diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,26647 @@
+{
+  "best_metric": 0.45416226983070374,
+  "best_model_checkpoint": "ai-light-dance_singing_ft_wav2vec2-large-lv60/checkpoint-26532",
+  "epoch": 10.0,
+  "global_step": 44220,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 37.1758,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.4e-07,
+      "loss": 33.7139,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-06,
+      "loss": 34.2276,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-06,
+      "loss": 33.332,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.58e-06,
+      "loss": 60.1396,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.18e-06,
+      "loss": 50.1667,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 29.6683,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3799999999999996e-06,
+      "loss": 34.8814,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 32.3991,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.58e-06,
+      "loss": 21.3376,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.18e-06,
+      "loss": 27.9194,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.78e-06,
+      "loss": 27.0382,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 14.7897,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.98e-06,
+      "loss": 19.5646,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.58e-06,
+      "loss": 12.8493,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.18e-06,
+      "loss": 19.8708,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 10.4222,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0379999999999999e-05,
+      "loss": 8.5441,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.098e-05,
+      "loss": 10.8924,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 16.0292,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 13.912,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.278e-05,
+      "loss": 7.436,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.338e-05,
+      "loss": 9.9503,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 8.6611,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.458e-05,
+      "loss": 5.4315,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.518e-05,
+      "loss": 5.8837,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.578e-05,
+      "loss": 6.534,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 7.3949,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.698e-05,
+      "loss": 5.3465,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.758e-05,
+      "loss": 6.0069,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.818e-05,
+      "loss": 4.97,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.878e-05,
+      "loss": 6.7921,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.938e-05,
+      "loss": 5.1235,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 4.119,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 4.1115,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.118e-05,
+      "loss": 5.2302,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.178e-05,
+      "loss": 4.7818,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.238e-05,
+      "loss": 3.2685,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.298e-05,
+      "loss": 3.1541,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.358e-05,
+      "loss": 4.516,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4180000000000002e-05,
+      "loss": 3.0714,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.478e-05,
+      "loss": 4.6763,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.538e-05,
+      "loss": 2.8357,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5980000000000002e-05,
+      "loss": 3.3214,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.658e-05,
+      "loss": 3.2299,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.718e-05,
+      "loss": 4.7229,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.778e-05,
+      "loss": 3.1254,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.838e-05,
+      "loss": 3.0209,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.898e-05,
+      "loss": 3.15,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.958e-05,
+      "loss": 2.5196,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9997941445562673e-05,
+      "loss": 2.6628,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9991079597438244e-05,
+      "loss": 3.1059,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9984217749313815e-05,
+      "loss": 2.6725,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997735590118939e-05,
+      "loss": 2.4818,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9970494053064958e-05,
+      "loss": 2.966,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996363220494053e-05,
+      "loss": 3.2046,
+      "step": 560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9956770356816104e-05,
+      "loss": 3.0149,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9949908508691672e-05,
+      "loss": 2.665,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9943046660567247e-05,
+      "loss": 3.1208,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9936184812442818e-05,
+      "loss": 2.723,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.992932296431839e-05,
+      "loss": 2.5166,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.992246111619396e-05,
+      "loss": 2.396,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9915599268069535e-05,
+      "loss": 2.2962,
+      "step": 630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9908737419945103e-05,
+      "loss": 2.453,
+      "step": 640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9901875571820678e-05,
+      "loss": 2.5502,
+      "step": 650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.989501372369625e-05,
+      "loss": 3.1002,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.988815187557182e-05,
+      "loss": 2.6122,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9881290027447392e-05,
+      "loss": 3.3551,
+      "step": 680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9874428179322967e-05,
+      "loss": 2.544,
+      "step": 690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9867566331198534e-05,
+      "loss": 2.5115,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.986070448307411e-05,
+      "loss": 2.4845,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.985384263494968e-05,
+      "loss": 2.5213,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9846980786825252e-05,
+      "loss": 2.578,
+      "step": 730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9840118938700823e-05,
+      "loss": 2.815,
+      "step": 740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9833257090576398e-05,
+      "loss": 2.5662,
+      "step": 750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9826395242451966e-05,
+      "loss": 2.4813,
+      "step": 760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.981953339432754e-05,
+      "loss": 2.8654,
+      "step": 770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.981267154620311e-05,
+      "loss": 2.4185,
+      "step": 780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9805809698078683e-05,
+      "loss": 2.5217,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9798947849954254e-05,
+      "loss": 2.2132,
+      "step": 800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.979208600182983e-05,
+      "loss": 2.4615,
+      "step": 810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9785224153705397e-05,
+      "loss": 2.3366,
+      "step": 820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.977836230558097e-05,
+      "loss": 2.1925,
+      "step": 830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9771500457456543e-05,
+      "loss": 2.24,
+      "step": 840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9764638609332114e-05,
+      "loss": 2.9538,
+      "step": 850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9757776761207686e-05,
+      "loss": 2.4215,
+      "step": 860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.975091491308326e-05,
+      "loss": 2.2419,
+      "step": 870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9744053064958828e-05,
+      "loss": 2.3682,
+      "step": 880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9737191216834403e-05,
+      "loss": 2.2048,
+      "step": 890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9730329368709974e-05,
+      "loss": 2.5094,
+      "step": 900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9723467520585545e-05,
+      "loss": 2.3766,
+      "step": 910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9716605672461117e-05,
+      "loss": 2.1605,
+      "step": 920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.970974382433669e-05,
+      "loss": 2.166,
+      "step": 930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.970288197621226e-05,
+      "loss": 2.2566,
+      "step": 940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9696020128087834e-05,
+      "loss": 2.2579,
+      "step": 950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9689158279963405e-05,
+      "loss": 2.2978,
+      "step": 960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9682296431838973e-05,
+      "loss": 2.2482,
+      "step": 970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9675434583714548e-05,
+      "loss": 2.4976,
+      "step": 980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.966857273559012e-05,
+      "loss": 2.2864,
+      "step": 990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.966171088746569e-05,
+      "loss": 2.2622,
+      "step": 1000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9654849039341262e-05,
+      "loss": 2.4167,
+      "step": 1010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9647987191216837e-05,
+      "loss": 2.2898,
+      "step": 1020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9641125343092404e-05,
+      "loss": 2.3912,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.963426349496798e-05,
+      "loss": 2.2157,
+      "step": 1040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.962740164684355e-05,
+      "loss": 2.2854,
+      "step": 1050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9620539798719122e-05,
+      "loss": 2.2625,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9613677950594693e-05,
+      "loss": 2.3351,
+      "step": 1070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9606816102470268e-05,
+      "loss": 2.3576,
+      "step": 1080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9599954254345836e-05,
+      "loss": 2.1645,
+      "step": 1090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.959309240622141e-05,
+      "loss": 2.2722,
+      "step": 1100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9586230558096982e-05,
+      "loss": 2.3357,
+      "step": 1110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9579368709972553e-05,
+      "loss": 2.2817,
+      "step": 1120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9572506861848124e-05,
+      "loss": 2.1896,
+      "step": 1130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.95656450137237e-05,
+      "loss": 2.4213,
+      "step": 1140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9558783165599267e-05,
+      "loss": 2.2778,
+      "step": 1150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.955192131747484e-05,
+      "loss": 2.2386,
+      "step": 1160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9545059469350413e-05,
+      "loss": 2.1875,
+      "step": 1170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9538197621225984e-05,
+      "loss": 2.4207,
+      "step": 1180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9531335773101556e-05,
+      "loss": 2.2113,
+      "step": 1190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.952447392497713e-05,
+      "loss": 2.2096,
+      "step": 1200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9517612076852698e-05,
+      "loss": 2.2508,
+      "step": 1210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9510750228728273e-05,
+      "loss": 2.1723,
+      "step": 1220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9503888380603844e-05,
+      "loss": 2.1498,
+      "step": 1230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9497026532479415e-05,
+      "loss": 2.1461,
+      "step": 1240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9490164684354987e-05,
+      "loss": 2.1665,
+      "step": 1250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.948330283623056e-05,
+      "loss": 2.6024,
+      "step": 1260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.947644098810613e-05,
+      "loss": 2.1082,
+      "step": 1270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9469579139981704e-05,
+      "loss": 2.096,
+      "step": 1280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9462717291857275e-05,
+      "loss": 2.4516,
+      "step": 1290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9455855443732847e-05,
+      "loss": 2.1172,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9448993595608418e-05,
+      "loss": 2.3277,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9442131747483993e-05,
+      "loss": 2.1226,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.943526989935956e-05,
+      "loss": 2.1378,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9428408051235135e-05,
+      "loss": 2.1837,
+      "step": 1340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9421546203110707e-05,
+      "loss": 2.0979,
+      "step": 1350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9414684354986275e-05,
+      "loss": 2.226,
+      "step": 1360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.940782250686185e-05,
+      "loss": 2.1714,
+      "step": 1370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.940096065873742e-05,
+      "loss": 2.1507,
+      "step": 1380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9394098810612992e-05,
+      "loss": 2.1953,
+      "step": 1390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9387236962488563e-05,
+      "loss": 2.1311,
+      "step": 1400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9380375114364138e-05,
+      "loss": 2.0973,
+      "step": 1410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9373513266239706e-05,
+      "loss": 2.1186,
+      "step": 1420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.936665141811528e-05,
+      "loss": 2.161,
+      "step": 1430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9359789569990852e-05,
+      "loss": 2.2864,
+      "step": 1440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9352927721866423e-05,
+      "loss": 2.2422,
+      "step": 1450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9346065873741994e-05,
+      "loss": 2.2116,
+      "step": 1460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.933920402561757e-05,
+      "loss": 2.1846,
+      "step": 1470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9332342177493137e-05,
+      "loss": 2.1567,
+      "step": 1480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9325480329368712e-05,
+      "loss": 2.1555,
+      "step": 1490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9318618481244283e-05,
+      "loss": 2.1352,
+      "step": 1500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9311756633119854e-05,
+      "loss": 2.095,
+      "step": 1510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9304894784995426e-05,
+      "loss": 2.1243,
+      "step": 1520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9298032936871e-05,
+      "loss": 2.1052,
+      "step": 1530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9291171088746568e-05,
+      "loss": 2.0929,
+      "step": 1540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9284309240622143e-05,
+      "loss": 2.0988,
+      "step": 1550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9277447392497714e-05,
+      "loss": 2.1647,
+      "step": 1560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9270585544373286e-05,
+      "loss": 2.1908,
+      "step": 1570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9263723696248857e-05,
+      "loss": 2.0981,
+      "step": 1580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.925686184812443e-05,
+      "loss": 2.0871,
+      "step": 1590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.925e-05,
+      "loss": 2.0703,
+      "step": 1600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9243138151875574e-05,
+      "loss": 2.1051,
+      "step": 1610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9236276303751145e-05,
+      "loss": 2.0527,
+      "step": 1620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9229414455626717e-05,
+      "loss": 2.0666,
+      "step": 1630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9222552607502288e-05,
+      "loss": 2.0749,
+      "step": 1640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.921569075937786e-05,
+      "loss": 2.0761,
+      "step": 1650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.920882891125343e-05,
+      "loss": 2.1017,
+      "step": 1660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9201967063129005e-05,
+      "loss": 2.3527,
+      "step": 1670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9195105215004573e-05,
+      "loss": 2.2014,
+      "step": 1680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9188243366880148e-05,
+      "loss": 2.081,
+      "step": 1690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.918138151875572e-05,
+      "loss": 2.1462,
+      "step": 1700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.917451967063129e-05,
+      "loss": 2.0499,
+      "step": 1710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9167657822506862e-05,
+      "loss": 2.0307,
+      "step": 1720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9160795974382437e-05,
+      "loss": 2.0292,
+      "step": 1730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9153934126258005e-05,
+      "loss": 2.2026,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.914707227813358e-05,
+      "loss": 2.1391,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.914021043000915e-05,
+      "loss": 2.0657,
+      "step": 1760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.913334858188472e-05,
+      "loss": 2.0181,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9126486733760293e-05,
+      "loss": 2.1043,
+      "step": 1780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9119624885635864e-05,
+      "loss": 2.0328,
+      "step": 1790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9112763037511436e-05,
+      "loss": 2.0506,
+      "step": 1800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9105901189387007e-05,
+      "loss": 2.0206,
+      "step": 1810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9099039341262582e-05,
+      "loss": 2.1034,
+      "step": 1820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.909217749313815e-05,
+      "loss": 2.0943,
+      "step": 1830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9085315645013724e-05,
+      "loss": 2.1201,
+      "step": 1840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9078453796889296e-05,
+      "loss": 2.0623,
+      "step": 1850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9071591948764867e-05,
+      "loss": 2.0372,
+      "step": 1860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9064730100640438e-05,
+      "loss": 2.0054,
+      "step": 1870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9057868252516013e-05,
+      "loss": 2.0497,
+      "step": 1880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.905100640439158e-05,
+      "loss": 2.0002,
+      "step": 1890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9044144556267156e-05,
+      "loss": 2.0797,
+      "step": 1900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9037282708142727e-05,
+      "loss": 2.1054,
+      "step": 1910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9030420860018298e-05,
+      "loss": 2.0184,
+      "step": 1920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.902355901189387e-05,
+      "loss": 2.0164,
+      "step": 1930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9016697163769444e-05,
+      "loss": 2.0079,
+      "step": 1940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9009835315645012e-05,
+      "loss": 2.0289,
+      "step": 1950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9002973467520587e-05,
+      "loss": 2.0082,
+      "step": 1960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8996111619396158e-05,
+      "loss": 1.9881,
+      "step": 1970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.898924977127173e-05,
+      "loss": 1.9745,
+      "step": 1980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.89823879231473e-05,
+      "loss": 1.9501,
+      "step": 1990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8975526075022875e-05,
+      "loss": 2.1461,
+      "step": 2000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8968664226898443e-05,
+      "loss": 1.9909,
+      "step": 2010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8961802378774018e-05,
+      "loss": 1.9796,
+      "step": 2020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.895494053064959e-05,
+      "loss": 2.0002,
+      "step": 2030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.894807868252516e-05,
+      "loss": 1.9989,
+      "step": 2040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8941216834400732e-05,
+      "loss": 1.9924,
+      "step": 2050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8934354986276307e-05,
+      "loss": 2.0759,
+      "step": 2060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8927493138151875e-05,
+      "loss": 1.9797,
+      "step": 2070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.892063129002745e-05,
+      "loss": 1.9897,
+      "step": 2080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.891376944190302e-05,
+      "loss": 1.9414,
+      "step": 2090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8906907593778592e-05,
+      "loss": 1.8993,
+      "step": 2100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8900045745654163e-05,
+      "loss": 1.9662,
+      "step": 2110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8893183897529738e-05,
+      "loss": 2.0106,
+      "step": 2120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8886322049405306e-05,
+      "loss": 1.9246,
+      "step": 2130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.887946020128088e-05,
+      "loss": 1.9744,
+      "step": 2140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8872598353156452e-05,
+      "loss": 2.1524,
+      "step": 2150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8865736505032023e-05,
+      "loss": 2.0603,
+      "step": 2160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8858874656907594e-05,
+      "loss": 1.9362,
+      "step": 2170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8852012808783166e-05,
+      "loss": 2.0504,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8845150960658737e-05,
+      "loss": 2.1456,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.883828911253431e-05,
+      "loss": 1.9578,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8831427264409883e-05,
+      "loss": 1.9269,
+      "step": 2210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.882456541628545e-05,
+      "loss": 1.8796,
+      "step": 2220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8817703568161026e-05,
+      "loss": 1.8765,
+      "step": 2230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8810841720036597e-05,
+      "loss": 1.9282,
+      "step": 2240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8803979871912168e-05,
+      "loss": 1.8994,
+      "step": 2250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.879711802378774e-05,
+      "loss": 2.0287,
+      "step": 2260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8790256175663314e-05,
+      "loss": 1.916,
+      "step": 2270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8783394327538882e-05,
+      "loss": 1.9699,
+      "step": 2280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8776532479414457e-05,
+      "loss": 1.9005,
+      "step": 2290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8769670631290028e-05,
+      "loss": 1.9542,
+      "step": 2300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.87628087831656e-05,
+      "loss": 1.9592,
+      "step": 2310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.875594693504117e-05,
+      "loss": 1.9094,
+      "step": 2320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8749085086916746e-05,
+      "loss": 1.8967,
+      "step": 2330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8742223238792313e-05,
+      "loss": 1.9243,
+      "step": 2340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8735361390667888e-05,
+      "loss": 1.939,
+      "step": 2350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.872849954254346e-05,
+      "loss": 1.8647,
+      "step": 2360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.872163769441903e-05,
+      "loss": 1.8912,
+      "step": 2370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8714775846294602e-05,
+      "loss": 1.9185,
+      "step": 2380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8707913998170177e-05,
+      "loss": 1.8188,
+      "step": 2390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8701052150045745e-05,
+      "loss": 1.8337,
+      "step": 2400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.869419030192132e-05,
+      "loss": 1.7287,
+      "step": 2410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.868732845379689e-05,
+      "loss": 1.6752,
+      "step": 2420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8680466605672462e-05,
+      "loss": 1.5658,
+      "step": 2430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8673604757548033e-05,
+      "loss": 1.6892,
+      "step": 2440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8666742909423608e-05,
+      "loss": 1.5935,
+      "step": 2450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8659881061299176e-05,
+      "loss": 1.6941,
+      "step": 2460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.865301921317475e-05,
+      "loss": 1.5151,
+      "step": 2470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8646157365050322e-05,
+      "loss": 1.6501,
+      "step": 2480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8639295516925893e-05,
+      "loss": 1.5354,
+      "step": 2490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8632433668801465e-05,
+      "loss": 1.4624,
+      "step": 2500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.862557182067704e-05,
+      "loss": 1.4882,
+      "step": 2510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8618709972552607e-05,
+      "loss": 1.5143,
+      "step": 2520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8611848124428182e-05,
+      "loss": 1.4429,
+      "step": 2530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8604986276303753e-05,
+      "loss": 1.4033,
+      "step": 2540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8598124428179324e-05,
+      "loss": 1.3851,
+      "step": 2550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8591262580054896e-05,
+      "loss": 1.3784,
+      "step": 2560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.858440073193047e-05,
+      "loss": 1.3039,
+      "step": 2570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.857753888380604e-05,
+      "loss": 1.392,
+      "step": 2580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.857067703568161e-05,
+      "loss": 1.5103,
+      "step": 2590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8563815187557184e-05,
+      "loss": 1.3928,
+      "step": 2600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8556953339432752e-05,
+      "loss": 1.2144,
+      "step": 2610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8550091491308327e-05,
+      "loss": 1.2911,
+      "step": 2620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8543229643183898e-05,
+      "loss": 1.2075,
+      "step": 2630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.853636779505947e-05,
+      "loss": 1.2228,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.852950594693504e-05,
+      "loss": 2.0613,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8522644098810616e-05,
+      "loss": 1.6446,
+      "step": 2660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8515782250686183e-05,
+      "loss": 1.1913,
+      "step": 2670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8508920402561758e-05,
+      "loss": 1.3027,
+      "step": 2680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.850205855443733e-05,
+      "loss": 1.1386,
+      "step": 2690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.84951967063129e-05,
+      "loss": 1.2568,
+      "step": 2700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8488334858188472e-05,
+      "loss": 1.133,
+      "step": 2710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8481473010064047e-05,
+      "loss": 1.2209,
+      "step": 2720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8474611161939615e-05,
+      "loss": 1.1303,
+      "step": 2730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.846774931381519e-05,
+      "loss": 1.8574,
+      "step": 2740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.846088746569076e-05,
+      "loss": 1.1937,
+      "step": 2750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8454025617566332e-05,
+      "loss": 1.209,
+      "step": 2760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8447163769441903e-05,
+      "loss": 1.1782,
+      "step": 2770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8440301921317478e-05,
+      "loss": 1.1959,
+      "step": 2780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8433440073193046e-05,
+      "loss": 1.1586,
+      "step": 2790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.842657822506862e-05,
+      "loss": 1.2659,
+      "step": 2800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8419716376944192e-05,
+      "loss": 1.0712,
+      "step": 2810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8412854528819763e-05,
+      "loss": 1.0084,
+      "step": 2820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8405992680695335e-05,
+      "loss": 1.06,
+      "step": 2830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.839913083257091e-05,
+      "loss": 1.008,
+      "step": 2840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8392268984446477e-05,
+      "loss": 1.1835,
+      "step": 2850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8385407136322052e-05,
+      "loss": 1.0131,
+      "step": 2860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8378545288197623e-05,
+      "loss": 1.168,
+      "step": 2870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8371683440073194e-05,
+      "loss": 1.0264,
+      "step": 2880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8364821591948766e-05,
+      "loss": 1.0821,
+      "step": 2890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8357959743824337e-05,
+      "loss": 1.1483,
+      "step": 2900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.835109789569991e-05,
+      "loss": 1.0999,
+      "step": 2910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8344236047575483e-05,
+      "loss": 1.0313,
+      "step": 2920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.833737419945105e-05,
+      "loss": 1.0776,
+      "step": 2930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8330512351326626e-05,
+      "loss": 1.1369,
+      "step": 2940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8323650503202197e-05,
+      "loss": 1.0765,
+      "step": 2950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.831678865507777e-05,
+      "loss": 1.0703,
+      "step": 2960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.830992680695334e-05,
+      "loss": 1.2575,
+      "step": 2970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.830306495882891e-05,
+      "loss": 1.0333,
+      "step": 2980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8296203110704482e-05,
+      "loss": 1.0887,
+      "step": 2990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8289341262580054e-05,
+      "loss": 1.0326,
+      "step": 3000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8282479414455628e-05,
+      "loss": 0.9706,
+      "step": 3010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8275617566331196e-05,
+      "loss": 1.346,
+      "step": 3020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.826875571820677e-05,
+      "loss": 0.944,
+      "step": 3030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8261893870082342e-05,
+      "loss": 1.0345,
+      "step": 3040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8255032021957913e-05,
+      "loss": 1.032,
+      "step": 3050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.8248170173833485e-05,
+      "loss": 1.0581,
+      "step": 3060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.824130832570906e-05,
+      "loss": 1.0615,
+      "step": 3070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8234446477584627e-05,
+      "loss": 1.0596,
+      "step": 3080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8227584629460202e-05,
+      "loss": 1.0249,
+      "step": 3090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8220722781335773e-05,
+      "loss": 0.9737,
+      "step": 3100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.8213860933211345e-05,
+      "loss": 1.6334,
+      "step": 3110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8206999085086916e-05,
+      "loss": 0.9849,
+      "step": 3120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.820013723696249e-05,
+      "loss": 1.0126,
+      "step": 3130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.819327538883806e-05,
+      "loss": 0.972,
+      "step": 3140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8186413540713633e-05,
+      "loss": 1.0346,
+      "step": 3150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8179551692589205e-05,
+      "loss": 0.9599,
+      "step": 3160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8172689844464776e-05,
+      "loss": 1.0754,
+      "step": 3170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8165827996340347e-05,
+      "loss": 1.0093,
+      "step": 3180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8158966148215922e-05,
+      "loss": 1.0385,
+      "step": 3190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.815210430009149e-05,
+      "loss": 1.1087,
+      "step": 3200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8145242451967065e-05,
+      "loss": 0.9592,
+      "step": 3210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8138380603842636e-05,
+      "loss": 1.0637,
+      "step": 3220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8131518755718207e-05,
+      "loss": 1.1238,
+      "step": 3230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.812465690759378e-05,
+      "loss": 0.9952,
+      "step": 3240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8117795059469353e-05,
+      "loss": 0.9996,
+      "step": 3250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.811093321134492e-05,
+      "loss": 0.9632,
+      "step": 3260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8104071363220496e-05,
+      "loss": 1.071,
+      "step": 3270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8097209515096067e-05,
+      "loss": 0.9251,
+      "step": 3280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.809034766697164e-05,
+      "loss": 0.932,
+      "step": 3290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.808348581884721e-05,
+      "loss": 0.9899,
+      "step": 3300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8076623970722784e-05,
+      "loss": 1.0929,
+      "step": 3310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8069762122598352e-05,
+      "loss": 0.936,
+      "step": 3320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8062900274473927e-05,
+      "loss": 0.9376,
+      "step": 3330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.80560384263495e-05,
+      "loss": 1.3437,
+      "step": 3340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.804917657822507e-05,
+      "loss": 0.9007,
+      "step": 3350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.804231473010064e-05,
+      "loss": 0.884,
+      "step": 3360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8035452881976216e-05,
+      "loss": 1.019,
+      "step": 3370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8028591033851784e-05,
+      "loss": 0.8682,
+      "step": 3380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8021729185727355e-05,
+      "loss": 0.9508,
+      "step": 3390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.801486733760293e-05,
+      "loss": 0.9011,
+      "step": 3400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8008005489478497e-05,
+      "loss": 0.9312,
+      "step": 3410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8001143641354072e-05,
+      "loss": 1.138,
+      "step": 3420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7994281793229643e-05,
+      "loss": 0.9147,
+      "step": 3430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7987419945105215e-05,
+      "loss": 1.0369,
+      "step": 3440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7980558096980786e-05,
+      "loss": 1.1523,
+      "step": 3450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.797369624885636e-05,
+      "loss": 0.8635,
+      "step": 3460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.796683440073193e-05,
+      "loss": 0.9153,
+      "step": 3470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7959972552607503e-05,
+      "loss": 0.893,
+      "step": 3480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7953110704483075e-05,
+      "loss": 0.8549,
+      "step": 3490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7946248856358646e-05,
+      "loss": 0.9991,
+      "step": 3500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7939387008234217e-05,
+      "loss": 1.0525,
+      "step": 3510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7932525160109792e-05,
+      "loss": 0.87,
+      "step": 3520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.792566331198536e-05,
+      "loss": 0.985,
+      "step": 3530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7918801463860935e-05,
+      "loss": 1.1302,
+      "step": 3540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7911939615736506e-05,
+      "loss": 0.9358,
+      "step": 3550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7905077767612077e-05,
+      "loss": 0.8556,
+      "step": 3560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.789821591948765e-05,
+      "loss": 0.8971,
+      "step": 3570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7891354071363223e-05,
+      "loss": 0.8796,
+      "step": 3580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.788449222323879e-05,
+      "loss": 0.893,
+      "step": 3590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7877630375114366e-05,
+      "loss": 0.9474,
+      "step": 3600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7870768526989937e-05,
+      "loss": 0.8998,
+      "step": 3610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.786390667886551e-05,
+      "loss": 0.9937,
+      "step": 3620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.785704483074108e-05,
+      "loss": 0.8461,
+      "step": 3630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7850182982616654e-05,
+      "loss": 1.0427,
+      "step": 3640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7843321134492222e-05,
+      "loss": 0.971,
+      "step": 3650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7836459286367797e-05,
+      "loss": 0.8469,
+      "step": 3660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.782959743824337e-05,
+      "loss": 0.8628,
+      "step": 3670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.782273559011894e-05,
+      "loss": 0.8588,
+      "step": 3680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.781587374199451e-05,
+      "loss": 0.856,
+      "step": 3690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7809011893870086e-05,
+      "loss": 0.9033,
+      "step": 3700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7802150045745654e-05,
+      "loss": 0.9318,
+      "step": 3710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7795288197621228e-05,
+      "loss": 0.967,
+      "step": 3720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.77884263494968e-05,
+      "loss": 0.9818,
+      "step": 3730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.778156450137237e-05,
+      "loss": 0.8874,
+      "step": 3740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7774702653247942e-05,
+      "loss": 0.8867,
+      "step": 3750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7767840805123517e-05,
+      "loss": 0.8005,
+      "step": 3760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7760978956999085e-05,
+      "loss": 0.7982,
+      "step": 3770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.775411710887466e-05,
+      "loss": 0.8159,
+      "step": 3780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.774725526075023e-05,
+      "loss": 0.9008,
+      "step": 3790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.77403934126258e-05,
+      "loss": 0.8821,
+      "step": 3800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7733531564501373e-05,
+      "loss": 0.9729,
+      "step": 3810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7726669716376945e-05,
+      "loss": 0.915,
+      "step": 3820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7719807868252516e-05,
+      "loss": 0.8572,
+      "step": 3830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7712946020128087e-05,
+      "loss": 0.7661,
+      "step": 3840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7706084172003662e-05,
+      "loss": 0.9187,
+      "step": 3850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.769922232387923e-05,
+      "loss": 0.8268,
+      "step": 3860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7692360475754805e-05,
+      "loss": 0.9062,
+      "step": 3870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7685498627630376e-05,
+      "loss": 0.8894,
+      "step": 3880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7678636779505947e-05,
+      "loss": 0.8746,
+      "step": 3890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.767177493138152e-05,
+      "loss": 0.9264,
+      "step": 3900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7664913083257093e-05,
+      "loss": 0.8538,
+      "step": 3910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.765805123513266e-05,
+      "loss": 1.7354,
+      "step": 3920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7651189387008236e-05,
+      "loss": 0.8216,
+      "step": 3930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7644327538883807e-05,
+      "loss": 0.8997,
+      "step": 3940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.763746569075938e-05,
+      "loss": 0.8912,
+      "step": 3950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.763060384263495e-05,
+      "loss": 0.8727,
+      "step": 3960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7623741994510525e-05,
+      "loss": 1.0729,
+      "step": 3970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7616880146386092e-05,
+      "loss": 1.01,
+      "step": 3980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7610018298261667e-05,
+      "loss": 0.7456,
+      "step": 3990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.760315645013724e-05,
+      "loss": 0.883,
+      "step": 4000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.759629460201281e-05,
+      "loss": 0.8301,
+      "step": 4010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.758943275388838e-05,
+      "loss": 0.8499,
+      "step": 4020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7582570905763956e-05,
+      "loss": 1.0551,
+      "step": 4030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7575709057639524e-05,
+      "loss": 0.8832,
+      "step": 4040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.75688472095151e-05,
+      "loss": 0.7827,
+      "step": 4050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.756198536139067e-05,
+      "loss": 0.8324,
+      "step": 4060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.755512351326624e-05,
+      "loss": 0.7954,
+      "step": 4070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7548261665141812e-05,
+      "loss": 0.9387,
+      "step": 4080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7541399817017387e-05,
+      "loss": 0.7394,
+      "step": 4090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7534537968892955e-05,
+      "loss": 0.9553,
+      "step": 4100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.752767612076853e-05,
+      "loss": 0.895,
+      "step": 4110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.75208142726441e-05,
+      "loss": 0.8401,
+      "step": 4120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7513952424519672e-05,
+      "loss": 0.8303,
+      "step": 4130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7507090576395244e-05,
+      "loss": 0.8984,
+      "step": 4140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7500228728270815e-05,
+      "loss": 0.8402,
+      "step": 4150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7493366880146386e-05,
+      "loss": 0.8464,
+      "step": 4160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.748650503202196e-05,
+      "loss": 0.8514,
+      "step": 4170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.747964318389753e-05,
+      "loss": 0.9106,
+      "step": 4180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7472781335773103e-05,
+      "loss": 1.019,
+      "step": 4190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7465919487648675e-05,
+      "loss": 0.847,
+      "step": 4200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7459057639524243e-05,
+      "loss": 0.8511,
+      "step": 4210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7452195791399817e-05,
+      "loss": 0.816,
+      "step": 4220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.744533394327539e-05,
+      "loss": 1.7753,
+      "step": 4230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.743847209515096e-05,
+      "loss": 0.8036,
+      "step": 4240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.743161024702653e-05,
+      "loss": 0.8818,
+      "step": 4250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7424748398902106e-05,
+      "loss": 0.7232,
+      "step": 4260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7417886550777674e-05,
+      "loss": 1.1432,
+      "step": 4270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.741102470265325e-05,
+      "loss": 0.8257,
+      "step": 4280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.740416285452882e-05,
+      "loss": 0.9215,
+      "step": 4290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.739730100640439e-05,
+      "loss": 0.9081,
+      "step": 4300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7390439158279963e-05,
+      "loss": 0.7445,
+      "step": 4310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7383577310155537e-05,
+      "loss": 0.7597,
+      "step": 4320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7376715462031105e-05,
+      "loss": 0.7473,
+      "step": 4330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.736985361390668e-05,
+      "loss": 0.7908,
+      "step": 4340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.736299176578225e-05,
+      "loss": 0.8619,
+      "step": 4350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7356129917657822e-05,
+      "loss": 0.7501,
+      "step": 4360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7349268069533394e-05,
+      "loss": 0.8016,
+      "step": 4370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.734240622140897e-05,
+      "loss": 0.8872,
+      "step": 4380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7335544373284536e-05,
+      "loss": 0.7989,
+      "step": 4390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.732868252516011e-05,
+      "loss": 0.7617,
+      "step": 4400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7321820677035682e-05,
+      "loss": 0.6984,
+      "step": 4410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7314958828911254e-05,
+      "loss": 0.7432,
+      "step": 4420
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.8938871622085571,
+      "eval_runtime": 301.5602,
+      "eval_samples_per_second": 6.516,
+      "eval_steps_per_second": 1.632,
+      "eval_wer": 0.6323393506253465,
+      "step": 4422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7308096980786825e-05,
+      "loss": 0.7489,
+      "step": 4430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.73012351326624e-05,
+      "loss": 0.6812,
+      "step": 4440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7294373284537968e-05,
+      "loss": 0.7218,
+      "step": 4450
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7287511436413542e-05,
+      "loss": 0.7712,
+      "step": 4460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7280649588289114e-05,
+      "loss": 0.8089,
+      "step": 4470
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7273787740164685e-05,
+      "loss": 0.7432,
+      "step": 4480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7266925892040256e-05,
+      "loss": 0.7153,
+      "step": 4490
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.726006404391583e-05,
+      "loss": 0.7779,
+      "step": 4500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.72532021957914e-05,
+      "loss": 0.7074,
+      "step": 4510
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7246340347666973e-05,
+      "loss": 0.8038,
+      "step": 4520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7239478499542545e-05,
+      "loss": 0.8192,
+      "step": 4530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.7232616651418116e-05,
+      "loss": 0.7207,
+      "step": 4540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.7225754803293687e-05,
+      "loss": 0.8382,
+      "step": 4550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.7218892955169262e-05,
+      "loss": 0.8091,
+      "step": 4560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.721203110704483e-05,
+      "loss": 0.9364,
+      "step": 4570
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7205169258920405e-05,
+      "loss": 0.7665,
+      "step": 4580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7198307410795976e-05,
+      "loss": 0.7893,
+      "step": 4590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7191445562671544e-05,
+      "loss": 0.7045,
+      "step": 4600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.718458371454712e-05,
+      "loss": 0.7499,
+      "step": 4610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.717772186642269e-05,
+      "loss": 0.749,
+      "step": 4620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.717086001829826e-05,
+      "loss": 0.7916,
+      "step": 4630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7163998170173833e-05,
+      "loss": 0.7996,
+      "step": 4640
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7157136322049407e-05,
+      "loss": 0.7692,
+      "step": 4650
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7150274473924975e-05,
+      "loss": 0.6438,
+      "step": 4660
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.714341262580055e-05,
+      "loss": 0.8061,
+      "step": 4670
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.713655077767612e-05,
+      "loss": 0.9341,
+      "step": 4680
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7129688929551692e-05,
+      "loss": 0.6602,
+      "step": 4690
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7122827081427264e-05,
+      "loss": 0.8671,
+      "step": 4700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.711596523330284e-05,
+      "loss": 0.6589,
+      "step": 4710
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7109103385178406e-05,
+      "loss": 0.7783,
+      "step": 4720
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.710224153705398e-05,
+      "loss": 0.826,
+      "step": 4730
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7095379688929552e-05,
+      "loss": 0.6231,
+      "step": 4740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7088517840805124e-05,
+      "loss": 0.7834,
+      "step": 4750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7081655992680695e-05,
+      "loss": 0.8228,
+      "step": 4760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.707479414455627e-05,
+      "loss": 0.7257,
+      "step": 4770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7067932296431838e-05,
+      "loss": 0.8811,
+      "step": 4780
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7061070448307412e-05,
+      "loss": 0.7349,
+      "step": 4790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7054208600182984e-05,
+      "loss": 0.9661,
+      "step": 4800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7047346752058555e-05,
+      "loss": 0.6678,
+      "step": 4810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7040484903934126e-05,
+      "loss": 0.6027,
+      "step": 4820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.70336230558097e-05,
+      "loss": 0.7965,
+      "step": 4830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.702676120768527e-05,
+      "loss": 0.664,
+      "step": 4840
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7019899359560844e-05,
+      "loss": 0.9666,
+      "step": 4850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7013037511436415e-05,
+      "loss": 0.69,
+      "step": 4860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7006175663311986e-05,
+      "loss": 0.6335,
+      "step": 4870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6999313815187557e-05,
+      "loss": 0.8758,
+      "step": 4880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6992451967063132e-05,
+      "loss": 0.7074,
+      "step": 4890
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.69855901189387e-05,
+      "loss": 0.7541,
+      "step": 4900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6978728270814275e-05,
+      "loss": 0.6441,
+      "step": 4910
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6971866422689846e-05,
+      "loss": 0.8988,
+      "step": 4920
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6965004574565417e-05,
+      "loss": 0.8444,
+      "step": 4930
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.695814272644099e-05,
+      "loss": 0.733,
+      "step": 4940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.6951280878316563e-05,
+      "loss": 0.6593,
+      "step": 4950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.694441903019213e-05,
+      "loss": 0.8311,
+      "step": 4960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.6937557182067706e-05,
+      "loss": 0.696,
+      "step": 4970
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.6930695333943277e-05,
+      "loss": 0.7854,
+      "step": 4980
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.692383348581885e-05,
+      "loss": 0.7372,
+      "step": 4990
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.691697163769442e-05,
+      "loss": 1.103,
+      "step": 5000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.691010978956999e-05,
+      "loss": 0.7543,
+      "step": 5010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6903247941445563e-05,
+      "loss": 0.6754,
+      "step": 5020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6896386093321134e-05,
+      "loss": 0.7466,
+      "step": 5030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.688952424519671e-05,
+      "loss": 0.8223,
+      "step": 5040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6882662397072276e-05,
+      "loss": 0.7737,
+      "step": 5050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.687580054894785e-05,
+      "loss": 0.6907,
+      "step": 5060
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.6868938700823422e-05,
+      "loss": 0.7119,
+      "step": 5070
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.6862076852698994e-05,
+      "loss": 0.6785,
+      "step": 5080
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.6855215004574565e-05,
+      "loss": 0.7005,
+      "step": 5090
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.684835315645014e-05,
+      "loss": 0.845,
+      "step": 5100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.6841491308325708e-05,
+      "loss": 0.7128,
+      "step": 5110
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.6834629460201282e-05,
+      "loss": 0.675,
+      "step": 5120
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.6827767612076854e-05,
+      "loss": 0.7055,
+      "step": 5130
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.6820905763952425e-05,
+      "loss": 0.7218,
+      "step": 5140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.6814043915827996e-05,
+      "loss": 0.799,
+      "step": 5150
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.680718206770357e-05,
+      "loss": 0.6957,
+      "step": 5160
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.680032021957914e-05,
+      "loss": 0.891,
+      "step": 5170
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.6793458371454714e-05,
+      "loss": 0.6047,
+      "step": 5180
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.6786596523330285e-05,
+      "loss": 0.722,
+      "step": 5190
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.6779734675205856e-05,
+      "loss": 1.0583,
+      "step": 5200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.6772872827081428e-05,
+      "loss": 0.7346,
+      "step": 5210
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.6766010978957002e-05,
+      "loss": 0.6657,
+      "step": 5220
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.675914913083257e-05,
+      "loss": 0.8309,
+      "step": 5230
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.6752287282708145e-05,
+      "loss": 0.7003,
+      "step": 5240
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.6745425434583716e-05,
+      "loss": 0.8258,
+      "step": 5250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.6738563586459287e-05,
+      "loss": 0.7789,
+      "step": 5260
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.673170173833486e-05,
+      "loss": 0.7073,
+      "step": 5270
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.6724839890210433e-05,
+      "loss": 0.624,
+      "step": 5280
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.6717978042086e-05,
+      "loss": 0.6846,
+      "step": 5290
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.6711116193961576e-05,
+      "loss": 0.7163,
+      "step": 5300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.6704254345837147e-05,
+      "loss": 0.7286,
+      "step": 5310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.669739249771272e-05,
+      "loss": 0.6046,
+      "step": 5320
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.669053064958829e-05,
+      "loss": 0.7511,
+      "step": 5330
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.6683668801463865e-05,
+      "loss": 0.6507,
+      "step": 5340
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.6676806953339433e-05,
+      "loss": 0.7332,
+      "step": 5350
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.6669945105215007e-05,
+      "loss": 0.6554,
+      "step": 5360
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.666308325709058e-05,
+      "loss": 0.6902,
+      "step": 5370
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.665622140896615e-05,
+      "loss": 1.8523,
+      "step": 5380
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.664935956084172e-05,
+      "loss": 0.8124,
+      "step": 5390
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.6642497712717293e-05,
+      "loss": 0.7052,
+      "step": 5400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.6635635864592864e-05,
+      "loss": 0.7936,
+      "step": 5410
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.6628774016468435e-05,
+      "loss": 0.9143,
+      "step": 5420
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.6621912168344006e-05,
+      "loss": 0.6593,
+      "step": 5430
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.6615050320219578e-05,
+      "loss": 0.6911,
+      "step": 5440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.6608188472095152e-05,
+      "loss": 0.7075,
+      "step": 5450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.660132662397072e-05,
+      "loss": 0.6413,
+      "step": 5460
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.6594464775846295e-05,
+      "loss": 0.693,
+      "step": 5470
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.6587602927721866e-05,
+      "loss": 0.8034,
+      "step": 5480
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.6580741079597438e-05,
+      "loss": 1.1163,
+      "step": 5490
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.657387923147301e-05,
+      "loss": 0.7228,
+      "step": 5500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.6567017383348584e-05,
+      "loss": 0.7995,
+      "step": 5510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.656015553522415e-05,
+      "loss": 0.7051,
+      "step": 5520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.6553293687099726e-05,
+      "loss": 0.6834,
+      "step": 5530
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.6546431838975298e-05,
+      "loss": 0.6808,
+      "step": 5540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.653956999085087e-05,
+      "loss": 0.6471,
+      "step": 5550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.653270814272644e-05,
+      "loss": 0.7569,
+      "step": 5560
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.6525846294602015e-05,
+      "loss": 0.7105,
+      "step": 5570
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.6518984446477583e-05,
+      "loss": 1.0915,
+      "step": 5580
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.6512122598353158e-05,
+      "loss": 0.6883,
+      "step": 5590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.650526075022873e-05,
+      "loss": 0.7543,
+      "step": 5600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.64983989021043e-05,
+      "loss": 0.7489,
+      "step": 5610
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.649153705397987e-05,
+      "loss": 0.902,
+      "step": 5620
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.6484675205855446e-05,
+      "loss": 0.7084,
+      "step": 5630
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6477813357731014e-05,
+      "loss": 0.6005,
+      "step": 5640
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.647095150960659e-05,
+      "loss": 0.642,
+      "step": 5650
