0143884892086333e-06,
+      "loss": 0.563,
+      "step": 13500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0010658140154546e-06,
+      "loss": 0.5317,
+      "step": 13510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9877431388222755e-06,
+      "loss": 0.5416,
+      "step": 13520
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.974420463629097e-06,
+      "loss": 0.59,
+      "step": 13530
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.961097788435918e-06,
+      "loss": 0.4728,
+      "step": 13540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9477751132427394e-06,
+      "loss": 0.6021,
+      "step": 13550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9344524380495607e-06,
+      "loss": 0.5801,
+      "step": 13560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9211297628563815e-06,
+      "loss": 0.5634,
+      "step": 13570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.907807087663203e-06,
+      "loss": 0.5021,
+      "step": 13580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8944844124700241e-06,
+      "loss": 0.5824,
+      "step": 13590
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8811617372768454e-06,
+      "loss": 0.6598,
+      "step": 13600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8678390620836665e-06,
+      "loss": 0.5233,
+      "step": 13610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8545163868904878e-06,
+      "loss": 0.7899,
+      "step": 13620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8411937116973089e-06,
+      "loss": 0.4576,
+      "step": 13630
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8278710365041302e-06,
+      "loss": 0.4726,
+      "step": 13640
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8145483613109512e-06,
+      "loss": 0.6278,
+      "step": 13650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8012256861177727e-06,
+      "loss": 0.4879,
+      "step": 13660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7879030109245938e-06,
+      "loss": 0.5082,
+      "step": 13670
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7745803357314151e-06,
+      "loss": 0.5767,
+      "step": 13680
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7612576605382362e-06,
+      "loss": 0.4782,
+      "step": 13690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7479349853450575e-06,
+      "loss": 0.5911,
+      "step": 13700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7346123101518786e-06,
+      "loss": 0.4495,
+      "step": 13710
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7212896349586999e-06,
+      "loss": 0.6191,
+      "step": 13720
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.707966959765521e-06,
+      "loss": 0.5123,
+      "step": 13730
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6946442845723425e-06,
+      "loss": 0.6759,
+      "step": 13740
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6813216093791635e-06,
+      "loss": 0.6946,
+      "step": 13750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6679989341859848e-06,
+      "loss": 0.8245,
+      "step": 13760
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.654676258992806e-06,
+      "loss": 0.4792,
+      "step": 13770
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6413535837996272e-06,
+      "loss": 0.4847,
+      "step": 13780
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6280309086064483e-06,
+      "loss": 0.5823,
+      "step": 13790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6147082334132696e-06,
+      "loss": 0.6636,
+      "step": 13800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6013855582200907e-06,
+      "loss": 0.4208,
+      "step": 13810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5880628830269117e-06,
+      "loss": 0.4308,
+      "step": 13820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5747402078337332e-06,
+      "loss": 0.6352,
+      "step": 13830
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5614175326405543e-06,
+      "loss": 0.4812,
+      "step": 13840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5480948574473756e-06,
+      "loss": 0.281,
+      "step": 13850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5347721822541967e-06,
+      "loss": 0.6616,
+      "step": 13860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.521449507061018e-06,
+      "loss": 0.6039,
+      "step": 13870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.508126831867839e-06,
+      "loss": 0.5124,
+      "step": 13880
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4948041566746604e-06,
+      "loss": 0.5828,
+      "step": 13890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 0.4895,
+      "step": 13900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.468158806288303e-06,
+      "loss": 0.5507,
+      "step": 13910
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.454836131095124e-06,
+      "loss": 0.3509,
+      "step": 13920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4415134559019453e-06,
+      "loss": 0.6344,
+      "step": 13930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4281907807087664e-06,
+      "loss": 0.435,
+      "step": 13940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4148681055155877e-06,
+      "loss": 0.5045,
+      "step": 13950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4015454303224088e-06,
+      "loss": 0.6366,
+      "step": 13960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.38822275512923e-06,
+      "loss": 0.5821,
+      "step": 13970
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3749000799360512e-06,
+      "loss": 0.3629,
+      "step": 13980
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3615774047428727e-06,
+      "loss": 0.4365,
+      "step": 13990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3482547295496938e-06,
+      "loss": 0.5178,
+      "step": 14000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.334932054356515e-06,