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.646408966148216e-05,
+      "loss": 0.7322,
+      "step": 5660
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.645722781335773e-05,
+      "loss": 0.6196,
+      "step": 5670
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6450365965233303e-05,
+      "loss": 0.7635,
+      "step": 5680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6443504117108877e-05,
+      "loss": 0.6072,
+      "step": 5690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6436642268984445e-05,
+      "loss": 0.7128,
+      "step": 5700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.642978042086002e-05,
+      "loss": 0.6604,
+      "step": 5710
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.642291857273559e-05,
+      "loss": 0.6864,
+      "step": 5720
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6416056724611163e-05,
+      "loss": 0.7495,
+      "step": 5730
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6409194876486734e-05,
+      "loss": 0.6803,
+      "step": 5740
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.640233302836231e-05,
+      "loss": 0.6045,
+      "step": 5750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6395471180237877e-05,
+      "loss": 0.7325,
+      "step": 5760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.638860933211345e-05,
+      "loss": 0.7188,
+      "step": 5770
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6381747483989023e-05,
+      "loss": 0.6841,
+      "step": 5780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6374885635864594e-05,
+      "loss": 0.6581,
+      "step": 5790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6368023787740165e-05,
+      "loss": 0.635,
+      "step": 5800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.636116193961574e-05,
+      "loss": 0.6708,
+      "step": 5810
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6354300091491308e-05,
+      "loss": 0.5686,
+      "step": 5820
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.634743824336688e-05,
+      "loss": 0.6745,
+      "step": 5830
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6340576395242454e-05,
+      "loss": 0.5867,
+      "step": 5840
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.633371454711802e-05,
+      "loss": 0.6521,
+      "step": 5850
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6326852698993596e-05,
+      "loss": 0.6731,
+      "step": 5860
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6319990850869168e-05,
+      "loss": 0.8174,
+      "step": 5870
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.631312900274474e-05,
+      "loss": 0.6252,
+      "step": 5880
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.630626715462031e-05,
+      "loss": 0.6335,
+      "step": 5890
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6299405306495885e-05,
+      "loss": 1.0509,
+      "step": 5900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6292543458371453e-05,
+      "loss": 0.651,
+      "step": 5910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6285681610247028e-05,
+      "loss": 0.6506,
+      "step": 5920
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.62788197621226e-05,
+      "loss": 0.6147,
+      "step": 5930
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.627195791399817e-05,
+      "loss": 0.5866,
+      "step": 5940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.626509606587374e-05,
+      "loss": 0.7034,
+      "step": 5950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6258234217749316e-05,
+      "loss": 0.6476,
+      "step": 5960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6251372369624884e-05,
+      "loss": 0.88,
+      "step": 5970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.624451052150046e-05,
+      "loss": 0.7263,
+      "step": 5980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.623764867337603e-05,
+      "loss": 0.8857,
+      "step": 5990
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.62307868252516e-05,
+      "loss": 0.6898,
+      "step": 6000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6223924977127173e-05,
+      "loss": 0.7084,
+      "step": 6010
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6217063129002747e-05,
+      "loss": 0.5825,
+      "step": 6020
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6210201280878315e-05,
+      "loss": 0.6783,
+      "step": 6030
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.620333943275389e-05,
+      "loss": 0.6471,
+      "step": 6040
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.619647758462946e-05,
+      "loss": 0.6936,
+      "step": 6050
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6189615736505033e-05,
+      "loss": 0.642,
+      "step": 6060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6182753888380604e-05,
+      "loss": 0.6936,
+      "step": 6070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.617589204025618e-05,
+      "loss": 0.6709,
+      "step": 6080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6169030192131747e-05,
+      "loss": 0.7081,
+      "step": 6090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.616216834400732e-05,
+      "loss": 0.7138,
+      "step": 6100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6155306495882893e-05,
+      "loss": 0.5869,
+      "step": 6110
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6148444647758464e-05,
+      "loss": 0.6317,
+      "step": 6120
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6141582799634035e-05,
+      "loss": 0.5808,
+      "step": 6130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.613472095150961e-05,
+      "loss": 0.639,
+      "step": 6140
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6127859103385178e-05,
+      "loss": 0.7126,
+      "step": 6150
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6120997255260752e-05,
+      "loss": 0.734,
+      "step": 6160
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6114135407136324e-05,
+      "loss": 0.7027,
+      "step": 6170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6107273559011895e-05,
+      "loss": 0.7745,
+      "step": 6180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6100411710887466e-05,
+      "loss": 0.6702,
+      "step": 6190
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.609354986276304e-05,
+      "loss": 0.5887,
+      "step": 6200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.608668801463861e-05,
+      "loss": 0.6791,
+      "step": 6210
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.607982616651418e-05,
+      "loss": 0.6405,
+      "step": 6220
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6072964318389755e-05,
+      "loss": 0.6764,
+      "step": 6230
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6066102470265323e-05,
+      "loss": 0.6759,
+      "step": 6240
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6059240622140898e-05,
+      "loss": 0.7085,
+      "step": 6250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.605237877401647e-05,
+      "loss": 0.5594,
+      "step": 6260
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.604551692589204e-05,
+      "loss": 0.5741,
+      "step": 6270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.603865507776761e-05,
+      "loss": 0.6195,
+      "step": 6280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6031793229643186e-05,
+      "loss": 0.5477,
+      "step": 6290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6024931381518754e-05,
+      "loss": 0.7658,
+      "step": 6300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.601806953339433e-05,
+      "loss": 0.6269,
+      "step": 6310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.60112076852699e-05,
+      "loss": 0.6129,
+      "step": 6320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.600434583714547e-05,
+      "loss": 0.737,
+      "step": 6330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5997483989021043e-05,
+      "loss": 0.6448,
+      "step": 6340
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5990622140896617e-05,
+      "loss": 0.6506,
+      "step": 6350
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5983760292772185e-05,
+      "loss": 0.7529,
+      "step": 6360
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.597689844464776e-05,
+      "loss": 0.741,
+      "step": 6370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.597003659652333e-05,
+      "loss": 0.6941,
+      "step": 6380
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5963174748398903e-05,
+      "loss": 0.6639,
+      "step": 6390
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5956312900274474e-05,
+      "loss": 0.6545,
+      "step": 6400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.594945105215005e-05,
+      "loss": 0.531,
+      "step": 6410
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5942589204025617e-05,
+      "loss": 0.6039,
+      "step": 6420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.593572735590119e-05,
+      "loss": 0.944,
+      "step": 6430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5928865507776763e-05,
+      "loss": 0.5585,
+      "step": 6440
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5922003659652334e-05,
+      "loss": 0.6576,
+      "step": 6450
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5915141811527905e-05,
+      "loss": 0.7944,
+      "step": 6460
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.590827996340348e-05,
+      "loss": 0.7899,
+      "step": 6470
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5901418115279048e-05,
+      "loss": 0.6054,
+      "step": 6480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5894556267154623e-05,
+      "loss": 0.6404,
+      "step": 6490
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5887694419030194e-05,
+      "loss": 0.5572,
+      "step": 6500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5880832570905765e-05,
+      "loss": 0.5664,
+      "step": 6510
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5873970722781336e-05,
+      "loss": 0.559,
+      "step": 6520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.586710887465691e-05,
+      "loss": 0.5821,
+      "step": 6530
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.586024702653248e-05,
+      "loss": 0.6293,
+      "step": 6540
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5853385178408054e-05,
+      "loss": 0.5967,
+      "step": 6550
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5846523330283625e-05,
+      "loss": 1.1112,
+      "step": 6560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5839661482159196e-05,
+      "loss": 0.7736,
+      "step": 6570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5832799634034768e-05,
+      "loss": 0.6468,
+      "step": 6580
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5825937785910342e-05,
+      "loss": 0.7239,
+      "step": 6590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.581907593778591e-05,
+      "loss": 0.6316,
+      "step": 6600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5812214089661485e-05,
+      "loss": 0.6805,
+      "step": 6610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5805352241537056e-05,
+      "loss": 0.6643,
+      "step": 6620
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5798490393412624e-05,
+      "loss": 0.5585,
+      "step": 6630
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.57916285452882e-05,
+      "loss": 0.7611,
+      "step": 6640
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.578476669716377e-05,
+      "loss": 0.6892,
+      "step": 6650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.577790484903934e-05,
+      "loss": 0.6668,
+      "step": 6660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5771043000914913e-05,
+      "loss": 0.6955,
+      "step": 6670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5764181152790484e-05,
+      "loss": 0.6631,
+      "step": 6680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5757319304666055e-05,
+      "loss": 0.5698,
+      "step": 6690
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.575045745654163e-05,
+      "loss": 0.6468,
+      "step": 6700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5743595608417198e-05,
+      "loss": 0.7247,
+      "step": 6710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5736733760292773e-05,
+      "loss": 0.5426,
+      "step": 6720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5729871912168344e-05,
+      "loss": 0.7148,
+      "step": 6730
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5723010064043915e-05,
+      "loss": 0.8091,
+      "step": 6740
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5716148215919487e-05,
+      "loss": 0.7559,
+      "step": 6750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.570928636779506e-05,
+      "loss": 0.5512,
+      "step": 6760
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.570242451967063e-05,
+      "loss": 0.691,
+      "step": 6770
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5695562671546204e-05,
+      "loss": 0.632,
+      "step": 6780
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5688700823421775e-05,
+      "loss": 0.6128,
+      "step": 6790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5681838975297347e-05,
+      "loss": 0.6156,
+      "step": 6800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5674977127172918e-05,
+      "loss": 0.5803,
+      "step": 6810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5668115279048493e-05,
+      "loss": 0.594,
+      "step": 6820
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.566125343092406e-05,
+      "loss": 0.6516,
+      "step": 6830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5654391582799635e-05,
+      "loss": 0.5892,
+      "step": 6840
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5647529734675207e-05,
+      "loss": 0.6403,
+      "step": 6850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5640667886550778e-05,
+      "loss": 0.61,
+      "step": 6860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.563380603842635e-05,
+      "loss": 0.6049,
+      "step": 6870
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5626944190301924e-05,
+      "loss": 0.7423,
+      "step": 6880
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5620082342177492e-05,
+      "loss": 0.6308,
+      "step": 6890
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5613220494053066e-05,
+      "loss": 0.5252,
+      "step": 6900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5606358645928638e-05,
+      "loss": 0.5669,
+      "step": 6910
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.559949679780421e-05,
+      "loss": 0.5994,
+      "step": 6920
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.559263494967978e-05,
+      "loss": 0.6102,
+      "step": 6930
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5585773101555355e-05,
+      "loss": 0.5946,
+      "step": 6940
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5578911253430923e-05,
+      "loss": 1.697,
+      "step": 6950
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5572049405306498e-05,
+      "loss": 0.6632,
+      "step": 6960
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.556518755718207e-05,
+      "loss": 0.5423,
+      "step": 6970
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.555832570905764e-05,
+      "loss": 0.5854,
+      "step": 6980
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.555146386093321e-05,
+      "loss": 0.609,
+      "step": 6990
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5544602012808786e-05,
+      "loss": 0.706,
+      "step": 7000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5537740164684354e-05,
+      "loss": 0.6794,
+      "step": 7010
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.553087831655993e-05,
+      "loss": 0.6046,
+      "step": 7020
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.55240164684355e-05,
+      "loss": 0.7173,
+      "step": 7030
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5517154620311068e-05,
+      "loss": 0.6721,
+      "step": 7040
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5510292772186643e-05,
+      "loss": 0.6735,
+      "step": 7050
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5503430924062214e-05,
+      "loss": 0.5755,
+      "step": 7060
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5496569075937785e-05,
+      "loss": 0.7356,
+      "step": 7070
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5489707227813357e-05,
+      "loss": 0.6502,
+      "step": 7080
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.548284537968893e-05,
+      "loss": 0.5867,
+      "step": 7090
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.54759835315645e-05,
+      "loss": 0.6005,
+      "step": 7100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5469121683440074e-05,
+      "loss": 0.6572,
+      "step": 7110
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5462259835315645e-05,
+      "loss": 0.6509,
+      "step": 7120
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5455397987191217e-05,
+      "loss": 0.6073,
+      "step": 7130
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5448536139066788e-05,
+      "loss": 0.5582,
+      "step": 7140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5441674290942363e-05,
+      "loss": 0.5881,
+      "step": 7150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.543481244281793e-05,
+      "loss": 0.627,
+      "step": 7160
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5427950594693505e-05,
+      "loss": 0.6544,
+      "step": 7170
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5421088746569077e-05,
+      "loss": 0.5875,
+      "step": 7180
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5414226898444648e-05,
+      "loss": 0.5573,
+      "step": 7190
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.540736505032022e-05,
+      "loss": 0.6325,
+      "step": 7200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5400503202195794e-05,
+      "loss": 0.6134,
+      "step": 7210
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5393641354071362e-05,
+      "loss": 0.629,
+      "step": 7220
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5386779505946937e-05,
+      "loss": 0.7547,
+      "step": 7230
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5379917657822508e-05,
+      "loss": 0.6518,
+      "step": 7240
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.537305580969808e-05,
+      "loss": 0.5721,
+      "step": 7250
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.536619396157365e-05,
+      "loss": 0.6864,
+      "step": 7260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5359332113449225e-05,
+      "loss": 0.6169,
+      "step": 7270
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5352470265324793e-05,
+      "loss": 0.5021,
+      "step": 7280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5345608417200368e-05,
+      "loss": 0.6499,
+      "step": 7290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.533874656907594e-05,
+      "loss": 0.5664,
+      "step": 7300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.533188472095151e-05,
+      "loss": 0.5213,
+      "step": 7310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.532502287282708e-05,
+      "loss": 0.6183,
+      "step": 7320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5318161024702656e-05,
+      "loss": 0.5832,
+      "step": 7330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5311299176578224e-05,
+      "loss": 0.681,
+      "step": 7340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.53044373284538e-05,
+      "loss": 0.7722,
+      "step": 7350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.529757548032937e-05,
+      "loss": 0.6952,
+      "step": 7360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.529071363220494e-05,
+      "loss": 0.6411,
+      "step": 7370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5283851784080513e-05,
+      "loss": 0.7237,
+      "step": 7380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5276989935956088e-05,
+      "loss": 0.6289,
+      "step": 7390
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5270128087831656e-05,
+      "loss": 0.552,
+      "step": 7400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.526326623970723e-05,
+      "loss": 0.6245,
+      "step": 7410
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.52564043915828e-05,
+      "loss": 0.5178,
+      "step": 7420
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.524954254345837e-05,
+      "loss": 0.7838,
+      "step": 7430
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5242680695333944e-05,
+      "loss": 0.6345,
+      "step": 7440
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5235818847209515e-05,
+      "loss": 0.7012,
+      "step": 7450
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5228956999085087e-05,
+      "loss": 0.6414,
+      "step": 7460
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5222095150960658e-05,
+      "loss": 0.6281,
+      "step": 7470
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5215233302836233e-05,
+      "loss": 0.5748,
+      "step": 7480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.52083714547118e-05,
+      "loss": 0.545,
+      "step": 7490
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.5201509606587375e-05,
+      "loss": 0.6683,
+      "step": 7500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.5194647758462947e-05,
+      "loss": 0.6055,
+      "step": 7510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.5187785910338518e-05,
+      "loss": 0.7,
+      "step": 7520
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.518092406221409e-05,
+      "loss": 0.5909,
+      "step": 7530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.5174062214089664e-05,
+      "loss": 0.5792,
+      "step": 7540
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.5167200365965232e-05,
+      "loss": 0.5504,
+      "step": 7550
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.5160338517840807e-05,
+      "loss": 0.5784,
+      "step": 7560
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.5153476669716378e-05,
+      "loss": 0.5384,
+      "step": 7570
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.514661482159195e-05,
+      "loss": 0.7527,
+      "step": 7580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.513975297346752e-05,
+      "loss": 0.6142,
+      "step": 7590
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.5132891125343095e-05,
+      "loss": 0.5995,
+      "step": 7600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.5126029277218663e-05,
+      "loss": 0.6336,
+      "step": 7610
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.5119167429094238e-05,
+      "loss": 0.9482,
+      "step": 7620
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.511230558096981e-05,
+      "loss": 0.6149,
+      "step": 7630
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.510544373284538e-05,
+      "loss": 0.671,
+      "step": 7640
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.5098581884720952e-05,
+      "loss": 0.5526,
+      "step": 7650
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.5091720036596526e-05,
+      "loss": 0.5387,
+      "step": 7660
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.5084858188472094e-05,
+      "loss": 0.5752,
+      "step": 7670
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.507799634034767e-05,
+      "loss": 0.6122,
+      "step": 7680
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.507113449222324e-05,
+      "loss": 0.5376,
+      "step": 7690
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.506427264409881e-05,
+      "loss": 0.5703,
+      "step": 7700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5057410795974383e-05,
+      "loss": 0.6286,
+      "step": 7710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5050548947849958e-05,
+      "loss": 0.5156,
+      "step": 7720
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5043687099725526e-05,
+      "loss": 0.7298,
+      "step": 7730
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.50368252516011e-05,
+      "loss": 0.5684,
+      "step": 7740
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.502996340347667e-05,
+      "loss": 0.8077,
+      "step": 7750
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5023101555352243e-05,
+      "loss": 0.721,
+      "step": 7760
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5016239707227814e-05,
+      "loss": 0.7268,
+      "step": 7770
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.500937785910339e-05,
+      "loss": 0.5803,
+      "step": 7780
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5002516010978957e-05,
+      "loss": 0.5072,
+      "step": 7790
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.499565416285453e-05,
+      "loss": 0.6524,
+      "step": 7800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4988792314730103e-05,
+      "loss": 0.6905,
+      "step": 7810
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4981930466605674e-05,
+      "loss": 0.592,
+      "step": 7820
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4975068618481245e-05,
+      "loss": 0.5522,
+      "step": 7830
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4968206770356817e-05,
+      "loss": 0.5837,
+      "step": 7840
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4961344922232388e-05,
+      "loss": 0.5655,
+      "step": 7850
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.495448307410796e-05,
+      "loss": 0.4803,
+      "step": 7860
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4947621225983534e-05,
+      "loss": 0.6077,
+      "step": 7870
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4940759377859102e-05,
+      "loss": 0.6003,
+      "step": 7880
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4933897529734677e-05,
+      "loss": 0.5603,
+      "step": 7890
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4927035681610248e-05,
+      "loss": 0.4799,
+      "step": 7900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.492017383348582e-05,
+      "loss": 0.505,
+      "step": 7910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.491331198536139e-05,
+      "loss": 0.5812,
+      "step": 7920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4906450137236962e-05,
+      "loss": 0.701,
+      "step": 7930
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4899588289112533e-05,
+      "loss": 0.5734,
+      "step": 7940
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4892726440988108e-05,
+      "loss": 0.6763,
+      "step": 7950
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4885864592863676e-05,
+      "loss": 0.5441,
+      "step": 7960
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.487900274473925e-05,
+      "loss": 0.5055,
+      "step": 7970
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4872140896614822e-05,
+      "loss": 0.5581,
+      "step": 7980
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4865279048490393e-05,
+      "loss": 0.707,
+      "step": 7990
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4858417200365964e-05,
+      "loss": 0.6531,
+      "step": 8000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.485155535224154e-05,
+      "loss": 0.5933,
+      "step": 8010
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4844693504117107e-05,
+      "loss": 0.4786,
+      "step": 8020
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4837831655992682e-05,
+      "loss": 0.5347,
+      "step": 8030
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4830969807868253e-05,
+      "loss": 0.545,
+      "step": 8040
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4824107959743824e-05,
+      "loss": 0.5753,
+      "step": 8050
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4817246111619396e-05,
+      "loss": 0.5671,
+      "step": 8060
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.481038426349497e-05,
+      "loss": 0.5829,
+      "step": 8070
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4803522415370538e-05,
+      "loss": 0.5425,
+      "step": 8080
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4796660567246113e-05,
+      "loss": 0.5468,
+      "step": 8090
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4789798719121684e-05,
+      "loss": 0.6799,
+      "step": 8100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4782936870997256e-05,
+      "loss": 0.5507,
+      "step": 8110
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4776075022872827e-05,
+      "loss": 0.6167,
+      "step": 8120
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.47692131747484e-05,
+      "loss": 0.4715,
+      "step": 8130
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.476235132662397e-05,
+      "loss": 0.5276,
+      "step": 8140
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4755489478499544e-05,
+      "loss": 0.5288,
+      "step": 8150
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4748627630375115e-05,
+      "loss": 0.5498,
+      "step": 8160
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4741765782250687e-05,
+      "loss": 0.4235,
+      "step": 8170
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4734903934126258e-05,
+      "loss": 0.5232,
+      "step": 8180
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4728042086001833e-05,
+      "loss": 0.5758,
+      "step": 8190
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.47211802378774e-05,
+      "loss": 0.4721,
+      "step": 8200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4714318389752975e-05,
+      "loss": 0.6065,
+      "step": 8210
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4707456541628547e-05,
+      "loss": 0.4996,
+      "step": 8220
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4700594693504118e-05,
+      "loss": 0.6101,
+      "step": 8230
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.469373284537969e-05,
+      "loss": 0.5236,
+      "step": 8240
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.468687099725526e-05,
+      "loss": 0.5328,
+      "step": 8250
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4680009149130832e-05,
+      "loss": 0.6184,
+      "step": 8260
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4673147301006403e-05,
+      "loss": 0.5228,
+      "step": 8270
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4666285452881978e-05,
+      "loss": 0.5192,
+      "step": 8280
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4659423604757546e-05,
+      "loss": 0.5657,
+      "step": 8290
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.465256175663312e-05,
+      "loss": 0.675,
+      "step": 8300
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4645699908508692e-05,
+      "loss": 0.5061,
+      "step": 8310
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4638838060384263e-05,
+      "loss": 0.4824,
+      "step": 8320
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4631976212259834e-05,
+      "loss": 0.5105,
+      "step": 8330
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.462511436413541e-05,
+      "loss": 0.8377,
+      "step": 8340
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4618252516010977e-05,
+      "loss": 0.5614,
+      "step": 8350
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4611390667886552e-05,
+      "loss": 0.5904,
+      "step": 8360
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4604528819762123e-05,
+      "loss": 0.4763,
+      "step": 8370
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4597666971637694e-05,
+      "loss": 0.5998,
+      "step": 8380
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4590805123513266e-05,
+      "loss": 0.6412,
+      "step": 8390
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.458394327538884e-05,
+      "loss": 0.5583,
+      "step": 8400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.457708142726441e-05,
+      "loss": 0.6376,
+      "step": 8410
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4570219579139983e-05,
+      "loss": 0.6403,
+      "step": 8420
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4563357731015554e-05,
+      "loss": 0.5479,
+      "step": 8430
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4556495882891126e-05,
+      "loss": 0.5946,
+      "step": 8440
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4549634034766697e-05,
+      "loss": 0.6569,
+      "step": 8450
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.454277218664227e-05,
+      "loss": 0.6844,
+      "step": 8460
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.453591033851784e-05,
+      "loss": 0.5039,
+      "step": 8470
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4529048490393414e-05,
+      "loss": 0.673,
+      "step": 8480
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4522186642268986e-05,
+      "loss": 0.45,
+      "step": 8490
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4515324794144557e-05,
+      "loss": 0.5678,
+      "step": 8500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4508462946020128e-05,
+      "loss": 0.4741,
+      "step": 8510
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.4501601097895703e-05,
+      "loss": 0.4443,
+      "step": 8520
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.449473924977127e-05,
+      "loss": 0.541,
+      "step": 8530
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.4487877401646845e-05,
+      "loss": 0.7118,
+      "step": 8540
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.4481015553522417e-05,
+      "loss": 0.7102,
+      "step": 8550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4474153705397988e-05,
+      "loss": 0.5733,
+      "step": 8560
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.446729185727356e-05,
+      "loss": 0.5627,
+      "step": 8570
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4460430009149134e-05,
+      "loss": 0.5939,
+      "step": 8580
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4453568161024702e-05,
+      "loss": 0.5868,
+      "step": 8590
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4446706312900277e-05,
+      "loss": 0.4732,
+      "step": 8600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.4439844464775848e-05,
+      "loss": 0.4843,
+      "step": 8610
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.443298261665142e-05,
+      "loss": 0.5152,
+      "step": 8620
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.442612076852699e-05,
+      "loss": 0.6337,
+      "step": 8630
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.4419258920402565e-05,
+      "loss": 0.5841,
+      "step": 8640
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4412397072278133e-05,
+      "loss": 0.5544,
+      "step": 8650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4405535224153705e-05,
+      "loss": 0.7199,
+      "step": 8660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.439867337602928e-05,
+      "loss": 0.5786,
+      "step": 8670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4391811527904847e-05,
+      "loss": 0.6505,
+      "step": 8680
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4384949679780422e-05,
+      "loss": 0.5758,
+      "step": 8690
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4378087831655993e-05,
+      "loss": 0.6264,
+      "step": 8700
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.437191216834401e-05,
+      "loss": 0.5698,
+      "step": 8710
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.436505032021958e-05,
+      "loss": 0.566,
+      "step": 8720
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4358188472095152e-05,
+      "loss": 0.4935,
+      "step": 8730
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4351326623970723e-05,
+      "loss": 0.6003,
+      "step": 8740
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4344464775846298e-05,
+      "loss": 0.6164,
+      "step": 8750
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4337602927721866e-05,
+      "loss": 0.5598,
+      "step": 8760
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4330741079597437e-05,
+      "loss": 0.56,
+      "step": 8770
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4323879231473012e-05,
+      "loss": 0.5638,
+      "step": 8780
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.431701738334858e-05,
+      "loss": 0.6267,
+      "step": 8790
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4310155535224155e-05,
+      "loss": 0.5238,
+      "step": 8800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4303293687099726e-05,
+      "loss": 0.5121,
+      "step": 8810
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4296431838975297e-05,
+      "loss": 0.5616,
+      "step": 8820
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.428956999085087e-05,
+      "loss": 0.5643,
+      "step": 8830
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.4282708142726443e-05,
+      "loss": 0.5484,
+      "step": 8840
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.6393277645111084,
+      "eval_runtime": 302.3673,
+      "eval_samples_per_second": 6.499,
+      "eval_steps_per_second": 1.627,
+      "eval_wer": 0.35573901792865503,
+      "step": 8844
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.427584629460201e-05,
+      "loss": 0.7161,
+      "step": 8850
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.4268984446477586e-05,
+      "loss": 0.5089,
+      "step": 8860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.4262122598353157e-05,
+      "loss": 0.4762,
+      "step": 8870
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.425526075022873e-05,
+      "loss": 0.5025,
+      "step": 8880
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.42483989021043e-05,
+      "loss": 0.6689,
+      "step": 8890
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.4241537053979874e-05,
+      "loss": 0.6802,
+      "step": 8900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.4234675205855442e-05,
+      "loss": 0.5225,
+      "step": 8910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.4227813357731017e-05,
+      "loss": 0.562,
+      "step": 8920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.4220951509606588e-05,
+      "loss": 0.4801,
+      "step": 8930
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.421408966148216e-05,
+      "loss": 0.522,
+      "step": 8940
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.420722781335773e-05,
+      "loss": 0.6107,
+      "step": 8950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4200365965233306e-05,
+      "loss": 0.5736,
+      "step": 8960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4193504117108873e-05,
+      "loss": 0.5538,
+      "step": 8970
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4186642268984448e-05,
+      "loss": 0.5092,
+      "step": 8980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.417978042086002e-05,
+      "loss": 0.5721,
+      "step": 8990
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.417291857273559e-05,
+      "loss": 0.5322,
+      "step": 9000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4166056724611162e-05,
+      "loss": 0.4785,
+      "step": 9010
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4159194876486737e-05,
+      "loss": 0.5281,
+      "step": 9020
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4152333028362305e-05,
+      "loss": 0.4317,
+      "step": 9030
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.414547118023788e-05,
+      "loss": 0.4991,
+      "step": 9040
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.413860933211345e-05,
+      "loss": 0.4799,
+      "step": 9050
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4131747483989022e-05,
+      "loss": 0.4652,
+      "step": 9060
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4124885635864593e-05,
+      "loss": 0.5136,
+      "step": 9070
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4118023787740168e-05,
+      "loss": 0.6252,
+      "step": 9080
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4111161939615736e-05,
+      "loss": 0.56,
+      "step": 9090
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.410430009149131e-05,
+      "loss": 0.49,
+      "step": 9100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4097438243366882e-05,
+      "loss": 0.5256,
+      "step": 9110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4090576395242453e-05,
+      "loss": 0.4827,
+      "step": 9120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4083714547118025e-05,
+      "loss": 0.497,
+      "step": 9130
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.40768526989936e-05,
+      "loss": 0.5063,
+      "step": 9140
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4069990850869167e-05,
+      "loss": 0.5313,
+      "step": 9150
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4063129002744742e-05,
+      "loss": 0.5933,
+      "step": 9160
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4056267154620313e-05,
+      "loss": 0.5236,
+      "step": 9170
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.404940530649588e-05,
+      "loss": 0.4886,
+      "step": 9180
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4042543458371456e-05,
+      "loss": 0.4789,
+      "step": 9190
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4035681610247027e-05,
+      "loss": 0.4856,
+      "step": 9200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.40288197621226e-05,
+      "loss": 0.507,
+      "step": 9210
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.402195791399817e-05,
+      "loss": 0.4735,
+      "step": 9220
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4015096065873744e-05,
+      "loss": 0.5281,
+      "step": 9230
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4008234217749312e-05,
+      "loss": 0.4241,
+      "step": 9240
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4001372369624887e-05,
+      "loss": 0.5284,
+      "step": 9250
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.399451052150046e-05,
+      "loss": 0.4866,
+      "step": 9260
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.398764867337603e-05,
+      "loss": 0.5027,
+      "step": 9270
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.39807868252516e-05,
+      "loss": 0.4774,
+      "step": 9280
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3973924977127176e-05,
+      "loss": 0.435,
+      "step": 9290
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3967063129002744e-05,
+      "loss": 0.5405,
+      "step": 9300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3960201280878318e-05,
+      "loss": 0.5142,
+      "step": 9310
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.395333943275389e-05,
+      "loss": 0.459,
+      "step": 9320
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.394647758462946e-05,
+      "loss": 0.4808,
+      "step": 9330
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3939615736505032e-05,
+      "loss": 0.6241,
+      "step": 9340
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3932753888380607e-05,
+      "loss": 0.4613,
+      "step": 9350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3925892040256175e-05,
+      "loss": 0.4705,
+      "step": 9360
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.391903019213175e-05,
+      "loss": 0.5457,
+      "step": 9370
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.391216834400732e-05,
+      "loss": 0.5516,
+      "step": 9380
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3905306495882892e-05,
+      "loss": 0.5496,
+      "step": 9390
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3898444647758463e-05,
+      "loss": 0.4701,
+      "step": 9400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3891582799634035e-05,
+      "loss": 0.4653,
+      "step": 9410
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3884720951509606e-05,
+      "loss": 0.4871,
+      "step": 9420
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.387785910338518e-05,
+      "loss": 0.4809,
+      "step": 9430
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.387099725526075e-05,
+      "loss": 0.5409,
+      "step": 9440
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3864135407136323e-05,
+      "loss": 0.5183,
+      "step": 9450
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3857273559011895e-05,
+      "loss": 0.409,
+      "step": 9460
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3850411710887466e-05,
+      "loss": 0.5297,
+      "step": 9470
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3843549862763037e-05,
+      "loss": 0.4661,
+      "step": 9480
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3836688014638612e-05,
+      "loss": 0.4452,
+      "step": 9490
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.382982616651418e-05,
+      "loss": 0.4437,
+      "step": 9500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3822964318389755e-05,
+      "loss": 0.5091,
+      "step": 9510
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3816102470265326e-05,
+      "loss": 0.4111,
+      "step": 9520
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3809240622140897e-05,
+      "loss": 0.4552,
+      "step": 9530
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.380237877401647e-05,
+      "loss": 0.4065,
+      "step": 9540
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3795516925892043e-05,
+      "loss": 0.5347,
+      "step": 9550
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.378865507776761e-05,
+      "loss": 0.5393,
+      "step": 9560
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3781793229643186e-05,
+      "loss": 0.4526,
+      "step": 9570
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3774931381518757e-05,
+      "loss": 0.9216,
+      "step": 9580
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3768069533394325e-05,
+      "loss": 0.5275,
+      "step": 9590
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.37612076852699e-05,
+      "loss": 0.5369,
+      "step": 9600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.375434583714547e-05,
+      "loss": 0.4072,
+      "step": 9610
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3747483989021042e-05,
+      "loss": 0.5441,
+      "step": 9620
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3740622140896614e-05,
+      "loss": 0.5381,
+      "step": 9630
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.373376029277219e-05,
+      "loss": 0.5863,
+      "step": 9640
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3726898444647756e-05,
+      "loss": 0.572,
+      "step": 9650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.372003659652333e-05,
+      "loss": 0.4387,
+      "step": 9660
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3713174748398902e-05,
+      "loss": 0.4603,
+      "step": 9670
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3706312900274474e-05,
+      "loss": 0.4423,
+      "step": 9680
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3699451052150045e-05,
+      "loss": 0.4845,
+      "step": 9690
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.369258920402562e-05,
+      "loss": 0.5257,
+      "step": 9700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.3685727355901187e-05,
+      "loss": 0.4691,
+      "step": 9710
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.3678865507776762e-05,
+      "loss": 0.4315,
+      "step": 9720
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.3672003659652333e-05,
+      "loss": 0.5982,
+      "step": 9730
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.3665141811527905e-05,
+      "loss": 0.5364,
+      "step": 9740
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.3658279963403476e-05,
+      "loss": 0.4846,
+      "step": 9750
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.365141811527905e-05,
+      "loss": 0.497,
+      "step": 9760
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.364455626715462e-05,
+      "loss": 0.4869,
+      "step": 9770
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3637694419030193e-05,
+      "loss": 0.4508,
+      "step": 9780
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3630832570905765e-05,
+      "loss": 0.4955,
+      "step": 9790
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3623970722781336e-05,
+      "loss": 0.5239,
+      "step": 9800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3617108874656907e-05,
+      "loss": 0.558,
+      "step": 9810
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3610247026532482e-05,
+      "loss": 0.4866,
+      "step": 9820
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.360338517840805e-05,
+      "loss": 0.5996,
+      "step": 9830
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.3596523330283625e-05,
+      "loss": 0.5101,
+      "step": 9840
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.3589661482159196e-05,
+      "loss": 0.4727,
+      "step": 9850
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.3582799634034767e-05,
+      "loss": 0.5666,
+      "step": 9860
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.357593778591034e-05,
+      "loss": 0.4839,
+      "step": 9870
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.3569075937785913e-05,
+      "loss": 0.4668,
+      "step": 9880
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.356221408966148e-05,
+      "loss": 0.5374,
+      "step": 9890
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.3555352241537056e-05,
+      "loss": 1.12,
+      "step": 9900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.3548490393412627e-05,
+      "loss": 0.4885,
+      "step": 9910
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.35416285452882e-05,
+      "loss": 0.3907,
+      "step": 9920
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.353476669716377e-05,
+      "loss": 0.5966,
+      "step": 9930
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.3527904849039344e-05,
+      "loss": 0.5696,
+      "step": 9940
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.3521043000914912e-05,
+      "loss": 0.5193,
+      "step": 9950
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.3514181152790487e-05,
+      "loss": 0.6053,
+      "step": 9960
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.350731930466606e-05,
+      "loss": 0.5621,
+      "step": 9970
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.350045745654163e-05,
+      "loss": 0.5008,
+      "step": 9980
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.34935956084172e-05,
+      "loss": 0.4819,
+      "step": 9990
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.3486733760292772e-05,
+      "loss": 0.4979,
+      "step": 10000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.3479871912168344e-05,
+      "loss": 0.4239,
+      "step": 10010
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3473010064043915e-05,
+      "loss": 0.6851,
+      "step": 10020
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.346614821591949e-05,
+      "loss": 0.5529,
+      "step": 10030
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3459286367795058e-05,
+      "loss": 0.4244,
+      "step": 10040
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3452424519670632e-05,
+      "loss": 0.5006,
+      "step": 10050
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3445562671546204e-05,
+      "loss": 0.4228,
+      "step": 10060
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3438700823421775e-05,
+      "loss": 0.4228,
+      "step": 10070
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3431838975297346e-05,
+      "loss": 0.4959,
+      "step": 10080
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.342497712717292e-05,
+      "loss": 0.4369,
+      "step": 10090
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.341811527904849e-05,
+      "loss": 0.4844,
+      "step": 10100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3411253430924063e-05,
+      "loss": 0.3923,
+      "step": 10110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3404391582799635e-05,
+      "loss": 0.5248,
+      "step": 10120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3397529734675206e-05,
+      "loss": 0.4367,
+      "step": 10130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3390667886550777e-05,
+      "loss": 0.5547,
+      "step": 10140
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3383806038426352e-05,
+      "loss": 0.4703,
+      "step": 10150
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.337694419030192e-05,
+      "loss": 0.4124,
+      "step": 10160
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3370082342177495e-05,
+      "loss": 0.5036,
+      "step": 10170
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3363220494053066e-05,
+      "loss": 0.4582,
+      "step": 10180
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3356358645928637e-05,
+      "loss": 0.4934,
+      "step": 10190
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.334949679780421e-05,
+      "loss": 0.4615,
+      "step": 10200
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3342634949679783e-05,
+      "loss": 0.4645,
+      "step": 10210
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.333577310155535e-05,
+      "loss": 0.4627,
+      "step": 10220
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3328911253430926e-05,
+      "loss": 0.4749,
+      "step": 10230
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3322049405306497e-05,
+      "loss": 0.4725,
+      "step": 10240
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.331518755718207e-05,
+      "loss": 0.612,
+      "step": 10250
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.330832570905764e-05,
+      "loss": 0.3749,
+      "step": 10260
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3301463860933215e-05,
+      "loss": 0.4033,
+      "step": 10270
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3294602012808782e-05,
+      "loss": 0.4448,
+      "step": 10280
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3287740164684357e-05,
+      "loss": 0.474,
+      "step": 10290
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.328087831655993e-05,
+      "loss": 0.5202,
+      "step": 10300
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.32740164684355e-05,
+      "loss": 0.422,
+      "step": 10310
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.326715462031107e-05,
+      "loss": 0.4925,
+      "step": 10320
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3260292772186646e-05,
+      "loss": 0.4705,
+      "step": 10330
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3253430924062214e-05,
+      "loss": 0.417,
+      "step": 10340
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.324656907593779e-05,
+      "loss": 0.4738,
+      "step": 10350
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.323970722781336e-05,
+      "loss": 0.4766,
+      "step": 10360
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.323284537968893e-05,
+      "loss": 1.1178,
+      "step": 10370
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3225983531564502e-05,
+      "loss": 0.51,
+      "step": 10380
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3219121683440074e-05,
+      "loss": 0.5296,
+      "step": 10390
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3212259835315645e-05,
+      "loss": 0.7555,
+      "step": 10400
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3205397987191216e-05,
+      "loss": 0.4784,
+      "step": 10410
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.319853613906679e-05,
+      "loss": 0.5036,
+      "step": 10420
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.319167429094236e-05,
+      "loss": 0.3982,
+      "step": 10430
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3184812442817934e-05,
+      "loss": 0.3837,
+      "step": 10440
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3177950594693505e-05,
+      "loss": 0.4588,
+      "step": 10450
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3171088746569076e-05,
+      "loss": 0.5688,
+      "step": 10460
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3164226898444647e-05,
+      "loss": 0.5478,
+      "step": 10470
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3157365050320222e-05,
+      "loss": 0.4479,
+      "step": 10480
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.315050320219579e-05,
+      "loss": 0.4933,
+      "step": 10490
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3143641354071365e-05,
+      "loss": 0.4521,
+      "step": 10500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3136779505946936e-05,
+      "loss": 0.5471,
+      "step": 10510
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3129917657822507e-05,
+      "loss": 0.5376,
+      "step": 10520
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.312305580969808e-05,
+      "loss": 0.5062,
+      "step": 10530
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3116193961573653e-05,
+      "loss": 0.5228,
+      "step": 10540
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.310933211344922e-05,
+      "loss": 0.4708,
+      "step": 10550
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.3102470265324796e-05,
+      "loss": 0.4746,
+      "step": 10560
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.3095608417200367e-05,
+      "loss": 0.36,
+      "step": 10570
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.308874656907594e-05,
+      "loss": 0.5012,
+      "step": 10580
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.308188472095151e-05,
+      "loss": 0.4187,
+      "step": 10590
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3075022872827085e-05,
+      "loss": 0.4409,
+      "step": 10600
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3068161024702652e-05,
+      "loss": 0.4642,
+      "step": 10610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3061299176578227e-05,
+      "loss": 0.4539,
+      "step": 10620
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.30544373284538e-05,
+      "loss": 0.4168,
+      "step": 10630
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.304757548032937e-05,
+      "loss": 0.4301,
+      "step": 10640
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.304071363220494e-05,
+      "loss": 0.5487,
+      "step": 10650
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.3033851784080512e-05,
+      "loss": 0.5605,
+      "step": 10660
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.3026989935956084e-05,
+      "loss": 0.4072,
+      "step": 10670
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.302012808783166e-05,
+      "loss": 0.4377,
+      "step": 10680
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.3013266239707226e-05,
+      "loss": 0.5189,
+      "step": 10690
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.30064043915828e-05,
+      "loss": 0.4658,
+      "step": 10700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.2999542543458372e-05,
+      "loss": 0.4703,
+      "step": 10710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.2992680695333944e-05,
+      "loss": 0.5065,
+      "step": 10720
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.2985818847209515e-05,
+      "loss": 0.4516,
+      "step": 10730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.297895699908509e-05,
+      "loss": 0.4614,
+      "step": 10740
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.2972095150960658e-05,
+      "loss": 0.4441,
+      "step": 10750
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.2965233302836232e-05,
+      "loss": 0.5839,
+      "step": 10760
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.2958371454711804e-05,
+      "loss": 0.4996,
+      "step": 10770
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.2951509606587375e-05,
+      "loss": 0.4515,
+      "step": 10780
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.2944647758462946e-05,
+      "loss": 0.4978,
+      "step": 10790
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.2937785910338517e-05,
+      "loss": 0.478,
+      "step": 10800
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.293092406221409e-05,
+      "loss": 0.4903,
+      "step": 10810
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.292406221408966e-05,
+      "loss": 0.3926,
+      "step": 10820
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.2917200365965235e-05,
+      "loss": 0.4751,
+      "step": 10830
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.2910338517840803e-05,
+      "loss": 0.3994,
+      "step": 10840
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.2903476669716377e-05,
+      "loss": 0.4401,
+      "step": 10850
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.289661482159195e-05,
+      "loss": 0.4519,
+      "step": 10860
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.288975297346752e-05,
+      "loss": 0.5174,
+      "step": 10870
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.288289112534309e-05,
+      "loss": 0.444,
+      "step": 10880
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.2876029277218666e-05,
+      "loss": 0.4851,
+      "step": 10890
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.2869167429094234e-05,
+      "loss": 0.396,
+      "step": 10900
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.286230558096981e-05,
+      "loss": 0.5417,
+      "step": 10910
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.285544373284538e-05,
+      "loss": 0.5179,
+      "step": 10920
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.284858188472095e-05,
+      "loss": 0.5348,
+      "step": 10930
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2841720036596523e-05,
+      "loss": 0.4357,
+      "step": 10940
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2834858188472097e-05,
+      "loss": 0.4163,
+      "step": 10950
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2827996340347665e-05,
+      "loss": 0.4978,
+      "step": 10960
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.282113449222324e-05,
+      "loss": 0.6329,
+      "step": 10970
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.281427264409881e-05,
+      "loss": 0.6548,
+      "step": 10980
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2807410795974382e-05,
+      "loss": 0.4133,
+      "step": 10990
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2800548947849954e-05,
+      "loss": 0.549,
+      "step": 11000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.279368709972553e-05,
+      "loss": 0.4433,
+      "step": 11010
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2786825251601096e-05,
+      "loss": 0.5768,
+      "step": 11020
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.277996340347667e-05,
+      "loss": 0.4742,
+      "step": 11030
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.2773101555352242e-05,
+      "loss": 0.6165,
+      "step": 11040
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.2766239707227814e-05,
+      "loss": 0.4569,
+      "step": 11050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.2759377859103385e-05,
+      "loss": 1.3119,
+      "step": 11060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.275251601097896e-05,
+      "loss": 0.507,
+      "step": 11070
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2745654162854528e-05,
+      "loss": 0.4307,
+      "step": 11080
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2738792314730102e-05,
+      "loss": 0.6127,
+      "step": 11090
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2731930466605674e-05,
+      "loss": 0.4536,
+      "step": 11100
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2725068618481245e-05,
+      "loss": 0.487,
+      "step": 11110
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2718206770356816e-05,
+      "loss": 0.4184,
+      "step": 11120
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.271134492223239e-05,
+      "loss": 0.4765,
+      "step": 11130
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.270448307410796e-05,
+      "loss": 0.5635,
+      "step": 11140
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.2697621225983534e-05,
+      "loss": 0.4393,
+      "step": 11150
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.2690759377859105e-05,
+      "loss": 0.5835,
+      "step": 11160
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2683897529734676e-05,
+      "loss": 0.4582,
+      "step": 11170
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2677035681610247e-05,
+      "loss": 0.5261,
+      "step": 11180
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2670173833485822e-05,
+      "loss": 0.4173,
+      "step": 11190
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.266331198536139e-05,
+      "loss": 0.6234,
+      "step": 11200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.265645013723696e-05,
+      "loss": 0.6344,
+      "step": 11210
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.2649588289112536e-05,
+      "loss": 0.4937,
+      "step": 11220
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.2642726440988104e-05,
+      "loss": 0.601,
+      "step": 11230
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.263586459286368e-05,
+      "loss": 0.4667,
+      "step": 11240
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.262900274473925e-05,
+      "loss": 0.4378,
+      "step": 11250
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.262214089661482e-05,
+      "loss": 0.6075,
+      "step": 11260
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2615279048490393e-05,
+      "loss": 0.4767,
+      "step": 11270
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2608417200365967e-05,
+      "loss": 0.5407,
+      "step": 11280
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2601555352241535e-05,
+      "loss": 0.4206,
+      "step": 11290
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.259469350411711e-05,
+      "loss": 0.4092,
+      "step": 11300
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.258783165599268e-05,
+      "loss": 0.3951,
+      "step": 11310
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2580969807868253e-05,
+      "loss": 0.488,
+      "step": 11320
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2574107959743824e-05,
+      "loss": 0.4993,
+      "step": 11330
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.25672461116194e-05,
+      "loss": 0.3814,
+      "step": 11340
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.2560384263494966e-05,
+      "loss": 0.5188,
+      "step": 11350
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.255352241537054e-05,
+      "loss": 0.4486,
+      "step": 11360
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.2546660567246112e-05,
+      "loss": 0.4009,
+      "step": 11370
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.2539798719121684e-05,
+      "loss": 0.4843,
+      "step": 11380
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2532936870997255e-05,
+      "loss": 0.5467,
+      "step": 11390
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.252607502287283e-05,
+      "loss": 0.4863,
+      "step": 11400
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2519213174748398e-05,
+      "loss": 0.3911,
+      "step": 11410
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2512351326623972e-05,
+      "loss": 0.4306,
+      "step": 11420
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2505489478499544e-05,
+      "loss": 0.4078,
+      "step": 11430
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2498627630375115e-05,
+      "loss": 0.4946,
+      "step": 11440
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2491765782250686e-05,
+      "loss": 0.3709,
+      "step": 11450
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.248490393412626e-05,
+      "loss": 0.4473,
+      "step": 11460
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.247804208600183e-05,
+      "loss": 0.4047,
+      "step": 11470
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.2471180237877404e-05,
+      "loss": 0.4226,
+      "step": 11480
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.2464318389752975e-05,
+      "loss": 0.389,
+      "step": 11490
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.2457456541628546e-05,
+      "loss": 0.4588,
+      "step": 11500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.2450594693504118e-05,
+      "loss": 0.4704,
+      "step": 11510
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.2443732845379692e-05,
+      "loss": 0.541,
+      "step": 11520
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.243687099725526e-05,
+      "loss": 0.4501,
+      "step": 11530
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.2430009149130835e-05,
+      "loss": 0.42,
+      "step": 11540
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.2423147301006406e-05,
+      "loss": 0.519,
+      "step": 11550
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.2416285452881977e-05,
+      "loss": 0.3681,
+      "step": 11560
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.240942360475755e-05,
+      "loss": 0.4613,
+      "step": 11570
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.2402561756633123e-05,
+      "loss": 0.4268,
+      "step": 11580
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.239569990850869e-05,
+      "loss": 0.4769,
+      "step": 11590
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.2388838060384263e-05,
+      "loss": 0.4649,
+      "step": 11600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2381976212259837e-05,
+      "loss": 0.5629,
+      "step": 11610
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2375114364135405e-05,
+      "loss": 0.411,
+      "step": 11620
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.236825251601098e-05,
+      "loss": 0.5389,
+      "step": 11630
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.236139066788655e-05,
+      "loss": 0.4145,
+      "step": 11640
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2354528819762123e-05,
+      "loss": 0.4674,
+      "step": 11650
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.2347666971637694e-05,
+      "loss": 0.4827,
+      "step": 11660
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.234080512351327e-05,
+      "loss": 0.5046,
+      "step": 11670
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.2333943275388837e-05,
+      "loss": 0.4411,
+      "step": 11680
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.232708142726441e-05,
+      "loss": 0.4961,
+      "step": 11690
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.2320219579139983e-05,
+      "loss": 0.4678,
+      "step": 11700
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.2313357731015554e-05,
+      "loss": 0.4787,
+      "step": 11710
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.2306495882891125e-05,
+      "loss": 0.3444,
+      "step": 11720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.22996340347667e-05,
+      "loss": 0.4261,
+      "step": 11730
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.2292772186642268e-05,
+      "loss": 0.394,
+      "step": 11740
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2285910338517842e-05,
+      "loss": 0.4132,
+      "step": 11750
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2279048490393414e-05,
+      "loss": 0.9416,
+      "step": 11760
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2272186642268985e-05,
+      "loss": 0.451,
+      "step": 11770
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2265324794144556e-05,
+      "loss": 0.6032,
+      "step": 11780
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.225846294602013e-05,
+      "loss": 0.6204,
+      "step": 11790
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.22516010978957e-05,
+      "loss": 0.5017,
+      "step": 11800
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2244739249771274e-05,
+      "loss": 0.4116,
+      "step": 11810
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2237877401646845e-05,
+      "loss": 0.4423,
+      "step": 11820
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2231015553522416e-05,
+      "loss": 0.5612,
+      "step": 11830
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2224153705397988e-05,
+      "loss": 0.442,
+      "step": 11840
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2217291857273562e-05,
+      "loss": 0.4319,
+      "step": 11850
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.221043000914913e-05,
+      "loss": 0.5644,
+      "step": 11860
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2203568161024705e-05,
+      "loss": 0.5717,
+      "step": 11870
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.2196706312900276e-05,
+      "loss": 0.4494,
+      "step": 11880
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.2189844464775848e-05,
+      "loss": 0.4763,
+      "step": 11890
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.218298261665142e-05,
+      "loss": 0.4524,
+      "step": 11900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.217612076852699e-05,
+      "loss": 0.4528,
+      "step": 11910
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.216925892040256e-05,
+      "loss": 0.595,
+      "step": 11920
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2162397072278136e-05,
+      "loss": 0.5191,
+      "step": 11930
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2155535224153704e-05,
+      "loss": 0.4749,
+      "step": 11940
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.214867337602928e-05,
+      "loss": 0.465,
+      "step": 11950
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.214181152790485e-05,
+      "loss": 0.3958,
+      "step": 11960
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.213494967978042e-05,
+      "loss": 0.4873,
+      "step": 11970
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2128087831655993e-05,
+      "loss": 0.4848,
+      "step": 11980
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2121225983531567e-05,
+      "loss": 0.4935,
+      "step": 11990
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2114364135407135e-05,
+      "loss": 0.5667,
+      "step": 12000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2107502287282707e-05,
+      "loss": 0.4255,
+      "step": 12010
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.210064043915828e-05,
+      "loss": 0.4808,
+      "step": 12020
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.209377859103385e-05,
+      "loss": 0.7455,
+      "step": 12030
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2086916742909424e-05,
+      "loss": 0.4527,
+      "step": 12040
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2080054894784995e-05,
+      "loss": 0.4426,
+      "step": 12050
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2073193046660567e-05,
+      "loss": 0.4417,
+      "step": 12060
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2066331198536138e-05,
+      "loss": 0.5207,
+      "step": 12070
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2059469350411713e-05,
+      "loss": 0.4288,
+      "step": 12080
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.205260750228728e-05,
+      "loss": 0.5003,
+      "step": 12090
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2045745654162855e-05,
+      "loss": 0.5115,
+      "step": 12100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2038883806038426e-05,
+      "loss": 0.5531,
+      "step": 12110
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2032021957913998e-05,
+      "loss": 0.4139,
+      "step": 12120
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.202516010978957e-05,
+      "loss": 0.7044,
+      "step": 12130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2018298261665144e-05,
+      "loss": 0.4026,
+      "step": 12140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.201143641354071e-05,
+      "loss": 0.5033,
+      "step": 12150
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2004574565416286e-05,
+      "loss": 0.5575,
+      "step": 12160
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.1997712717291858e-05,
+      "loss": 0.4287,
+      "step": 12170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.199085086916743e-05,
+      "loss": 0.4433,
+      "step": 12180
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1983989021043e-05,
+      "loss": 0.4342,
+      "step": 12190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1977127172918575e-05,
+      "loss": 0.4385,
+      "step": 12200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1970265324794143e-05,
+      "loss": 0.4303,
+      "step": 12210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1963403476669718e-05,
+      "loss": 0.4819,
+      "step": 12220
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.195654162854529e-05,
+      "loss": 0.4496,
+      "step": 12230
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.194967978042086e-05,
+      "loss": 0.4958,
+      "step": 12240
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.194281793229643e-05,
+      "loss": 0.4648,
+      "step": 12250
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1935956084172006e-05,
+      "loss": 0.4495,
+      "step": 12260
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1929094236047574e-05,
+      "loss": 0.4075,
+      "step": 12270
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.192223238792315e-05,
+      "loss": 0.4687,
+      "step": 12280
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.191537053979872e-05,
+      "loss": 0.4748,
+      "step": 12290
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.190850869167429e-05,
+      "loss": 0.4753,
+      "step": 12300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1901646843549863e-05,
+      "loss": 0.3619,
+      "step": 12310
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1894784995425437e-05,
+      "loss": 0.5797,
+      "step": 12320
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1887923147301005e-05,
+      "loss": 0.4644,
+      "step": 12330
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.188106129917658e-05,
+      "loss": 0.4668,
+      "step": 12340
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.187419945105215e-05,
+      "loss": 0.6164,
+      "step": 12350
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1867337602927723e-05,
+      "loss": 0.519,
+      "step": 12360
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1860475754803294e-05,
+      "loss": 0.5485,
+      "step": 12370
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.185361390667887e-05,
+      "loss": 0.4159,
+      "step": 12380
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1846752058554437e-05,
+      "loss": 0.4154,
+      "step": 12390
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.183989021043001e-05,
+      "loss": 0.3626,
+      "step": 12400
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1833028362305583e-05,
+      "loss": 0.4285,
+      "step": 12410
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.182616651418115e-05,
+      "loss": 0.5165,
+      "step": 12420
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1819304666056725e-05,
+      "loss": 0.404,
+      "step": 12430
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1812442817932296e-05,
+      "loss": 0.3934,
+      "step": 12440
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1805580969807868e-05,
+      "loss": 0.4984,
+      "step": 12450
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.179871912168344e-05,
+      "loss": 0.5308,
+      "step": 12460
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1791857273559014e-05,
+      "loss": 0.4134,
+      "step": 12470
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1784995425434582e-05,
+      "loss": 0.4629,
+      "step": 12480
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1778133577310156e-05,
+      "loss": 0.4229,
+      "step": 12490
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1771271729185728e-05,
+      "loss": 0.5307,
+      "step": 12500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.17644098810613e-05,
+      "loss": 0.3967,
+      "step": 12510
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.175754803293687e-05,
+      "loss": 0.4152,
+      "step": 12520
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1750686184812445e-05,
+      "loss": 0.3714,
+      "step": 12530
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1743824336688013e-05,
+      "loss": 0.4006,
+      "step": 12540
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1736962488563588e-05,
+      "loss": 0.4157,
+      "step": 12550
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.173010064043916e-05,
+      "loss": 0.4106,
+      "step": 12560
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.172323879231473e-05,
+      "loss": 0.4192,
+      "step": 12570
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.17163769441903e-05,
+      "loss": 0.487,
+      "step": 12580
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1709515096065876e-05,
+      "loss": 0.4523,
+      "step": 12590
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1702653247941444e-05,
+      "loss": 0.4683,
+      "step": 12600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.169579139981702e-05,
+      "loss": 0.3965,
+      "step": 12610
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.168892955169259e-05,
+      "loss": 0.4629,
+      "step": 12620
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.168206770356816e-05,
+      "loss": 0.5595,
+      "step": 12630
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1675205855443733e-05,
+      "loss": 0.4581,
+      "step": 12640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1668344007319307e-05,
+      "loss": 0.459,
+      "step": 12650
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1661482159194875e-05,
+      "loss": 0.4152,
+      "step": 12660
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.165462031107045e-05,
+      "loss": 0.4664,
+      "step": 12670
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.164775846294602e-05,
+      "loss": 0.4106,
+      "step": 12680
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1640896614821593e-05,
+      "loss": 0.5461,
+      "step": 12690
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1634034766697164e-05,
+      "loss": 0.5084,
+      "step": 12700
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.162717291857274e-05,
+      "loss": 0.324,
+      "step": 12710
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1620311070448307e-05,
+      "loss": 0.478,
+      "step": 12720
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.161344922232388e-05,
+      "loss": 0.5101,
+      "step": 12730
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1606587374199453e-05,
+      "loss": 0.3632,
+      "step": 12740
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1599725526075024e-05,
+      "loss": 0.4473,
+      "step": 12750
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1592863677950595e-05,
+      "loss": 0.4203,
+      "step": 12760
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.158600182982617e-05,
+      "loss": 0.5095,
+      "step": 12770
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1579139981701738e-05,
+      "loss": 0.5079,
+      "step": 12780
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1572278133577313e-05,
+      "loss": 0.5178,
+      "step": 12790
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1565416285452884e-05,
+      "loss": 0.4144,
+      "step": 12800
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1558554437328455e-05,
+      "loss": 0.3774,
+      "step": 12810
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1551692589204026e-05,
+      "loss": 0.4005,
+      "step": 12820
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1544830741079598e-05,
+      "loss": 0.4439,
+      "step": 12830
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.153796889295517e-05,
+      "loss": 0.4598,
+      "step": 12840
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.153110704483074e-05,
+      "loss": 0.4572,
+      "step": 12850
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1524245196706315e-05,
+      "loss": 0.544,
+      "step": 12860
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1517383348581883e-05,
+      "loss": 0.3503,
+      "step": 12870
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1510521500457458e-05,
+      "loss": 0.4139,
+      "step": 12880
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.150365965233303e-05,
+      "loss": 0.3508,
+      "step": 12890
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.14967978042086e-05,
+      "loss": 0.4696,
+      "step": 12900
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.148993595608417e-05,
+      "loss": 0.4277,
+      "step": 12910
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1483074107959746e-05,
+      "loss": 0.3519,
+      "step": 12920
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1476212259835314e-05,
+      "loss": 0.3626,
+      "step": 12930
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.146935041171089e-05,
+      "loss": 0.3953,
+      "step": 12940
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.146248856358646e-05,
+      "loss": 0.741,
+      "step": 12950
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.145562671546203e-05,
+      "loss": 0.4741,
+      "step": 12960
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.1448764867337603e-05,
+      "loss": 0.4142,
+      "step": 12970
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1441903019213178e-05,
+      "loss": 0.3972,
+      "step": 12980
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1435041171088745e-05,
+      "loss": 0.4456,
+      "step": 12990
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.142817932296432e-05,
+      "loss": 0.4148,
+      "step": 13000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.142131747483989e-05,
+      "loss": 0.363,
+      "step": 13010
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1414455626715463e-05,
+      "loss": 0.4989,
+      "step": 13020
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1407593778591034e-05,
+      "loss": 0.4711,
+      "step": 13030
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.140073193046661e-05,
+      "loss": 0.5193,
+      "step": 13040
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1393870082342177e-05,
+      "loss": 0.382,
+      "step": 13050
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.138700823421775e-05,
+      "loss": 0.423,
+      "step": 13060
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1380146386093323e-05,
+      "loss": 0.4382,
+      "step": 13070
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1373284537968894e-05,
+      "loss": 0.5377,
+      "step": 13080
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1366422689844465e-05,
+      "loss": 0.4917,
+      "step": 13090
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.135956084172004e-05,
+      "loss": 0.312,
+      "step": 13100
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1352698993595608e-05,
+      "loss": 0.4307,
+      "step": 13110
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1345837145471183e-05,
+      "loss": 0.4343,
+      "step": 13120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1338975297346754e-05,
+      "loss": 0.5318,
+      "step": 13130
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1332113449222325e-05,
+      "loss": 0.4058,
+      "step": 13140
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1325251601097897e-05,
+      "loss": 0.4868,
+      "step": 13150
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1318389752973468e-05,
+      "loss": 0.5004,
+      "step": 13160
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.131152790484904e-05,
+      "loss": 0.5072,
+      "step": 13170
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1304666056724614e-05,
+      "loss": 0.4262,
+      "step": 13180
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1297804208600182e-05,
+      "loss": 0.5674,
+      "step": 13190
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1290942360475756e-05,
+      "loss": 0.4162,
+      "step": 13200
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1284080512351328e-05,
+      "loss": 0.5244,
+      "step": 13210
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1277218664226896e-05,
+      "loss": 0.4652,
+      "step": 13220
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.127035681610247e-05,
+      "loss": 0.4915,
+      "step": 13230
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.126349496797804e-05,
+      "loss": 0.5509,
+      "step": 13240
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1256633119853613e-05,
+      "loss": 0.4472,
+      "step": 13250
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1249771271729184e-05,
+      "loss": 0.3919,
+      "step": 13260
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.5314621925354004,
+      "eval_runtime": 299.6146,
+      "eval_samples_per_second": 6.558,
+      "eval_steps_per_second": 1.642,
+      "eval_wer": 0.2833466822746596,
+      "step": 13266
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.124290942360476e-05,
+      "loss": 0.4211,
+      "step": 13270
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1236047575480327e-05,
+      "loss": 0.4488,
+      "step": 13280
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.12291857273559e-05,
+      "loss": 0.4027,
+      "step": 13290
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1223010064043918e-05,
+      "loss": 0.438,
+      "step": 13300
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.121614821591949e-05,
+      "loss": 0.4416,
+      "step": 13310
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.120928636779506e-05,
+      "loss": 0.4342,
+      "step": 13320
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1202424519670635e-05,
+      "loss": 0.3563,
+      "step": 13330
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.1195562671546203e-05,
+      "loss": 0.4036,
+      "step": 13340
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.1188700823421774e-05,
+      "loss": 0.3554,
+      "step": 13350
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.118183897529735e-05,
+      "loss": 0.4647,
+      "step": 13360
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.1174977127172917e-05,
+      "loss": 0.3352,
+      "step": 13370
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.116811527904849e-05,
+      "loss": 0.524,
+      "step": 13380
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.1161253430924063e-05,
+      "loss": 0.3928,
+      "step": 13390
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.1154391582799634e-05,
+      "loss": 0.3712,
+      "step": 13400
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.1147529734675206e-05,
+      "loss": 0.3854,
+      "step": 13410
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.1141354071363222e-05,
+      "loss": 0.4129,
+      "step": 13420
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.1134492223238793e-05,
+      "loss": 0.3982,
+      "step": 13430
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.1127630375114368e-05,
+      "loss": 0.5115,
+      "step": 13440
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.1120768526989936e-05,
+      "loss": 0.3984,
+      "step": 13450
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.111390667886551e-05,
+      "loss": 0.3045,
+      "step": 13460
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.1107044830741082e-05,
+      "loss": 0.4645,
+      "step": 13470
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.110018298261665e-05,
+      "loss": 0.4177,
+      "step": 13480
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.1093321134492224e-05,
+      "loss": 0.5587,
+      "step": 13490
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.1086459286367796e-05,
+      "loss": 0.3906,
+      "step": 13500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.1079597438243367e-05,
+      "loss": 0.3412,
+      "step": 13510
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.1072735590118938e-05,
+      "loss": 0.4927,
+      "step": 13520
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.1065873741994513e-05,
+      "loss": 0.4442,
+      "step": 13530
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.105901189387008e-05,
+      "loss": 0.3697,
+      "step": 13540
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.1052150045745656e-05,
+      "loss": 0.5073,
+      "step": 13550
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.1045288197621227e-05,
+      "loss": 0.377,
+      "step": 13560
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.1038426349496798e-05,
+      "loss": 0.4236,
+      "step": 13570
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.103156450137237e-05,
+      "loss": 0.3966,
+      "step": 13580
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.1024702653247944e-05,
+      "loss": 0.4558,
+      "step": 13590
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.1017840805123512e-05,
+      "loss": 0.339,
+      "step": 13600
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.1010978956999087e-05,
+      "loss": 0.3902,
+      "step": 13610
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.1004117108874658e-05,
+      "loss": 0.3925,
+      "step": 13620
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.099725526075023e-05,
+      "loss": 0.4682,
+      "step": 13630
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.09903934126258e-05,
+      "loss": 0.4159,
+      "step": 13640
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.0983531564501372e-05,
+      "loss": 0.4732,
+      "step": 13650
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.0976669716376943e-05,
+      "loss": 0.3902,
+      "step": 13660
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.0969807868252518e-05,
+      "loss": 0.3943,
+      "step": 13670
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.0962946020128086e-05,
+      "loss": 0.4261,
+      "step": 13680
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.095608417200366e-05,
+      "loss": 0.3761,
+      "step": 13690
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.0949222323879232e-05,
+      "loss": 0.3864,
+      "step": 13700
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.0942360475754803e-05,
+      "loss": 0.3123,
+      "step": 13710
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.0935498627630375e-05,
+      "loss": 0.5097,
+      "step": 13720
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.092863677950595e-05,
+      "loss": 0.3006,
+      "step": 13730
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.0921774931381517e-05,
+      "loss": 0.4041,
+      "step": 13740
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.0914913083257092e-05,
+      "loss": 0.3305,
+      "step": 13750
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.0908051235132663e-05,
+      "loss": 0.3921,
+      "step": 13760
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.0901189387008234e-05,
+      "loss": 0.4477,
+      "step": 13770
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.0894327538883806e-05,
+      "loss": 0.385,
+      "step": 13780
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.088746569075938e-05,
+      "loss": 0.842,
+      "step": 13790
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.088060384263495e-05,
+      "loss": 0.4066,
+      "step": 13800
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.0873741994510523e-05,
+      "loss": 0.4404,
+      "step": 13810
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0866880146386094e-05,
+      "loss": 0.3461,
+      "step": 13820
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0860018298261666e-05,
+      "loss": 0.3484,
+      "step": 13830
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0853156450137237e-05,
+      "loss": 0.3284,
+      "step": 13840
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.084629460201281e-05,
+      "loss": 0.3926,
+      "step": 13850
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.083943275388838e-05,
+      "loss": 0.3821,
+      "step": 13860
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.083257090576395e-05,
+      "loss": 0.4603,
+      "step": 13870
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0825709057639526e-05,
+      "loss": 0.4254,
+      "step": 13880
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0818847209515094e-05,
+      "loss": 0.3468,
+      "step": 13890
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0811985361390668e-05,
+      "loss": 0.3438,
+      "step": 13900
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.080512351326624e-05,
+      "loss": 0.3742,
+      "step": 13910
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.079826166514181e-05,
+      "loss": 0.4392,
+      "step": 13920
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0791399817017382e-05,
+      "loss": 0.3135,
+      "step": 13930
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0784537968892957e-05,
+      "loss": 0.4203,
+      "step": 13940
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0777676120768525e-05,
+      "loss": 0.4214,
+      "step": 13950
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.07708142726441e-05,
+      "loss": 0.2921,
+      "step": 13960
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.076395242451967e-05,
+      "loss": 0.4641,
+      "step": 13970
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.0757090576395242e-05,
+      "loss": 0.4387,
+      "step": 13980
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.0750228728270813e-05,
+      "loss": 0.4093,
+      "step": 13990
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.0743366880146388e-05,
+      "loss": 0.3138,
+      "step": 14000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.0736505032021956e-05,
+      "loss": 0.3188,
+      "step": 14010
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.072964318389753e-05,
+      "loss": 0.4575,
+      "step": 14020
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.0722781335773102e-05,
+      "loss": 0.3834,
+      "step": 14030
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.0715919487648673e-05,
+      "loss": 0.4177,
+      "step": 14040
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.0709057639524245e-05,
+      "loss": 0.3889,
+      "step": 14050
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.070219579139982e-05,
+      "loss": 0.2905,
+      "step": 14060
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.0695333943275387e-05,
+      "loss": 0.4377,
+      "step": 14070
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.0688472095150962e-05,
+      "loss": 0.4243,
+      "step": 14080
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0681610247026533e-05,
+      "loss": 0.365,
+      "step": 14090
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0674748398902105e-05,
+      "loss": 0.4318,
+      "step": 14100
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0667886550777676e-05,
+      "loss": 0.5247,
+      "step": 14110
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.066102470265325e-05,
+      "loss": 0.3744,
+      "step": 14120
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.065416285452882e-05,
+      "loss": 0.4636,
+      "step": 14130
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0647301006404393e-05,
+      "loss": 0.452,
+      "step": 14140
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0640439158279964e-05,
+      "loss": 0.4399,
+      "step": 14150
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0633577310155536e-05,
+      "loss": 0.3518,
+      "step": 14160
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0626715462031107e-05,
+      "loss": 0.3486,
+      "step": 14170
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0619853613906682e-05,
+      "loss": 0.3876,
+      "step": 14180
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.061299176578225e-05,
+      "loss": 0.3565,
+      "step": 14190
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0606129917657824e-05,
+      "loss": 0.4252,
+      "step": 14200
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0599268069533396e-05,
+      "loss": 0.399,
+      "step": 14210
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0592406221408967e-05,
+      "loss": 0.3921,
+      "step": 14220
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0585544373284538e-05,
+      "loss": 0.3342,
+      "step": 14230
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0578682525160113e-05,
+      "loss": 0.3972,
+      "step": 14240
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.057182067703568e-05,
+      "loss": 0.3482,
+      "step": 14250
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0564958828911256e-05,
+      "loss": 0.4452,
+      "step": 14260
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.0558096980786827e-05,
+      "loss": 0.3718,
+      "step": 14270
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.0551235132662395e-05,
+      "loss": 0.3894,
+      "step": 14280
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.054437328453797e-05,
+      "loss": 0.4193,
+      "step": 14290
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.053751143641354e-05,
+      "loss": 0.4639,
+      "step": 14300
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.0530649588289112e-05,
+      "loss": 0.3309,
+      "step": 14310
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.0523787740164683e-05,
+      "loss": 0.3791,
+      "step": 14320
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.0516925892040258e-05,
+      "loss": 0.3401,
+      "step": 14330
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.0510064043915826e-05,
+      "loss": 0.312,
+      "step": 14340
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.05032021957914e-05,
+      "loss": 0.6453,
+      "step": 14350
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.0496340347666972e-05,
+      "loss": 0.4543,
+      "step": 14360
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.0489478499542543e-05,
+      "loss": 0.3914,
+      "step": 14370
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.0482616651418115e-05,
+      "loss": 0.4661,
+      "step": 14380
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.047575480329369e-05,
+      "loss": 0.4116,
+      "step": 14390
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.0468892955169257e-05,
+      "loss": 0.4174,
+      "step": 14400
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.0462031107044832e-05,
+      "loss": 0.4854,
+      "step": 14410
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.0455169258920403e-05,
+      "loss": 0.3896,
+      "step": 14420
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.0448307410795975e-05,
+      "loss": 0.3732,
+      "step": 14430
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0441445562671546e-05,
+      "loss": 0.3731,
+      "step": 14440
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.043526989935956e-05,
+      "loss": 1.0973,
+      "step": 14450
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0428408051235133e-05,
+      "loss": 0.3464,
+      "step": 14460
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0421546203110705e-05,
+      "loss": 0.3832,
+      "step": 14470
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0414684354986276e-05,
+      "loss": 0.311,
+      "step": 14480
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0407822506861847e-05,
+      "loss": 0.4894,
+      "step": 14490
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0400960658737422e-05,
+      "loss": 0.3553,
+      "step": 14500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.039409881061299e-05,
+      "loss": 0.3187,
+      "step": 14510
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0387236962488565e-05,
+      "loss": 0.3952,
+      "step": 14520
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.0380375114364136e-05,
+      "loss": 0.3706,
+      "step": 14530
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.0373513266239707e-05,
+      "loss": 0.4099,
+      "step": 14540
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.036665141811528e-05,
+      "loss": 0.3903,
+      "step": 14550
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.0359789569990853e-05,
+      "loss": 0.5459,
+      "step": 14560
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.035292772186642e-05,
+      "loss": 0.4401,
+      "step": 14570
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.0346065873741996e-05,
+      "loss": 0.3227,
+      "step": 14580
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.0339204025617567e-05,
+      "loss": 0.3964,
+      "step": 14590
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.033234217749314e-05,
+      "loss": 0.3932,
+      "step": 14600
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.032548032936871e-05,
+      "loss": 0.4678,
+      "step": 14610
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.0318618481244284e-05,
+      "loss": 0.6736,
+      "step": 14620
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.0311756633119852e-05,
+      "loss": 0.4525,
+      "step": 14630
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.0304894784995427e-05,
+      "loss": 0.3661,
+      "step": 14640
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.0298032936871e-05,
+      "loss": 0.338,
+      "step": 14650
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.029117108874657e-05,
+      "loss": 0.4076,
+      "step": 14660
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.028430924062214e-05,
+      "loss": 0.3558,
+      "step": 14670
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.0277447392497716e-05,
+      "loss": 0.5346,
+      "step": 14680
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.0270585544373284e-05,
+      "loss": 0.4037,
+      "step": 14690
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.026372369624886e-05,
+      "loss": 0.4243,
+      "step": 14700
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.025686184812443e-05,
+      "loss": 0.376,
+      "step": 14710
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.025e-05,
+      "loss": 0.4554,
+      "step": 14720
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.0243138151875572e-05,
+      "loss": 0.3271,
+      "step": 14730
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.0236276303751147e-05,
+      "loss": 0.6163,
+      "step": 14740
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.0229414455626715e-05,
+      "loss": 0.3967,
+      "step": 14750
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.022255260750229e-05,
+      "loss": 0.385,
+      "step": 14760
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.021569075937786e-05,
+      "loss": 0.453,
+      "step": 14770
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.0208828911253432e-05,
+      "loss": 0.3353,
+      "step": 14780
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.0201967063129003e-05,
+      "loss": 0.4231,
+      "step": 14790
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.0195105215004578e-05,
+      "loss": 0.4803,
+      "step": 14800
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.0188243366880146e-05,
+      "loss": 0.3866,
+      "step": 14810
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.0181381518755717e-05,
+      "loss": 0.4082,
+      "step": 14820
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.0174519670631292e-05,
+      "loss": 0.4028,
+      "step": 14830
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.016765782250686e-05,
+      "loss": 0.3799,
+      "step": 14840
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.0160795974382435e-05,
+      "loss": 0.4221,
+      "step": 14850
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.0153934126258006e-05,
+      "loss": 0.4051,
+      "step": 14860
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.0147072278133577e-05,
+      "loss": 0.3152,
+      "step": 14870
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.014021043000915e-05,
+      "loss": 0.4045,
+      "step": 14880
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.0133348581884723e-05,
+      "loss": 0.4497,
+      "step": 14890
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.012648673376029e-05,
+      "loss": 0.4382,
+      "step": 14900
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.0119624885635866e-05,
+      "loss": 0.4319,
+      "step": 14910
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.0112763037511437e-05,
+      "loss": 0.3969,
+      "step": 14920
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.010590118938701e-05,
+      "loss": 0.3669,
+      "step": 14930
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.009903934126258e-05,
+      "loss": 0.401,
+      "step": 14940
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.0092177493138155e-05,
+      "loss": 0.3112,
+      "step": 14950
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.0085315645013722e-05,
+      "loss": 0.3017,
+      "step": 14960
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.0078453796889297e-05,
+      "loss": 0.5401,
+      "step": 14970
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.007159194876487e-05,
+      "loss": 0.419,
+      "step": 14980
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.006473010064044e-05,
+      "loss": 0.7084,
+      "step": 14990
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.005786825251601e-05,
+      "loss": 0.3665,
+      "step": 15000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.0051006404391586e-05,
+      "loss": 0.3687,
+      "step": 15010
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.0044144556267154e-05,
+      "loss": 0.4699,
+      "step": 15020
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.003728270814273e-05,
+      "loss": 0.4026,
+      "step": 15030
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.00304208600183e-05,
+      "loss": 0.3661,
+      "step": 15040
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.002355901189387e-05,
+      "loss": 0.3706,
+      "step": 15050
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.0016697163769442e-05,
+      "loss": 0.3769,
+      "step": 15060
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.0009835315645017e-05,
+      "loss": 0.3336,
+      "step": 15070
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.0002973467520585e-05,
+      "loss": 0.2834,
+      "step": 15080
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.999611161939616e-05,
+      "loss": 0.3693,
+      "step": 15090
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.998924977127173e-05,
+      "loss": 0.3624,
+      "step": 15100
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9982387923147302e-05,
+      "loss": 0.4136,
+      "step": 15110
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9975526075022874e-05,
+      "loss": 0.3934,
+      "step": 15120
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9968664226898445e-05,
+      "loss": 0.3373,
+      "step": 15130
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9961802378774016e-05,
+      "loss": 0.3012,
+      "step": 15140
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.995494053064959e-05,
+      "loss": 0.3893,
+      "step": 15150
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.994807868252516e-05,
+      "loss": 0.3431,
+      "step": 15160
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.9941216834400733e-05,
+      "loss": 0.3873,
+      "step": 15170
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.9934354986276305e-05,
+      "loss": 0.4615,
+      "step": 15180
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.9927493138151876e-05,
+      "loss": 0.413,
+      "step": 15190
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.9920631290027447e-05,
+      "loss": 0.3134,
+      "step": 15200
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.991376944190302e-05,
+      "loss": 0.3988,
+      "step": 15210
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.990690759377859e-05,
+      "loss": 0.3607,
+      "step": 15220
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.990004574565416e-05,
+      "loss": 0.3682,
+      "step": 15230
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.9893183897529736e-05,
+      "loss": 0.3882,
+      "step": 15240
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.9886322049405304e-05,
+      "loss": 0.3122,
+      "step": 15250
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.987946020128088e-05,
+      "loss": 0.2643,
+      "step": 15260
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.987259835315645e-05,
+      "loss": 0.461,
+      "step": 15270
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.986573650503202e-05,
+      "loss": 0.3983,
+      "step": 15280
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.9858874656907593e-05,
+      "loss": 0.5172,
+      "step": 15290
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.9852012808783167e-05,
+      "loss": 0.4463,
+      "step": 15300
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.9845150960658735e-05,
+      "loss": 0.3893,
+      "step": 15310
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.983828911253431e-05,
+      "loss": 0.3495,
+      "step": 15320
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.983142726440988e-05,
+      "loss": 0.4307,
+      "step": 15330
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.9824565416285452e-05,
+      "loss": 0.3994,
+      "step": 15340
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.9817703568161024e-05,
+      "loss": 0.4049,
+      "step": 15350
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.98108417200366e-05,
+      "loss": 0.3914,
+      "step": 15360
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.9803979871912166e-05,
+      "loss": 0.448,
+      "step": 15370
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.979711802378774e-05,
+      "loss": 0.2797,
+      "step": 15380
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.9790256175663312e-05,
+      "loss": 0.5462,
+      "step": 15390
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.9783394327538884e-05,
+      "loss": 0.2932,
+      "step": 15400
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.9776532479414455e-05,
+      "loss": 0.4411,
+      "step": 15410
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.976967063129003e-05,
+      "loss": 0.4236,
+      "step": 15420
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.9762808783165598e-05,
+      "loss": 0.3453,
+      "step": 15430
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.9755946935041172e-05,
+      "loss": 0.3664,
+      "step": 15440
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.9749085086916744e-05,
+      "loss": 0.4532,
+      "step": 15450
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.9742223238792315e-05,
+      "loss": 0.4104,
+      "step": 15460
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.9735361390667886e-05,
+      "loss": 0.3464,
+      "step": 15470
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.972849954254346e-05,
+      "loss": 0.307,
+      "step": 15480
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.972163769441903e-05,
+      "loss": 0.3621,
+      "step": 15490
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9714775846294604e-05,
+      "loss": 0.3116,
+      "step": 15500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9707913998170175e-05,
+      "loss": 0.3544,
+      "step": 15510
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9701052150045746e-05,
+      "loss": 0.4239,
+      "step": 15520
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9694190301921317e-05,
+      "loss": 0.3414,
+      "step": 15530
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9687328453796892e-05,
+      "loss": 0.4225,
+      "step": 15540
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.968046660567246e-05,
+      "loss": 0.3266,
+      "step": 15550
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.9673604757548035e-05,
+      "loss": 0.494,
+      "step": 15560
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.9666742909423606e-05,
+      "loss": 0.2909,
+      "step": 15570
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.9659881061299177e-05,
+      "loss": 0.4275,
+      "step": 15580
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.965301921317475e-05,
+      "loss": 0.3029,
+      "step": 15590
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.9646157365050323e-05,
+      "loss": 0.3145,
+      "step": 15600
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.963929551692589e-05,
+      "loss": 0.4645,
+      "step": 15610
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.9632433668801463e-05,
+      "loss": 0.3365,
+      "step": 15620
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.9625571820677037e-05,
+      "loss": 0.309,
+      "step": 15630
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.9618709972552605e-05,
+      "loss": 0.3832,
+      "step": 15640
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.961184812442818e-05,
+      "loss": 0.3142,
+      "step": 15650
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.960498627630375e-05,
+      "loss": 1.0136,
+      "step": 15660
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.9598124428179323e-05,
+      "loss": 0.3024,
+      "step": 15670
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.9591262580054894e-05,
+      "loss": 0.3115,
+      "step": 15680
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.958440073193047e-05,
+      "loss": 0.4038,
+      "step": 15690
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.9577538883806036e-05,
+      "loss": 0.3436,
+      "step": 15700
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.957067703568161e-05,
+      "loss": 0.3175,
+      "step": 15710
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.9563815187557182e-05,
+      "loss": 0.5586,
+      "step": 15720
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.9556953339432754e-05,
+      "loss": 0.4197,
+      "step": 15730
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.9550091491308325e-05,
+      "loss": 0.369,
+      "step": 15740
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.95432296431839e-05,
+      "loss": 0.3628,
+      "step": 15750
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.9536367795059468e-05,
+      "loss": 0.3389,
+      "step": 15760
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9529505946935042e-05,
+      "loss": 0.3075,
+      "step": 15770
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9522644098810614e-05,
+      "loss": 0.4434,
+      "step": 15780
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9515782250686185e-05,
+      "loss": 0.44,
+      "step": 15790
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9508920402561756e-05,
+      "loss": 0.3323,
+      "step": 15800
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.950205855443733e-05,
+      "loss": 0.3214,
+      "step": 15810
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.94951967063129e-05,
+      "loss": 0.3923,
+      "step": 15820
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9488334858188474e-05,
+      "loss": 0.3153,
+      "step": 15830
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9481473010064045e-05,
+      "loss": 0.3998,
+      "step": 15840
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9474611161939616e-05,
+      "loss": 0.8393,
+      "step": 15850
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9467749313815187e-05,
+      "loss": 0.3901,
+      "step": 15860
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9460887465690762e-05,
+      "loss": 0.4211,
+      "step": 15870
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.945402561756633e-05,
+      "loss": 0.4632,
+      "step": 15880
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9447163769441905e-05,
+      "loss": 0.3611,
+      "step": 15890
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9440301921317476e-05,
+      "loss": 0.413,
+      "step": 15900
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9433440073193047e-05,
+      "loss": 0.3564,
+      "step": 15910
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.942657822506862e-05,
+      "loss": 0.331,
+      "step": 15920
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9419716376944193e-05,
+      "loss": 0.3195,
+      "step": 15930
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.941285452881976e-05,
+      "loss": 0.4204,
+      "step": 15940
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.9405992680695336e-05,
+      "loss": 0.4273,
+      "step": 15950
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.9399130832570907e-05,
+      "loss": 0.4167,
+      "step": 15960
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.939226898444648e-05,
+      "loss": 0.3138,
+      "step": 15970
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.938540713632205e-05,
+      "loss": 0.4198,
+      "step": 15980
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9378545288197625e-05,
+      "loss": 0.4698,
+      "step": 15990
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9371683440073193e-05,
+      "loss": 0.4301,
+      "step": 16000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9364821591948767e-05,
+      "loss": 0.3905,
+      "step": 16010
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.935795974382434e-05,
+      "loss": 0.3582,
+      "step": 16020
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9351097895699906e-05,
+      "loss": 0.3801,
+      "step": 16030
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.934423604757548e-05,
+      "loss": 0.4185,
+      "step": 16040
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9337374199451052e-05,
+      "loss": 0.4606,
+      "step": 16050
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9330512351326624e-05,
+      "loss": 0.3779,
+      "step": 16060
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9323650503202195e-05,
+      "loss": 0.3775,
+      "step": 16070
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.931678865507777e-05,
+      "loss": 0.4428,
+      "step": 16080
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9309926806953338e-05,
+      "loss": 0.4193,
+      "step": 16090
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9303064958828912e-05,
+      "loss": 0.3136,
+      "step": 16100
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9296203110704484e-05,
+      "loss": 0.5986,
+      "step": 16110
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.9289341262580055e-05,
+      "loss": 0.453,
+      "step": 16120
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.9282479414455626e-05,
+      "loss": 0.8498,
+      "step": 16130
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.92756175663312e-05,
+      "loss": 0.4313,
+      "step": 16140
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.926875571820677e-05,
+      "loss": 0.4078,
+      "step": 16150
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.9261893870082344e-05,
+      "loss": 0.3489,
+      "step": 16160
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.9255032021957915e-05,
+      "loss": 0.4316,
+      "step": 16170
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.9248170173833486e-05,
+      "loss": 0.2886,
+      "step": 16180
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.9241308325709058e-05,
+      "loss": 0.2787,
+      "step": 16190
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.9234446477584632e-05,
+      "loss": 0.4065,
+      "step": 16200
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.92275846294602e-05,
+      "loss": 0.3639,
+      "step": 16210
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.9220722781335775e-05,
+      "loss": 0.4446,
+      "step": 16220
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.9213860933211346e-05,
+      "loss": 0.4535,
+      "step": 16230
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.9206999085086917e-05,
+      "loss": 0.5061,
+      "step": 16240
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.920013723696249e-05,
+      "loss": 0.2505,
+      "step": 16250
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.9193275388838063e-05,
+      "loss": 0.4103,
+      "step": 16260
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.918641354071363e-05,
+      "loss": 0.3509,
+      "step": 16270
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.9179551692589206e-05,
+      "loss": 0.3003,
+      "step": 16280
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.9172689844464777e-05,
+      "loss": 0.3976,
+      "step": 16290
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.916582799634035e-05,
+      "loss": 0.3102,
+      "step": 16300
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.915896614821592e-05,
+      "loss": 0.3005,
+      "step": 16310
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.9152104300091495e-05,
+      "loss": 0.3709,
+      "step": 16320
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.9145242451967063e-05,
+      "loss": 0.3941,
+      "step": 16330
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.9138380603842637e-05,
+      "loss": 0.4138,
+      "step": 16340
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.913151875571821e-05,
+      "loss": 0.3681,
+      "step": 16350
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.912465690759378e-05,
+      "loss": 0.3561,
+      "step": 16360
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.911779505946935e-05,
+      "loss": 0.4263,
+      "step": 16370
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.9110933211344923e-05,
+      "loss": 0.4012,
+      "step": 16380
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.9104071363220494e-05,
+      "loss": 0.4239,
+      "step": 16390
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.909720951509607e-05,
+      "loss": 0.2959,
+      "step": 16400
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.9090347666971636e-05,
+      "loss": 0.3194,
+      "step": 16410
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.908348581884721e-05,
+      "loss": 0.416,
+      "step": 16420
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.9076623970722782e-05,
+      "loss": 0.7281,
+      "step": 16430