+      "loss": 0.7401,
+      "step": 14010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3216093791633361e-06,
+      "loss": 0.59,
+      "step": 14020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3082867039701574e-06,
+      "loss": 0.5952,
+      "step": 14030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2949640287769785e-06,
+      "loss": 0.7081,
+      "step": 14040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2816413535837996e-06,
+      "loss": 0.3996,
+      "step": 14050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2683186783906209e-06,
+      "loss": 0.537,
+      "step": 14060
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.254996003197442e-06,
+      "loss": 0.4258,
+      "step": 14070
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2416733280042635e-06,
+      "loss": 0.3974,
+      "step": 14080
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2283506528110845e-06,
+      "loss": 0.5042,
+      "step": 14090
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2150279776179058e-06,
+      "loss": 0.5661,
+      "step": 14100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.201705302424727e-06,
+      "loss": 0.457,
+      "step": 14110
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1883826272315482e-06,
+      "loss": 0.6158,
+      "step": 14120
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1750599520383695e-06,
+      "loss": 0.5112,
+      "step": 14130
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1617372768451906e-06,
+      "loss": 0.7427,
+      "step": 14140
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1484146016520119e-06,
+      "loss": 0.548,
+      "step": 14150
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1350919264588332e-06,
+      "loss": 0.4385,
+      "step": 14160
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1217692512656543e-06,
+      "loss": 0.6332,
+      "step": 14170
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1084465760724753e-06,
+      "loss": 0.6839,
+      "step": 14180
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0951239008792966e-06,
+      "loss": 0.5741,
+      "step": 14190
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0818012256861177e-06,
+      "loss": 0.4525,
+      "step": 14200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.068478550492939e-06,
+      "loss": 0.4109,
+      "step": 14210
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0551558752997603e-06,
+      "loss": 0.4207,
+      "step": 14220
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0418332001065814e-06,
+      "loss": 0.5708,
+      "step": 14230
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0285105249134027e-06,
+      "loss": 0.5641,
+      "step": 14240
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.015187849720224e-06,
+      "loss": 0.5639,
+      "step": 14250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.001865174527045e-06,
+      "loss": 0.5215,
+      "step": 14260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.885424993338663e-07,
+      "loss": 0.5582,
+      "step": 14270
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.752198241406874e-07,
+      "loss": 0.6286,
+      "step": 14280
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.618971489475087e-07,
+      "loss": 0.4788,
+      "step": 14290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.485744737543299e-07,
+      "loss": 0.5799,
+      "step": 14300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.352517985611512e-07,
+      "loss": 0.7389,
+      "step": 14310
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.219291233679724e-07,
+      "loss": 0.6456,
+      "step": 14320
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.086064481747936e-07,
+      "loss": 0.4753,
+      "step": 14330
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.952837729816148e-07,
+      "loss": 0.5516,
+      "step": 14340
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.819610977884361e-07,
+      "loss": 0.6144,
+      "step": 14350
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.686384225952572e-07,
+      "loss": 0.5399,
+      "step": 14360
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.553157474020784e-07,
+      "loss": 0.4359,
+      "step": 14370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.419930722088996e-07,
+      "loss": 0.6362,
+      "step": 14380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.286703970157209e-07,
+      "loss": 0.6311,
+      "step": 14390
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.153477218225421e-07,
+      "loss": 0.528,
+      "step": 14400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.020250466293632e-07,
+      "loss": 0.4934,
+      "step": 14410
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.887023714361844e-07,
+      "loss": 0.3794,
+      "step": 14420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.753796962430056e-07,
+      "loss": 0.622,
+      "step": 14430
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.620570210498268e-07,
+      "loss": 0.542,
+      "step": 14440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.48734345856648e-07,
+      "loss": 0.5141,
+      "step": 14450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.354116706634692e-07,
+      "loss": 0.6752,
+      "step": 14460
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.220889954702904e-07,
+      "loss": 0.5327,
+      "step": 14470
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.087663202771117e-07,
+      "loss": 0.4935,
+      "step": 14480
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.954436450839329e-07,
+      "loss": 0.53,
+      "step": 14490
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.821209698907541e-07,
+      "loss": 0.5226,
+      "step": 14500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.687982946975753e-07,
+      "loss": 0.4801,
+      "step": 14510
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.554756195043966e-07,