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.906976212259835e-05,
+      "loss": 0.3582,
+      "step": 16440
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.9062900274473925e-05,
+      "loss": 0.42,
+      "step": 16450
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.9056038426349496e-05,
+      "loss": 0.3148,
+      "step": 16460
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.9049176578225068e-05,
+      "loss": 0.3628,
+      "step": 16470
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.904231473010064e-05,
+      "loss": 0.4192,
+      "step": 16480
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.9035452881976214e-05,
+      "loss": 0.4363,
+      "step": 16490
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.902859103385178e-05,
+      "loss": 0.441,
+      "step": 16500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.9021729185727356e-05,
+      "loss": 0.3375,
+      "step": 16510
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.9014867337602928e-05,
+      "loss": 0.3293,
+      "step": 16520
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.90080054894785e-05,
+      "loss": 0.3499,
+      "step": 16530
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.900114364135407e-05,
+      "loss": 0.263,
+      "step": 16540
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8994281793229645e-05,
+      "loss": 0.4719,
+      "step": 16550
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8987419945105213e-05,
+      "loss": 0.4389,
+      "step": 16560
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8980558096980788e-05,
+      "loss": 0.2995,
+      "step": 16570
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.897369624885636e-05,
+      "loss": 0.3905,
+      "step": 16580
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.896683440073193e-05,
+      "loss": 0.4461,
+      "step": 16590
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.89599725526075e-05,
+      "loss": 0.2967,
+      "step": 16600
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8953110704483076e-05,
+      "loss": 0.5792,
+      "step": 16610
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8946248856358644e-05,
+      "loss": 0.3595,
+      "step": 16620
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.893938700823422e-05,
+      "loss": 0.3737,
+      "step": 16630
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.893252516010979e-05,
+      "loss": 0.465,
+      "step": 16640
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.892566331198536e-05,
+      "loss": 0.3811,
+      "step": 16650
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8918801463860933e-05,
+      "loss": 0.3472,
+      "step": 16660
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8911939615736507e-05,
+      "loss": 0.5727,
+      "step": 16670
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8905077767612075e-05,
+      "loss": 0.4421,
+      "step": 16680
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.889821591948765e-05,
+      "loss": 0.3711,
+      "step": 16690
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.889135407136322e-05,
+      "loss": 0.4288,
+      "step": 16700
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8884492223238793e-05,
+      "loss": 0.4993,
+      "step": 16710
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8877630375114364e-05,
+      "loss": 0.3993,
+      "step": 16720
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.887076852698994e-05,
+      "loss": 0.3368,
+      "step": 16730
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8863906678865507e-05,
+      "loss": 0.3595,
+      "step": 16740
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.885704483074108e-05,
+      "loss": 0.2943,
+      "step": 16750
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8850182982616653e-05,
+      "loss": 0.4597,
+      "step": 16760
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8843321134492224e-05,
+      "loss": 0.4347,
+      "step": 16770
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8836459286367795e-05,
+      "loss": 0.3142,
+      "step": 16780
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.882959743824337e-05,
+      "loss": 0.3061,
+      "step": 16790
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.8822735590118938e-05,
+      "loss": 0.4478,
+      "step": 16800
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.8815873741994512e-05,
+      "loss": 0.2756,
+      "step": 16810
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.8809011893870084e-05,
+      "loss": 0.3785,
+      "step": 16820
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.880215004574565e-05,
+      "loss": 0.3937,
+      "step": 16830
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.8795288197621226e-05,
+      "loss": 0.2975,
+      "step": 16840
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.8788426349496798e-05,
+      "loss": 0.3604,
+      "step": 16850
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.878156450137237e-05,
+      "loss": 0.4051,
+      "step": 16860
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.877470265324794e-05,
+      "loss": 0.4154,
+      "step": 16870
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.8767840805123515e-05,
+      "loss": 0.3598,
+      "step": 16880
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.8760978956999083e-05,
+      "loss": 0.4833,
+      "step": 16890
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.8754117108874658e-05,
+      "loss": 0.2735,
+      "step": 16900
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.874725526075023e-05,
+      "loss": 0.3085,
+      "step": 16910
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.87403934126258e-05,
+      "loss": 0.4832,
+      "step": 16920
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.873353156450137e-05,
+      "loss": 0.3495,
+      "step": 16930
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8726669716376946e-05,
+      "loss": 0.5057,
+      "step": 16940
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8719807868252514e-05,
+      "loss": 0.3358,
+      "step": 16950
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.871294602012809e-05,
+      "loss": 0.4288,
+      "step": 16960
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.870608417200366e-05,
+      "loss": 0.3979,
+      "step": 16970
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.869922232387923e-05,
+      "loss": 0.3278,
+      "step": 16980
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.8692360475754803e-05,
+      "loss": 0.3366,
+      "step": 16990
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.8685498627630377e-05,
+      "loss": 0.3821,
+      "step": 17000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8678636779505945e-05,
+      "loss": 0.3745,
+      "step": 17010
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.867177493138152e-05,
+      "loss": 0.4552,
+      "step": 17020
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.866491308325709e-05,
+      "loss": 0.3901,
+      "step": 17030
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8658051235132663e-05,
+      "loss": 0.3886,
+      "step": 17040
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8651189387008234e-05,
+      "loss": 0.371,
+      "step": 17050
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.864432753888381e-05,
+      "loss": 0.429,
+      "step": 17060
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8637465690759377e-05,
+      "loss": 0.4329,
+      "step": 17070
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.863060384263495e-05,
+      "loss": 0.2932,
+      "step": 17080
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8623741994510523e-05,
+      "loss": 0.4014,
+      "step": 17090
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8616880146386094e-05,
+      "loss": 0.3765,
+      "step": 17100
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8610018298261665e-05,
+      "loss": 0.5275,
+      "step": 17110
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.860315645013724e-05,
+      "loss": 0.4925,
+      "step": 17120
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8596294602012808e-05,
+      "loss": 0.4724,
+      "step": 17130
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8589432753888383e-05,
+      "loss": 0.3734,
+      "step": 17140
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8582570905763954e-05,
+      "loss": 0.3537,
+      "step": 17150
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8575709057639525e-05,
+      "loss": 0.2628,
+      "step": 17160
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8568847209515096e-05,
+      "loss": 0.3907,
+      "step": 17170
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.856198536139067e-05,
+      "loss": 0.4061,
+      "step": 17180
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.855512351326624e-05,
+      "loss": 0.3569,
+      "step": 17190
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8548261665141814e-05,
+      "loss": 0.3695,
+      "step": 17200
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8541399817017385e-05,
+      "loss": 0.3485,
+      "step": 17210
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8534537968892956e-05,
+      "loss": 0.3295,
+      "step": 17220
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.8527676120768528e-05,
+      "loss": 0.395,
+      "step": 17230
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.85208142726441e-05,
+      "loss": 0.3563,
+      "step": 17240
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.851395242451967e-05,
+      "loss": 0.3388,
+      "step": 17250
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.850709057639524e-05,
+      "loss": 0.3486,
+      "step": 17260
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.8500228728270816e-05,
+      "loss": 0.3891,
+      "step": 17270
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.8493366880146384e-05,
+      "loss": 0.3462,
+      "step": 17280
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.848650503202196e-05,
+      "loss": 0.3859,
+      "step": 17290
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.847964318389753e-05,
+      "loss": 0.3334,
+      "step": 17300
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.84727813357731e-05,
+      "loss": 0.3229,
+      "step": 17310
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8465919487648673e-05,
+      "loss": 0.3293,
+      "step": 17320
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8459057639524248e-05,
+      "loss": 0.3131,
+      "step": 17330
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8452195791399815e-05,
+      "loss": 0.346,
+      "step": 17340
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.844533394327539e-05,
+      "loss": 0.3297,
+      "step": 17350
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.843847209515096e-05,
+      "loss": 0.3588,
+      "step": 17360
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.8431610247026533e-05,
+      "loss": 0.4069,
+      "step": 17370
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.8424748398902104e-05,
+      "loss": 0.4967,
+      "step": 17380
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.841788655077768e-05,
+      "loss": 0.3683,
+      "step": 17390
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.8411024702653247e-05,
+      "loss": 0.4104,
+      "step": 17400
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.840416285452882e-05,
+      "loss": 0.415,
+      "step": 17410
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8397301006404393e-05,
+      "loss": 0.3159,
+      "step": 17420
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8390439158279964e-05,
+      "loss": 0.3804,
+      "step": 17430
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8383577310155535e-05,
+      "loss": 0.2718,
+      "step": 17440
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.837671546203111e-05,
+      "loss": 0.4164,
+      "step": 17450
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.8369853613906678e-05,
+      "loss": 0.3319,
+      "step": 17460
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.8362991765782253e-05,
+      "loss": 0.3105,
+      "step": 17470
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.8356129917657824e-05,
+      "loss": 0.4299,
+      "step": 17480
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.8349268069533395e-05,
+      "loss": 0.4007,
+      "step": 17490
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.8342406221408967e-05,
+      "loss": 0.3969,
+      "step": 17500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.833554437328454e-05,
+      "loss": 0.5271,
+      "step": 17510
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.832868252516011e-05,
+      "loss": 0.3948,
+      "step": 17520
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.8321820677035684e-05,
+      "loss": 0.3802,
+      "step": 17530
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.8314958828911255e-05,
+      "loss": 0.3396,
+      "step": 17540
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.8308096980786826e-05,
+      "loss": 0.3371,
+      "step": 17550
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.8301235132662398e-05,
+      "loss": 0.338,
+      "step": 17560
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.8294373284537972e-05,
+      "loss": 0.4585,
+      "step": 17570
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.828751143641354e-05,
+      "loss": 0.3508,
+      "step": 17580
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.8280649588289115e-05,
+      "loss": 0.4639,
+      "step": 17590
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.8273787740164686e-05,
+      "loss": 0.4571,
+      "step": 17600
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.8266925892040258e-05,
+      "loss": 0.3066,
+      "step": 17610
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.826006404391583e-05,
+      "loss": 0.3944,
+      "step": 17620
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.82532021957914e-05,
+      "loss": 0.4384,
+      "step": 17630
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.824634034766697e-05,
+      "loss": 0.4364,
+      "step": 17640
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.8239478499542543e-05,
+      "loss": 0.352,
+      "step": 17650
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.8232616651418114e-05,
+      "loss": 0.3311,
+      "step": 17660
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8225754803293685e-05,
+      "loss": 0.3804,
+      "step": 17670
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.821889295516926e-05,
+      "loss": 0.421,
+      "step": 17680
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.5233563184738159,
+      "eval_runtime": 304.3702,
+      "eval_samples_per_second": 6.456,
+      "eval_steps_per_second": 1.616,
+      "eval_wer": 0.2521717700696199,
+      "step": 17688
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8212031107044828e-05,
+      "loss": 0.5453,
+      "step": 17690
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8205169258920403e-05,
+      "loss": 0.2927,
+      "step": 17700
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8198307410795974e-05,
+      "loss": 0.3392,
+      "step": 17710
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.8191445562671545e-05,
+      "loss": 0.3916,
+      "step": 17720
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.8184583714547117e-05,
+      "loss": 0.3198,
+      "step": 17730
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.817772186642269e-05,
+      "loss": 0.3473,
+      "step": 17740
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.817086001829826e-05,
+      "loss": 0.3119,
+      "step": 17750
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.8163998170173834e-05,
+      "loss": 0.3457,
+      "step": 17760
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.8157136322049405e-05,
+      "loss": 0.3063,
+      "step": 17770
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.8150274473924977e-05,
+      "loss": 0.4029,
+      "step": 17780
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.8143412625800548e-05,
+      "loss": 0.3472,
+      "step": 17790
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8136550777676123e-05,
+      "loss": 0.3133,
+      "step": 17800
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.812968892955169e-05,
+      "loss": 0.2918,
+      "step": 17810
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8122827081427265e-05,
+      "loss": 0.3515,
+      "step": 17820
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8115965233302837e-05,
+      "loss": 0.3123,
+      "step": 17830
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8109103385178408e-05,
+      "loss": 0.2834,
+      "step": 17840
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.810224153705398e-05,
+      "loss": 0.3306,
+      "step": 17850
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8095379688929554e-05,
+      "loss": 0.3205,
+      "step": 17860
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8088517840805122e-05,
+      "loss": 0.4033,
+      "step": 17870
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8081655992680696e-05,
+      "loss": 0.2648,
+      "step": 17880
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8074794144556268e-05,
+      "loss": 0.4834,
+      "step": 17890
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.806793229643184e-05,
+      "loss": 0.3235,
+      "step": 17900
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.806107044830741e-05,
+      "loss": 0.3425,
+      "step": 17910
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8054208600182985e-05,
+      "loss": 0.2956,
+      "step": 17920
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8047346752058553e-05,
+      "loss": 0.2976,
+      "step": 17930
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8040484903934128e-05,
+      "loss": 0.2933,
+      "step": 17940
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.80336230558097e-05,
+      "loss": 0.2794,
+      "step": 17950
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.802676120768527e-05,
+      "loss": 0.3067,
+      "step": 17960
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.801989935956084e-05,
+      "loss": 0.2847,
+      "step": 17970
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.8013037511436416e-05,
+      "loss": 0.5393,
+      "step": 17980
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.8006175663311984e-05,
+      "loss": 0.4717,
+      "step": 17990
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.799931381518756e-05,
+      "loss": 0.2925,
+      "step": 18000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.799245196706313e-05,
+      "loss": 0.3735,
+      "step": 18010
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.79855901189387e-05,
+      "loss": 0.2953,
+      "step": 18020
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7978728270814273e-05,
+      "loss": 0.3207,
+      "step": 18030
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7971866422689844e-05,
+      "loss": 0.4612,
+      "step": 18040
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7965004574565415e-05,
+      "loss": 0.3412,
+      "step": 18050
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7958142726440987e-05,
+      "loss": 0.2675,
+      "step": 18060
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.795128087831656e-05,
+      "loss": 0.2972,
+      "step": 18070
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.794441903019213e-05,
+      "loss": 0.329,
+      "step": 18080
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7937557182067704e-05,
+      "loss": 0.2705,
+      "step": 18090
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7930695333943275e-05,
+      "loss": 0.2926,
+      "step": 18100
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7923833485818847e-05,
+      "loss": 0.3243,
+      "step": 18110
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7916971637694418e-05,
+      "loss": 0.3524,
+      "step": 18120
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7910109789569993e-05,
+      "loss": 0.3205,
+      "step": 18130
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.790324794144556e-05,
+      "loss": 0.3633,
+      "step": 18140
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7896386093321135e-05,
+      "loss": 0.333,
+      "step": 18150
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7889524245196707e-05,
+      "loss": 0.2775,
+      "step": 18160
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7882662397072278e-05,
+      "loss": 0.317,
+      "step": 18170
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.787580054894785e-05,
+      "loss": 0.3322,
+      "step": 18180
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7868938700823424e-05,
+      "loss": 0.3304,
+      "step": 18190
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7862076852698992e-05,
+      "loss": 0.2732,
+      "step": 18200
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7855215004574567e-05,
+      "loss": 0.2908,
+      "step": 18210
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7848353156450138e-05,
+      "loss": 0.326,
+      "step": 18220
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.784149130832571e-05,
+      "loss": 0.2926,
+      "step": 18230
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.783462946020128e-05,
+      "loss": 0.3605,
+      "step": 18240
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7827767612076855e-05,
+      "loss": 0.3172,
+      "step": 18250
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7820905763952423e-05,
+      "loss": 0.3292,
+      "step": 18260
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7814043915827998e-05,
+      "loss": 0.3032,
+      "step": 18270
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.780718206770357e-05,
+      "loss": 0.3825,
+      "step": 18280
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.780032021957914e-05,
+      "loss": 0.3394,
+      "step": 18290
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7793458371454712e-05,
+      "loss": 0.3262,
+      "step": 18300
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7786596523330286e-05,
+      "loss": 0.2628,
+      "step": 18310
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7779734675205854e-05,
+      "loss": 0.3642,
+      "step": 18320
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.777287282708143e-05,
+      "loss": 0.3435,
+      "step": 18330
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7766010978957e-05,
+      "loss": 0.3547,
+      "step": 18340
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.775914913083257e-05,
+      "loss": 0.3128,
+      "step": 18350
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7752287282708143e-05,
+      "loss": 0.3998,
+      "step": 18360
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7745425434583718e-05,
+      "loss": 0.3131,
+      "step": 18370
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7738563586459286e-05,
+      "loss": 0.4655,
+      "step": 18380
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.773170173833486e-05,
+      "loss": 0.2992,
+      "step": 18390
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.772483989021043e-05,
+      "loss": 0.2968,
+      "step": 18400
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7717978042086003e-05,
+      "loss": 0.2693,
+      "step": 18410
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7711116193961574e-05,
+      "loss": 0.3015,
+      "step": 18420
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.770425434583715e-05,
+      "loss": 0.3748,
+      "step": 18430
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7697392497712717e-05,
+      "loss": 0.7822,
+      "step": 18440
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7690530649588288e-05,
+      "loss": 0.2598,
+      "step": 18450
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7683668801463863e-05,
+      "loss": 0.3495,
+      "step": 18460
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.767680695333943e-05,
+      "loss": 0.4256,
+      "step": 18470
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7669945105215005e-05,
+      "loss": 0.2942,
+      "step": 18480
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7663083257090577e-05,
+      "loss": 0.2481,
+      "step": 18490
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7656221408966148e-05,
+      "loss": 0.3022,
+      "step": 18500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.764935956084172e-05,
+      "loss": 0.3818,
+      "step": 18510
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7643183897529736e-05,
+      "loss": 0.344,
+      "step": 18520
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7636322049405307e-05,
+      "loss": 0.3306,
+      "step": 18530
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.762946020128088e-05,
+      "loss": 0.3032,
+      "step": 18540
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.762259835315645e-05,
+      "loss": 0.3148,
+      "step": 18550
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7615736505032024e-05,
+      "loss": 0.3458,
+      "step": 18560
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7608874656907595e-05,
+      "loss": 0.2581,
+      "step": 18570
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7602012808783163e-05,
+      "loss": 0.2838,
+      "step": 18580
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7595150960658738e-05,
+      "loss": 0.3244,
+      "step": 18590
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.758828911253431e-05,
+      "loss": 0.2244,
+      "step": 18600
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.758142726440988e-05,
+      "loss": 0.3354,
+      "step": 18610
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7574565416285452e-05,
+      "loss": 0.253,
+      "step": 18620
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7567703568161027e-05,
+      "loss": 0.2981,
+      "step": 18630
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7560841720036595e-05,
+      "loss": 0.2593,
+      "step": 18640
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.755397987191217e-05,
+      "loss": 0.3724,
+      "step": 18650
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.754711802378774e-05,
+      "loss": 0.335,
+      "step": 18660
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7540256175663312e-05,
+      "loss": 0.2454,
+      "step": 18670
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7533394327538883e-05,
+      "loss": 0.2865,
+      "step": 18680
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7526532479414458e-05,
+      "loss": 0.5088,
+      "step": 18690
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7519670631290026e-05,
+      "loss": 0.3922,
+      "step": 18700
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.75128087831656e-05,
+      "loss": 0.3017,
+      "step": 18710
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7505946935041172e-05,
+      "loss": 0.3115,
+      "step": 18720
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7499085086916743e-05,
+      "loss": 0.3404,
+      "step": 18730
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7492223238792314e-05,
+      "loss": 0.302,
+      "step": 18740
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.748536139066789e-05,
+      "loss": 0.344,
+      "step": 18750
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7478499542543457e-05,
+      "loss": 0.2813,
+      "step": 18760
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7471637694419032e-05,
+      "loss": 0.2784,
+      "step": 18770
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7464775846294603e-05,
+      "loss": 0.2694,
+      "step": 18780
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7457913998170174e-05,
+      "loss": 0.3065,
+      "step": 18790
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7451052150045746e-05,
+      "loss": 0.2433,
+      "step": 18800
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.744419030192132e-05,
+      "loss": 0.2461,
+      "step": 18810
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7437328453796888e-05,
+      "loss": 0.291,
+      "step": 18820
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7430466605672463e-05,
+      "loss": 0.638,
+      "step": 18830
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7423604757548034e-05,
+      "loss": 0.4554,
+      "step": 18840
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7416742909423606e-05,
+      "loss": 0.2864,
+      "step": 18850
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7409881061299177e-05,
+      "loss": 0.3634,
+      "step": 18860
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.740301921317475e-05,
+      "loss": 0.3347,
+      "step": 18870
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.739615736505032e-05,
+      "loss": 0.3253,
+      "step": 18880
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7389295516925894e-05,
+      "loss": 0.2845,
+      "step": 18890
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7382433668801465e-05,
+      "loss": 0.4574,
+      "step": 18900
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7375571820677037e-05,
+      "loss": 0.3239,
+      "step": 18910
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7368709972552608e-05,
+      "loss": 0.3669,
+      "step": 18920
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7361848124428183e-05,
+      "loss": 0.3177,
+      "step": 18930
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.735498627630375e-05,
+      "loss": 0.3034,
+      "step": 18940
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7348124428179325e-05,
+      "loss": 0.307,
+      "step": 18950
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7341262580054897e-05,
+      "loss": 0.3091,
+      "step": 18960
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7334400731930468e-05,
+      "loss": 0.3768,
+      "step": 18970
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.732753888380604e-05,
+      "loss": 0.3127,
+      "step": 18980
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.732067703568161e-05,
+      "loss": 0.3006,
+      "step": 18990
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7313815187557182e-05,
+      "loss": 0.3845,
+      "step": 19000
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7306953339432753e-05,
+      "loss": 0.2394,
+      "step": 19010
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7300091491308328e-05,
+      "loss": 0.3306,
+      "step": 19020
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7293229643183896e-05,
+      "loss": 0.3833,
+      "step": 19030
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.728636779505947e-05,
+      "loss": 0.3264,
+      "step": 19040
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7279505946935042e-05,
+      "loss": 0.345,
+      "step": 19050
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7272644098810613e-05,
+      "loss": 0.3864,
+      "step": 19060
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7265782250686184e-05,
+      "loss": 0.4164,
+      "step": 19070
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.725892040256176e-05,
+      "loss": 0.3749,
+      "step": 19080
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7252058554437327e-05,
+      "loss": 0.3643,
+      "step": 19090
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7245196706312902e-05,
+      "loss": 0.4085,
+      "step": 19100
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7238334858188473e-05,
+      "loss": 0.3778,
+      "step": 19110
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7231473010064044e-05,
+      "loss": 0.4101,
+      "step": 19120
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7224611161939616e-05,
+      "loss": 0.5395,
+      "step": 19130
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7217749313815187e-05,
+      "loss": 0.3942,
+      "step": 19140
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.721088746569076e-05,
+      "loss": 0.3107,
+      "step": 19150
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7204025617566333e-05,
+      "loss": 0.3429,
+      "step": 19160
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.71971637694419e-05,
+      "loss": 0.2475,
+      "step": 19170
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7190301921317476e-05,
+      "loss": 0.2452,
+      "step": 19180
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7183440073193047e-05,
+      "loss": 0.3901,
+      "step": 19190
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7176578225068618e-05,
+      "loss": 0.3631,
+      "step": 19200
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.716971637694419e-05,
+      "loss": 0.3111,
+      "step": 19210
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.7162854528819764e-05,
+      "loss": 0.4351,
+      "step": 19220
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.7155992680695332e-05,
+      "loss": 0.3133,
+      "step": 19230
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.7149130832570907e-05,
+      "loss": 0.5005,
+      "step": 19240
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.7142268984446478e-05,
+      "loss": 0.2807,
+      "step": 19250
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.713540713632205e-05,
+      "loss": 0.3228,
+      "step": 19260
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.712854528819762e-05,
+      "loss": 0.3045,
+      "step": 19270
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.7121683440073195e-05,
+      "loss": 0.3291,
+      "step": 19280
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.7114821591948763e-05,
+      "loss": 0.382,
+      "step": 19290
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.7107959743824338e-05,
+      "loss": 0.2573,
+      "step": 19300
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.710109789569991e-05,
+      "loss": 0.3371,
+      "step": 19310
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.709423604757548e-05,
+      "loss": 0.2756,
+      "step": 19320
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.7087374199451052e-05,
+      "loss": 0.2761,
+      "step": 19330
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.7080512351326627e-05,
+      "loss": 0.3508,
+      "step": 19340
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7073650503202195e-05,
+      "loss": 0.301,
+      "step": 19350
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.706678865507777e-05,
+      "loss": 0.4034,
+      "step": 19360
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.705992680695334e-05,
+      "loss": 0.3371,
+      "step": 19370
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.705306495882891e-05,
+      "loss": 0.3171,
+      "step": 19380
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7046203110704483e-05,
+      "loss": 0.271,
+      "step": 19390
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7039341262580055e-05,
+      "loss": 0.3784,
+      "step": 19400
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7032479414455626e-05,
+      "loss": 0.3063,
+      "step": 19410
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7025617566331197e-05,
+      "loss": 0.3392,
+      "step": 19420
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7018755718206772e-05,
+      "loss": 0.3373,
+      "step": 19430
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.701189387008234e-05,
+      "loss": 0.3332,
+      "step": 19440
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.7005032021957914e-05,
+      "loss": 0.2706,
+      "step": 19450
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6998170173833486e-05,
+      "loss": 0.3493,
+      "step": 19460
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6991308325709057e-05,
+      "loss": 0.3459,
+      "step": 19470
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.698444647758463e-05,
+      "loss": 0.3389,
+      "step": 19480
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6977584629460203e-05,
+      "loss": 0.2906,
+      "step": 19490
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.697072278133577e-05,
+      "loss": 0.3688,
+      "step": 19500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6963860933211346e-05,
+      "loss": 0.3414,
+      "step": 19510
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6956999085086917e-05,
+      "loss": 0.2897,
+      "step": 19520
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.695013723696249e-05,
+      "loss": 0.4392,
+      "step": 19530
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.694327538883806e-05,
+      "loss": 0.4308,
+      "step": 19540
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6936413540713634e-05,
+      "loss": 0.3869,
+      "step": 19550
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6929551692589202e-05,
+      "loss": 0.2993,
+      "step": 19560
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6922689844464777e-05,
+      "loss": 0.2599,
+      "step": 19570
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6915827996340348e-05,
+      "loss": 0.3331,
+      "step": 19580
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.690896614821592e-05,
+      "loss": 0.3341,
+      "step": 19590
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.690210430009149e-05,
+      "loss": 0.2193,
+      "step": 19600
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6895242451967066e-05,
+      "loss": 0.2924,
+      "step": 19610
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6888380603842633e-05,
+      "loss": 0.2944,
+      "step": 19620
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6881518755718208e-05,
+      "loss": 0.2586,
+      "step": 19630
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.687465690759378e-05,
+      "loss": 0.3596,
+      "step": 19640
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.686779505946935e-05,
+      "loss": 0.3253,
+      "step": 19650
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6860933211344922e-05,
+      "loss": 0.301,
+      "step": 19660
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6854071363220497e-05,
+      "loss": 0.3745,
+      "step": 19670
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6847209515096065e-05,
+      "loss": 0.2986,
+      "step": 19680
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.684034766697164e-05,
+      "loss": 0.393,
+      "step": 19690
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.683348581884721e-05,
+      "loss": 0.3419,
+      "step": 19700
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6826623970722782e-05,
+      "loss": 0.2334,
+      "step": 19710
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6819762122598353e-05,
+      "loss": 0.3187,
+      "step": 19720
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6812900274473928e-05,
+      "loss": 0.3064,
+      "step": 19730
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6806038426349496e-05,
+      "loss": 0.313,
+      "step": 19740
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.679917657822507e-05,
+      "loss": 0.2861,
+      "step": 19750
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6792314730100642e-05,
+      "loss": 0.3406,
+      "step": 19760
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6785452881976213e-05,
+      "loss": 0.3475,
+      "step": 19770
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6778591033851785e-05,
+      "loss": 0.2726,
+      "step": 19780
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6771729185727356e-05,
+      "loss": 0.4028,
+      "step": 19790
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6764867337602927e-05,
+      "loss": 0.234,
+      "step": 19800
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.67580054894785e-05,
+      "loss": 0.2911,
+      "step": 19810
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6751143641354073e-05,
+      "loss": 0.3906,
+      "step": 19820
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.674428179322964e-05,
+      "loss": 0.3624,
+      "step": 19830
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6737419945105216e-05,
+      "loss": 0.2336,
+      "step": 19840
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6730558096980787e-05,
+      "loss": 0.2887,
+      "step": 19850
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.672369624885636e-05,
+      "loss": 0.3374,
+      "step": 19860
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.671683440073193e-05,
+      "loss": 0.2989,
+      "step": 19870
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6709972552607504e-05,
+      "loss": 0.238,
+      "step": 19880
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6703110704483072e-05,
+      "loss": 0.3462,
+      "step": 19890
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6696248856358647e-05,
+      "loss": 0.2656,
+      "step": 19900
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6689387008234218e-05,
+      "loss": 0.2825,
+      "step": 19910
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.668252516010979e-05,
+      "loss": 0.3876,
+      "step": 19920
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.667566331198536e-05,
+      "loss": 0.2631,
+      "step": 19930
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6668801463860936e-05,
+      "loss": 0.2411,
+      "step": 19940
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6661939615736504e-05,
+      "loss": 0.3137,
+      "step": 19950
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6655077767612078e-05,
+      "loss": 0.3442,
+      "step": 19960
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.664821591948765e-05,
+      "loss": 0.31,
+      "step": 19970
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.664135407136322e-05,
+      "loss": 0.3342,
+      "step": 19980
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.6634492223238792e-05,
+      "loss": 0.3847,
+      "step": 19990
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.6627630375114367e-05,
+      "loss": 0.3553,
+      "step": 20000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6620768526989935e-05,
+      "loss": 0.3186,
+      "step": 20010
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.661390667886551e-05,
+      "loss": 0.3855,
+      "step": 20020
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.660704483074108e-05,
+      "loss": 0.3371,
+      "step": 20030
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6600182982616652e-05,
+      "loss": 0.305,
+      "step": 20040
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6593321134492223e-05,
+      "loss": 0.3048,
+      "step": 20050
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6586459286367798e-05,
+      "loss": 0.384,
+      "step": 20060
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6579597438243366e-05,
+      "loss": 0.3476,
+      "step": 20070
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.657273559011894e-05,
+      "loss": 0.3314,
+      "step": 20080
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6565873741994512e-05,
+      "loss": 0.4846,
+      "step": 20090
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6559011893870083e-05,
+      "loss": 0.3786,
+      "step": 20100
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6552150045745655e-05,
+      "loss": 0.2958,
+      "step": 20110
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.654528819762123e-05,
+      "loss": 0.3325,
+      "step": 20120
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6538426349496797e-05,
+      "loss": 0.2837,
+      "step": 20130
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6531564501372372e-05,
+      "loss": 0.2231,
+      "step": 20140
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6524702653247943e-05,
+      "loss": 0.2246,
+      "step": 20150
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6517840805123515e-05,
+      "loss": 0.3621,
+      "step": 20160
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6510978956999086e-05,
+      "loss": 0.2959,
+      "step": 20170
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.650411710887466e-05,
+      "loss": 0.3282,
+      "step": 20180
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.649725526075023e-05,
+      "loss": 0.3973,
+      "step": 20190
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.64903934126258e-05,
+      "loss": 0.3326,
+      "step": 20200
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6483531564501374e-05,
+      "loss": 0.2392,
+      "step": 20210
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6476669716376942e-05,
+      "loss": 0.3905,
+      "step": 20220
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6469807868252517e-05,
+      "loss": 0.238,
+      "step": 20230
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.646294602012809e-05,
+      "loss": 0.2746,
+      "step": 20240
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.645608417200366e-05,
+      "loss": 0.3567,
+      "step": 20250
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.644922232387923e-05,
+      "loss": 0.3556,
+      "step": 20260
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6442360475754806e-05,
+      "loss": 0.3104,
+      "step": 20270
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6435498627630374e-05,
+      "loss": 0.2731,
+      "step": 20280
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6428636779505948e-05,
+      "loss": 0.2696,
+      "step": 20290
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.642177493138152e-05,
+      "loss": 0.3393,
+      "step": 20300
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.641491308325709e-05,
+      "loss": 0.3235,
+      "step": 20310
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6408051235132662e-05,
+      "loss": 0.3441,
+      "step": 20320
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6401189387008237e-05,
+      "loss": 0.3094,
+      "step": 20330
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6394327538883805e-05,
+      "loss": 0.3419,
+      "step": 20340
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.638746569075938e-05,
+      "loss": 0.411,
+      "step": 20350
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.638060384263495e-05,
+      "loss": 0.2794,
+      "step": 20360
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6373741994510522e-05,
+      "loss": 0.316,
+      "step": 20370
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6366880146386093e-05,
+      "loss": 0.2331,
+      "step": 20380
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6360018298261665e-05,
+      "loss": 0.2986,
+      "step": 20390
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6353156450137236e-05,
+      "loss": 0.3148,
+      "step": 20400
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.634629460201281e-05,
+      "loss": 0.2906,
+      "step": 20410
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.633943275388838e-05,
+      "loss": 0.3955,
+      "step": 20420
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6332570905763953e-05,
+      "loss": 0.2363,
+      "step": 20430
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6325709057639525e-05,
+      "loss": 1.3877,
+      "step": 20440
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6318847209515096e-05,
+      "loss": 0.3234,
+      "step": 20450
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6311985361390667e-05,
+      "loss": 0.268,
+      "step": 20460
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6305123513266242e-05,
+      "loss": 0.2665,
+      "step": 20470
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.629826166514181e-05,
+      "loss": 0.3164,
+      "step": 20480
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6291399817017385e-05,
+      "loss": 0.2639,
+      "step": 20490
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6284537968892956e-05,
+      "loss": 0.2324,
+      "step": 20500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6277676120768527e-05,
+      "loss": 0.3857,
+      "step": 20510
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.62708142726441e-05,
+      "loss": 0.4068,
+      "step": 20520
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6263952424519673e-05,
+      "loss": 0.4079,
+      "step": 20530
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.625709057639524e-05,
+      "loss": 0.2822,
+      "step": 20540
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6250228728270816e-05,
+      "loss": 0.2945,
+      "step": 20550
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6243366880146387e-05,
+      "loss": 0.3271,
+      "step": 20560
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.623650503202196e-05,
+      "loss": 0.4126,
+      "step": 20570
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.622964318389753e-05,
+      "loss": 0.5001,
+      "step": 20580
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6222781335773104e-05,
+      "loss": 0.3258,
+      "step": 20590
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6215919487648672e-05,
+      "loss": 0.3585,
+      "step": 20600
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6209057639524244e-05,
+      "loss": 0.3048,
+      "step": 20610
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.620219579139982e-05,
+      "loss": 0.2748,
+      "step": 20620
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6195333943275386e-05,
+      "loss": 0.2479,
+      "step": 20630
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.618847209515096e-05,
+      "loss": 0.3416,
+      "step": 20640
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6181610247026532e-05,
+      "loss": 0.3122,
+      "step": 20650
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6174748398902104e-05,
+      "loss": 0.3835,
+      "step": 20660
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6167886550777675e-05,
+      "loss": 0.3111,
+      "step": 20670
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.616102470265325e-05,
+      "loss": 0.3067,
+      "step": 20680
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.6154162854528817e-05,
+      "loss": 0.3034,
+      "step": 20690
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.6147301006404392e-05,
+      "loss": 0.3865,
+      "step": 20700
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.6140439158279963e-05,
+      "loss": 0.3181,
+      "step": 20710
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6133577310155535e-05,
+      "loss": 0.3732,
+      "step": 20720
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6126715462031106e-05,
+      "loss": 0.3449,
+      "step": 20730
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.611985361390668e-05,
+      "loss": 0.3052,
+      "step": 20740
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.611299176578225e-05,
+      "loss": 0.2483,
+      "step": 20750
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6106129917657823e-05,
+      "loss": 0.3364,
+      "step": 20760
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6099268069533395e-05,
+      "loss": 0.3005,
+      "step": 20770
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6092406221408966e-05,
+      "loss": 0.3143,
+      "step": 20780
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6085544373284537e-05,
+      "loss": 0.2662,
+      "step": 20790
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6078682525160112e-05,
+      "loss": 0.2497,
+      "step": 20800
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.607182067703568e-05,
+      "loss": 0.4175,
+      "step": 20810
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6064958828911255e-05,
+      "loss": 0.3143,
+      "step": 20820
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6058096980786826e-05,
+      "loss": 0.2849,
+      "step": 20830
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6051235132662397e-05,
+      "loss": 0.2209,
+      "step": 20840
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.604437328453797e-05,
+      "loss": 0.3223,
+      "step": 20850
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6037511436413543e-05,
+      "loss": 0.2816,
+      "step": 20860
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.603064958828911e-05,
+      "loss": 0.3119,
+      "step": 20870
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6023787740164686e-05,
+      "loss": 0.3671,
+      "step": 20880
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6016925892040257e-05,
+      "loss": 0.3052,
+      "step": 20890
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.601006404391583e-05,
+      "loss": 0.4074,
+      "step": 20900
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.60032021957914e-05,
+      "loss": 0.3133,
+      "step": 20910
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5996340347666974e-05,
+      "loss": 0.2607,
+      "step": 20920
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5989478499542542e-05,
+      "loss": 0.2751,
+      "step": 20930
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.5982616651418117e-05,
+      "loss": 0.3539,
+      "step": 20940
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.597575480329369e-05,
+      "loss": 0.2798,
+      "step": 20950
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.596889295516926e-05,
+      "loss": 0.2772,
+      "step": 20960
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.596203110704483e-05,
+      "loss": 0.3907,
+      "step": 20970
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.5955169258920406e-05,
+      "loss": 0.2858,
+      "step": 20980
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5948307410795974e-05,
+      "loss": 0.3289,
+      "step": 20990
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5941445562671545e-05,
+      "loss": 0.2612,
+      "step": 21000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.593458371454712e-05,
+      "loss": 0.2732,
+      "step": 21010
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5927721866422688e-05,
+      "loss": 0.2935,
+      "step": 21020
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5920860018298262e-05,
+      "loss": 0.3668,
+      "step": 21030
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5913998170173834e-05,
+      "loss": 0.3081,
+      "step": 21040
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5907136322049405e-05,
+      "loss": 0.2766,
+      "step": 21050
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5900274473924976e-05,
+      "loss": 0.2636,
+      "step": 21060
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.589341262580055e-05,
+      "loss": 0.2728,
+      "step": 21070
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.588655077767612e-05,
+      "loss": 0.3511,
+      "step": 21080
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5879688929551693e-05,
+      "loss": 0.2988,
+      "step": 21090
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5872827081427265e-05,
+      "loss": 0.44,
+      "step": 21100
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5865965233302836e-05,
+      "loss": 0.2759,
+      "step": 21110
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5859103385178407e-05,
+      "loss": 0.338,
+      "step": 21120
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5852241537053982e-05,
+      "loss": 0.3357,
+      "step": 21130
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.584537968892955e-05,
+      "loss": 0.4261,
+      "step": 21140
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5838517840805125e-05,
+      "loss": 0.3245,
+      "step": 21150
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5831655992680696e-05,
+      "loss": 0.4086,
+      "step": 21160
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5824794144556267e-05,
+      "loss": 0.3233,
+      "step": 21170
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.581793229643184e-05,
+      "loss": 0.2743,
+      "step": 21180
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5811070448307413e-05,
+      "loss": 0.3543,
+      "step": 21190
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.580420860018298e-05,
+      "loss": 0.2789,
+      "step": 21200
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5797346752058556e-05,
+      "loss": 0.2897,
+      "step": 21210
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5790484903934127e-05,
+      "loss": 0.2737,
+      "step": 21220
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.57836230558097e-05,
+      "loss": 0.3039,
+      "step": 21230
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.577676120768527e-05,
+      "loss": 0.3771,
+      "step": 21240
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5769899359560845e-05,
+      "loss": 0.238,
+      "step": 21250
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5763037511436412e-05,
+      "loss": 0.4485,
+      "step": 21260
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5756175663311987e-05,
+      "loss": 0.2433,
+      "step": 21270
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.574931381518756e-05,
+      "loss": 0.2657,
+      "step": 21280
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.574245196706313e-05,
+      "loss": 0.2549,
+      "step": 21290
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.57355901189387e-05,
+      "loss": 0.5503,
+      "step": 21300
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5728728270814276e-05,
+      "loss": 0.2801,
+      "step": 21310
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5721866422689844e-05,
+      "loss": 0.2686,
+      "step": 21320
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.571500457456542e-05,
+      "loss": 0.3172,
+      "step": 21330
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.570814272644099e-05,
+      "loss": 0.4264,
+      "step": 21340
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.570128087831656e-05,
+      "loss": 0.389,
+      "step": 21350
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5694419030192132e-05,
+      "loss": 0.3565,
+      "step": 21360
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5687557182067707e-05,
+      "loss": 0.3073,
+      "step": 21370
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5680695333943275e-05,
+      "loss": 0.2694,
+      "step": 21380
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.567383348581885e-05,
+      "loss": 0.3077,
+      "step": 21390
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.566697163769442e-05,
+      "loss": 0.3554,
+      "step": 21400
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.566010978956999e-05,
+      "loss": 0.3595,
+      "step": 21410
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5653247941445564e-05,
+      "loss": 0.2793,
+      "step": 21420
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5646386093321135e-05,
+      "loss": 0.2241,
+      "step": 21430
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5639524245196706e-05,
+      "loss": 0.2953,
+      "step": 21440
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5632662397072277e-05,
+      "loss": 0.3126,
+      "step": 21450
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5625800548947852e-05,
+      "loss": 0.3017,
+      "step": 21460
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.561893870082342e-05,
+      "loss": 0.3509,
+      "step": 21470
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.5612076852698995e-05,
+      "loss": 0.2922,
+      "step": 21480
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.5605215004574566e-05,
+      "loss": 0.374,
+      "step": 21490
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.5598353156450137e-05,
+      "loss": 0.3895,
+      "step": 21500
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.559149130832571e-05,
+      "loss": 0.3348,
+      "step": 21510
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5584629460201283e-05,
+      "loss": 0.3259,
+      "step": 21520
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.557776761207685e-05,
+      "loss": 0.6164,
+      "step": 21530
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5570905763952426e-05,
+      "loss": 0.3481,
+      "step": 21540
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5564043915827997e-05,
+      "loss": 0.3528,
+      "step": 21550
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.555718206770357e-05,
+      "loss": 0.3416,
+      "step": 21560
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.555032021957914e-05,
+      "loss": 0.2711,
+      "step": 21570
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5543458371454715e-05,
+      "loss": 0.2613,
+      "step": 21580
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5536596523330283e-05,
+      "loss": 0.3429,
+      "step": 21590
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5529734675205857e-05,
+      "loss": 0.3628,
+      "step": 21600
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.552287282708143e-05,
+      "loss": 0.2927,
+      "step": 21610
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5516010978957e-05,
+      "loss": 0.2716,
+      "step": 21620
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.550914913083257e-05,
+      "loss": 0.3721,
+      "step": 21630
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5502287282708142e-05,
+      "loss": 0.3009,
+      "step": 21640
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5495425434583714e-05,
+      "loss": 0.3056,
+      "step": 21650
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.548856358645929e-05,
+      "loss": 0.2728,
+      "step": 21660
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5481701738334856e-05,
+      "loss": 0.319,
+      "step": 21670
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.547483989021043e-05,
+      "loss": 0.3002,
+      "step": 21680
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5467978042086002e-05,
+      "loss": 0.3812,
+      "step": 21690
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5461116193961574e-05,
+      "loss": 0.351,
+      "step": 21700
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5454254345837145e-05,
+      "loss": 0.2987,
+      "step": 21710
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.544739249771272e-05,
+      "loss": 0.3318,
+      "step": 21720
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5440530649588288e-05,
+      "loss": 0.2329,
+      "step": 21730
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5433668801463862e-05,
+      "loss": 0.3341,
+      "step": 21740
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5426806953339434e-05,
+      "loss": 0.3333,
+      "step": 21750
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5419945105215005e-05,
+      "loss": 0.4045,
+      "step": 21760
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5413083257090576e-05,
+      "loss": 0.2459,
+      "step": 21770
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.540622140896615e-05,
+      "loss": 0.2064,
+      "step": 21780
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.539935956084172e-05,
+      "loss": 0.3609,
+      "step": 21790
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5392497712717294e-05,
+      "loss": 0.258,
+      "step": 21800
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5385635864592865e-05,
+      "loss": 0.2492,
+      "step": 21810
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5378774016468433e-05,
+      "loss": 0.3917,
+      "step": 21820
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5371912168344007e-05,
+      "loss": 0.2351,
+      "step": 21830
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.536505032021958e-05,
+      "loss": 0.3251,
+      "step": 21840
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5358874656907595e-05,
+      "loss": 0.2121,
+      "step": 21850
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5352012808783166e-05,
+      "loss": 0.2808,
+      "step": 21860
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5345150960658738e-05,
+      "loss": 0.2862,
+      "step": 21870
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.533828911253431e-05,
+      "loss": 0.4142,
+      "step": 21880
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5331427264409884e-05,
+      "loss": 0.3919,
+      "step": 21890
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.532456541628545e-05,
+      "loss": 0.3544,
+      "step": 21900
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5317703568161026e-05,
+      "loss": 0.3052,
+      "step": 21910
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5310841720036597e-05,
+      "loss": 0.3186,
+      "step": 21920
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5303979871912165e-05,
+      "loss": 0.2119,
+      "step": 21930
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.529711802378774e-05,
+      "loss": 0.3137,
+      "step": 21940
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.529025617566331e-05,
+      "loss": 0.3247,
+      "step": 21950
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5283394327538883e-05,
+      "loss": 0.4101,
+      "step": 21960
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5276532479414454e-05,
+      "loss": 0.248,
+      "step": 21970
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.526967063129003e-05,
+      "loss": 0.3042,
+      "step": 21980
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.526349496797804e-05,
+      "loss": 0.3784,
+      "step": 21990
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5256633119853616e-05,
+      "loss": 0.2369,
+      "step": 22000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5249771271729186e-05,
+      "loss": 0.3899,
+      "step": 22010
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5242909423604759e-05,
+      "loss": 0.3036,
+      "step": 22020
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.523604757548033e-05,
+      "loss": 0.3017,
+      "step": 22030
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5229185727355903e-05,
+      "loss": 0.3846,
+      "step": 22040
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5222323879231473e-05,
+      "loss": 0.3105,
+      "step": 22050
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5215462031107044e-05,
+      "loss": 0.3863,
+      "step": 22060
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5208600182982617e-05,
+      "loss": 0.3769,
+      "step": 22070
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5201738334858187e-05,
+      "loss": 0.3714,
+      "step": 22080
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5194876486733761e-05,
+      "loss": 0.2282,
+      "step": 22090
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5188014638609331e-05,
+      "loss": 0.3147,
+      "step": 22100
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5181152790484904e-05,
+      "loss": 0.3957,
+      "step": 22110
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.5124755501747131,
+      "eval_runtime": 299.9679,
+      "eval_samples_per_second": 6.551,
+      "eval_steps_per_second": 1.64,
+      "eval_wer": 0.2246626825211016,
+      "step": 22110
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5174290942360475e-05,
+      "loss": 0.2208,
+      "step": 22120
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5167429094236048e-05,
+      "loss": 0.2422,
+      "step": 22130
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5160567246111618e-05,
+      "loss": 0.327,
+      "step": 22140
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5153705397987193e-05,
+      "loss": 0.2468,
+      "step": 22150
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5146843549862762e-05,
+      "loss": 0.3082,
+      "step": 22160
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5139981701738335e-05,
+      "loss": 0.2689,
+      "step": 22170
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.5133119853613907e-05,
+      "loss": 0.3262,
+      "step": 22180
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.512625800548948e-05,
+      "loss": 0.2407,
+      "step": 22190
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.5119396157365049e-05,
+      "loss": 0.2792,
+      "step": 22200
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.5112534309240624e-05,
+      "loss": 0.3211,
+      "step": 22210
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.5105672461116193e-05,
+      "loss": 0.295,
+      "step": 22220
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.5098810612991766e-05,
+      "loss": 0.2343,
+      "step": 22230
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.5091948764867338e-05,
+      "loss": 0.2613,
+      "step": 22240
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.508508691674291e-05,
+      "loss": 0.3758,
+      "step": 22250
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.507822506861848e-05,
+      "loss": 0.2862,
+      "step": 22260
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5071363220494055e-05,
+      "loss": 0.325,
+      "step": 22270
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5064501372369625e-05,
+      "loss": 0.3075,
+      "step": 22280
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5057639524245198e-05,
+      "loss": 0.345,
+      "step": 22290
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5050777676120769e-05,
+      "loss": 0.2638,
+      "step": 22300
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5043915827996342e-05,
+      "loss": 0.2805,
+      "step": 22310
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5037053979871912e-05,
+      "loss": 0.2959,
+      "step": 22320
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5030192131747486e-05,
+      "loss": 0.2754,
+      "step": 22330
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5023330283623056e-05,
+      "loss": 0.3353,
+      "step": 22340
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5016468435498629e-05,
+      "loss": 0.25,
+      "step": 22350
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.50096065873742e-05,
+      "loss": 0.2426,
+      "step": 22360
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.5002744739249773e-05,
+      "loss": 0.3323,
+      "step": 22370
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.4995882891125343e-05,
+      "loss": 0.3221,
+      "step": 22380
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.4989021043000916e-05,
+      "loss": 0.2949,
+      "step": 22390
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4982159194876487e-05,
+      "loss": 0.2585,
+      "step": 22400
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4975297346752058e-05,
+      "loss": 0.253,
+      "step": 22410
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4968435498627631e-05,
+      "loss": 0.2991,
+      "step": 22420
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4961573650503203e-05,
+      "loss": 0.2295,
+      "step": 22430
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4954711802378774e-05,
+      "loss": 0.4382,
+      "step": 22440
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.4947849954254345e-05,
+      "loss": 0.2196,
+      "step": 22450
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.4940988106129918e-05,
+      "loss": 0.2605,
+      "step": 22460
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.493412625800549e-05,
+      "loss": 0.3686,
+      "step": 22470
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.4927264409881061e-05,
+      "loss": 0.3084,
+      "step": 22480
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.4920402561756634e-05,
+      "loss": 0.2664,
+      "step": 22490
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.4913540713632205e-05,
+      "loss": 0.2617,
+      "step": 22500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.4906678865507777e-05,
+      "loss": 0.3434,
+      "step": 22510
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.489981701738335e-05,
+      "loss": 0.262,
+      "step": 22520
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.4892955169258921e-05,
+      "loss": 0.2242,
+      "step": 22530
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.4886093321134492e-05,
+      "loss": 0.3188,
+      "step": 22540
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.4879231473010065e-05,
+      "loss": 0.2857,
+      "step": 22550
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.4872369624885637e-05,
+      "loss": 0.3105,
+      "step": 22560
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.4865507776761208e-05,
+      "loss": 0.3256,
+      "step": 22570
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.485864592863678e-05,
+      "loss": 0.3354,
+      "step": 22580
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.4851784080512352e-05,
+      "loss": 0.2763,
+      "step": 22590
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.4844922232387923e-05,
+      "loss": 0.2557,
+      "step": 22600
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.4838060384263496e-05,
+      "loss": 0.3017,
+      "step": 22610
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4831198536139068e-05,
+      "loss": 0.2798,
+      "step": 22620
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4824336688014639e-05,
+      "loss": 0.2955,
+      "step": 22630
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4817474839890212e-05,
+      "loss": 0.3092,
+      "step": 22640
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4810612991765783e-05,
+      "loss": 0.4072,
+      "step": 22650
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4803751143641353e-05,
+      "loss": 0.3267,
+      "step": 22660
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4796889295516926e-05,
+      "loss": 0.27,
+      "step": 22670
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4790027447392497e-05,
+      "loss": 0.221,
+      "step": 22680
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4783165599268069e-05,
+      "loss": 0.2384,
+      "step": 22690
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4776303751143642e-05,
+      "loss": 0.2339,
+      "step": 22700
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4769441903019213e-05,
+      "loss": 0.2922,
+      "step": 22710
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4762580054894784e-05,
+      "loss": 0.2645,
+      "step": 22720
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4755718206770357e-05,
+      "loss": 0.2955,
+      "step": 22730
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4748856358645929e-05,
+      "loss": 0.3289,
+      "step": 22740
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.47419945105215e-05,
+      "loss": 0.372,
+      "step": 22750
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.4735132662397073e-05,
+      "loss": 0.2393,
+      "step": 22760
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.4728270814272644e-05,
+      "loss": 0.3707,
+      "step": 22770
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.4721408966148215e-05,
+      "loss": 0.3257,
+      "step": 22780
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.4714547118023788e-05,
+      "loss": 0.246,
+      "step": 22790
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.470768526989936e-05,
+      "loss": 0.2664,
+      "step": 22800
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.4700823421774931e-05,
+      "loss": 0.3922,
+      "step": 22810
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.4693961573650504e-05,
+      "loss": 0.3313,
+      "step": 22820
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.4687099725526075e-05,
+      "loss": 0.2703,
+      "step": 22830
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4680237877401647e-05,
+      "loss": 0.3452,
+      "step": 22840
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.467337602927722e-05,
+      "loss": 0.2631,
+      "step": 22850
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4666514181152791e-05,
+      "loss": 0.3225,
+      "step": 22860
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4659652333028362e-05,
+      "loss": 0.24,
+      "step": 22870
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4652790484903935e-05,
+      "loss": 0.3245,
+      "step": 22880
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4645928636779507e-05,
+      "loss": 0.1694,
+      "step": 22890
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4639066788655078e-05,
+      "loss": 0.2421,
+      "step": 22900
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4632204940530651e-05,
+      "loss": 0.273,
+      "step": 22910
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4625343092406222e-05,
+      "loss": 0.2551,
+      "step": 22920
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4618481244281793e-05,
+      "loss": 0.2744,
+      "step": 22930
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4611619396157366e-05,
+      "loss": 0.3506,
+      "step": 22940
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4604757548032938e-05,
+      "loss": 0.2815,
+      "step": 22950
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4597895699908509e-05,
+      "loss": 0.2783,
+      "step": 22960
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4591033851784082e-05,
+      "loss": 0.2104,
+      "step": 22970
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4584172003659653e-05,
+      "loss": 0.2731,
+      "step": 22980
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4577310155535225e-05,
+      "loss": 0.2999,
+      "step": 22990
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4570448307410798e-05,
+      "loss": 0.2318,
+      "step": 23000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4563586459286369e-05,
+      "loss": 0.2744,
+      "step": 23010
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.455672461116194e-05,
+      "loss": 0.2913,
+      "step": 23020
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.4549862763037513e-05,
+      "loss": 0.2676,
+      "step": 23030
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.4543000914913085e-05,
+      "loss": 0.2491,
+      "step": 23040
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.4536139066788656e-05,
+      "loss": 0.2467,
+      "step": 23050
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.4529277218664227e-05,
+      "loss": 0.3296,
+      "step": 23060
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.4522415370539799e-05,
+      "loss": 0.2552,
+      "step": 23070
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.451555352241537e-05,
+      "loss": 0.3747,
+      "step": 23080
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.4508691674290941e-05,
+      "loss": 0.2939,
+      "step": 23090
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.4501829826166514e-05,
+      "loss": 0.1815,
+      "step": 23100
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.4494967978042085e-05,
+      "loss": 0.2361,
+      "step": 23110
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.4488106129917657e-05,
+      "loss": 0.2483,
+      "step": 23120
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.448124428179323e-05,
+      "loss": 0.2737,
+      "step": 23130
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.4474382433668801e-05,
+      "loss": 0.2811,
+      "step": 23140
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.4467520585544372e-05,
+      "loss": 0.274,
+      "step": 23150
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.4460658737419945e-05,
+      "loss": 0.2912,
+      "step": 23160
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.4453796889295517e-05,
+      "loss": 0.3884,
+      "step": 23170
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.4446935041171088e-05,
+      "loss": 0.221,
+      "step": 23180
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.4440073193046661e-05,
+      "loss": 0.2085,
+      "step": 23190
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4433211344922232e-05,
+      "loss": 0.2954,
+      "step": 23200
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4426349496797804e-05,
+      "loss": 0.3657,
+      "step": 23210
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4419487648673377e-05,
+      "loss": 0.2711,
+      "step": 23220
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4412625800548948e-05,
+      "loss": 0.2379,
+      "step": 23230
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.440576395242452e-05,
+      "loss": 0.2958,
+      "step": 23240
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4398902104300092e-05,
+      "loss": 0.2394,
+      "step": 23250
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4392040256175664e-05,
+      "loss": 0.2201,
+      "step": 23260
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4385178408051235e-05,
+      "loss": 0.2676,
+      "step": 23270
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4378316559926808e-05,
+      "loss": 0.257,
+      "step": 23280
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.437145471180238e-05,
+      "loss": 0.2199,
+      "step": 23290
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.436459286367795e-05,
+      "loss": 0.3363,
+      "step": 23300
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4357731015553523e-05,
+      "loss": 0.2417,
+      "step": 23310
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4350869167429095e-05,
+      "loss": 0.2995,
+      "step": 23320
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4344007319304666e-05,
+      "loss": 0.2634,
+      "step": 23330
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4337145471180239e-05,
+      "loss": 0.2271,
+      "step": 23340
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.433028362305581e-05,
+      "loss": 0.3389,
+      "step": 23350
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4323421774931382e-05,
+      "loss": 0.311,
+      "step": 23360
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4316559926806955e-05,
+      "loss": 0.2676,
+      "step": 23370
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.4309698078682526e-05,
+      "loss": 0.1806,
+      "step": 23380
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.4302836230558097e-05,
+      "loss": 0.2565,
+      "step": 23390
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.429597438243367e-05,
+      "loss": 0.3406,
+      "step": 23400
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.4289112534309242e-05,
+      "loss": 0.3151,
+      "step": 23410
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.4282250686184813e-05,
+      "loss": 0.327,
+      "step": 23420
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.4275388838060386e-05,
+      "loss": 0.2328,
+      "step": 23430
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.4268526989935957e-05,
+      "loss": 0.2604,
+      "step": 23440
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.4261665141811529e-05,
+      "loss": 0.2066,
+      "step": 23450
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4254803293687102e-05,
+      "loss": 0.2994,
+      "step": 23460
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4247941445562671e-05,
+      "loss": 0.3216,
+      "step": 23470
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4241079597438242e-05,
+      "loss": 0.3044,
+      "step": 23480
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4234217749313815e-05,
+      "loss": 0.2721,
+      "step": 23490
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4227355901189387e-05,
+      "loss": 0.2316,
+      "step": 23500
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4220494053064958e-05,
+      "loss": 0.2882,
+      "step": 23510
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4213632204940531e-05,
+      "loss": 0.2426,
+      "step": 23520
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4206770356816102e-05,
+      "loss": 0.2187,
+      "step": 23530
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4199908508691674e-05,
+      "loss": 0.2359,
+      "step": 23540
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4193046660567247e-05,
+      "loss": 0.2061,
+      "step": 23550
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4186184812442818e-05,
+      "loss": 0.286,
+      "step": 23560
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.417932296431839e-05,
+      "loss": 0.2901,
+      "step": 23570
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4172461116193962e-05,
+      "loss": 0.269,
+      "step": 23580
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4165599268069534e-05,
+      "loss": 0.3128,
+      "step": 23590
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4158737419945105e-05,
+      "loss": 0.2728,
+      "step": 23600
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4151875571820678e-05,
+      "loss": 0.311,
+      "step": 23610
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.414501372369625e-05,
+      "loss": 0.2074,
+      "step": 23620
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.413815187557182e-05,
+      "loss": 0.2881,
+      "step": 23630
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4131290027447394e-05,
+      "loss": 0.5094,
+      "step": 23640
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4124428179322965e-05,
+      "loss": 0.322,
+      "step": 23650
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4117566331198536e-05,
+      "loss": 0.2567,
+      "step": 23660
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4110704483074109e-05,
+      "loss": 0.324,
+      "step": 23670
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.410384263494968e-05,
+      "loss": 0.3633,
+      "step": 23680
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.4096980786825252e-05,
+      "loss": 0.2601,
+      "step": 23690
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.4090118938700823e-05,
+      "loss": 0.2599,
+      "step": 23700
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.4083257090576396e-05,
+      "loss": 0.2355,
+      "step": 23710
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.4076395242451967e-05,
+      "loss": 0.2838,
+      "step": 23720
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.4069533394327539e-05,
+      "loss": 0.249,
+      "step": 23730
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.4062671546203112e-05,
+      "loss": 0.2473,
+      "step": 23740
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.4055809698078683e-05,
+      "loss": 0.2785,
+      "step": 23750
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.4048947849954254e-05,
+      "loss": 0.2338,
+      "step": 23760
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.4042086001829827e-05,
+      "loss": 0.2874,
+      "step": 23770
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.4035224153705399e-05,
+      "loss": 0.2757,
+      "step": 23780
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.402836230558097e-05,
+      "loss": 0.2723,
+      "step": 23790
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.4021500457456543e-05,
+      "loss": 0.2413,
+      "step": 23800
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.4014638609332114e-05,
+      "loss": 0.3381,
+      "step": 23810
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.4007776761207686e-05,
+      "loss": 0.2852,
+      "step": 23820
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.4000914913083259e-05,
+      "loss": 0.2636,
+      "step": 23830
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.399405306495883e-05,
+      "loss": 0.2566,
+      "step": 23840
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.3987191216834401e-05,
+      "loss": 0.2034,
+      "step": 23850
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3980329368709974e-05,
+      "loss": 0.3982,
+      "step": 23860
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3973467520585545e-05,
+      "loss": 0.282,
+      "step": 23870
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3966605672461115e-05,
+      "loss": 0.2715,
+      "step": 23880
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3959743824336688e-05,
+      "loss": 0.2103,
+      "step": 23890
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.395288197621226e-05,
+      "loss": 0.3184,
+      "step": 23900
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.394602012808783e-05,
+      "loss": 0.2887,
+      "step": 23910
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.3939158279963404e-05,
+      "loss": 0.3421,
+      "step": 23920
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.3932296431838975e-05,
+      "loss": 0.2307,
+      "step": 23930
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.3925434583714546e-05,
+      "loss": 0.2549,
+      "step": 23940
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.391857273559012e-05,
+      "loss": 0.2315,
+      "step": 23950
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.391171088746569e-05,
+      "loss": 0.2613,
+      "step": 23960
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.3904849039341262e-05,
+      "loss": 0.2319,
+      "step": 23970
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.3897987191216835e-05,
+      "loss": 0.3219,
+      "step": 23980
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.3891125343092406e-05,
+      "loss": 0.2519,
+      "step": 23990
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.3884263494967978e-05,
+      "loss": 0.975,
+      "step": 24000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.387740164684355e-05,
+      "loss": 0.2607,
+      "step": 24010
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.3870539798719122e-05,
+      "loss": 0.326,
+      "step": 24020
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.3863677950594693e-05,
+      "loss": 0.2285,
+      "step": 24030
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.3856816102470266e-05,
+      "loss": 0.2073,
+      "step": 24040
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.3849954254345837e-05,
+      "loss": 0.1696,
+      "step": 24050
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.3843092406221409e-05,
+      "loss": 0.3503,
+      "step": 24060
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.3836230558096982e-05,
+      "loss": 0.2535,
+      "step": 24070
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3829368709972553e-05,
+      "loss": 0.2742,
+      "step": 24080
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3822506861848124e-05,
+      "loss": 0.2546,
+      "step": 24090
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3815645013723697e-05,
+      "loss": 0.2271,
+      "step": 24100
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3808783165599269e-05,
+      "loss": 0.2973,
+      "step": 24110
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.380192131747484e-05,
+      "loss": 0.2254,
+      "step": 24120
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3795059469350413e-05,
+      "loss": 0.232,
+      "step": 24130
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3788197621225984e-05,
+      "loss": 0.2284,
+      "step": 24140
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3781335773101556e-05,
+      "loss": 0.2013,
+      "step": 24150
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3774473924977129e-05,
+      "loss": 0.2345,
+      "step": 24160
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.37676120768527e-05,
+      "loss": 0.2857,
+      "step": 24170
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.3760750228728271e-05,
+      "loss": 0.2066,
+      "step": 24180
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.3753888380603844e-05,
+      "loss": 0.2726,
+      "step": 24190
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.3747026532479416e-05,
+      "loss": 0.2312,
+      "step": 24200
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.3740164684354987e-05,
+      "loss": 0.2831,
+      "step": 24210
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.373330283623056e-05,
+      "loss": 0.2464,
+      "step": 24220
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.3726440988106131e-05,
+      "loss": 0.2695,
+      "step": 24230
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.3719579139981702e-05,
+      "loss": 0.24,
+      "step": 24240
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.3712717291857275e-05,
+      "loss": 0.3235,
+      "step": 24250
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.3705855443732847e-05,
+      "loss": 0.2352,
+      "step": 24260
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.3698993595608418e-05,
+      "loss": 0.2775,
+      "step": 24270
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.369213174748399e-05,
+      "loss": 0.2687,
+      "step": 24280
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.368526989935956e-05,
+      "loss": 0.2516,
+      "step": 24290
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.3678408051235132e-05,
+      "loss": 0.2498,
+      "step": 24300
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.3671546203110705e-05,
+      "loss": 0.2855,
+      "step": 24310
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.3664684354986276e-05,
+      "loss": 0.3311,
+      "step": 24320
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.3657822506861848e-05,
+      "loss": 0.3002,
+      "step": 24330
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.3650960658737419e-05,
+      "loss": 0.2772,
+      "step": 24340
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3644098810612992e-05,
+      "loss": 0.2758,
+      "step": 24350
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3637236962488563e-05,
+      "loss": 0.245,
+      "step": 24360
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3630375114364135e-05,
+      "loss": 0.2638,
+      "step": 24370
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3623513266239708e-05,
+      "loss": 0.24,
+      "step": 24380
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.3616651418115279e-05,
+      "loss": 0.2562,
+      "step": 24390
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.360978956999085e-05,
+      "loss": 0.2974,
+      "step": 24400
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.3602927721866423e-05,
+      "loss": 0.2994,
+      "step": 24410
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.3596065873741994e-05,
+      "loss": 0.3778,
+      "step": 24420
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.3589204025617566e-05,
+      "loss": 0.3742,
+      "step": 24430
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.3582342177493139e-05,
+      "loss": 0.2114,
+      "step": 24440
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.357548032936871e-05,
+      "loss": 0.271,
+      "step": 24450
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.3568618481244281e-05,
+      "loss": 0.3057,
+      "step": 24460
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.3561756633119854e-05,
+      "loss": 0.2297,
+      "step": 24470
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.3554894784995426e-05,
+      "loss": 0.2785,
+      "step": 24480
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.3548032936870997e-05,
+      "loss": 0.1941,
+      "step": 24490
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.354117108874657e-05,
+      "loss": 0.3101,
+      "step": 24500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.3534309240622141e-05,
+      "loss": 0.3719,
+      "step": 24510
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.3527447392497713e-05,
+      "loss": 0.3381,
+      "step": 24520
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.3520585544373286e-05,
+      "loss": 0.2038,
+      "step": 24530
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.3513723696248857e-05,
+      "loss": 0.3289,
+      "step": 24540
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.3506861848124428e-05,
+      "loss": 0.2547,
+      "step": 24550
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.2061,
+      "step": 24560
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.3493138151875572e-05,
+      "loss": 0.3412,
+      "step": 24570
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.3486276303751144e-05,
+      "loss": 0.2721,
+      "step": 24580
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.3479414455626717e-05,
+      "loss": 0.2227,
+      "step": 24590
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.3472552607502288e-05,
+      "loss": 0.3343,
+      "step": 24600
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.346569075937786e-05,
+      "loss": 0.2869,
+      "step": 24610
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.3458828911253432e-05,
+      "loss": 0.2512,
+      "step": 24620
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.3451967063129004e-05,
+      "loss": 0.3016,
+      "step": 24630
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.3445105215004575e-05,
+      "loss": 0.288,
+      "step": 24640
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.3438243366880148e-05,
+      "loss": 0.3445,
+      "step": 24650
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.343138151875572e-05,
+      "loss": 0.2925,
+      "step": 24660
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.342451967063129e-05,
+      "loss": 0.2445,
+      "step": 24670
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.3417657822506864e-05,
+      "loss": 0.2526,
+      "step": 24680
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.3410795974382433e-05,
+      "loss": 0.2703,
+      "step": 24690
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3403934126258005e-05,
+      "loss": 0.2671,
+      "step": 24700
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3397072278133578e-05,
+      "loss": 0.272,
+      "step": 24710
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3390210430009149e-05,
+      "loss": 0.2032,
+      "step": 24720
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.338334858188472e-05,
+      "loss": 0.2399,
+      "step": 24730
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3376486733760293e-05,
+      "loss": 0.2401,
+      "step": 24740
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3369624885635864e-05,
+      "loss": 0.2577,
+      "step": 24750
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3362763037511436e-05,
+      "loss": 0.2299,
+      "step": 24760
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3355901189387009e-05,
+      "loss": 0.2544,
+      "step": 24770
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.334903934126258e-05,
+      "loss": 0.3282,
+      "step": 24780
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.3342177493138151e-05,
+      "loss": 0.2849,
+      "step": 24790
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.3335315645013724e-05,
+      "loss": 0.262,
+      "step": 24800
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.3328453796889296e-05,
+      "loss": 0.2528,
+      "step": 24810
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.3321591948764867e-05,
+      "loss": 0.2976,
+      "step": 24820
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.331473010064044e-05,
+      "loss": 0.3744,
+      "step": 24830
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.3307868252516011e-05,
+      "loss": 0.2194,
+      "step": 24840
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.3301006404391583e-05,
+      "loss": 0.2777,
+      "step": 24850
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.3294144556267156e-05,
+      "loss": 0.245,
+      "step": 24860
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.3287282708142727e-05,
+      "loss": 0.1919,
+      "step": 24870
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.3280420860018298e-05,
+      "loss": 0.2316,
+      "step": 24880
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.3273559011893871e-05,
+      "loss": 0.2619,
+      "step": 24890
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.3266697163769443e-05,
+      "loss": 0.1817,
+      "step": 24900
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.3259835315645014e-05,
+      "loss": 0.2743,
+      "step": 24910
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.3252973467520587e-05,
+      "loss": 0.2865,
+      "step": 24920
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.3246111619396158e-05,
+      "loss": 0.2108,
+      "step": 24930
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.323924977127173e-05,
+      "loss": 0.2084,
+      "step": 24940
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.32323879231473e-05,
+      "loss": 0.2421,
+      "step": 24950
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.3225526075022874e-05,
+      "loss": 0.3251,
+      "step": 24960
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.3218664226898445e-05,
+      "loss": 0.2371,
+      "step": 24970
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.3211802378774016e-05,
+      "loss": 0.221,
+      "step": 24980
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.320494053064959e-05,
+      "loss": 0.2424,
+      "step": 24990
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.319807868252516e-05,
+      "loss": 0.254,
+      "step": 25000
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3191216834400732e-05,
+      "loss": 0.4231,
+      "step": 25010
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3184354986276305e-05,
+      "loss": 0.2252,
+      "step": 25020
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3177493138151876e-05,
+      "loss": 0.225,
+      "step": 25030
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3170631290027448e-05,
+      "loss": 0.2894,
+      "step": 25040
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.316376944190302e-05,
+      "loss": 0.2373,
+      "step": 25050
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.3156907593778592e-05,
+      "loss": 0.3581,
+      "step": 25060
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.3150045745654163e-05,
+      "loss": 0.2315,
+      "step": 25070
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.3143183897529736e-05,
+      "loss": 0.2785,
+      "step": 25080
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.3136322049405306e-05,
+      "loss": 0.3271,
+      "step": 25090
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.3129460201280877e-05,
+      "loss": 0.2361,
+      "step": 25100
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.312259835315645e-05,
+      "loss": 0.2527,
+      "step": 25110
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.3115736505032021e-05,
+      "loss": 0.2459,
+      "step": 25120
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.3108874656907593e-05,
+      "loss": 0.1802,
+      "step": 25130
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3102012808783166e-05,
+      "loss": 0.2553,
+      "step": 25140
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3095150960658737e-05,
+      "loss": 0.2544,
+      "step": 25150
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3088289112534308e-05,
+      "loss": 0.3026,
+      "step": 25160
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3081427264409881e-05,
+      "loss": 0.3168,
+      "step": 25170
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3074565416285453e-05,
+      "loss": 0.3067,
+      "step": 25180
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.3067703568161024e-05,
+      "loss": 0.3135,
+      "step": 25190
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.3060841720036597e-05,
+      "loss": 0.3589,
+      "step": 25200
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.3053979871912168e-05,
+      "loss": 0.3213,
+      "step": 25210
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.304711802378774e-05,
+      "loss": 0.287,
+      "step": 25220
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.3040256175663313e-05,
+      "loss": 0.2166,
+      "step": 25230
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.3033394327538884e-05,
+      "loss": 0.265,
+      "step": 25240
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.3026532479414455e-05,
+      "loss": 0.3425,
+      "step": 25250
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.3019670631290028e-05,
+      "loss": 0.3805,
+      "step": 25260
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.30128087831656e-05,
+      "loss": 0.2256,
+      "step": 25270
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.3005946935041171e-05,
+      "loss": 0.2341,
+      "step": 25280
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.2999085086916744e-05,
+      "loss": 0.2956,
+      "step": 25290
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.2992223238792315e-05,
+      "loss": 0.2556,
+      "step": 25300
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.2985361390667886e-05,
+      "loss": 0.258,
+      "step": 25310
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.297849954254346e-05,
+      "loss": 0.2194,
+      "step": 25320
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.297163769441903e-05,
+      "loss": 0.3181,
+      "step": 25330
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.2964775846294602e-05,
+      "loss": 0.3104,
+      "step": 25340
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.2957913998170175e-05,
+      "loss": 0.2143,
+      "step": 25350
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.2951052150045746e-05,
+      "loss": 0.2686,
+      "step": 25360
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.2944190301921318e-05,
+      "loss": 0.2409,
+      "step": 25370
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.293732845379689e-05,
+      "loss": 0.3012,
+      "step": 25380
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.2930466605672462e-05,
+      "loss": 0.2826,
+      "step": 25390
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.2923604757548033e-05,
+      "loss": 0.2853,
+      "step": 25400
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2916742909423606e-05,
+      "loss": 0.2722,
+      "step": 25410
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2909881061299178e-05,
+      "loss": 0.2919,
+      "step": 25420
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2903019213174749e-05,
+      "loss": 0.2963,
+      "step": 25430
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2896157365050322e-05,
+      "loss": 0.2318,
+      "step": 25440
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2889295516925893e-05,
+      "loss": 0.2424,
+      "step": 25450
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2882433668801465e-05,
+      "loss": 0.2835,
+      "step": 25460
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2875571820677038e-05,
+      "loss": 0.2613,
+      "step": 25470
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2868709972552609e-05,
+      "loss": 0.3164,
+      "step": 25480
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.286184812442818e-05,
+      "loss": 0.2565,
+      "step": 25490
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.2854986276303751e-05,
+      "loss": 0.2435,
+      "step": 25500
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.2848124428179323e-05,
+      "loss": 0.2558,
+      "step": 25510
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.2841262580054894e-05,
+      "loss": 0.2599,
+      "step": 25520
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.2834400731930467e-05,
+      "loss": 0.2138,
+      "step": 25530
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.2827538883806038e-05,
+      "loss": 0.2933,
+      "step": 25540
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.282067703568161e-05,
+      "loss": 0.2583,
+      "step": 25550
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.2813815187557183e-05,
+      "loss": 0.2734,
+      "step": 25560
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.2806953339432754e-05,
+      "loss": 0.2867,
+      "step": 25570
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.2800091491308325e-05,
+      "loss": 0.2812,
+      "step": 25580
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.2793229643183897e-05,
+      "loss": 0.2266,
+      "step": 25590
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.278636779505947e-05,
+      "loss": 0.2549,
+      "step": 25600
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.2779505946935041e-05,
+      "loss": 0.2877,
+      "step": 25610
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.2772644098810612e-05,
+      "loss": 0.2802,
+      "step": 25620
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.2765782250686185e-05,
+      "loss": 0.261,
+      "step": 25630
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.2758920402561757e-05,
+      "loss": 0.2036,
+      "step": 25640
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.2752058554437328e-05,
+      "loss": 0.2439,
+      "step": 25650
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.27451967063129e-05,
+      "loss": 0.24,
+      "step": 25660
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2738334858188472e-05,
+      "loss": 0.2547,
+      "step": 25670
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2731473010064043e-05,
+      "loss": 0.3133,
+      "step": 25680
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2724611161939616e-05,
+      "loss": 0.3013,
+      "step": 25690
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2717749313815188e-05,
+      "loss": 0.2126,
+      "step": 25700
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2710887465690759e-05,
+      "loss": 0.2914,
+      "step": 25710
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.2704025617566332e-05,
+      "loss": 0.2126,
+      "step": 25720
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.2697163769441903e-05,
+      "loss": 0.2611,
+      "step": 25730
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.2690301921317475e-05,
+      "loss": 0.2485,
+      "step": 25740
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.2683440073193048e-05,
+      "loss": 0.3766,
+      "step": 25750
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.2676578225068619e-05,
+      "loss": 0.2854,
+      "step": 25760
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.266971637694419e-05,
+      "loss": 0.2716,
+      "step": 25770
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.2662854528819763e-05,
+      "loss": 0.2622,
+      "step": 25780
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.2655992680695335e-05,
+      "loss": 0.3213,
+      "step": 25790
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.2649130832570906e-05,
+      "loss": 0.3524,
+      "step": 25800
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.2642268984446479e-05,
+      "loss": 0.2326,
+      "step": 25810
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.263540713632205e-05,
+      "loss": 0.2997,
+      "step": 25820
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.2628545288197622e-05,
+      "loss": 0.2393,
+      "step": 25830
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.2621683440073195e-05,
+      "loss": 0.2647,
+      "step": 25840
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2614821591948766e-05,
+      "loss": 0.3467,
+      "step": 25850
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2607959743824337e-05,
+      "loss": 0.2329,
+      "step": 25860
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.260109789569991e-05,
+      "loss": 0.2395,
+      "step": 25870
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2594236047575481e-05,
+      "loss": 0.2762,
+      "step": 25880
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2587374199451053e-05,
+      "loss": 0.1967,
+      "step": 25890
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.2580512351326624e-05,
+      "loss": 0.3013,
+      "step": 25900
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.2573650503202195e-05,
+      "loss": 0.3517,
+      "step": 25910
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.2566788655077767e-05,
+      "loss": 0.2232,
+      "step": 25920
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.255992680695334e-05,
+      "loss": 0.2486,
+      "step": 25930
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.2553064958828911e-05,
+      "loss": 0.2766,
+      "step": 25940
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.2546203110704482e-05,
+      "loss": 0.2448,
+      "step": 25950
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.2539341262580055e-05,
+      "loss": 0.3476,
+      "step": 25960
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.2532479414455627e-05,
+      "loss": 0.2965,
+      "step": 25970
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.2525617566331198e-05,
+      "loss": 0.2716,
+      "step": 25980
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.2518755718206771e-05,
+      "loss": 0.2604,
+      "step": 25990
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.2511893870082342e-05,
+      "loss": 0.2237,
+      "step": 26000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.2505032021957914e-05,
+      "loss": 0.3,
+      "step": 26010
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.2498170173833487e-05,
+      "loss": 0.2422,
+      "step": 26020
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.2491308325709058e-05,
+      "loss": 0.2432,
+      "step": 26030
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.2484446477584629e-05,
+      "loss": 0.2265,
+      "step": 26040
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.2477584629460202e-05,
+      "loss": 0.3739,
+      "step": 26050
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.2470722781335773e-05,
+      "loss": 0.2315,
+      "step": 26060
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.2463860933211345e-05,
+      "loss": 0.2566,
+      "step": 26070
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.2456999085086918e-05,
+      "loss": 0.2587,
+      "step": 26080
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.2450137236962489e-05,
+      "loss": 0.2177,
+      "step": 26090
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.244327538883806e-05,
+      "loss": 0.2132,
+      "step": 26100
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.2436413540713633e-05,
+      "loss": 0.241,
+      "step": 26110
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2429551692589205e-05,
+      "loss": 0.2872,
+      "step": 26120
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2422689844464776e-05,
+      "loss": 0.2821,
+      "step": 26130
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2415827996340349e-05,
+      "loss": 0.2516,
+      "step": 26140
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.240896614821592e-05,
+      "loss": 0.2835,
+      "step": 26150
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2402104300091492e-05,
+      "loss": 0.28,
+      "step": 26160
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2395242451967065e-05,
+      "loss": 0.2593,
+      "step": 26170
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2388380603842636e-05,
+      "loss": 0.1928,
+      "step": 26180
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2381518755718207e-05,
+      "loss": 0.248,
+      "step": 26190
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2374656907593778e-05,
+      "loss": 0.2445,
+      "step": 26200
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2367795059469351e-05,
+      "loss": 0.2486,
+      "step": 26210
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2360933211344923e-05,
+      "loss": 0.2629,
+      "step": 26220
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2354071363220494e-05,
+      "loss": 0.267,
+      "step": 26230
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2347209515096067e-05,
+      "loss": 0.2611,
+      "step": 26240
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.2340347666971638e-05,
+      "loss": 0.4266,
+      "step": 26250
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.233348581884721e-05,
+      "loss": 0.2646,
+      "step": 26260
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.2327310155535224e-05,
+      "loss": 0.3029,
+      "step": 26270
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.2320448307410797e-05,
+      "loss": 0.2523,
+      "step": 26280
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.2313586459286369e-05,
+      "loss": 0.232,
+      "step": 26290
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.230672461116194e-05,
+      "loss": 0.3591,
+      "step": 26300
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.2299862763037513e-05,
+      "loss": 0.3298,
+      "step": 26310
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.2293000914913084e-05,
+      "loss": 0.2328,
+      "step": 26320
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.2286139066788655e-05,
+      "loss": 0.3151,
+      "step": 26330
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.2279277218664228e-05,
+      "loss": 0.2516,
+      "step": 26340
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.22724153705398e-05,
+      "loss": 0.2813,
+      "step": 26350
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.2265553522415371e-05,
+      "loss": 0.3616,
+      "step": 26360
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.2258691674290944e-05,
+      "loss": 0.3013,
+      "step": 26370
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.2251829826166515e-05,
+      "loss": 0.3194,
+      "step": 26380
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.2244967978042087e-05,
+      "loss": 0.3502,
+      "step": 26390
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.223810612991766e-05,
+      "loss": 0.3087,
+      "step": 26400
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.2231244281793231e-05,
+      "loss": 0.2568,
+      "step": 26410
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.2224382433668802e-05,
+      "loss": 0.2809,
+      "step": 26420
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2217520585544374e-05,
+      "loss": 0.2154,
+      "step": 26430
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2210658737419945e-05,
+      "loss": 0.3177,
+      "step": 26440
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2203796889295516e-05,
+      "loss": 0.2454,
+      "step": 26450
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2196935041171088e-05,
+      "loss": 0.2403,
+      "step": 26460
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.219007319304666e-05,
+      "loss": 0.3954,
+      "step": 26470
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2183211344922232e-05,
+      "loss": 0.2798,
+      "step": 26480
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2176349496797803e-05,
+      "loss": 0.23,
+      "step": 26490
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2169487648673376e-05,
+      "loss": 0.2602,
+      "step": 26500
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2162625800548947e-05,
+      "loss": 0.3387,
+      "step": 26510
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2155763952424519e-05,
+      "loss": 0.2224,
+      "step": 26520
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2148902104300092e-05,
+      "loss": 0.3228,
+      "step": 26530
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 0.45416226983070374,
+      "eval_runtime": 307.9514,
+      "eval_samples_per_second": 6.381,
+      "eval_steps_per_second": 1.598,
+      "eval_wer": 0.20876717392643707,
+      "step": 26532
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2142040256175663e-05,
+      "loss": 0.319,
+      "step": 26540
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2135178408051234e-05,
+      "loss": 0.158,
+      "step": 26550
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.2128316559926807e-05,
+      "loss": 0.1849,
+      "step": 26560
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.2121454711802379e-05,
+      "loss": 0.2535,
+      "step": 26570
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.211459286367795e-05,
+      "loss": 0.2378,
+      "step": 26580
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.2107731015553523e-05,
+      "loss": 0.2917,
+      "step": 26590
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.2100869167429094e-05,
+      "loss": 0.2509,
+      "step": 26600
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.2094007319304666e-05,
+      "loss": 0.2276,
+      "step": 26610
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.2087145471180239e-05,
+      "loss": 0.2279,
+      "step": 26620
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.208028362305581e-05,
+      "loss": 0.2301,
+      "step": 26630
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.2073421774931381e-05,
+      "loss": 0.29,
+      "step": 26640
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.2066559926806954e-05,
+      "loss": 0.1777,
+      "step": 26650
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.2059698078682526e-05,
+      "loss": 0.2117,
+      "step": 26660
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.2052836230558097e-05,
+      "loss": 0.2477,
+      "step": 26670
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.204597438243367e-05,
+      "loss": 0.2284,
+      "step": 26680
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.2039112534309241e-05,
+      "loss": 0.1947,
+      "step": 26690
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.2032250686184812e-05,
+      "loss": 0.23,
+      "step": 26700
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.2025388838060385e-05,
+      "loss": 0.2041,
+      "step": 26710
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.2018526989935957e-05,
+      "loss": 0.2348,
+      "step": 26720
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.2011665141811528e-05,
+      "loss": 0.2265,
+      "step": 26730
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.2004803293687101e-05,
+      "loss": 0.3213,
+      "step": 26740
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.1997941445562672e-05,
+      "loss": 0.2631,
+      "step": 26750
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.1991079597438244e-05,
+      "loss": 0.1725,
+      "step": 26760
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.1984217749313817e-05,
+      "loss": 0.2382,
+      "step": 26770
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.1977355901189388e-05,
+      "loss": 0.3674,
+      "step": 26780
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.197049405306496e-05,
+      "loss": 0.2528,
+      "step": 26790
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.1963632204940532e-05,
+      "loss": 0.3718,
+      "step": 26800
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.1956770356816104e-05,
+      "loss": 0.1687,
+      "step": 26810
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1949908508691675e-05,
+      "loss": 0.2343,
+      "step": 26820
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1943046660567246e-05,
+      "loss": 0.2129,
+      "step": 26830
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1936184812442818e-05,
+      "loss": 0.2471,
+      "step": 26840
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1929322964318389e-05,
+      "loss": 0.1755,
+      "step": 26850
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1922461116193962e-05,
+      "loss": 0.3465,
+      "step": 26860
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.1915599268069533e-05,
+      "loss": 0.238,
+      "step": 26870
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.1908737419945104e-05,
+      "loss": 0.245,
+      "step": 26880
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.1901875571820677e-05,
+      "loss": 0.2506,
+      "step": 26890
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.1895013723696249e-05,
+      "loss": 0.1839,
+      "step": 26900
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.188815187557182e-05,
+      "loss": 0.1949,
+      "step": 26910
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.1881290027447393e-05,
+      "loss": 0.2371,
+      "step": 26920
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.1874428179322964e-05,
+      "loss": 0.2159,
+      "step": 26930
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.1867566331198536e-05,
+      "loss": 0.2182,
+      "step": 26940
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.1860704483074109e-05,
+      "loss": 0.2356,
+      "step": 26950
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.185384263494968e-05,
+      "loss": 0.2329,
+      "step": 26960
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.1846980786825251e-05,
+      "loss": 0.3306,
+      "step": 26970
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.1840118938700824e-05,
+      "loss": 0.2393,
+      "step": 26980
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.1833257090576396e-05,
+      "loss": 0.1907,
+      "step": 26990
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1826395242451967e-05,
+      "loss": 0.228,
+      "step": 27000
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.181953339432754e-05,
+      "loss": 0.3278,
+      "step": 27010
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1812671546203111e-05,
+      "loss": 0.2243,
+      "step": 27020
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1805809698078683e-05,
+      "loss": 0.2363,
+      "step": 27030
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1798947849954256e-05,
+      "loss": 0.2481,
+      "step": 27040
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.1792086001829827e-05,
+      "loss": 0.3157,
+      "step": 27050
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.1785224153705398e-05,
+      "loss": 0.266,
+      "step": 27060
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.177836230558097e-05,
+      "loss": 0.2788,
+      "step": 27070
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.1771500457456542e-05,
+      "loss": 0.2632,
+      "step": 27080
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.1764638609332114e-05,
+      "loss": 0.2538,
+      "step": 27090
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.1757776761207685e-05,
+      "loss": 0.1796,
+      "step": 27100
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.1750914913083258e-05,
+      "loss": 0.2638,
+      "step": 27110
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.174405306495883e-05,
+      "loss": 0.2687,
+      "step": 27120
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.17371912168344e-05,
+      "loss": 0.3337,
+      "step": 27130
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.1730329368709974e-05,