+      "loss": 0.5048,
+      "step": 14520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.421529443112177e-07,
+      "loss": 0.5826,
+      "step": 14530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.288302691180389e-07,
+      "loss": 0.553,
+      "step": 14540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.155075939248602e-07,
+      "loss": 0.592,
+      "step": 14550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.021849187316814e-07,
+      "loss": 0.5556,
+      "step": 14560
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.888622435385026e-07,
+      "loss": 0.5731,
+      "step": 14570
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.755395683453238e-07,
+      "loss": 0.5511,
+      "step": 14580
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.62216893152145e-07,
+      "loss": 0.6793,
+      "step": 14590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.488942179589662e-07,
+      "loss": 0.4778,
+      "step": 14600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.355715427657874e-07,
+      "loss": 0.5263,
+      "step": 14610
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.222488675726086e-07,
+      "loss": 0.5329,
+      "step": 14620
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.089261923794298e-07,
+      "loss": 0.4903,
+      "step": 14630
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.95603517186251e-07,
+      "loss": 0.5053,
+      "step": 14640
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.822808419930722e-07,
+      "loss": 0.6906,
+      "step": 14650
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6895816679989345e-07,
+      "loss": 0.4313,
+      "step": 14660
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.556354916067147e-07,
+      "loss": 0.3458,
+      "step": 14670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.423128164135359e-07,
+      "loss": 0.4378,
+      "step": 14680
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.289901412203571e-07,
+      "loss": 0.6225,
+      "step": 14690
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.156674660271783e-07,
+      "loss": 0.4633,
+      "step": 14700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0234479083399944e-07,
+      "loss": 0.5472,
+      "step": 14710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.890221156408207e-07,
+      "loss": 0.4989,
+      "step": 14720
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.756994404476419e-07,
+      "loss": 0.6646,
+      "step": 14730
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.623767652544631e-07,
+      "loss": 0.5993,
+      "step": 14740
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4905409006128435e-07,
+      "loss": 0.5009,
+      "step": 14750
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.3573141486810554e-07,
+      "loss": 0.4229,
+      "step": 14760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.224087396749268e-07,
+      "loss": 0.7085,
+      "step": 14770
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0908606448174796e-07,
+      "loss": 0.4241,
+      "step": 14780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.957633892885692e-07,
+      "loss": 0.5699,
+      "step": 14790
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.824407140953904e-07,
+      "loss": 0.5307,
+      "step": 14800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.691180389022116e-07,
+      "loss": 0.4612,
+      "step": 14810
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5579536370903277e-07,
+      "loss": 0.5194,
+      "step": 14820
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.42472688515854e-07,
+      "loss": 0.4469,
+      "step": 14830
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2915001332267522e-07,
+      "loss": 0.3941,
+      "step": 14840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1582733812949643e-07,
+      "loss": 0.4714,
+      "step": 14850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.0250466293631765e-07,
+      "loss": 0.5364,
+      "step": 14860
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8918198774313884e-07,
+      "loss": 0.4054,
+      "step": 14870
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7585931254996005e-07,
+      "loss": 0.4351,
+      "step": 14880
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6253663735678126e-07,
+      "loss": 0.4373,
+      "step": 14890
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4921396216360248e-07,
+      "loss": 0.5169,
+      "step": 14900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3589128697042367e-07,
+      "loss": 0.4695,
+      "step": 14910
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2256861177724488e-07,
+      "loss": 0.6149,
+      "step": 14920
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0924593658406609e-07,
+      "loss": 0.6501,
+      "step": 14930
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.59232613908873e-08,
+      "loss": 0.6889,
+      "step": 14940
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.260058619770851e-08,
+      "loss": 0.5285,
+      "step": 14950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.927791100452971e-08,
+      "loss": 0.5547,
+      "step": 14960
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.595523581135092e-08,
+      "loss": 0.3699,
+      "step": 14970
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.263256061817213e-08,
+      "loss": 0.4134,
+      "step": 14980
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.930988542499334e-08,
+      "loss": 0.7625,
+      "step": 14990
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5987210231814548e-08,
+      "loss": 0.546,
+      "step": 15000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.664535038635758e-09,
+      "loss": 0.4524,
+      "step": 15010
+    }
+  ],
+  "max_steps": 15012,
+  "num_train_epochs": 2,
+  "total_flos": 1.1146687779176448e+17,
+  "trial_name": null,
+  "trial_params": null
+}