+      "loss": 0.2117,
+      "step": 27140
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.1723467520585545e-05,
+      "loss": 0.2142,
+      "step": 27150
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.1716605672461116e-05,
+      "loss": 0.2082,
+      "step": 27160
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.170974382433669e-05,
+      "loss": 0.2282,
+      "step": 27170
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.170288197621226e-05,
+      "loss": 0.2457,
+      "step": 27180
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.1696020128087832e-05,
+      "loss": 0.2808,
+      "step": 27190
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.1689158279963405e-05,
+      "loss": 0.1944,
+      "step": 27200
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.1682296431838976e-05,
+      "loss": 0.185,
+      "step": 27210
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.1675434583714548e-05,
+      "loss": 0.2589,
+      "step": 27220
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.166857273559012e-05,
+      "loss": 0.3067,
+      "step": 27230
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.166171088746569e-05,
+      "loss": 0.1944,
+      "step": 27240
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.1654849039341261e-05,
+      "loss": 0.2785,
+      "step": 27250
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.1647987191216834e-05,
+      "loss": 0.2244,
+      "step": 27260
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.1641125343092406e-05,
+      "loss": 0.2562,
+      "step": 27270
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.1634263494967977e-05,
+      "loss": 0.2723,
+      "step": 27280
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.162740164684355e-05,
+      "loss": 0.2167,
+      "step": 27290
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.1620539798719121e-05,
+      "loss": 0.2151,
+      "step": 27300
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1613677950594693e-05,
+      "loss": 0.2046,
+      "step": 27310
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1606816102470266e-05,
+      "loss": 0.1642,
+      "step": 27320
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1599954254345837e-05,
+      "loss": 0.2209,
+      "step": 27330
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1593092406221408e-05,
+      "loss": 0.2179,
+      "step": 27340
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1586230558096981e-05,
+      "loss": 0.2114,
+      "step": 27350
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1579368709972553e-05,
+      "loss": 0.2525,
+      "step": 27360
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1572506861848124e-05,
+      "loss": 0.2482,
+      "step": 27370
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1565645013723697e-05,
+      "loss": 0.1834,
+      "step": 27380
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1558783165599268e-05,
+      "loss": 0.2648,
+      "step": 27390
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.155192131747484e-05,
+      "loss": 0.2312,
+      "step": 27400
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.1545059469350412e-05,
+      "loss": 0.1859,
+      "step": 27410
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.1538197621225984e-05,
+      "loss": 0.2311,
+      "step": 27420
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.1531335773101555e-05,
+      "loss": 0.3427,
+      "step": 27430
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.1524473924977128e-05,
+      "loss": 0.2395,
+      "step": 27440
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.15176120768527e-05,
+      "loss": 0.1641,
+      "step": 27450
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.151075022872827e-05,
+      "loss": 0.2747,
+      "step": 27460
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.1503888380603844e-05,
+      "loss": 0.2337,
+      "step": 27470
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.1497026532479415e-05,
+      "loss": 0.2099,
+      "step": 27480
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.1490164684354986e-05,
+      "loss": 0.2039,
+      "step": 27490
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.148330283623056e-05,
+      "loss": 0.2723,
+      "step": 27500
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.147644098810613e-05,
+      "loss": 0.2515,
+      "step": 27510
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.1469579139981702e-05,
+      "loss": 0.224,
+      "step": 27520
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1462717291857275e-05,
+      "loss": 0.2301,
+      "step": 27530
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1455855443732846e-05,
+      "loss": 0.1906,
+      "step": 27540
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1448993595608418e-05,
+      "loss": 0.2309,
+      "step": 27550
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.144213174748399e-05,
+      "loss": 0.2653,
+      "step": 27560
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1435269899359562e-05,
+      "loss": 0.2406,
+      "step": 27570
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.1428408051235133e-05,
+      "loss": 0.2137,
+      "step": 27580
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.1421546203110706e-05,
+      "loss": 0.3244,
+      "step": 27590
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.1414684354986277e-05,
+      "loss": 0.76,
+      "step": 27600
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.1407822506861849e-05,
+      "loss": 0.2181,
+      "step": 27610
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1400960658737422e-05,
+      "loss": 0.1965,
+      "step": 27620
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1394098810612993e-05,
+      "loss": 0.2269,
+      "step": 27630
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1387236962488563e-05,
+      "loss": 0.2867,
+      "step": 27640
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1380375114364136e-05,
+      "loss": 0.1943,
+      "step": 27650
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1373513266239707e-05,
+      "loss": 0.1973,
+      "step": 27660
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1366651418115278e-05,
+      "loss": 0.1564,
+      "step": 27670
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1359789569990851e-05,
+      "loss": 0.1921,
+      "step": 27680
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1352927721866423e-05,
+      "loss": 0.2227,
+      "step": 27690
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1346065873741994e-05,
+      "loss": 0.227,
+      "step": 27700
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.1339204025617565e-05,
+      "loss": 0.3191,
+      "step": 27710
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.1332342177493138e-05,
+      "loss": 0.2104,
+      "step": 27720
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.132548032936871e-05,
+      "loss": 0.2563,
+      "step": 27730
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.1318618481244281e-05,
+      "loss": 0.2471,
+      "step": 27740
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.1311756633119854e-05,
+      "loss": 0.2591,
+      "step": 27750
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.1304894784995425e-05,
+      "loss": 0.2226,
+      "step": 27760
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.1298032936870996e-05,
+      "loss": 0.1984,
+      "step": 27770
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.129117108874657e-05,
+      "loss": 0.2579,
+      "step": 27780
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.128430924062214e-05,
+      "loss": 0.1839,
+      "step": 27790
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.1277447392497712e-05,
+      "loss": 0.1986,
+      "step": 27800
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.1270585544373285e-05,
+      "loss": 0.2172,
+      "step": 27810
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.1263723696248856e-05,
+      "loss": 0.2411,
+      "step": 27820
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.1256861848124428e-05,
+      "loss": 0.2058,
+      "step": 27830
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.125e-05,
+      "loss": 0.282,
+      "step": 27840
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.1243138151875572e-05,
+      "loss": 0.2613,
+      "step": 27850
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.1236276303751143e-05,
+      "loss": 0.27,
+      "step": 27860
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.1229414455626716e-05,
+      "loss": 0.1837,
+      "step": 27870
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.1222552607502288e-05,
+      "loss": 0.2337,
+      "step": 27880
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.1215690759377859e-05,
+      "loss": 0.2318,
+      "step": 27890
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.1208828911253432e-05,
+      "loss": 0.2762,
+      "step": 27900
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.1201967063129003e-05,
+      "loss": 0.2263,
+      "step": 27910
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.1195105215004575e-05,
+      "loss": 0.2319,
+      "step": 27920
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.1188243366880148e-05,
+      "loss": 0.2977,
+      "step": 27930
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.1181381518755719e-05,
+      "loss": 0.2499,
+      "step": 27940
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.117451967063129e-05,
+      "loss": 0.1986,
+      "step": 27950
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.1167657822506863e-05,
+      "loss": 0.1993,
+      "step": 27960
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.1160795974382434e-05,
+      "loss": 0.3146,
+      "step": 27970
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.1153934126258006e-05,
+      "loss": 0.2171,
+      "step": 27980
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.1147072278133579e-05,
+      "loss": 0.1962,
+      "step": 27990
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.114021043000915e-05,
+      "loss": 0.2332,
+      "step": 28000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.1133348581884721e-05,
+      "loss": 0.226,
+      "step": 28010
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.1126486733760294e-05,
+      "loss": 0.2047,
+      "step": 28020
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.1119624885635866e-05,
+      "loss": 0.2034,
+      "step": 28030
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.1112763037511437e-05,
+      "loss": 0.2293,
+      "step": 28040
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.1105901189387008e-05,
+      "loss": 0.2137,
+      "step": 28050
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.109903934126258e-05,
+      "loss": 0.2053,
+      "step": 28060
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.1092177493138151e-05,
+      "loss": 0.2588,
+      "step": 28070
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.1085315645013724e-05,
+      "loss": 0.2405,
+      "step": 28080
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.1078453796889295e-05,
+      "loss": 0.2182,
+      "step": 28090
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.1071591948764867e-05,
+      "loss": 0.247,
+      "step": 28100
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.106473010064044e-05,
+      "loss": 0.2059,
+      "step": 28110
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.1057868252516011e-05,
+      "loss": 0.2535,
+      "step": 28120
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.1051006404391582e-05,
+      "loss": 0.2457,
+      "step": 28130
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.1044144556267155e-05,
+      "loss": 0.2386,
+      "step": 28140
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.1037282708142726e-05,
+      "loss": 0.2029,
+      "step": 28150
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.1030420860018298e-05,
+      "loss": 0.1981,
+      "step": 28160
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.102355901189387e-05,
+      "loss": 0.1695,
+      "step": 28170
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.1016697163769442e-05,
+      "loss": 0.2073,
+      "step": 28180
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.1009835315645013e-05,
+      "loss": 0.2306,
+      "step": 28190
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.1002973467520586e-05,
+      "loss": 0.2585,
+      "step": 28200
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.0996111619396158e-05,
+      "loss": 0.2609,
+      "step": 28210
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.0989249771271729e-05,
+      "loss": 0.2384,
+      "step": 28220
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.0982387923147302e-05,
+      "loss": 0.3525,
+      "step": 28230
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0975526075022873e-05,
+      "loss": 0.2866,
+      "step": 28240
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0968664226898445e-05,
+      "loss": 0.1858,
+      "step": 28250
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0961802378774018e-05,
+      "loss": 0.2762,
+      "step": 28260
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0954940530649589e-05,
+      "loss": 0.1858,
+      "step": 28270
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.094807868252516e-05,
+      "loss": 0.1931,
+      "step": 28280
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.0941216834400733e-05,
+      "loss": 0.2871,
+      "step": 28290
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.0934354986276305e-05,
+      "loss": 0.2389,
+      "step": 28300
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.0927493138151876e-05,
+      "loss": 0.2827,
+      "step": 28310
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.0920631290027447e-05,
+      "loss": 0.2227,
+      "step": 28320
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.091376944190302e-05,
+      "loss": 0.2754,
+      "step": 28330
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.0906907593778591e-05,
+      "loss": 0.2365,
+      "step": 28340
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.0900045745654163e-05,
+      "loss": 0.2072,
+      "step": 28350
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.0893183897529736e-05,
+      "loss": 0.2227,
+      "step": 28360
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0886322049405307e-05,
+      "loss": 0.2298,
+      "step": 28370
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0879460201280878e-05,
+      "loss": 0.2196,
+      "step": 28380
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0872598353156451e-05,
+      "loss": 0.2012,
+      "step": 28390
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0865736505032023e-05,
+      "loss": 0.216,
+      "step": 28400
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0858874656907594e-05,
+      "loss": 0.199,
+      "step": 28410
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.0852012808783167e-05,
+      "loss": 0.3041,
+      "step": 28420
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.0845150960658738e-05,
+      "loss": 0.1992,
+      "step": 28430
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.083828911253431e-05,
+      "loss": 0.2923,
+      "step": 28440
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.0831427264409881e-05,
+      "loss": 0.2323,
+      "step": 28450
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0824565416285452e-05,
+      "loss": 0.1941,
+      "step": 28460
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0817703568161024e-05,
+      "loss": 0.1768,
+      "step": 28470
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0810841720036597e-05,
+      "loss": 0.206,
+      "step": 28480
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0803979871912168e-05,
+      "loss": 0.2365,
+      "step": 28490
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0797118023787739e-05,
+      "loss": 0.1436,
+      "step": 28500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0790256175663312e-05,
+      "loss": 0.2132,
+      "step": 28510
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0783394327538883e-05,
+      "loss": 0.17,
+      "step": 28520
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0776532479414455e-05,
+      "loss": 0.2417,
+      "step": 28530
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0769670631290028e-05,
+      "loss": 0.238,
+      "step": 28540
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.0762808783165599e-05,
+      "loss": 0.2546,
+      "step": 28550
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.075594693504117e-05,
+      "loss": 0.2886,
+      "step": 28560
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.0749085086916743e-05,
+      "loss": 0.207,
+      "step": 28570
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.0742223238792315e-05,
+      "loss": 0.2554,
+      "step": 28580
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.0735361390667886e-05,
+      "loss": 0.2451,
+      "step": 28590
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.0728499542543459e-05,
+      "loss": 0.198,
+      "step": 28600
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.072163769441903e-05,
+      "loss": 0.2703,
+      "step": 28610
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.0714775846294602e-05,
+      "loss": 0.2449,
+      "step": 28620
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.0707913998170175e-05,
+      "loss": 0.2517,
+      "step": 28630
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.0701052150045746e-05,
+      "loss": 0.2336,
+      "step": 28640
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.0694190301921317e-05,
+      "loss": 0.1198,
+      "step": 28650
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.068732845379689e-05,
+      "loss": 0.2737,
+      "step": 28660
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.0680466605672462e-05,
+      "loss": 0.176,
+      "step": 28670
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0673604757548033e-05,
+      "loss": 0.2327,
+      "step": 28680
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0666742909423606e-05,
+      "loss": 0.1922,
+      "step": 28690
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0659881061299177e-05,
+      "loss": 0.1985,
+      "step": 28700
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0653019213174748e-05,
+      "loss": 0.256,
+      "step": 28710
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0646157365050321e-05,
+      "loss": 0.4075,
+      "step": 28720
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0639295516925893e-05,
+      "loss": 0.2035,
+      "step": 28730
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0632433668801464e-05,
+      "loss": 0.2608,
+      "step": 28740
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0625571820677037e-05,
+      "loss": 0.1873,
+      "step": 28750
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0618709972552608e-05,
+      "loss": 0.1775,
+      "step": 28760
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.061184812442818e-05,
+      "loss": 0.2103,
+      "step": 28770
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.0604986276303753e-05,
+      "loss": 0.2146,
+      "step": 28780
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.0598124428179324e-05,
+      "loss": 0.332,
+      "step": 28790
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.0591262580054895e-05,
+      "loss": 0.2032,
+      "step": 28800
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.0584400731930468e-05,
+      "loss": 0.2457,
+      "step": 28810
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.057753888380604e-05,
+      "loss": 0.3389,
+      "step": 28820
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.0570677035681611e-05,
+      "loss": 0.2731,
+      "step": 28830
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.0563815187557184e-05,
+      "loss": 0.2689,
+      "step": 28840
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.0556953339432755e-05,
+      "loss": 0.2128,
+      "step": 28850
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.0550091491308325e-05,
+      "loss": 0.2398,
+      "step": 28860
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.0543229643183898e-05,
+      "loss": 0.2383,
+      "step": 28870
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.0536367795059469e-05,
+      "loss": 0.1965,
+      "step": 28880
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.052950594693504e-05,
+      "loss": 0.2833,
+      "step": 28890
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.0522644098810613e-05,
+      "loss": 0.3237,
+      "step": 28900
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.0515782250686185e-05,
+      "loss": 0.2052,
+      "step": 28910
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.0508920402561756e-05,
+      "loss": 0.157,
+      "step": 28920
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.0502058554437329e-05,
+      "loss": 0.2562,
+      "step": 28930
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.04951967063129e-05,
+      "loss": 0.1887,
+      "step": 28940
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.0488334858188472e-05,
+      "loss": 0.2343,
+      "step": 28950
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.0481473010064043e-05,
+      "loss": 0.2191,
+      "step": 28960
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.0474611161939616e-05,
+      "loss": 0.2693,
+      "step": 28970
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.0467749313815187e-05,
+      "loss": 0.1874,
+      "step": 28980
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0460887465690759e-05,
+      "loss": 0.3065,
+      "step": 28990
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0454025617566332e-05,
+      "loss": 0.1901,
+      "step": 29000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0447163769441903e-05,
+      "loss": 0.1773,
+      "step": 29010
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0440301921317474e-05,
+      "loss": 0.2211,
+      "step": 29020
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0433440073193047e-05,
+      "loss": 0.2058,
+      "step": 29030
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.0426578225068618e-05,
+      "loss": 0.2145,
+      "step": 29040
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.041971637694419e-05,
+      "loss": 0.2553,
+      "step": 29050
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.0412854528819763e-05,
+      "loss": 0.2656,
+      "step": 29060
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.0406678865507777e-05,
+      "loss": 0.2232,
+      "step": 29070
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0399817017383349e-05,
+      "loss": 0.2107,
+      "step": 29080
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.039295516925892e-05,
+      "loss": 0.3049,
+      "step": 29090
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0386093321134493e-05,
+      "loss": 0.2093,
+      "step": 29100
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0379231473010064e-05,
+      "loss": 0.2007,
+      "step": 29110
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.0372369624885636e-05,
+      "loss": 0.2161,
+      "step": 29120
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.0365507776761209e-05,
+      "loss": 0.1922,
+      "step": 29130
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.035864592863678e-05,
+      "loss": 0.1598,
+      "step": 29140
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.0351784080512351e-05,
+      "loss": 0.2793,
+      "step": 29150
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.0344922232387924e-05,
+      "loss": 0.2412,
+      "step": 29160
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.0338060384263495e-05,
+      "loss": 0.2658,
+      "step": 29170
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.0331198536139067e-05,
+      "loss": 0.1916,
+      "step": 29180
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.0324336688014638e-05,
+      "loss": 0.2402,
+      "step": 29190
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.0317474839890211e-05,
+      "loss": 0.2075,
+      "step": 29200
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.0310612991765782e-05,
+      "loss": 0.2264,
+      "step": 29210
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.0303751143641354e-05,
+      "loss": 0.1554,
+      "step": 29220
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.0296889295516927e-05,
+      "loss": 0.2428,
+      "step": 29230
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.0290027447392498e-05,
+      "loss": 0.2499,
+      "step": 29240
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.028316559926807e-05,
+      "loss": 0.2028,
+      "step": 29250
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.0276303751143642e-05,
+      "loss": 0.2796,
+      "step": 29260
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.0269441903019214e-05,
+      "loss": 0.217,
+      "step": 29270
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.0262580054894785e-05,
+      "loss": 0.2293,
+      "step": 29280
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.0255718206770358e-05,
+      "loss": 0.2348,
+      "step": 29290
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.024885635864593e-05,
+      "loss": 0.2216,
+      "step": 29300
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.02419945105215e-05,
+      "loss": 0.2293,
+      "step": 29310
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.0235132662397074e-05,
+      "loss": 0.21,
+      "step": 29320
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.0228270814272645e-05,
+      "loss": 0.183,
+      "step": 29330
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.0221408966148216e-05,
+      "loss": 0.2235,
+      "step": 29340
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.021454711802379e-05,
+      "loss": 0.2619,
+      "step": 29350
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.020768526989936e-05,
+      "loss": 0.1947,
+      "step": 29360
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.0200823421774932e-05,
+      "loss": 0.1633,
+      "step": 29370
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.0193961573650503e-05,
+      "loss": 0.2614,
+      "step": 29380
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.0187099725526074e-05,
+      "loss": 0.2369,
+      "step": 29390
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.0180237877401646e-05,
+      "loss": 0.1936,
+      "step": 29400
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.0173376029277219e-05,
+      "loss": 0.2234,
+      "step": 29410
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.016651418115279e-05,
+      "loss": 0.2242,
+      "step": 29420
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.0159652333028361e-05,
+      "loss": 0.2741,
+      "step": 29430
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.0152790484903934e-05,
+      "loss": 0.2085,
+      "step": 29440
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.0145928636779506e-05,
+      "loss": 0.2069,
+      "step": 29450
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.0139066788655077e-05,
+      "loss": 0.1924,
+      "step": 29460
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.013220494053065e-05,
+      "loss": 0.2071,
+      "step": 29470
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.0125343092406221e-05,
+      "loss": 0.338,
+      "step": 29480
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.0118481244281793e-05,
+      "loss": 0.214,
+      "step": 29490
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.0111619396157366e-05,
+      "loss": 0.2589,
+      "step": 29500
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.0104757548032937e-05,
+      "loss": 0.2335,
+      "step": 29510
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.0097895699908508e-05,
+      "loss": 0.2086,
+      "step": 29520
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.0091033851784081e-05,
+      "loss": 0.232,
+      "step": 29530
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.0084172003659652e-05,
+      "loss": 0.2267,
+      "step": 29540
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.0077310155535224e-05,
+      "loss": 0.2621,
+      "step": 29550
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.0070448307410797e-05,
+      "loss": 0.2459,
+      "step": 29560
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.0063586459286368e-05,
+      "loss": 0.2195,
+      "step": 29570
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.005672461116194e-05,
+      "loss": 0.2645,
+      "step": 29580
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.0049862763037512e-05,
+      "loss": 0.2347,
+      "step": 29590
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.0043000914913084e-05,
+      "loss": 0.1973,
+      "step": 29600
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.0036139066788655e-05,
+      "loss": 0.2413,
+      "step": 29610
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.0029277218664228e-05,
+      "loss": 0.2294,
+      "step": 29620
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.00224153705398e-05,
+      "loss": 0.206,
+      "step": 29630
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.001555352241537e-05,
+      "loss": 0.1798,
+      "step": 29640
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.0008691674290944e-05,
+      "loss": 0.2125,
+      "step": 29650
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.0001829826166515e-05,
+      "loss": 0.2276,
+      "step": 29660
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.994967978042086e-06,
+      "loss": 0.1599,
+      "step": 29670
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.98810612991766e-06,
+      "loss": 0.2974,
+      "step": 29680
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.98124428179323e-06,
+      "loss": 0.219,
+      "step": 29690
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.974382433668802e-06,
+      "loss": 0.1673,
+      "step": 29700
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.967520585544375e-06,
+      "loss": 0.2062,
+      "step": 29710
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.960658737419946e-06,
+      "loss": 0.1991,
+      "step": 29720
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.953796889295517e-06,
+      "loss": 0.3136,
+      "step": 29730
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.94693504117109e-06,
+      "loss": 0.1922,
+      "step": 29740
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.940073193046662e-06,
+      "loss": 0.2093,
+      "step": 29750
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.933211344922233e-06,
+      "loss": 0.215,
+      "step": 29760
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.926349496797806e-06,
+      "loss": 0.2481,
+      "step": 29770
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.919487648673377e-06,
+      "loss": 0.2164,
+      "step": 29780
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.912625800548947e-06,
+      "loss": 0.2693,
+      "step": 29790
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.90576395242452e-06,
+      "loss": 0.3059,
+      "step": 29800
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.898902104300091e-06,
+      "loss": 0.1549,
+      "step": 29810
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.892040256175663e-06,
+      "loss": 0.2824,
+      "step": 29820
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.885178408051234e-06,
+      "loss": 0.262,
+      "step": 29830
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.878316559926807e-06,
+      "loss": 0.2158,
+      "step": 29840
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.871454711802378e-06,
+      "loss": 0.2682,
+      "step": 29850
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.86459286367795e-06,
+      "loss": 0.2099,
+      "step": 29860
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.857731015553523e-06,
+      "loss": 0.2188,
+      "step": 29870
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.850869167429094e-06,
+      "loss": 0.2746,
+      "step": 29880
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.844007319304665e-06,
+      "loss": 0.2251,
+      "step": 29890
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.837145471180238e-06,
+      "loss": 0.2515,
+      "step": 29900
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.83028362305581e-06,
+      "loss": 0.1752,
+      "step": 29910
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.82342177493138e-06,
+      "loss": 0.2215,
+      "step": 29920
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.816559926806954e-06,
+      "loss": 0.2509,
+      "step": 29930
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.809698078682525e-06,
+      "loss": 0.2489,
+      "step": 29940
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.802836230558096e-06,
+      "loss": 0.1992,
+      "step": 29950
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.79597438243367e-06,
+      "loss": 0.2405,
+      "step": 29960
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.78911253430924e-06,
+      "loss": 0.2434,
+      "step": 29970
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.782250686184812e-06,
+      "loss": 0.2723,
+      "step": 29980
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.775388838060385e-06,
+      "loss": 0.2038,
+      "step": 29990
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.768526989935956e-06,
+      "loss": 0.2719,
+      "step": 30000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.761665141811528e-06,
+      "loss": 0.2382,
+      "step": 30010
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.7548032936871e-06,
+      "loss": 0.2578,
+      "step": 30020
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.747941445562672e-06,
+      "loss": 0.2252,
+      "step": 30030
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.741079597438243e-06,
+      "loss": 0.1777,
+      "step": 30040
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.734217749313816e-06,
+      "loss": 0.2855,
+      "step": 30050
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.727355901189388e-06,
+      "loss": 0.255,
+      "step": 30060
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.720494053064959e-06,
+      "loss": 0.1812,
+      "step": 30070
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.713632204940532e-06,
+      "loss": 0.3766,
+      "step": 30080
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.706770356816103e-06,
+      "loss": 0.3375,
+      "step": 30090
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.699908508691674e-06,
+      "loss": 0.2214,
+      "step": 30100
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.693046660567247e-06,
+      "loss": 0.3241,
+      "step": 30110
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.686184812442819e-06,
+      "loss": 0.2902,
+      "step": 30120
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.67932296431839e-06,
+      "loss": 0.2394,
+      "step": 30130
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.672461116193963e-06,
+      "loss": 0.239,
+      "step": 30140
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.665599268069534e-06,
+      "loss": 0.2125,
+      "step": 30150
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.658737419945106e-06,
+      "loss": 0.1614,
+      "step": 30160
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.651875571820679e-06,
+      "loss": 0.2292,
+      "step": 30170
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.64501372369625e-06,
+      "loss": 0.2553,
+      "step": 30180
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.63815187557182e-06,
+      "loss": 0.2381,
+      "step": 30190
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.631290027447393e-06,
+      "loss": 0.2451,
+      "step": 30200
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.624428179322964e-06,
+      "loss": 0.1767,
+      "step": 30210
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.617566331198535e-06,
+      "loss": 0.2384,
+      "step": 30220
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.610704483074108e-06,
+      "loss": 0.1639,
+      "step": 30230
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.60384263494968e-06,
+      "loss": 0.2388,
+      "step": 30240
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.59698078682525e-06,
+      "loss": 0.1861,
+      "step": 30250
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.590118938700824e-06,
+      "loss": 0.286,
+      "step": 30260
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.583257090576395e-06,
+      "loss": 0.2771,
+      "step": 30270
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.576395242451966e-06,
+      "loss": 0.2815,
+      "step": 30280
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.56953339432754e-06,
+      "loss": 0.2453,
+      "step": 30290
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.56267154620311e-06,
+      "loss": 0.2722,
+      "step": 30300
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.555809698078682e-06,
+      "loss": 0.1862,
+      "step": 30310
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.548947849954255e-06,
+      "loss": 0.1806,
+      "step": 30320
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.542086001829826e-06,
+      "loss": 0.2421,
+      "step": 30330
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.535224153705398e-06,
+      "loss": 0.2089,
+      "step": 30340
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.52836230558097e-06,
+      "loss": 0.1459,
+      "step": 30350
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.521500457456542e-06,
+      "loss": 0.2075,
+      "step": 30360
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.514638609332113e-06,
+      "loss": 0.2222,
+      "step": 30370
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.507776761207686e-06,
+      "loss": 0.1915,
+      "step": 30380
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.500914913083258e-06,
+      "loss": 0.2289,
+      "step": 30390
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.494053064958829e-06,
+      "loss": 0.2402,
+      "step": 30400
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.487191216834402e-06,
+      "loss": 0.2109,
+      "step": 30410
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.480329368709973e-06,
+      "loss": 0.2296,
+      "step": 30420
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.473467520585544e-06,
+      "loss": 0.2232,
+      "step": 30430
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.466605672461116e-06,
+      "loss": 0.1935,
+      "step": 30440
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.459743824336689e-06,
+      "loss": 0.2643,
+      "step": 30450
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.45288197621226e-06,
+      "loss": 0.1998,
+      "step": 30460
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.446020128087831e-06,
+      "loss": 0.1839,
+      "step": 30470
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.439158279963404e-06,
+      "loss": 0.1771,
+      "step": 30480
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.432296431838976e-06,
+      "loss": 0.265,
+      "step": 30490
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.425434583714547e-06,
+      "loss": 0.1589,
+      "step": 30500
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.41857273559012e-06,
+      "loss": 0.2222,
+      "step": 30510
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.411710887465691e-06,
+      "loss": 0.1984,
+      "step": 30520
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.404849039341263e-06,
+      "loss": 0.199,
+      "step": 30530
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.397987191216836e-06,
+      "loss": 0.1583,
+      "step": 30540
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.391125343092407e-06,
+      "loss": 0.2136,
+      "step": 30550
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.384263494967978e-06,
+      "loss": 0.1978,
+      "step": 30560
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.377401646843551e-06,
+      "loss": 0.1804,
+      "step": 30570
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.370539798719123e-06,
+      "loss": 0.2453,
+      "step": 30580
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.363677950594694e-06,
+      "loss": 0.279,
+      "step": 30590
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.356816102470265e-06,
+      "loss": 0.3549,
+      "step": 30600
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.349954254345836e-06,
+      "loss": 0.3055,
+      "step": 30610
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.343092406221408e-06,
+      "loss": 0.2449,
+      "step": 30620
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.33623055809698e-06,
+      "loss": 0.1633,
+      "step": 30630
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.329368709972552e-06,
+      "loss": 0.2481,
+      "step": 30640
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.322506861848123e-06,
+      "loss": 0.1858,
+      "step": 30650
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.315645013723696e-06,
+      "loss": 0.2704,
+      "step": 30660
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.308783165599268e-06,
+      "loss": 0.214,
+      "step": 30670
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.301921317474839e-06,
+      "loss": 0.2434,
+      "step": 30680
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.295059469350412e-06,
+      "loss": 0.2454,
+      "step": 30690
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.288197621225983e-06,
+      "loss": 0.2767,
+      "step": 30700
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.281335773101555e-06,
+      "loss": 0.2202,
+      "step": 30710
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.274473924977128e-06,
+      "loss": 0.1538,
+      "step": 30720
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.267612076852699e-06,
+      "loss": 0.1985,
+      "step": 30730
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.26075022872827e-06,
+      "loss": 0.2077,
+      "step": 30740
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.253888380603843e-06,
+      "loss": 0.2826,
+      "step": 30750
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.247026532479415e-06,
+      "loss": 0.2069,
+      "step": 30760
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.240164684354986e-06,
+      "loss": 0.2149,
+      "step": 30770
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.233302836230559e-06,
+      "loss": 0.2009,
+      "step": 30780
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.22644098810613e-06,
+      "loss": 0.2267,
+      "step": 30790
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.219579139981701e-06,
+      "loss": 0.1813,
+      "step": 30800
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.212717291857274e-06,
+      "loss": 0.1497,
+      "step": 30810
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.205855443732846e-06,
+      "loss": 0.1957,
+      "step": 30820
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.198993595608417e-06,
+      "loss": 0.1674,
+      "step": 30830
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.19213174748399e-06,
+      "loss": 0.2289,
+      "step": 30840
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.185269899359561e-06,
+      "loss": 0.2175,
+      "step": 30850
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.178408051235133e-06,
+      "loss": 0.1944,
+      "step": 30860
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.171546203110706e-06,
+      "loss": 0.2684,
+      "step": 30870
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.164684354986277e-06,
+      "loss": 0.2056,
+      "step": 30880
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.157822506861848e-06,
+      "loss": 0.2203,
+      "step": 30890
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.150960658737421e-06,
+      "loss": 0.2542,
+      "step": 30900
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.144098810612993e-06,
+      "loss": 0.3163,
+      "step": 30910
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.137236962488564e-06,
+      "loss": 0.2467,
+      "step": 30920
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.130375114364137e-06,
+      "loss": 0.2332,
+      "step": 30930
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.123513266239708e-06,
+      "loss": 0.1645,
+      "step": 30940
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.11665141811528e-06,
+      "loss": 0.346,
+      "step": 30950
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 0.4672754108905792,
+      "eval_runtime": 305.7392,
+      "eval_samples_per_second": 6.427,
+      "eval_steps_per_second": 1.609,
+      "eval_wer": 0.19971043065738403,
+      "step": 30954
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.109789569990853e-06,
+      "loss": 0.1958,
+      "step": 30960
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.102927721866424e-06,
+      "loss": 0.243,
+      "step": 30970
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.096065873741995e-06,
+      "loss": 0.204,
+      "step": 30980
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.089204025617568e-06,
+      "loss": 0.211,
+      "step": 30990
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.082342177493138e-06,
+      "loss": 0.2007,
+      "step": 31000
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.075480329368709e-06,
+      "loss": 0.2649,
+      "step": 31010
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.068618481244282e-06,
+      "loss": 0.1549,
+      "step": 31020
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.061756633119853e-06,
+      "loss": 0.2227,
+      "step": 31030
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.054894784995425e-06,
+      "loss": 0.1785,
+      "step": 31040
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.048032936870998e-06,
+      "loss": 0.1842,
+      "step": 31050
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.041171088746569e-06,
+      "loss": 0.1863,
+      "step": 31060
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.03430924062214e-06,
+      "loss": 0.1886,
+      "step": 31070
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.027447392497712e-06,
+      "loss": 0.2246,
+      "step": 31080
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.020585544373285e-06,
+      "loss": 0.1668,
+      "step": 31090
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.013723696248856e-06,
+      "loss": 0.217,
+      "step": 31100
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.006861848124427e-06,
+      "loss": 0.1858,
+      "step": 31110
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9e-06,
+      "loss": 0.1856,
+      "step": 31120
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 8.993138151875572e-06,
+      "loss": 0.2384,
+      "step": 31130
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 8.986276303751143e-06,
+      "loss": 0.2019,
+      "step": 31140
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 8.979414455626716e-06,
+      "loss": 0.186,
+      "step": 31150
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.972552607502287e-06,
+      "loss": 0.1332,
+      "step": 31160
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.965690759377858e-06,
+      "loss": 0.2219,
+      "step": 31170
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.958828911253431e-06,
+      "loss": 0.2206,
+      "step": 31180
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.951967063129003e-06,
+      "loss": 0.2458,
+      "step": 31190
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.945105215004574e-06,
+      "loss": 0.1901,
+      "step": 31200
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.938243366880147e-06,
+      "loss": 0.2357,
+      "step": 31210
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.931381518755718e-06,
+      "loss": 0.1956,
+      "step": 31220
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.92451967063129e-06,
+      "loss": 0.1902,
+      "step": 31230
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.917657822506863e-06,
+      "loss": 0.1556,
+      "step": 31240
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.910795974382434e-06,
+      "loss": 0.1915,
+      "step": 31250
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.903934126258005e-06,
+      "loss": 0.226,
+      "step": 31260
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.897072278133578e-06,
+      "loss": 0.184,
+      "step": 31270
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.89021043000915e-06,
+      "loss": 0.2624,
+      "step": 31280
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.883348581884721e-06,
+      "loss": 0.1888,
+      "step": 31290
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.876486733760294e-06,
+      "loss": 0.2106,
+      "step": 31300
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.870311070448307e-06,
+      "loss": 0.1955,
+      "step": 31310
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.86344922232388e-06,
+      "loss": 0.1607,
+      "step": 31320
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.856587374199451e-06,
+      "loss": 0.1511,
+      "step": 31330
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.849725526075022e-06,
+      "loss": 0.1826,
+      "step": 31340
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.842863677950595e-06,
+      "loss": 0.2042,
+      "step": 31350
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.836001829826167e-06,
+      "loss": 0.1249,
+      "step": 31360
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.829139981701738e-06,
+      "loss": 0.2343,
+      "step": 31370
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.822278133577311e-06,
+      "loss": 0.158,
+      "step": 31380
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.815416285452882e-06,
+      "loss": 0.163,
+      "step": 31390
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.808554437328454e-06,
+      "loss": 0.1514,
+      "step": 31400
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.801692589204027e-06,
+      "loss": 0.1893,
+      "step": 31410
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.794830741079598e-06,
+      "loss": 0.3025,
+      "step": 31420
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.78796889295517e-06,
+      "loss": 0.1721,
+      "step": 31430
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.781107044830742e-06,
+      "loss": 0.1795,
+      "step": 31440
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.774245196706313e-06,
+      "loss": 0.2211,
+      "step": 31450
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.767383348581885e-06,
+      "loss": 0.1694,
+      "step": 31460
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.760521500457458e-06,
+      "loss": 0.1669,
+      "step": 31470
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.753659652333029e-06,
+      "loss": 0.1929,
+      "step": 31480
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.7467978042086e-06,
+      "loss": 0.2082,
+      "step": 31490
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.739935956084173e-06,
+      "loss": 0.1852,
+      "step": 31500
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.733074107959745e-06,
+      "loss": 0.2133,
+      "step": 31510
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.726212259835316e-06,
+      "loss": 0.1863,
+      "step": 31520
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.719350411710887e-06,
+      "loss": 0.2106,
+      "step": 31530
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.712488563586459e-06,
+      "loss": 0.1735,
+      "step": 31540
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.70562671546203e-06,
+      "loss": 0.1358,
+      "step": 31550
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.698764867337603e-06,
+      "loss": 0.2182,
+      "step": 31560
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.691903019213174e-06,
+      "loss": 0.2399,
+      "step": 31570
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.685041171088746e-06,
+      "loss": 0.1985,
+      "step": 31580
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.678179322964319e-06,
+      "loss": 0.1302,
+      "step": 31590
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.67131747483989e-06,
+      "loss": 0.1545,
+      "step": 31600
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.664455626715461e-06,
+      "loss": 0.2363,
+      "step": 31610
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.657593778591034e-06,
+      "loss": 0.187,
+      "step": 31620
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.650731930466605e-06,
+      "loss": 0.1955,
+      "step": 31630
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.643870082342177e-06,
+      "loss": 0.2249,
+      "step": 31640
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.63700823421775e-06,
+      "loss": 0.1658,
+      "step": 31650
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.630146386093321e-06,
+      "loss": 0.2486,
+      "step": 31660
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.623284537968892e-06,
+      "loss": 0.2263,
+      "step": 31670
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.616422689844465e-06,
+      "loss": 0.1969,
+      "step": 31680
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.609560841720037e-06,
+      "loss": 0.1947,
+      "step": 31690
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.602698993595608e-06,
+      "loss": 0.2021,
+      "step": 31700
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.595837145471181e-06,
+      "loss": 0.211,
+      "step": 31710
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.588975297346752e-06,
+      "loss": 0.1624,
+      "step": 31720
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.582113449222324e-06,
+      "loss": 0.2524,
+      "step": 31730
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.575251601097897e-06,
+      "loss": 0.213,
+      "step": 31740
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.568389752973468e-06,
+      "loss": 0.1461,
+      "step": 31750
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.56152790484904e-06,
+      "loss": 0.2767,
+      "step": 31760
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.554666056724612e-06,
+      "loss": 0.2114,
+      "step": 31770
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.547804208600184e-06,
+      "loss": 0.2783,
+      "step": 31780
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.540942360475755e-06,
+      "loss": 0.1614,
+      "step": 31790
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.534080512351328e-06,
+      "loss": 0.1712,
+      "step": 31800
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.5272186642269e-06,
+      "loss": 0.2861,
+      "step": 31810
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.52035681610247e-06,
+      "loss": 0.2062,
+      "step": 31820
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.513494967978043e-06,
+      "loss": 0.1521,
+      "step": 31830
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.506633119853615e-06,
+      "loss": 0.2397,
+      "step": 31840
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.499771271729186e-06,
+      "loss": 0.1523,
+      "step": 31850
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.492909423604759e-06,
+      "loss": 0.2631,
+      "step": 31860
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.48604757548033e-06,
+      "loss": 0.1767,
+      "step": 31870
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.479185727355902e-06,
+      "loss": 0.1493,
+      "step": 31880
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.472323879231475e-06,
+      "loss": 0.1724,
+      "step": 31890
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.465462031107046e-06,
+      "loss": 0.1576,
+      "step": 31900
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.458600182982617e-06,
+      "loss": 0.2418,
+      "step": 31910
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.451738334858189e-06,
+      "loss": 0.1493,
+      "step": 31920
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.444876486733762e-06,
+      "loss": 0.2008,
+      "step": 31930
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.438014638609331e-06,
+      "loss": 0.1744,
+      "step": 31940
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.431152790484903e-06,
+      "loss": 0.2044,
+      "step": 31950
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.424290942360476e-06,
+      "loss": 0.2149,
+      "step": 31960
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.417429094236047e-06,
+      "loss": 0.2436,
+      "step": 31970
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.410567246111618e-06,
+      "loss": 0.2301,
+      "step": 31980
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.403705397987191e-06,
+      "loss": 0.1529,
+      "step": 31990
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.396843549862762e-06,
+      "loss": 0.2286,
+      "step": 32000
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.389981701738334e-06,
+      "loss": 0.3041,
+      "step": 32010
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.383119853613907e-06,
+      "loss": 0.1862,
+      "step": 32020
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.376258005489478e-06,
+      "loss": 0.214,
+      "step": 32030
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.36939615736505e-06,
+      "loss": 0.1419,
+      "step": 32040
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.362534309240622e-06,
+      "loss": 0.1191,
+      "step": 32050
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.355672461116194e-06,
+      "loss": 0.1925,
+      "step": 32060
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.348810612991765e-06,
+      "loss": 0.2456,
+      "step": 32070
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.341948764867338e-06,
+      "loss": 0.2147,
+      "step": 32080
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.33508691674291e-06,
+      "loss": 0.1938,
+      "step": 32090
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.32822506861848e-06,
+      "loss": 0.1718,
+      "step": 32100
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.321363220494054e-06,
+      "loss": 0.2117,
+      "step": 32110
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.314501372369625e-06,
+      "loss": 0.2267,
+      "step": 32120
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.307639524245196e-06,
+      "loss": 0.2062,
+      "step": 32130
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.30077767612077e-06,
+      "loss": 0.2798,
+      "step": 32140
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.29391582799634e-06,
+      "loss": 0.1686,
+      "step": 32150
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.287053979871912e-06,
+      "loss": 0.2032,
+      "step": 32160
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.280192131747485e-06,
+      "loss": 0.1821,
+      "step": 32170
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.273330283623056e-06,
+      "loss": 0.2165,
+      "step": 32180
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.266468435498627e-06,
+      "loss": 0.2211,
+      "step": 32190
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.2596065873742e-06,
+      "loss": 0.2141,
+      "step": 32200
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.252744739249772e-06,
+      "loss": 0.1821,
+      "step": 32210
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.245882891125343e-06,
+      "loss": 0.1844,
+      "step": 32220
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.239021043000916e-06,
+      "loss": 0.1704,
+      "step": 32230
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.232159194876487e-06,
+      "loss": 0.2166,
+      "step": 32240
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.225297346752059e-06,
+      "loss": 0.2503,
+      "step": 32250
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.218435498627632e-06,
+      "loss": 0.2659,
+      "step": 32260
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.211573650503203e-06,
+      "loss": 0.2019,
+      "step": 32270
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.204711802378774e-06,
+      "loss": 0.2226,
+      "step": 32280
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.197849954254347e-06,
+      "loss": 0.2465,
+      "step": 32290
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.190988106129919e-06,
+      "loss": 0.1495,
+      "step": 32300
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.18412625800549e-06,
+      "loss": 0.2183,
+      "step": 32310
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.177264409881063e-06,
+      "loss": 0.2211,
+      "step": 32320
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.170402561756634e-06,
+      "loss": 0.2264,
+      "step": 32330
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.163540713632204e-06,
+      "loss": 0.157,
+      "step": 32340
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.156678865507777e-06,
+      "loss": 0.2003,
+      "step": 32350
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.149817017383348e-06,
+      "loss": 0.2368,
+      "step": 32360
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.14295516925892e-06,
+      "loss": 0.1668,
+      "step": 32370
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.136093321134492e-06,
+      "loss": 0.2735,
+      "step": 32380
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.129231473010064e-06,
+      "loss": 0.1345,
+      "step": 32390
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.122369624885635e-06,
+      "loss": 0.2367,
+      "step": 32400
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.115507776761208e-06,
+      "loss": 0.2028,
+      "step": 32410
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.10864592863678e-06,
+      "loss": 0.184,
+      "step": 32420
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.10178408051235e-06,
+      "loss": 0.1757,
+      "step": 32430
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.094922232387924e-06,
+      "loss": 0.2158,
+      "step": 32440
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.088060384263495e-06,
+      "loss": 0.1849,
+      "step": 32450
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.081198536139066e-06,
+      "loss": 0.1802,
+      "step": 32460
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.07433668801464e-06,
+      "loss": 0.2045,
+      "step": 32470
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.06747483989021e-06,
+      "loss": 0.2037,
+      "step": 32480
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.060612991765782e-06,
+      "loss": 0.185,
+      "step": 32490
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.053751143641355e-06,
+      "loss": 0.1899,
+      "step": 32500
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.046889295516926e-06,
+      "loss": 0.1493,
+      "step": 32510
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.040027447392498e-06,
+      "loss": 0.2063,
+      "step": 32520
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.03316559926807e-06,
+      "loss": 0.1829,
+      "step": 32530
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.026303751143642e-06,
+      "loss": 0.1101,
+      "step": 32540
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.019441903019213e-06,
+      "loss": 0.1757,
+      "step": 32550
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.012580054894784e-06,
+      "loss": 0.2094,
+      "step": 32560
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.005718206770357e-06,
+      "loss": 0.2304,
+      "step": 32570
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.998856358645929e-06,
+      "loss": 0.2641,
+      "step": 32580
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.9919945105215e-06,
+      "loss": 0.166,
+      "step": 32590
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.985132662397073e-06,
+      "loss": 0.2086,
+      "step": 32600
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.978270814272644e-06,
+      "loss": 0.18,
+      "step": 32610
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.971408966148216e-06,
+      "loss": 0.1881,
+      "step": 32620
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.964547118023789e-06,
+      "loss": 0.1644,
+      "step": 32630
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.95768526989936e-06,
+      "loss": 0.2264,
+      "step": 32640
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.950823421774931e-06,
+      "loss": 0.1285,
+      "step": 32650
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.943961573650504e-06,
+      "loss": 0.2177,
+      "step": 32660
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.937099725526076e-06,
+      "loss": 0.1955,
+      "step": 32670
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.930237877401647e-06,
+      "loss": 0.3307,
+      "step": 32680
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.92337602927722e-06,
+      "loss": 0.1877,
+      "step": 32690
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.916514181152791e-06,
+      "loss": 0.1361,
+      "step": 32700
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.909652333028363e-06,
+      "loss": 0.2078,
+      "step": 32710
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.902790484903936e-06,
+      "loss": 0.1959,
+      "step": 32720
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.895928636779507e-06,
+      "loss": 0.1627,
+      "step": 32730
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.889066788655078e-06,
+      "loss": 0.2289,
+      "step": 32740
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.88220494053065e-06,
+      "loss": 0.2079,
+      "step": 32750
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.87534309240622e-06,
+      "loss": 0.2219,
+      "step": 32760
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.868481244281792e-06,
+      "loss": 0.223,
+      "step": 32770
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.861619396157365e-06,
+      "loss": 0.2145,
+      "step": 32780
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.854757548032936e-06,
+      "loss": 0.2396,
+      "step": 32790
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.847895699908508e-06,
+      "loss": 0.1979,
+      "step": 32800
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.84103385178408e-06,
+      "loss": 0.1723,
+      "step": 32810
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.834172003659652e-06,
+      "loss": 0.2779,
+      "step": 32820
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.827310155535223e-06,
+      "loss": 0.2006,
+      "step": 32830
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.820448307410796e-06,
+      "loss": 0.2467,
+      "step": 32840
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.813586459286368e-06,
+      "loss": 0.1544,
+      "step": 32850
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.806724611161939e-06,
+      "loss": 0.1618,
+      "step": 32860
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.799862763037512e-06,
+      "loss": 0.2579,
+      "step": 32870
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.793000914913083e-06,
+      "loss": 0.2167,
+      "step": 32880
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.786139066788655e-06,
+      "loss": 0.2047,
+      "step": 32890
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.779277218664228e-06,
+      "loss": 0.1659,
+      "step": 32900
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.772415370539799e-06,
+      "loss": 0.248,
+      "step": 32910
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.76555352241537e-06,
+      "loss": 0.2349,
+      "step": 32920
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.758691674290943e-06,
+      "loss": 0.2173,
+      "step": 32930
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.751829826166514e-06,
+      "loss": 0.2071,
+      "step": 32940
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.744967978042086e-06,
+      "loss": 0.196,
+      "step": 32950
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.738106129917659e-06,
+      "loss": 0.1618,
+      "step": 32960
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.73124428179323e-06,
+      "loss": 0.152,
+      "step": 32970
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.724382433668801e-06,
+      "loss": 0.1733,
+      "step": 32980
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.717520585544374e-06,
+      "loss": 0.166,
+      "step": 32990
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.710658737419946e-06,
+      "loss": 0.1874,
+      "step": 33000
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.703796889295517e-06,
+      "loss": 0.1439,
+      "step": 33010
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.69693504117109e-06,
+      "loss": 0.1783,
+      "step": 33020
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.690073193046661e-06,
+      "loss": 0.2511,
+      "step": 33030
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.683211344922233e-06,
+      "loss": 0.1644,
+      "step": 33040
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.676349496797806e-06,
+      "loss": 0.1266,
+      "step": 33050
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.669487648673377e-06,
+      "loss": 0.2047,
+      "step": 33060
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.662625800548948e-06,
+      "loss": 0.1567,
+      "step": 33070
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.655763952424521e-06,
+      "loss": 0.1777,
+      "step": 33080
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.648902104300093e-06,
+      "loss": 0.298,
+      "step": 33090
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.642040256175664e-06,
+      "loss": 0.3215,
+      "step": 33100
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.635178408051237e-06,
+      "loss": 0.1722,
+      "step": 33110
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.628316559926808e-06,
+      "loss": 0.1289,
+      "step": 33120
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.621454711802379e-06,
+      "loss": 0.1872,
+      "step": 33130
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.6145928636779516e-06,
+      "loss": 0.2006,
+      "step": 33140
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.607731015553522e-06,
+      "loss": 0.2227,
+      "step": 33150
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.600869167429093e-06,
+      "loss": 0.4038,
+      "step": 33160
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.5940073193046655e-06,
+      "loss": 0.1624,
+      "step": 33170
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.587145471180238e-06,
+      "loss": 0.2187,
+      "step": 33180
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.580283623055809e-06,
+      "loss": 0.2348,
+      "step": 33190
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.573421774931381e-06,
+      "loss": 0.1281,
+      "step": 33200
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.566559926806953e-06,
+      "loss": 0.1691,
+      "step": 33210
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.5596980786825246e-06,
+      "loss": 0.1557,
+      "step": 33220
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.552836230558097e-06,
+      "loss": 0.2015,
+      "step": 33230
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.545974382433669e-06,
+      "loss": 0.1826,
+      "step": 33240
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.53911253430924e-06,
+      "loss": 0.2411,
+      "step": 33250
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.532250686184812e-06,
+      "loss": 0.2521,
+      "step": 33260
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.5253888380603845e-06,
+      "loss": 0.1325,
+      "step": 33270
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.518526989935956e-06,
+      "loss": 0.2048,
+      "step": 33280
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.511665141811528e-06,
+      "loss": 0.2402,
+      "step": 33290
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.5048032936871e-06,
+      "loss": 0.2156,
+      "step": 33300
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.497941445562671e-06,
+      "loss": 0.2542,
+      "step": 33310
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.4910795974382436e-06,
+      "loss": 0.1651,
+      "step": 33320
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.484217749313816e-06,
+      "loss": 0.1714,
+      "step": 33330
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.477355901189387e-06,
+      "loss": 0.1599,
+      "step": 33340
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.470494053064959e-06,
+      "loss": 0.1943,
+      "step": 33350
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.4636322049405305e-06,
+      "loss": 0.217,
+      "step": 33360
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.456770356816103e-06,
+      "loss": 0.2344,
+      "step": 33370
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.449908508691675e-06,
+      "loss": 0.1684,
+      "step": 33380
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.443046660567246e-06,
+      "loss": 0.1903,
+      "step": 33390
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.436184812442818e-06,
+      "loss": 0.2313,
+      "step": 33400
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.42932296431839e-06,
+      "loss": 0.1908,
+      "step": 33410
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.422461116193962e-06,
+      "loss": 0.2482,
+      "step": 33420
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.415599268069534e-06,
+      "loss": 0.2242,
+      "step": 33430
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.408737419945106e-06,
+      "loss": 0.1636,
+      "step": 33440
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.4018755718206765e-06,
+      "loss": 0.2369,
+      "step": 33450
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.395013723696249e-06,
+      "loss": 0.2346,
+      "step": 33460
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.388151875571821e-06,
+      "loss": 0.1829,
+      "step": 33470
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.381290027447392e-06,
+      "loss": 0.1437,
+      "step": 33480
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.374428179322964e-06,
+      "loss": 0.2071,
+      "step": 33490
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.367566331198536e-06,
+      "loss": 0.2142,
+      "step": 33500
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.360704483074108e-06,
+      "loss": 0.2565,
+      "step": 33510
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.35384263494968e-06,
+      "loss": 0.1828,
+      "step": 33520
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.346980786825252e-06,
+      "loss": 0.2543,
+      "step": 33530
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.340118938700823e-06,
+      "loss": 0.2204,
+      "step": 33540
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.3332570905763955e-06,
+      "loss": 0.1102,
+      "step": 33550
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.326395242451968e-06,
+      "loss": 0.2764,
+      "step": 33560
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.319533394327539e-06,
+      "loss": 0.2547,
+      "step": 33570
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.312671546203111e-06,
+      "loss": 0.2329,
+      "step": 33580
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.305809698078683e-06,
+      "loss": 0.34,
+      "step": 33590
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.2989478499542546e-06,
+      "loss": 0.2626,
+      "step": 33600
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.292086001829827e-06,
+      "loss": 0.2078,
+      "step": 33610
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.285224153705399e-06,
+      "loss": 0.1842,
+      "step": 33620
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.27836230558097e-06,
+      "loss": 0.2142,
+      "step": 33630
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.271500457456542e-06,
+      "loss": 0.1809,
+      "step": 33640
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.264638609332114e-06,
+      "loss": 0.2441,
+      "step": 33650
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.257776761207685e-06,
+      "loss": 0.1817,
+      "step": 33660
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.250914913083257e-06,
+      "loss": 0.2123,
+      "step": 33670
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.244053064958828e-06,
+      "loss": 0.1947,
+      "step": 33680
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.2371912168344005e-06,
+      "loss": 0.227,
+      "step": 33690
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.230329368709973e-06,
+      "loss": 0.19,
+      "step": 33700
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.223467520585544e-06,
+      "loss": 0.1402,
+      "step": 33710
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.216605672461116e-06,
+      "loss": 0.1821,
+      "step": 33720
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.209743824336688e-06,
+      "loss": 0.1802,
+      "step": 33730
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.20288197621226e-06,
+      "loss": 0.2102,
+      "step": 33740
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.196020128087832e-06,
+      "loss": 0.1894,
+      "step": 33750
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.189158279963404e-06,
+      "loss": 0.1715,
+      "step": 33760
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.182296431838975e-06,
+      "loss": 0.179,
+      "step": 33770
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.175434583714547e-06,
+      "loss": 0.2505,
+      "step": 33780
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.1685727355901195e-06,
+      "loss": 0.2965,
+      "step": 33790
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.161710887465691e-06,
+      "loss": 0.2152,
+      "step": 33800
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.154849039341263e-06,
+      "loss": 0.1998,
+      "step": 33810
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.147987191216835e-06,
+      "loss": 0.2739,
+      "step": 33820
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.1411253430924065e-06,
+      "loss": 0.1407,
+      "step": 33830
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.134263494967979e-06,
+      "loss": 0.1883,
+      "step": 33840
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.127401646843551e-06,
+      "loss": 0.2018,
+      "step": 33850
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.120539798719121e-06,
+      "loss": 0.255,
+      "step": 33860
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.113677950594693e-06,
+      "loss": 0.1824,
+      "step": 33870
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.1068161024702655e-06,
+      "loss": 0.1465,
+      "step": 33880
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.099954254345837e-06,
+      "loss": 0.1512,
+      "step": 33890
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.093092406221409e-06,
+      "loss": 0.1537,
+      "step": 33900
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.086230558096981e-06,
+      "loss": 0.3078,
+      "step": 33910
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.0793687099725525e-06,
+      "loss": 0.1759,
+      "step": 33920
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.072506861848125e-06,
+      "loss": 0.1954,
+      "step": 33930
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.065645013723697e-06,
+      "loss": 0.1973,
+      "step": 33940
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.058783165599268e-06,
+      "loss": 0.2289,
+      "step": 33950
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.05192131747484e-06,
+      "loss": 0.2425,
+      "step": 33960
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.0450594693504115e-06,
+      "loss": 0.1437,
+      "step": 33970
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.038197621225984e-06,
+      "loss": 0.2208,
+      "step": 33980
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.031335773101556e-06,
+      "loss": 0.1571,
+      "step": 33990
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.024473924977127e-06,
+      "loss": 0.2357,
+      "step": 34000
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.017612076852699e-06,
+      "loss": 0.1647,
+      "step": 34010
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.0107502287282715e-06,
+      "loss": 0.1468,
+      "step": 34020
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.003888380603843e-06,
+      "loss": 0.2431,
+      "step": 34030
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.997026532479415e-06,
+      "loss": 0.2198,
+      "step": 34040
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.990164684354987e-06,
+      "loss": 0.1617,
+      "step": 34050
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.9833028362305575e-06,
+      "loss": 0.1887,
+      "step": 34060
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.97644098810613e-06,
+      "loss": 0.163,
+      "step": 34070
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.969579139981702e-06,
+      "loss": 0.163,
+      "step": 34080
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.962717291857273e-06,
+      "loss": 0.225,
+      "step": 34090
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.955855443732845e-06,
+      "loss": 0.2101,
+      "step": 34100
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.9489935956084175e-06,
+      "loss": 0.1956,
+      "step": 34110
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.942131747483989e-06,
+      "loss": 0.2421,
+      "step": 34120
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.935269899359561e-06,
+      "loss": 0.172,
+      "step": 34130
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.928408051235133e-06,
+      "loss": 0.2032,
+      "step": 34140
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.921546203110704e-06,
+      "loss": 0.1925,
+      "step": 34150
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.9146843549862765e-06,
+      "loss": 0.2591,
+      "step": 34160
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.907822506861849e-06,
+      "loss": 0.1576,
+      "step": 34170
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.90096065873742e-06,
+      "loss": 0.1682,
+      "step": 34180
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.894098810612992e-06,
+      "loss": 0.1411,
+      "step": 34190
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.887236962488564e-06,
+      "loss": 0.1998,
+      "step": 34200
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.880375114364136e-06,
+      "loss": 0.2171,
+      "step": 34210
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.873513266239708e-06,
+      "loss": 0.1699,
+      "step": 34220
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.86665141811528e-06,
+      "loss": 0.1501,
+      "step": 34230
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.859789569990851e-06,
+      "loss": 0.1655,
+      "step": 34240
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.852927721866423e-06,
+      "loss": 0.1217,
+      "step": 34250
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.846065873741995e-06,
+      "loss": 0.1782,
+      "step": 34260
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.839204025617566e-06,
+      "loss": 0.1709,
+      "step": 34270
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.832342177493138e-06,
+      "loss": 0.206,
+      "step": 34280
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.8254803293687094e-06,
+      "loss": 0.9682,
+      "step": 34290
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.818618481244282e-06,
+      "loss": 0.202,
+      "step": 34300
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.811756633119854e-06,
+      "loss": 0.1803,
+      "step": 34310
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.804894784995425e-06,
+      "loss": 0.3094,
+      "step": 34320
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.798032936870997e-06,
+      "loss": 0.1938,
+      "step": 34330
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.791171088746569e-06,
+      "loss": 0.1873,
+      "step": 34340
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.784309240622141e-06,
+      "loss": 0.1659,
+      "step": 34350
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.777447392497713e-06,
+      "loss": 0.1781,
+      "step": 34360
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.770585544373285e-06,
+      "loss": 0.1741,
+      "step": 34370
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.763723696248856e-06,
+      "loss": 0.2579,
+      "step": 34380
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.7568618481244284e-06,
+      "loss": 0.1743,
+      "step": 34390
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 0.2433,
+      "step": 34400
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.743138151875572e-06,
+      "loss": 0.2302,
+      "step": 34410
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.736276303751144e-06,
+      "loss": 0.2081,
+      "step": 34420
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.729414455626716e-06,
+      "loss": 0.1804,
+      "step": 34430
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.7225526075022875e-06,
+      "loss": 0.1947,
+      "step": 34440
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.71569075937786e-06,
+      "loss": 0.2728,
+      "step": 34450
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.708828911253432e-06,
+      "loss": 0.1531,
+      "step": 34460
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.701967063129002e-06,
+      "loss": 0.1827,
+      "step": 34470
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.6951052150045744e-06,
+      "loss": 0.222,
+      "step": 34480
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.688243366880147e-06,
+      "loss": 0.2782,
+      "step": 34490
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.681381518755718e-06,
+      "loss": 0.1257,
+      "step": 34500
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.67451967063129e-06,
+      "loss": 0.2131,
+      "step": 34510
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.667657822506862e-06,
+      "loss": 0.2205,
+      "step": 34520
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.6607959743824335e-06,
+      "loss": 0.1824,
+      "step": 34530
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.653934126258006e-06,
+      "loss": 0.1649,
+      "step": 34540
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.647072278133578e-06,
+      "loss": 0.2218,
+      "step": 34550
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.640210430009149e-06,
+      "loss": 0.204,
+      "step": 34560
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.633348581884721e-06,
+      "loss": 0.1807,
+      "step": 34570
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.6264867337602934e-06,
+      "loss": 0.1485,
+      "step": 34580
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.619624885635865e-06,
+      "loss": 0.1595,
+      "step": 34590
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.612763037511437e-06,
+      "loss": 0.1913,
+      "step": 34600
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.605901189387008e-06,
+      "loss": 0.2664,
+      "step": 34610
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.59903934126258e-06,
+      "loss": 0.2581,
+      "step": 34620
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.5921774931381525e-06,
+      "loss": 0.2341,
+      "step": 34630
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.585315645013724e-06,
+      "loss": 0.1176,
+      "step": 34640
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.578453796889296e-06,
+      "loss": 0.2918,
+      "step": 34650
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.571591948764868e-06,
+      "loss": 0.1978,
+      "step": 34660
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.564730100640439e-06,
+      "loss": 0.1465,
+      "step": 34670
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.557868252516011e-06,
+      "loss": 0.2302,
+      "step": 34680
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.551006404391583e-06,
+      "loss": 0.1866,
+      "step": 34690
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.544144556267154e-06,
+      "loss": 0.1639,
+      "step": 34700
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.537282708142726e-06,
+      "loss": 0.1706,
+      "step": 34710
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.5304208600182985e-06,
+      "loss": 0.1458,
+      "step": 34720
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.52355901189387e-06,
+      "loss": 0.1987,
+      "step": 34730
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.516697163769442e-06,
+      "loss": 0.1419,
+      "step": 34740
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.509835315645014e-06,
+      "loss": 0.1614,
+      "step": 34750
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.5029734675205854e-06,
+      "loss": 0.2977,
+      "step": 34760
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.496111619396158e-06,
+      "loss": 0.1899,
+      "step": 34770
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.48924977127173e-06,
+      "loss": 0.2396,
+      "step": 34780
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.482387923147301e-06,
+      "loss": 0.2268,
+      "step": 34790
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.475526075022873e-06,
+      "loss": 0.1679,
+      "step": 34800
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.468664226898445e-06,
+      "loss": 0.1826,
+      "step": 34810
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.461802378774017e-06,
+      "loss": 0.2445,
+      "step": 34820
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.454940530649589e-06,
+      "loss": 0.1942,
+      "step": 34830
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.448078682525161e-06,
+      "loss": 0.1397,
+      "step": 34840
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.441216834400732e-06,
+      "loss": 0.1567,
+      "step": 34850
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.434354986276304e-06,
+      "loss": 0.1604,
+      "step": 34860
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.427493138151876e-06,
+      "loss": 0.1846,
+      "step": 34870
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.420631290027447e-06,
+      "loss": 0.1912,
+      "step": 34880
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.413769441903019e-06,
+      "loss": 0.1163,
+      "step": 34890
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.406907593778591e-06,
+      "loss": 0.1297,
+      "step": 34900
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.400045745654163e-06,
+      "loss": 0.1371,
+      "step": 34910
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.393183897529735e-06,
+      "loss": 0.1437,
+      "step": 34920
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.386322049405306e-06,
+      "loss": 0.1805,
+      "step": 34930
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.379460201280878e-06,
+      "loss": 0.166,
+      "step": 34940
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.37259835315645e-06,
+      "loss": 0.1942,
+      "step": 34950
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.365736505032022e-06,
+      "loss": 0.1805,
+      "step": 34960
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.358874656907594e-06,
+      "loss": 0.1927,
+      "step": 34970
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.352012808783166e-06,
+      "loss": 0.1394,
+      "step": 34980
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.345150960658737e-06,
+      "loss": 0.1819,
+      "step": 34990
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.3382891125343095e-06,
+      "loss": 0.188,
+      "step": 35000
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.331427264409882e-06,
+      "loss": 0.1912,
+      "step": 35010
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.324565416285453e-06,
+      "loss": 0.2114,
+      "step": 35020
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.317703568161025e-06,
+      "loss": 0.1867,
+      "step": 35030
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.310841720036597e-06,
+      "loss": 0.148,
+      "step": 35040
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.3039798719121686e-06,
+      "loss": 0.1769,
+      "step": 35050
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.297118023787741e-06,
+      "loss": 0.2356,
+      "step": 35060
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.290256175663312e-06,
+      "loss": 0.19,
+      "step": 35070
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.283394327538883e-06,
+      "loss": 0.197,
+      "step": 35080
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.2765324794144555e-06,
+      "loss": 0.1858,
+      "step": 35090
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.269670631290028e-06,
+      "loss": 0.2624,
+      "step": 35100
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.262808783165599e-06,
+      "loss": 0.2221,
+      "step": 35110
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.255946935041171e-06,
+      "loss": 0.1832,
+      "step": 35120
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.249085086916743e-06,
+      "loss": 0.2237,
+      "step": 35130
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.2422232387923146e-06,
+      "loss": 0.1924,
+      "step": 35140
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.235361390667887e-06,
+      "loss": 0.2554,
+      "step": 35150
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.228499542543459e-06,
+      "loss": 0.1962,
+      "step": 35160
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.22163769441903e-06,
+      "loss": 0.1864,
+      "step": 35170
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.214775846294602e-06,
+      "loss": 0.2805,
+      "step": 35180
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.2079139981701745e-06,
+      "loss": 0.1907,
+      "step": 35190
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.201052150045746e-06,
+      "loss": 0.199,
+      "step": 35200
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.194190301921318e-06,
+      "loss": 0.1596,
+      "step": 35210
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.187328453796889e-06,
+      "loss": 0.2198,
+      "step": 35220
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.180466605672461e-06,
+      "loss": 0.1431,
+      "step": 35230
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.1736047575480336e-06,
+      "loss": 0.2504,
+      "step": 35240
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.166742909423605e-06,
+      "loss": 0.221,
+      "step": 35250
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.159881061299177e-06,
+      "loss": 0.217,
+      "step": 35260
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.153019213174749e-06,
+      "loss": 0.2016,
+      "step": 35270
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.14615736505032e-06,
+      "loss": 0.1789,
+      "step": 35280
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.139295516925892e-06,
+      "loss": 0.1477,
+      "step": 35290
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.132433668801464e-06,
+      "loss": 0.2234,
+      "step": 35300
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.125571820677035e-06,
+      "loss": 0.1323,
+      "step": 35310
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.118709972552607e-06,
+      "loss": 0.1551,
+      "step": 35320
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.1118481244281796e-06,
+      "loss": 0.2966,
+      "step": 35330
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.104986276303751e-06,
+      "loss": 0.1824,
+      "step": 35340
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.098124428179323e-06,
+      "loss": 0.1559,
+      "step": 35350
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.091262580054895e-06,
+      "loss": 0.1873,
+      "step": 35360
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.0844007319304665e-06,
+      "loss": 0.1637,
+      "step": 35370
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 0.4582900106906891,
+      "eval_runtime": 303.0304,
+      "eval_samples_per_second": 6.484,
+      "eval_steps_per_second": 1.624,
+      "eval_wer": 0.19096173988047563,
+      "step": 35376
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.077538883806039e-06,
+      "loss": 0.2064,
+      "step": 35380
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.070677035681611e-06,
+      "loss": 0.1978,
+      "step": 35390
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.063815187557182e-06,
+      "loss": 0.1581,
+      "step": 35400
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.056953339432754e-06,
+      "loss": 0.146,
+      "step": 35410
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.050091491308326e-06,
+      "loss": 0.1787,
+      "step": 35420
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.043229643183898e-06,
+      "loss": 0.1324,
+      "step": 35430
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.03636779505947e-06,
+      "loss": 0.1484,
+      "step": 35440
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.029505946935042e-06,
+      "loss": 0.1725,
+      "step": 35450
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.022644098810613e-06,
+      "loss": 0.1991,
+      "step": 35460
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.0157822506861855e-06,
+      "loss": 0.1892,
+      "step": 35470
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.008920402561757e-06,
+      "loss": 0.1763,
+      "step": 35480
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.002058554437328e-06,
+      "loss": 0.1893,
+      "step": 35490
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.9951967063129e-06,
+      "loss": 0.1576,
+      "step": 35500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.988334858188472e-06,
+      "loss": 0.162,
+      "step": 35510
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.981473010064044e-06,
+      "loss": 0.199,
+      "step": 35520
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.974611161939616e-06,
+      "loss": 0.1856,
+      "step": 35530
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.967749313815187e-06,
+      "loss": 0.1523,
+      "step": 35540
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.960887465690759e-06,
+      "loss": 0.1695,
+      "step": 35550
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.9540256175663315e-06,
+      "loss": 0.2563,
+      "step": 35560
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.947163769441903e-06,
+      "loss": 0.1529,
+      "step": 35570
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.940301921317475e-06,
+      "loss": 0.1881,
+      "step": 35580
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.933440073193047e-06,
+      "loss": 0.2358,
+      "step": 35590
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.926578225068618e-06,
+      "loss": 0.1455,
+      "step": 35600
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.9197163769441905e-06,
+      "loss": 0.1651,
+      "step": 35610
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.912854528819763e-06,
+      "loss": 0.2426,
+      "step": 35620
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.905992680695334e-06,
+      "loss": 0.1574,
+      "step": 35630
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.899130832570906e-06,
+      "loss": 0.1894,
+      "step": 35640
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.892268984446478e-06,
+      "loss": 0.1932,
+      "step": 35650
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.88540713632205e-06,
+      "loss": 0.1633,
+      "step": 35660
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.878545288197622e-06,
+      "loss": 0.1881,
+      "step": 35670
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.871683440073193e-06,
+      "loss": 0.1404,
+      "step": 35680
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.864821591948764e-06,
+      "loss": 0.1449,
+      "step": 35690
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.8579597438243365e-06,
+      "loss": 0.1547,
+      "step": 35700
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.851097895699909e-06,
+      "loss": 0.2717,
+      "step": 35710
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.84423604757548e-06,
+      "loss": 0.1491,
+      "step": 35720
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.837374199451052e-06,
+      "loss": 0.2079,
+      "step": 35730
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.830512351326624e-06,
+      "loss": 0.1282,
+      "step": 35740
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.823650503202196e-06,
+      "loss": 0.2286,
+      "step": 35750
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.816788655077768e-06,
+      "loss": 0.1622,
+      "step": 35760
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.80992680695334e-06,
+      "loss": 0.1652,
+      "step": 35770
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.803064958828911e-06,
+      "loss": 0.1757,
+      "step": 35780
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.796203110704483e-06,
+      "loss": 0.162,
+      "step": 35790
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.7893412625800555e-06,
+      "loss": 0.2181,
+      "step": 35800
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.782479414455627e-06,
+      "loss": 0.1435,
+      "step": 35810
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.775617566331199e-06,
+      "loss": 0.1874,
+      "step": 35820
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.768755718206771e-06,
+      "loss": 0.1927,
+      "step": 35830
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.7618938700823425e-06,
+      "loss": 0.1408,
+      "step": 35840
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.755032021957915e-06,
+      "loss": 0.2473,
+      "step": 35850
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.748170173833486e-06,
+      "loss": 0.1391,
+      "step": 35860
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.741308325709058e-06,
+      "loss": 0.1975,
+      "step": 35870
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.734446477584629e-06,
+      "loss": 0.1671,
+      "step": 35880
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.728270814272644e-06,
+      "loss": 0.2111,
+      "step": 35890
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.721408966148216e-06,
+      "loss": 0.1942,
+      "step": 35900
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.714547118023788e-06,
+      "loss": 0.1505,
+      "step": 35910
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.7076852698993595e-06,
+      "loss": 0.1601,
+      "step": 35920
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.700823421774932e-06,
+      "loss": 0.1986,
+      "step": 35930
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.693961573650504e-06,
+      "loss": 0.1908,
+      "step": 35940
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.687099725526075e-06,
+      "loss": 0.2239,
+      "step": 35950
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.680237877401647e-06,
+      "loss": 0.1734,
+      "step": 35960
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.673376029277219e-06,
+      "loss": 0.1847,
+      "step": 35970
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.666514181152791e-06,
+      "loss": 0.2041,
+      "step": 35980
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.659652333028363e-06,
+      "loss": 0.1841,
+      "step": 35990
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.652790484903935e-06,
+      "loss": 0.1177,
+      "step": 36000
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.6459286367795055e-06,
+      "loss": 0.1465,
+      "step": 36010
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.639066788655078e-06,
+      "loss": 0.1663,
+      "step": 36020
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.63220494053065e-06,
+      "loss": 0.162,
+      "step": 36030
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.625343092406221e-06,
+      "loss": 0.1889,
+      "step": 36040
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.618481244281793e-06,
+      "loss": 0.2089,
+      "step": 36050
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.611619396157365e-06,
+      "loss": 0.1175,
+      "step": 36060
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.604757548032937e-06,
+      "loss": 0.2177,
+      "step": 36070
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.597895699908509e-06,
+      "loss": 0.1756,
+      "step": 36080
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.59103385178408e-06,
+      "loss": 0.1573,
+      "step": 36090
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.584172003659652e-06,
+      "loss": 0.2408,
+      "step": 36100
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.5773101555352245e-06,
+      "loss": 0.14,
+      "step": 36110
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.570448307410796e-06,
+      "loss": 0.1539,
+      "step": 36120
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.563586459286368e-06,
+      "loss": 0.1572,
+      "step": 36130
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.55672461116194e-06,
+      "loss": 0.1999,
+      "step": 36140
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.549862763037511e-06,
+      "loss": 0.1484,
+      "step": 36150
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.5430009149130836e-06,
+      "loss": 0.3285,
+      "step": 36160
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.536139066788656e-06,
+      "loss": 0.1595,
+      "step": 36170
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.529277218664227e-06,
+      "loss": 0.1631,
+      "step": 36180
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.522415370539799e-06,
+      "loss": 0.159,
+      "step": 36190
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.515553522415371e-06,
+      "loss": 0.1141,
+      "step": 36200
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.508691674290942e-06,
+      "loss": 0.186,
+      "step": 36210
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.501829826166514e-06,
+      "loss": 0.1553,
+      "step": 36220
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.494967978042086e-06,
+      "loss": 0.1847,
+      "step": 36230
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.488106129917657e-06,
+      "loss": 0.1606,
+      "step": 36240
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.4812442817932296e-06,
+      "loss": 0.1958,
+      "step": 36250
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.474382433668802e-06,
+      "loss": 0.2162,
+      "step": 36260
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.467520585544373e-06,
+      "loss": 0.1584,
+      "step": 36270
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.460658737419945e-06,
+      "loss": 0.2105,
+      "step": 36280
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.453796889295517e-06,
+      "loss": 0.137,
+      "step": 36290
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.446935041171089e-06,
+      "loss": 0.1395,
+      "step": 36300
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.440073193046661e-06,
+      "loss": 0.1103,
+      "step": 36310
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.433211344922233e-06,
+      "loss": 0.2138,
+      "step": 36320
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.426349496797804e-06,
+      "loss": 0.2044,
+      "step": 36330
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.419487648673376e-06,
+      "loss": 0.1699,
+      "step": 36340
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.4126258005489486e-06,
+      "loss": 0.1371,
+      "step": 36350
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.40576395242452e-06,
+      "loss": 0.1792,
+      "step": 36360
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.398902104300092e-06,
+      "loss": 0.1871,
+      "step": 36370
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.392040256175664e-06,
+      "loss": 0.198,
+      "step": 36380
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.3851784080512355e-06,
+      "loss": 0.1584,
+      "step": 36390
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.378316559926808e-06,
+      "loss": 0.17,
+      "step": 36400
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.371454711802378e-06,
+      "loss": 0.1815,
+      "step": 36410
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.36459286367795e-06,
+      "loss": 0.2417,
+      "step": 36420
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.357731015553522e-06,
+      "loss": 0.1896,
+      "step": 36430
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.350869167429094e-06,
+      "loss": 0.1477,
+      "step": 36440
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.344007319304666e-06,
+      "loss": 0.1507,
+      "step": 36450
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.337145471180238e-06,
+      "loss": 0.232,
+      "step": 36460
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.330283623055809e-06,
+      "loss": 0.1338,
+      "step": 36470
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.3234217749313815e-06,
+      "loss": 0.1331,
+      "step": 36480
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.316559926806954e-06,
+      "loss": 0.2258,
+      "step": 36490
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.309698078682525e-06,
+      "loss": 0.1652,
+      "step": 36500
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.302836230558097e-06,
+      "loss": 0.1247,
+      "step": 36510
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.295974382433669e-06,
+      "loss": 0.1699,
+      "step": 36520
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.2891125343092406e-06,
+      "loss": 0.1326,
+      "step": 36530
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.282250686184813e-06,
+      "loss": 0.1817,
+      "step": 36540
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.275388838060385e-06,
+      "loss": 0.1661,
+      "step": 36550
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.268526989935956e-06,
+      "loss": 0.2227,
+      "step": 36560
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.261665141811528e-06,
+      "loss": 0.1551,
+      "step": 36570
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.2548032936871005e-06,
+      "loss": 0.1613,
+      "step": 36580
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.247941445562672e-06,
+      "loss": 0.2249,
+      "step": 36590
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.241079597438244e-06,
+      "loss": 0.1694,
+      "step": 36600
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.234217749313815e-06,
+      "loss": 0.1342,
+      "step": 36610
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.2273559011893865e-06,
+      "loss": 0.2176,
+      "step": 36620
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.220494053064959e-06,
+      "loss": 0.15,
+      "step": 36630
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.213632204940531e-06,
+      "loss": 0.2507,
+      "step": 36640
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.206770356816102e-06,
+      "loss": 0.1604,
+      "step": 36650
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.199908508691674e-06,
+      "loss": 0.1693,
+      "step": 36660
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.1930466605672465e-06,
+      "loss": 0.1383,
+      "step": 36670
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.186184812442818e-06,
+      "loss": 0.1587,
+      "step": 36680
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.17932296431839e-06,
+      "loss": 0.1149,
+      "step": 36690
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.172461116193962e-06,
+      "loss": 0.1754,
+      "step": 36700
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.165599268069533e-06,
+      "loss": 0.1623,
+      "step": 36710
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.1587374199451055e-06,
+      "loss": 0.1494,
+      "step": 36720
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.151875571820677e-06,
+      "loss": 0.2062,
+      "step": 36730
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.145013723696249e-06,
+      "loss": 0.1842,
+      "step": 36740
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.138151875571821e-06,
+      "loss": 0.1557,
+      "step": 36750
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.131976212259835e-06,
+      "loss": 0.153,
+      "step": 36760
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.125114364135407e-06,
+      "loss": 0.1822,
+      "step": 36770
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.118252516010979e-06,
+      "loss": 0.1415,
+      "step": 36780
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.1113906678865504e-06,
+      "loss": 0.1411,
+      "step": 36790
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.104528819762123e-06,
+      "loss": 0.2861,
+      "step": 36800
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.097666971637695e-06,
+      "loss": 0.1518,
+      "step": 36810
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.090805123513266e-06,
+      "loss": 0.1408,
+      "step": 36820
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.083943275388838e-06,
+      "loss": 0.2161,
+      "step": 36830
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.07708142726441e-06,
+      "loss": 0.1295,
+      "step": 36840
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.070219579139982e-06,
+      "loss": 0.1481,
+      "step": 36850
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.063357731015554e-06,
+      "loss": 0.1523,
+      "step": 36860
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.056495882891126e-06,
+      "loss": 0.1584,
+      "step": 36870
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.049634034766697e-06,
+      "loss": 0.135,
+      "step": 36880
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.0427721866422694e-06,
+      "loss": 0.1442,
+      "step": 36890
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.035910338517842e-06,
+      "loss": 0.1879,
+      "step": 36900
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.029048490393413e-06,
+      "loss": 0.159,
+      "step": 36910
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.022186642268985e-06,
+      "loss": 0.2,
+      "step": 36920
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.015324794144557e-06,
+      "loss": 0.2588,
+      "step": 36930
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.008462946020128e-06,
+      "loss": 0.1652,
+      "step": 36940
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.0016010978957e-06,
+      "loss": 0.1574,
+      "step": 36950
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.994739249771272e-06,
+      "loss": 0.1646,
+      "step": 36960
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.987877401646843e-06,
+      "loss": 0.1646,
+      "step": 36970
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.9810155535224154e-06,
+      "loss": 0.1666,
+      "step": 36980
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.974153705397987e-06,
+      "loss": 0.2045,
+      "step": 36990
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.967291857273559e-06,
+      "loss": 0.1601,
+      "step": 37000
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.960430009149131e-06,
+      "loss": 0.1297,
+      "step": 37010
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.953568161024702e-06,
+      "loss": 0.1919,
+      "step": 37020
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.9467063129002745e-06,
+      "loss": 0.1865,
+      "step": 37030
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.939844464775847e-06,
+      "loss": 0.2165,
+      "step": 37040
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.932982616651418e-06,
+      "loss": 0.2292,
+      "step": 37050
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.92612076852699e-06,
+      "loss": 0.1823,
+      "step": 37060
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.919258920402562e-06,
+      "loss": 0.1727,
+      "step": 37070
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.912397072278134e-06,
+      "loss": 0.1625,
+      "step": 37080
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.905535224153706e-06,
+      "loss": 0.1567,
+      "step": 37090
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.898673376029278e-06,
+      "loss": 0.1408,
+      "step": 37100
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.891811527904849e-06,
+      "loss": 0.1901,
+      "step": 37110
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.884949679780421e-06,
+      "loss": 0.1603,
+      "step": 37120
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.8780878316559935e-06,
+      "loss": 0.2222,
+      "step": 37130
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.871225983531564e-06,
+      "loss": 0.1938,
+      "step": 37140
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.864364135407136e-06,
+      "loss": 0.1914,
+      "step": 37150
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.857502287282708e-06,
+      "loss": 0.2185,
+      "step": 37160
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.8506404391582796e-06,
+      "loss": 0.1456,
+      "step": 37170
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.843778591033852e-06,
+      "loss": 0.1558,
+      "step": 37180
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.836916742909424e-06,
+      "loss": 0.1737,
+      "step": 37190
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.830054894784995e-06,
+      "loss": 0.1355,
+      "step": 37200
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.823193046660567e-06,
+      "loss": 0.1635,
+      "step": 37210
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.8163311985361395e-06,
+      "loss": 0.1495,
+      "step": 37220
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.809469350411711e-06,
+      "loss": 0.1834,
+      "step": 37230
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.802607502287283e-06,
+      "loss": 0.2022,
+      "step": 37240
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.795745654162855e-06,
+      "loss": 0.1781,
+      "step": 37250
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.788883806038426e-06,
+      "loss": 0.153,
+      "step": 37260
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.7820219579139986e-06,
+      "loss": 0.3676,
+      "step": 37270
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.77516010978957e-06,
+      "loss": 0.2094,
+      "step": 37280
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.768298261665142e-06,
+      "loss": 0.1447,
+      "step": 37290
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.761436413540714e-06,
+      "loss": 0.2102,
+      "step": 37300
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.7545745654162855e-06,
+      "loss": 0.1801,
+      "step": 37310
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.747712717291858e-06,
+      "loss": 0.1611,
+      "step": 37320
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.74085086916743e-06,
+      "loss": 0.168,
+      "step": 37330
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.733989021043e-06,
+      "loss": 0.2372,
+      "step": 37340
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.727127172918572e-06,
+      "loss": 0.1767,
+      "step": 37350
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.7202653247941446e-06,
+      "loss": 0.1715,
+      "step": 37360
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.713403476669716e-06,
+      "loss": 0.2235,
+      "step": 37370
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.706541628545288e-06,
+      "loss": 0.1868,
+      "step": 37380
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.69967978042086e-06,
+      "loss": 0.1375,
+      "step": 37390
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.6928179322964315e-06,
+      "loss": 0.2106,
+      "step": 37400
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.685956084172004e-06,
+      "loss": 0.1668,
+      "step": 37410
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.679094236047576e-06,
+      "loss": 0.1418,
+      "step": 37420
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.672232387923147e-06,
+      "loss": 0.1994,
+      "step": 37430
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.665370539798719e-06,
+      "loss": 0.1803,
+      "step": 37440
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.658508691674291e-06,
+      "loss": 0.1788,
+      "step": 37450
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.651646843549863e-06,
+      "loss": 0.2061,
+      "step": 37460
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.644784995425435e-06,
+      "loss": 0.1832,
+      "step": 37470
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.637923147301007e-06,
+      "loss": 0.2021,
+      "step": 37480
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.631061299176578e-06,
+      "loss": 0.1437,
+      "step": 37490
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.6241994510521505e-06,
+      "loss": 0.1564,
+      "step": 37500
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.617337602927723e-06,
+      "loss": 0.2422,
+      "step": 37510
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.610475754803294e-06,
+      "loss": 0.2035,
+      "step": 37520
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.603613906678866e-06,
+      "loss": 0.184,
+      "step": 37530
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.596752058554437e-06,
+      "loss": 0.1715,
+      "step": 37540
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.589890210430009e-06,
+      "loss": 0.1935,
+      "step": 37550
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.583028362305581e-06,
+      "loss": 0.1513,
+      "step": 37560
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.576166514181153e-06,
+      "loss": 0.2144,
+      "step": 37570
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.569304666056724e-06,
+      "loss": 0.2006,
+      "step": 37580
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.5624428179322965e-06,
+      "loss": 0.1957,
+      "step": 37590
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.555580969807868e-06,
+      "loss": 0.1679,
+      "step": 37600
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.54871912168344e-06,
+      "loss": 0.1609,
+      "step": 37610
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.541857273559012e-06,
+      "loss": 0.1878,
+      "step": 37620
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.534995425434583e-06,
+      "loss": 0.1294,
+      "step": 37630
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.5281335773101556e-06,
+      "loss": 0.1353,
+      "step": 37640
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.521271729185728e-06,
+      "loss": 0.1302,
+      "step": 37650
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.514409881061299e-06,
+      "loss": 0.181,
+      "step": 37660
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.507548032936871e-06,
+      "loss": 0.183,
+      "step": 37670
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.500686184812443e-06,
+      "loss": 0.1118,
+      "step": 37680
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.493824336688015e-06,
+      "loss": 0.1598,
+      "step": 37690
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.486962488563587e-06,
+      "loss": 0.1816,
+      "step": 37700
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.480100640439159e-06,
+      "loss": 0.1432,
+      "step": 37710
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.47323879231473e-06,
+      "loss": 0.1208,
+      "step": 37720
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.466376944190302e-06,
+      "loss": 0.2006,
+      "step": 37730
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.4595150960658746e-06,
+      "loss": 0.1666,
+      "step": 37740
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.452653247941445e-06,
+      "loss": 0.154,
+      "step": 37750
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.445791399817017e-06,
+      "loss": 0.2218,
+      "step": 37760
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.438929551692589e-06,
+      "loss": 0.1331,
+      "step": 37770
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.432067703568161e-06,
+      "loss": 0.1293,
+      "step": 37780
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.425205855443733e-06,
+      "loss": 0.1816,
+      "step": 37790
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.418344007319305e-06,
+      "loss": 0.1305,
+      "step": 37800
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.411482159194876e-06,
+      "loss": 0.2515,
+      "step": 37810
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.404620311070448e-06,
+      "loss": 0.1845,
+      "step": 37820
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.3977584629460205e-06,
+      "loss": 0.1485,
+      "step": 37830
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.390896614821592e-06,
+      "loss": 0.1553,
+      "step": 37840
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.384034766697164e-06,
+      "loss": 0.1564,
+      "step": 37850
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.377172918572736e-06,
+      "loss": 0.1818,
+      "step": 37860
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.3703110704483075e-06,
+      "loss": 0.1876,
+      "step": 37870
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.36344922232388e-06,
+      "loss": 0.1671,
+      "step": 37880
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.356587374199452e-06,
+      "loss": 0.1802,
+      "step": 37890
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.349725526075023e-06,
+      "loss": 0.1479,
+      "step": 37900
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.342863677950595e-06,
+      "loss": 0.2072,
+      "step": 37910
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.3360018298261665e-06,
+      "loss": 0.1887,
+      "step": 37920
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.329139981701739e-06,
+      "loss": 0.1137,
+      "step": 37930
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.322278133577311e-06,
+      "loss": 0.2773,
+      "step": 37940
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.315416285452881e-06,
+      "loss": 0.1964,
+      "step": 37950
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.3085544373284535e-06,
+      "loss": 0.1511,
+      "step": 37960
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.301692589204026e-06,
+      "loss": 0.1836,
+      "step": 37970
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.294830741079597e-06,
+      "loss": 0.1929,
+      "step": 37980
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.287968892955169e-06,
+      "loss": 0.15,
+      "step": 37990
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.281107044830741e-06,
+      "loss": 0.1108,
+      "step": 38000
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.2742451967063125e-06,
+      "loss": 0.1477,
+      "step": 38010
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.267383348581885e-06,
+      "loss": 0.1851,
+      "step": 38020
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.260521500457457e-06,
+      "loss": 0.2483,
+      "step": 38030
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.253659652333028e-06,
+      "loss": 0.1881,
+      "step": 38040
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.2467978042086e-06,
+      "loss": 0.1619,
+      "step": 38050
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.2399359560841725e-06,
+      "loss": 0.1498,
+      "step": 38060
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.233074107959744e-06,
+      "loss": 0.115,
+      "step": 38070
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.226212259835316e-06,
+      "loss": 0.1537,
+      "step": 38080
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.219350411710888e-06,
+      "loss": 0.2071,
+      "step": 38090
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.212488563586459e-06,
+      "loss": 0.1519,
+      "step": 38100
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.2056267154620315e-06,
+      "loss": 0.1724,
+      "step": 38110
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.198764867337604e-06,
+      "loss": 0.1482,
+      "step": 38120
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.191903019213175e-06,
+      "loss": 0.1236,
+      "step": 38130
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.185041171088747e-06,
+      "loss": 0.1865,
+      "step": 38140
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.1781793229643185e-06,
+      "loss": 0.1065,
+      "step": 38150
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.17131747483989e-06,
+      "loss": 0.1501,
+      "step": 38160
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.164455626715462e-06,
+      "loss": 0.2265,
+      "step": 38170
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.157593778591034e-06,
+      "loss": 0.1753,
+      "step": 38180
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.150731930466605e-06,
+      "loss": 0.1909,
+      "step": 38190
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.1438700823421775e-06,
+      "loss": 0.1572,
+      "step": 38200
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.13700823421775e-06,
+      "loss": 0.1402,
+      "step": 38210
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.130146386093321e-06,
+      "loss": 0.1738,
+      "step": 38220
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.123284537968893e-06,
+      "loss": 0.1381,
+      "step": 38230
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.1164226898444644e-06,
+      "loss": 0.1554,
+      "step": 38240
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.109560841720037e-06,
+      "loss": 0.1481,
+      "step": 38250
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.102698993595609e-06,
+      "loss": 0.1627,
+      "step": 38260
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.09583714547118e-06,
+      "loss": 0.2619,
+      "step": 38270
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.088975297346752e-06,
+      "loss": 0.2164,
+      "step": 38280
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.082113449222324e-06,
+      "loss": 0.1701,
+      "step": 38290
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.075251601097896e-06,
+      "loss": 0.1946,
+      "step": 38300
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.068389752973468e-06,
+      "loss": 0.2273,
+      "step": 38310
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.06152790484904e-06,
+      "loss": 0.1739,
+      "step": 38320
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.054666056724611e-06,
+      "loss": 0.1528,
+      "step": 38330
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.0478042086001834e-06,
+      "loss": 0.1741,
+      "step": 38340
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.040942360475755e-06,
+      "loss": 0.163,
+      "step": 38350
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.034080512351326e-06,
+      "loss": 0.2066,
+      "step": 38360
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.027218664226898e-06,
+      "loss": 0.2344,
+      "step": 38370
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.02035681610247e-06,
+      "loss": 0.1744,
+      "step": 38380
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.013494967978042e-06,
+      "loss": 0.1856,
+      "step": 38390
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.006633119853614e-06,
+      "loss": 0.1312,
+      "step": 38400
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.999771271729186e-06,
+      "loss": 0.192,
+      "step": 38410
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.992909423604757e-06,
+      "loss": 0.2508,
+      "step": 38420
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.9860475754803294e-06,
+      "loss": 0.1723,
+      "step": 38430
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.979185727355902e-06,
+      "loss": 0.1323,
+      "step": 38440
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.972323879231473e-06,
+      "loss": 0.1438,
+      "step": 38450
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.965462031107045e-06,
+      "loss": 0.1487,
+      "step": 38460
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.958600182982617e-06,
+      "loss": 0.1304,
+      "step": 38470
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.9517383348581885e-06,
+      "loss": 0.1283,
+      "step": 38480
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.944876486733761e-06,
+      "loss": 0.1903,
+      "step": 38490
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.938014638609333e-06,
+      "loss": 0.1417,
+      "step": 38500
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.931152790484904e-06,
+      "loss": 0.1569,
+      "step": 38510
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.924290942360476e-06,
+      "loss": 0.1346,
+      "step": 38520
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.917429094236048e-06,
+      "loss": 0.1268,
+      "step": 38530
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.91056724611162e-06,
+      "loss": 0.1703,
+      "step": 38540
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.903705397987192e-06,
+      "loss": 0.1456,
+      "step": 38550
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.896843549862762e-06,
+      "loss": 0.2489,
+      "step": 38560
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.8899817017383345e-06,
+      "loss": 0.0994,
+      "step": 38570
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.883119853613907e-06,
+      "loss": 0.1774,
+      "step": 38580
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.876258005489478e-06,
+      "loss": 0.1049,
+      "step": 38590
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.86939615736505e-06,
+      "loss": 0.1709,
+      "step": 38600
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.862534309240622e-06,
+      "loss": 0.1157,
+      "step": 38610
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.855672461116194e-06,
+      "loss": 0.1579,
+      "step": 38620
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.848810612991766e-06,
+      "loss": 0.0919,
+      "step": 38630
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.841948764867338e-06,
+      "loss": 0.1486,
+      "step": 38640
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.835086916742909e-06,
+      "loss": 0.2016,
+      "step": 38650
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.828225068618481e-06,
+      "loss": 0.1427,
+      "step": 38660
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.8213632204940535e-06,
+      "loss": 0.1532,
+      "step": 38670
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.8145013723696252e-06,
+      "loss": 0.2173,
+      "step": 38680
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.807639524245197e-06,
+      "loss": 0.1934,
+      "step": 38690
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.8007776761207687e-06,
+      "loss": 0.1522,
+      "step": 38700
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.793915827996341e-06,
+      "loss": 0.1508,
+      "step": 38710
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.7870539798719126e-06,
+      "loss": 0.2074,
+      "step": 38720
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.7801921317474843e-06,
+      "loss": 0.2807,
+      "step": 38730
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.7733302836230565e-06,
+      "loss": 0.157,
+      "step": 38740
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.766468435498628e-06,
+      "loss": 0.145,
+      "step": 38750
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.759606587374199e-06,
+      "loss": 0.2311,
+      "step": 38760
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.752744739249771e-06,
+      "loss": 0.1449,
+      "step": 38770
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.7458828911253434e-06,
+      "loss": 0.2026,
+      "step": 38780
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.739021043000915e-06,
+      "loss": 0.1548,
+      "step": 38790
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.732159194876487e-06,
+      "loss": 0.093,
+      "step": 38800
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.7252973467520586e-06,
+      "loss": 0.1535,
+      "step": 38810
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.7184354986276303e-06,
+      "loss": 0.1676,
+      "step": 38820
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.711573650503202e-06,
+      "loss": 0.1665,
+      "step": 38830
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.704711802378774e-06,
+      "loss": 0.2119,
+      "step": 38840
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.697849954254346e-06,
+      "loss": 0.1963,
+      "step": 38850
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.6909881061299177e-06,
+      "loss": 0.1601,
+      "step": 38860
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.68412625800549e-06,
+      "loss": 0.1495,
+      "step": 38870
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.6772644098810615e-06,
+      "loss": 0.1747,
+      "step": 38880
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.6704025617566333e-06,
+      "loss": 0.1382,
+      "step": 38890
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.6635407136322054e-06,
+      "loss": 0.1619,
+      "step": 38900
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.6566788655077767e-06,
+      "loss": 0.1732,
+      "step": 38910
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.6498170173833485e-06,
+      "loss": 0.1557,
+      "step": 38920
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.64295516925892e-06,
+      "loss": 0.2118,
+      "step": 38930
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.6360933211344923e-06,
+      "loss": 0.16,
+      "step": 38940
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.629231473010064e-06,
+      "loss": 0.1421,
+      "step": 38950
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.622369624885636e-06,
+      "loss": 0.1072,
+      "step": 38960
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.615507776761208e-06,
+      "loss": 0.1531,
+      "step": 38970
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.6086459286367797e-06,
+      "loss": 0.166,
+      "step": 38980
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.6017840805123514e-06,
+      "loss": 0.1395,
+      "step": 38990
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5949222323879236e-06,
+      "loss": 0.1409,
+      "step": 39000
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.588060384263495e-06,
+      "loss": 0.1066,
+      "step": 39010
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5811985361390666e-06,
+      "loss": 0.1876,
+      "step": 39020
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5743366880146388e-06,
+      "loss": 0.1781,
+      "step": 39030
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5674748398902105e-06,
+      "loss": 0.1835,
+      "step": 39040
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5606129917657822e-06,
+      "loss": 0.1694,
+      "step": 39050
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5537511436413544e-06,
+      "loss": 0.1974,
+      "step": 39060
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.546889295516926e-06,
+      "loss": 0.1579,
+      "step": 39070
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.540027447392498e-06,
+      "loss": 0.1971,
+      "step": 39080
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.5331655992680696e-06,
+      "loss": 0.1532,
+      "step": 39090
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.5263037511436417e-06,
+      "loss": 0.146,
+      "step": 39100
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.519441903019213e-06,
+      "loss": 0.1374,
+      "step": 39110
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.5125800548947848e-06,
+      "loss": 0.1126,
+      "step": 39120
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.505718206770357e-06,
+      "loss": 0.1096,
+      "step": 39130
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.4988563586459286e-06,
+      "loss": 0.1529,
+      "step": 39140
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.4919945105215004e-06,
+      "loss": 0.1432,
+      "step": 39150
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.4851326623970725e-06,
+      "loss": 0.1351,
+      "step": 39160
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.4782708142726443e-06,
+      "loss": 0.1354,
+      "step": 39170
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.471408966148216e-06,
+      "loss": 0.2001,
+      "step": 39180
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.464547118023788e-06,
+      "loss": 0.1598,
+      "step": 39190
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.45768526989936e-06,
+      "loss": 0.1586,
+      "step": 39200
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.450823421774931e-06,
+      "loss": 0.1508,
+      "step": 39210
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.4439615736505033e-06,
+      "loss": 0.1804,
+      "step": 39220
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.437099725526075e-06,
+      "loss": 0.1972,
+      "step": 39230
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.430237877401647e-06,
+      "loss": 0.2794,
+      "step": 39240
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.4233760292772185e-06,
+      "loss": 0.1106,
+      "step": 39250
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.4165141811527907e-06,
+      "loss": 0.2102,
+      "step": 39260
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.4096523330283624e-06,
+      "loss": 0.1379,
+      "step": 39270
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.402790484903934e-06,
+      "loss": 0.1256,
+      "step": 39280
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.3959286367795063e-06,
+      "loss": 0.2704,
+      "step": 39290
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.389066788655078e-06,
+      "loss": 0.1371,
+      "step": 39300
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.3822049405306497e-06,
+      "loss": 0.1644,
+      "step": 39310
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.3753430924062215e-06,
+      "loss": 0.1718,
+      "step": 39320
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.368481244281793e-06,
+      "loss": 0.2702,
+      "step": 39330
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.361619396157365e-06,
+      "loss": 0.1466,
+      "step": 39340
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.354757548032937e-06,
+      "loss": 0.1334,
+      "step": 39350
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.347895699908509e-06,
+      "loss": 0.2229,
+      "step": 39360
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.3410338517840806e-06,
+      "loss": 0.1398,
+      "step": 39370
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.3341720036596527e-06,
+      "loss": 0.1787,
+      "step": 39380
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.3273101555352244e-06,
+      "loss": 0.1459,
+      "step": 39390
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.320448307410796e-06,
+      "loss": 0.1791,
+      "step": 39400
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.313586459286368e-06,
+      "loss": 0.2263,
+      "step": 39410
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.3067246111619396e-06,
+      "loss": 0.1408,
+      "step": 39420
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.2998627630375114e-06,
+      "loss": 0.1396,
+      "step": 39430
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.293000914913083e-06,
+      "loss": 0.1803,
+      "step": 39440
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.2861390667886552e-06,
+      "loss": 0.1475,
+      "step": 39450
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.279277218664227e-06,
+      "loss": 0.1224,
+      "step": 39460
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.2724153705397987e-06,
+      "loss": 0.1116,
+      "step": 39470
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.265553522415371e-06,
+      "loss": 0.1569,
+      "step": 39480
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.2586916742909426e-06,
+      "loss": 0.1485,
+      "step": 39490
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.2518298261665143e-06,
+      "loss": 0.1288,
+      "step": 39500
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.2449679780420865e-06,
+      "loss": 0.2148,
+      "step": 39510
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.2381061299176578e-06,
+      "loss": 0.1491,
+      "step": 39520
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.2312442817932295e-06,
+      "loss": 0.2386,
+      "step": 39530
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.2243824336688017e-06,
+      "loss": 0.2023,
+      "step": 39540
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.2175205855443734e-06,
+      "loss": 0.1965,
+      "step": 39550
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.210658737419945e-06,
+      "loss": 0.1883,
+      "step": 39560
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.203796889295517e-06,
+      "loss": 0.1945,
+      "step": 39570
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.196935041171089e-06,
+      "loss": 0.1791,
+      "step": 39580
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.1900731930466607e-06,
+      "loss": 0.1445,
+      "step": 39590
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.1832113449222325e-06,
+      "loss": 0.1694,
+      "step": 39600
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.1763494967978046e-06,
+      "loss": 0.2543,
+      "step": 39610
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.169487648673376e-06,
+      "loss": 0.2076,
+      "step": 39620
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.1626258005489477e-06,
+      "loss": 0.2197,
+      "step": 39630
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.15576395242452e-06,
+      "loss": 0.2347,
+      "step": 39640
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.1489021043000915e-06,
+      "loss": 0.1304,
+      "step": 39650
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.1420402561756633e-06,
+      "loss": 0.1479,
+      "step": 39660
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.1351784080512354e-06,
+      "loss": 0.1771,
+      "step": 39670
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.128316559926807e-06,
+      "loss": 0.1617,
+      "step": 39680
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.121454711802379e-06,
+      "loss": 0.1422,
+      "step": 39690
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.1145928636779506e-06,
+      "loss": 0.2626,
+      "step": 39700
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.1077310155535228e-06,
+      "loss": 0.2131,
+      "step": 39710
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.100869167429094e-06,
+      "loss": 0.1535,
+      "step": 39720
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.094693504117109e-06,
+      "loss": 0.3441,
+      "step": 39730
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.0878316559926807e-06,
+      "loss": 0.7051,
+      "step": 39740
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.0809698078682525e-06,
+      "loss": 0.1847,
+      "step": 39750
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.074107959743824e-06,
+      "loss": 0.2071,
+      "step": 39760
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.0672461116193964e-06,
+      "loss": 0.1661,
+      "step": 39770
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.060384263494968e-06,
+      "loss": 0.1425,
+      "step": 39780
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.05352241537054e-06,
+      "loss": 0.1508,
+      "step": 39790
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 0.4623459279537201,
+      "eval_runtime": 304.2897,
+      "eval_samples_per_second": 6.458,
+      "eval_steps_per_second": 1.617,
+      "eval_wer": 0.18366089581664716,
+      "step": 39798
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0466605672461115e-06,
+      "loss": 0.1626,
+      "step": 39800
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0397987191216837e-06,
+      "loss": 0.1554,
+      "step": 39810
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0329368709972554e-06,
+      "loss": 0.163,
+      "step": 39820
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.026075022872827e-06,
+      "loss": 0.1386,
+      "step": 39830
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.019213174748399e-06,
+      "loss": 0.2019,
+      "step": 39840
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.0123513266239706e-06,
+      "loss": 0.1803,
+      "step": 39850
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.0054894784995424e-06,
+      "loss": 0.166,
+      "step": 39860
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.9986276303751145e-06,
+      "loss": 0.142,
+      "step": 39870
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.9917657822506862e-06,
+      "loss": 0.1531,
+      "step": 39880
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.984903934126258e-06,
+      "loss": 0.1511,
+      "step": 39890
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.97804208600183e-06,
+      "loss": 0.1892,
+      "step": 39900
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.971180237877402e-06,
+      "loss": 0.1883,
+      "step": 39910
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.9643183897529736e-06,
+      "loss": 0.1245,
+      "step": 39920
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.9574565416285457e-06,
+      "loss": 0.1678,
+      "step": 39930
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.950594693504117e-06,
+      "loss": 0.1204,
+      "step": 39940
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.9437328453796888e-06,
+      "loss": 0.2174,
+      "step": 39950
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.9368709972552605e-06,
+      "loss": 0.134,
+      "step": 39960
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.9300091491308327e-06,
+      "loss": 0.2057,
+      "step": 39970
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.9231473010064044e-06,
+      "loss": 0.1009,
+      "step": 39980
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.916285452881976e-06,
+      "loss": 0.1652,
+      "step": 39990
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.9094236047575483e-06,
+      "loss": 0.1436,
+      "step": 40000
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.90256175663312e-06,
+      "loss": 0.1585,
+      "step": 40010
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.8956999085086917e-06,
+      "loss": 0.1562,
+      "step": 40020
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.888838060384264e-06,
+      "loss": 0.2104,
+      "step": 40030
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.8819762122598356e-06,
+      "loss": 0.1644,
+      "step": 40040
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.875114364135407e-06,
+      "loss": 0.1875,
+      "step": 40050
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.868252516010979e-06,
+      "loss": 0.136,
+      "step": 40060
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.861390667886551e-06,
+      "loss": 0.1464,
+      "step": 40070
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.8545288197621225e-06,
+      "loss": 0.1316,
+      "step": 40080
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.8476669716376947e-06,
+      "loss": 0.1247,
+      "step": 40090
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.8408051235132664e-06,
+      "loss": 0.1237,
+      "step": 40100
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.833943275388838e-06,
+      "loss": 0.1342,
+      "step": 40110
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.82708142726441e-06,
+      "loss": 0.1708,
+      "step": 40120
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.820219579139982e-06,
+      "loss": 0.2037,
+      "step": 40130
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.8133577310155538e-06,
+      "loss": 0.1107,
+      "step": 40140
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.806495882891125e-06,
+      "loss": 0.1199,
+      "step": 40150
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.7996340347666972e-06,
+      "loss": 0.2299,
+      "step": 40160
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.792772186642269e-06,
+      "loss": 0.1828,
+      "step": 40170
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7859103385178407e-06,
+      "loss": 0.1638,
+      "step": 40180
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.779048490393413e-06,
+      "loss": 0.1845,
+      "step": 40190
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7728728270814274e-06,
+      "loss": 0.1212,
+      "step": 40200
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.766010978956999e-06,
+      "loss": 0.1611,
+      "step": 40210
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.759149130832571e-06,
+      "loss": 0.0931,
+      "step": 40220
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.752287282708143e-06,
+      "loss": 0.1213,
+      "step": 40230
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.7454254345837147e-06,
+      "loss": 0.1874,
+      "step": 40240
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.7385635864592864e-06,
+      "loss": 0.1056,
+      "step": 40250
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.7317017383348586e-06,
+      "loss": 0.1185,
+      "step": 40260
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.72483989021043e-06,
+      "loss": 0.1081,
+      "step": 40270
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.7179780420860016e-06,
+      "loss": 0.1454,
+      "step": 40280
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.7111161939615738e-06,
+      "loss": 0.163,
+      "step": 40290
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.7042543458371455e-06,
+      "loss": 0.2258,
+      "step": 40300
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.6973924977127172e-06,
+      "loss": 0.1654,
+      "step": 40310
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.6905306495882894e-06,
+      "loss": 0.1483,
+      "step": 40320
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.683668801463861e-06,
+      "loss": 0.1133,
+      "step": 40330
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.676806953339433e-06,
+      "loss": 0.1557,
+      "step": 40340
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.669945105215005e-06,
+      "loss": 0.171,
+      "step": 40350
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.6630832570905767e-06,
+      "loss": 0.1417,
+      "step": 40360
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.656221408966148e-06,
+      "loss": 0.1341,
+      "step": 40370
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.6493595608417198e-06,
+      "loss": 0.1633,
+      "step": 40380
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.642497712717292e-06,
+      "loss": 0.1373,
+      "step": 40390
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.6356358645928637e-06,
+      "loss": 0.1497,
+      "step": 40400
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.6287740164684354e-06,
+      "loss": 0.2094,
+      "step": 40410
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.6219121683440075e-06,
+      "loss": 0.1357,
+      "step": 40420
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.6150503202195793e-06,
+      "loss": 0.1306,
+      "step": 40430
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.608188472095151e-06,
+      "loss": 0.1508,
+      "step": 40440
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.601326623970723e-06,
+      "loss": 0.1482,
+      "step": 40450
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.594464775846295e-06,
+      "loss": 0.144,
+      "step": 40460
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.587602927721866e-06,
+      "loss": 0.1473,
+      "step": 40470
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.5807410795974383e-06,
+      "loss": 0.1467,
+      "step": 40480
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.57387923147301e-06,
+      "loss": 0.1307,
+      "step": 40490
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.567017383348582e-06,
+      "loss": 0.1244,
+      "step": 40500
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.5601555352241535e-06,
+      "loss": 0.2222,
+      "step": 40510
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.5532936870997257e-06,
+      "loss": 0.1244,
+      "step": 40520
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.5464318389752974e-06,
+      "loss": 0.1276,
+      "step": 40530
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.539569990850869e-06,
+      "loss": 0.1342,
+      "step": 40540
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.5327081427264413e-06,
+      "loss": 0.2215,
+      "step": 40550
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.525846294602013e-06,
+      "loss": 0.2152,
+      "step": 40560
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.5189844464775843e-06,
+      "loss": 0.1894,
+      "step": 40570
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.5121225983531565e-06,
+      "loss": 0.1464,
+      "step": 40580
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.5052607502287282e-06,
+      "loss": 0.1809,
+      "step": 40590
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.4983989021043e-06,
+      "loss": 0.1532,
+      "step": 40600
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.491537053979872e-06,
+      "loss": 0.124,
+      "step": 40610
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.484675205855444e-06,
+      "loss": 0.1488,
+      "step": 40620
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.4778133577310156e-06,
+      "loss": 0.1647,
+      "step": 40630
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.4709515096065877e-06,
+      "loss": 0.1625,
+      "step": 40640
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.4640896614821594e-06,
+      "loss": 0.1262,
+      "step": 40650
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.457227813357731e-06,
+      "loss": 0.1589,
+      "step": 40660
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.4503659652333025e-06,
+      "loss": 0.1392,
+      "step": 40670
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.4435041171088746e-06,
+      "loss": 0.1554,
+      "step": 40680
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.4366422689844464e-06,
+      "loss": 0.141,
+      "step": 40690
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.429780420860018e-06,
+      "loss": 0.1689,
+      "step": 40700
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.4229185727355903e-06,
+      "loss": 0.1396,
+      "step": 40710
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.416056724611162e-06,
+      "loss": 0.1219,
+      "step": 40720
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.4091948764867337e-06,
+      "loss": 0.1453,
+      "step": 40730
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.402333028362306e-06,
+      "loss": 0.1446,
+      "step": 40740
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.3954711802378776e-06,
+      "loss": 0.1631,
+      "step": 40750
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.3886093321134493e-06,
+      "loss": 0.1397,
+      "step": 40760
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.381747483989021e-06,
+      "loss": 0.1276,
+      "step": 40770
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.3748856358645928e-06,
+      "loss": 0.0893,
+      "step": 40780
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.3680237877401645e-06,
+      "loss": 0.2543,
+      "step": 40790
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.3611619396157367e-06,
+      "loss": 0.1257,
+      "step": 40800
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.3543000914913084e-06,
+      "loss": 0.1376,
+      "step": 40810
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.34743824336688e-06,
+      "loss": 0.1442,
+      "step": 40820
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.340576395242452e-06,
+      "loss": 0.1748,
+      "step": 40830
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.333714547118024e-06,
+      "loss": 0.1659,
+      "step": 40840
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.3268526989935957e-06,
+      "loss": 0.1609,
+      "step": 40850
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.3199908508691675e-06,
+      "loss": 0.1806,
+      "step": 40860
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.3131290027447396e-06,
+      "loss": 0.1714,
+      "step": 40870
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.306267154620311e-06,
+      "loss": 0.166,
+      "step": 40880
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.2994053064958827e-06,
+      "loss": 0.176,
+      "step": 40890
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.292543458371455e-06,
+      "loss": 0.1422,
+      "step": 40900
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.2856816102470266e-06,
+      "loss": 0.1325,
+      "step": 40910
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.2788197621225983e-06,
+      "loss": 0.1767,
+      "step": 40920
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.2719579139981704e-06,
+      "loss": 0.3011,
+      "step": 40930
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.265096065873742e-06,
+      "loss": 0.1869,
+      "step": 40940
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.258234217749314e-06,
+      "loss": 0.1275,
+      "step": 40950
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.251372369624886e-06,
+      "loss": 0.1676,
+      "step": 40960
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.2445105215004578e-06,
+      "loss": 0.1104,
+      "step": 40970
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.237648673376029e-06,
+      "loss": 0.1524,
+      "step": 40980
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.230786825251601e-06,
+      "loss": 0.1508,
+      "step": 40990
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.223924977127173e-06,
+      "loss": 0.1242,
+      "step": 41000
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.2170631290027447e-06,
+      "loss": 0.3309,
+      "step": 41010
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.2102012808783164e-06,
+      "loss": 0.1577,
+      "step": 41020
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.2033394327538886e-06,
+      "loss": 0.1307,
+      "step": 41030
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.1964775846294603e-06,
+      "loss": 0.1165,
+      "step": 41040
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.189615736505032e-06,
+      "loss": 0.1838,
+      "step": 41050
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.182753888380604e-06,
+      "loss": 0.2003,
+      "step": 41060
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.175892040256176e-06,
+      "loss": 0.1733,
+      "step": 41070
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.1690301921317472e-06,
+      "loss": 0.2,
+      "step": 41080
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.1621683440073194e-06,
+      "loss": 0.1413,
+      "step": 41090
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.155306495882891e-06,
+      "loss": 0.13,
+      "step": 41100
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.148444647758463e-06,
+      "loss": 0.1669,
+      "step": 41110
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.141582799634035e-06,
+      "loss": 0.1364,
+      "step": 41120
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.1347209515096067e-06,
+      "loss": 0.1727,
+      "step": 41130
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.1278591033851785e-06,
+      "loss": 0.1304,
+      "step": 41140
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.12099725526075e-06,
+      "loss": 0.1485,
+      "step": 41150
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.1141354071363223e-06,
+      "loss": 0.1559,
+      "step": 41160
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.107273559011894e-06,
+      "loss": 0.1714,
+      "step": 41170
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.1004117108874654e-06,
+      "loss": 0.1832,
+      "step": 41180
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.0935498627630375e-06,
+      "loss": 0.1328,
+      "step": 41190
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.0866880146386093e-06,
+      "loss": 0.1648,
+      "step": 41200
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.079826166514181e-06,
+      "loss": 0.1243,
+      "step": 41210
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.072964318389753e-06,
+      "loss": 0.2134,
+      "step": 41220
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.066102470265325e-06,
+      "loss": 0.1721,
+      "step": 41230
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.0592406221408966e-06,
+      "loss": 0.1461,
+      "step": 41240
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.0523787740164688e-06,
+      "loss": 0.1059,
+      "step": 41250
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.0455169258920405e-06,
+      "loss": 0.1119,
+      "step": 41260
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.0386550777676122e-06,
+      "loss": 0.1768,
+      "step": 41270
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.031793229643184e-06,
+      "loss": 0.144,
+      "step": 41280
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.0249313815187557e-06,
+      "loss": 0.0998,
+      "step": 41290
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.0180695333943274e-06,
+      "loss": 0.1342,
+      "step": 41300
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.011207685269899e-06,
+      "loss": 0.1034,
+      "step": 41310
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.0043458371454713e-06,
+      "loss": 0.1823,
+      "step": 41320
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.997483989021043e-06,
+      "loss": 0.1805,
+      "step": 41330
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.9906221408966148e-06,
+      "loss": 0.1507,
+      "step": 41340
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.983760292772187e-06,
+      "loss": 0.1522,
+      "step": 41350
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.9768984446477586e-06,
+      "loss": 0.0977,
+      "step": 41360
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.9700365965233304e-06,
+      "loss": 0.0777,
+      "step": 41370
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.963174748398902e-06,
+      "loss": 0.2539,
+      "step": 41380
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.956312900274474e-06,
+      "loss": 0.1326,
+      "step": 41390
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.9494510521500456e-06,
+      "loss": 0.175,
+      "step": 41400
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.9425892040256177e-06,
+      "loss": 0.1962,
+      "step": 41410
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.9357273559011895e-06,
+      "loss": 0.1817,
+      "step": 41420
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.928865507776761e-06,
+      "loss": 0.157,
+      "step": 41430
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.9220036596523333e-06,
+      "loss": 0.1429,
+      "step": 41440
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.915141811527905e-06,
+      "loss": 0.1131,
+      "step": 41450
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.908279963403477e-06,
+      "loss": 0.0956,
+      "step": 41460
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.9014181152790487e-06,
+      "loss": 0.1739,
+      "step": 41470
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8945562671546203e-06,
+      "loss": 0.1301,
+      "step": 41480
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.887694419030192e-06,
+      "loss": 0.1101,
+      "step": 41490
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.880832570905764e-06,
+      "loss": 0.1461,
+      "step": 41500
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8739707227813357e-06,
+      "loss": 0.135,
+      "step": 41510
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8671088746569076e-06,
+      "loss": 0.1542,
+      "step": 41520
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8602470265324795e-06,
+      "loss": 0.1517,
+      "step": 41530
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8533851784080513e-06,
+      "loss": 0.144,
+      "step": 41540
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8465233302836232e-06,
+      "loss": 0.1917,
+      "step": 41550
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.839661482159195e-06,
+      "loss": 0.1861,
+      "step": 41560
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8327996340347667e-06,
+      "loss": 0.1872,
+      "step": 41570
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8259377859103386e-06,
+      "loss": 0.1508,
+      "step": 41580
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8190759377859103e-06,
+      "loss": 0.2123,
+      "step": 41590
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8122140896614823e-06,
+      "loss": 0.1536,
+      "step": 41600
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.805352241537054e-06,
+      "loss": 0.1418,
+      "step": 41610
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.7984903934126258e-06,
+      "loss": 0.1349,
+      "step": 41620
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.7916285452881977e-06,
+      "loss": 0.1213,
+      "step": 41630
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.7847666971637696e-06,
+      "loss": 0.1389,
+      "step": 41640
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.7779048490393414e-06,
+      "loss": 0.1547,
+      "step": 41650
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.771043000914913e-06,
+      "loss": 0.1138,
+      "step": 41660
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.7641811527904848e-06,
+      "loss": 0.1549,
+      "step": 41670
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7573193046660568e-06,
+      "loss": 0.1309,
+      "step": 41680
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7504574565416287e-06,
+      "loss": 0.1187,
+      "step": 41690
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7435956084172004e-06,
+      "loss": 0.1721,
+      "step": 41700
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7367337602927722e-06,
+      "loss": 0.1536,
+      "step": 41710
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7298719121683441e-06,
+      "loss": 0.1458,
+      "step": 41720
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.7230100640439158e-06,
+      "loss": 0.1092,
+      "step": 41730
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.7161482159194878e-06,
+      "loss": 0.1433,
+      "step": 41740
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.7092863677950595e-06,
+      "loss": 0.1327,
+      "step": 41750
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.7024245196706312e-06,
+      "loss": 0.164,
+      "step": 41760
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6955626715462032e-06,
+      "loss": 0.1394,
+      "step": 41770
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.688700823421775e-06,
+      "loss": 0.1275,
+      "step": 41780
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6818389752973469e-06,
+      "loss": 0.2761,
+      "step": 41790
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6749771271729186e-06,
+      "loss": 0.1954,
+      "step": 41800
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6681152790484903e-06,
+      "loss": 0.1245,
+      "step": 41810
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.6612534309240623e-06,
+      "loss": 0.1333,
+      "step": 41820
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.654391582799634e-06,
+      "loss": 0.153,
+      "step": 41830
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.647529734675206e-06,
+      "loss": 0.1382,
+      "step": 41840
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.6406678865507779e-06,
+      "loss": 0.1292,
+      "step": 41850
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.6338060384263494e-06,
+      "loss": 0.1897,
+      "step": 41860
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.6269441903019213e-06,
+      "loss": 0.2398,
+      "step": 41870
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.620082342177493e-06,
+      "loss": 0.1474,
+      "step": 41880
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.613220494053065e-06,
+      "loss": 0.0904,
+      "step": 41890
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.606358645928637e-06,
+      "loss": 0.1538,
+      "step": 41900
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.5994967978042085e-06,
+      "loss": 0.1819,
+      "step": 41910
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.5926349496797804e-06,
+      "loss": 0.1664,
+      "step": 41920
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.5857731015553524e-06,
+      "loss": 0.1567,
+      "step": 41930
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.578911253430924e-06,
+      "loss": 0.2483,
+      "step": 41940
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.572049405306496e-06,
+      "loss": 0.1688,
+      "step": 41950
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.5651875571820675e-06,
+      "loss": 0.1057,
+      "step": 41960
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.5583257090576395e-06,
+      "loss": 0.1093,
+      "step": 41970
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.5514638609332114e-06,
+      "loss": 0.1489,
+      "step": 41980
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.5446020128087832e-06,
+      "loss": 0.1775,
+      "step": 41990
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.537740164684355e-06,
+      "loss": 0.1281,
+      "step": 42000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.5308783165599268e-06,
+      "loss": 0.1204,
+      "step": 42010
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.5240164684354986e-06,
+      "loss": 0.1593,
+      "step": 42020
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.5171546203110705e-06,
+      "loss": 0.1072,
+      "step": 42030
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.5102927721866422e-06,
+      "loss": 0.1034,
+      "step": 42040
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.5034309240622142e-06,
+      "loss": 0.1214,
+      "step": 42050
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.4965690759377861e-06,
+      "loss": 0.1719,
+      "step": 42060
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.4897072278133576e-06,
+      "loss": 0.1806,
+      "step": 42070
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.4828453796889296e-06,
+      "loss": 0.1666,
+      "step": 42080
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.4759835315645015e-06,
+      "loss": 0.1361,
+      "step": 42090
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.4691216834400732e-06,
+      "loss": 0.1779,
+      "step": 42100
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.4622598353156452e-06,
+      "loss": 0.2348,
+      "step": 42110
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4553979871912167e-06,
+      "loss": 0.0948,
+      "step": 42120
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4485361390667886e-06,
+      "loss": 0.1552,
+      "step": 42130
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4416742909423606e-06,
+      "loss": 0.1719,
+      "step": 42140
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4348124428179323e-06,
+      "loss": 0.1367,
+      "step": 42150
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4279505946935043e-06,
+      "loss": 0.1836,
+      "step": 42160
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.421088746569076e-06,
+      "loss": 0.1085,
+      "step": 42170
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.4142268984446477e-06,
+      "loss": 0.2896,
+      "step": 42180
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.4073650503202197e-06,
+      "loss": 0.1519,
+      "step": 42190
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.4005032021957914e-06,
+      "loss": 0.1607,
+      "step": 42200
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3936413540713633e-06,
+      "loss": 0.1744,
+      "step": 42210
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.386779505946935e-06,
+      "loss": 0.1369,
+      "step": 42220
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3799176578225068e-06,
+      "loss": 0.1484,
+      "step": 42230
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3730558096980787e-06,
+      "loss": 0.1034,
+      "step": 42240
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3661939615736507e-06,
+      "loss": 0.167,
+      "step": 42250
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3593321134492224e-06,
+      "loss": 0.1174,
+      "step": 42260
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3524702653247941e-06,
+      "loss": 0.2337,
+      "step": 42270
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3456084172003659e-06,
+      "loss": 0.1437,
+      "step": 42280
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3387465690759378e-06,
+      "loss": 0.1517,
+      "step": 42290
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3318847209515098e-06,
+      "loss": 0.1449,
+      "step": 42300
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3250228728270815e-06,
+      "loss": 0.1842,
+      "step": 42310
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3181610247026532e-06,
+      "loss": 0.1893,
+      "step": 42320
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3112991765782252e-06,
+      "loss": 0.1955,
+      "step": 42330
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3044373284537969e-06,
+      "loss": 0.1471,
+      "step": 42340
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2975754803293688e-06,
+      "loss": 0.2205,
+      "step": 42350
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2907136322049406e-06,
+      "loss": 0.1095,
+      "step": 42360
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2838517840805123e-06,
+      "loss": 0.1716,
+      "step": 42370
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2769899359560842e-06,
+      "loss": 0.1233,
+      "step": 42380
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.270128087831656e-06,
+      "loss": 0.1394,
+      "step": 42390
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.263266239707228e-06,
+      "loss": 0.2064,
+      "step": 42400
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.2564043915827998e-06,
+      "loss": 0.1643,
+      "step": 42410
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.2495425434583714e-06,
+      "loss": 0.1914,
+      "step": 42420
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.2426806953339433e-06,
+      "loss": 0.1525,
+      "step": 42430
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.235818847209515e-06,
+      "loss": 0.1284,
+      "step": 42440
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.228956999085087e-06,
+      "loss": 0.161,
+      "step": 42450
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.222095150960659e-06,
+      "loss": 0.1519,
+      "step": 42460
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.2152333028362304e-06,
+      "loss": 0.1344,
+      "step": 42470
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2083714547118024e-06,
+      "loss": 0.094,
+      "step": 42480
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2015096065873743e-06,
+      "loss": 0.1187,
+      "step": 42490
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.194647758462946e-06,
+      "loss": 0.1632,
+      "step": 42500
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1884720951509608e-06,
+      "loss": 0.1746,
+      "step": 42510
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1816102470265325e-06,
+      "loss": 0.1717,
+      "step": 42520
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1747483989021045e-06,
+      "loss": 0.1326,
+      "step": 42530
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.167886550777676e-06,
+      "loss": 0.153,
+      "step": 42540
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.161024702653248e-06,
+      "loss": 0.1379,
+      "step": 42550
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1541628545288199e-06,
+      "loss": 0.2042,
+      "step": 42560
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1473010064043916e-06,
+      "loss": 0.1214,
+      "step": 42570
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1404391582799635e-06,
+      "loss": 0.1195,
+      "step": 42580
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1335773101555353e-06,
+      "loss": 0.1884,
+      "step": 42590
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.126715462031107e-06,
+      "loss": 0.1503,
+      "step": 42600
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.119853613906679e-06,
+      "loss": 0.1431,
+      "step": 42610
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1129917657822507e-06,
+      "loss": 0.1532,
+      "step": 42620
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1061299176578226e-06,
+      "loss": 0.1221,
+      "step": 42630
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.0992680695333943e-06,
+      "loss": 0.2427,
+      "step": 42640
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.092406221408966e-06,
+      "loss": 0.2012,
+      "step": 42650
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.085544373284538e-06,
+      "loss": 0.0821,
+      "step": 42660
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.07868252516011e-06,
+      "loss": 0.1689,
+      "step": 42670
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.0718206770356817e-06,
+      "loss": 0.2131,
+      "step": 42680
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.0649588289112534e-06,
+      "loss": 0.1803,
+      "step": 42690
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.0580969807868251e-06,
+      "loss": 0.2291,
+      "step": 42700
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.051235132662397e-06,
+      "loss": 0.1245,
+      "step": 42710
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.044373284537969e-06,
+      "loss": 0.1196,
+      "step": 42720
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.0375114364135408e-06,
+      "loss": 0.228,
+      "step": 42730
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0306495882891125e-06,
+      "loss": 0.16,
+      "step": 42740
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0237877401646844e-06,
+      "loss": 0.2078,
+      "step": 42750
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0169258920402562e-06,
+      "loss": 0.0989,
+      "step": 42760
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.010064043915828e-06,
+      "loss": 0.1413,
+      "step": 42770
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0032021957913998e-06,
+      "loss": 0.1253,
+      "step": 42780
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.963403476669716e-07,
+      "loss": 0.1436,
+      "step": 42790
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.894784995425435e-07,
+      "loss": 0.1909,
+      "step": 42800
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.826166514181152e-07,
+      "loss": 0.202,
+      "step": 42810
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.757548032936872e-07,
+      "loss": 0.1455,
+      "step": 42820
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.68892955169259e-07,
+      "loss": 0.165,
+      "step": 42830
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.620311070448308e-07,
+      "loss": 0.1621,
+      "step": 42840
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.551692589204026e-07,
+      "loss": 0.1366,
+      "step": 42850
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.483074107959744e-07,
+      "loss": 0.1423,
+      "step": 42860
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.414455626715462e-07,
+      "loss": 0.1898,
+      "step": 42870
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.34583714547118e-07,
+      "loss": 0.1809,
+      "step": 42880
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.277218664226898e-07,
+      "loss": 0.2187,
+      "step": 42890
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.208600182982618e-07,
+      "loss": 0.0967,
+      "step": 42900
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.139981701738335e-07,
+      "loss": 0.1236,
+      "step": 42910
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.071363220494053e-07,
+      "loss": 0.1297,
+      "step": 42920
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.00274473924977e-07,
+      "loss": 0.1672,
+      "step": 42930
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.93412625800549e-07,
+      "loss": 0.1449,
+      "step": 42940
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.865507776761208e-07,
+      "loss": 0.134,
+      "step": 42950
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.796889295516926e-07,
+      "loss": 0.1658,
+      "step": 42960
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.728270814272644e-07,
+      "loss": 0.1414,
+      "step": 42970
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.659652333028362e-07,
+      "loss": 0.196,
+      "step": 42980
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.591033851784081e-07,
+      "loss": 0.1555,
+      "step": 42990
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.522415370539799e-07,
+      "loss": 0.1245,
+      "step": 43000
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.453796889295516e-07,
+      "loss": 0.1392,
+      "step": 43010
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.385178408051236e-07,
+      "loss": 0.1414,
+      "step": 43020
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.316559926806954e-07,
+      "loss": 0.1133,
+      "step": 43030
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.247941445562671e-07,
+      "loss": 0.1159,
+      "step": 43040
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.17932296431839e-07,
+      "loss": 0.1472,
+      "step": 43050
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.110704483074108e-07,
+      "loss": 0.4037,
+      "step": 43060
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.042086001829827e-07,
+      "loss": 0.1693,
+      "step": 43070
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 7.973467520585545e-07,
+      "loss": 0.2167,
+      "step": 43080
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 7.904849039341262e-07,
+      "loss": 0.1345,
+      "step": 43090
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.836230558096982e-07,
+      "loss": 0.1492,
+      "step": 43100
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.767612076852699e-07,
+      "loss": 0.1377,
+      "step": 43110
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.698993595608417e-07,
+      "loss": 0.201,
+      "step": 43120
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.630375114364136e-07,
+      "loss": 0.1344,
+      "step": 43130
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.561756633119854e-07,
+      "loss": 0.1257,
+      "step": 43140
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.493138151875572e-07,
+      "loss": 0.1335,
+      "step": 43150
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.42451967063129e-07,
+      "loss": 0.1658,
+      "step": 43160
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.355901189387008e-07,
+      "loss": 0.1397,
+      "step": 43170
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.287282708142727e-07,
+      "loss": 0.1585,
+      "step": 43180
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.218664226898445e-07,
+      "loss": 0.1702,
+      "step": 43190
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.150045745654163e-07,
+      "loss": 0.1825,
+      "step": 43200
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.08142726440988e-07,
+      "loss": 0.2041,
+      "step": 43210
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.0128087831656e-07,
+      "loss": 0.1474,
+      "step": 43220
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.944190301921318e-07,
+      "loss": 0.1047,
+      "step": 43230
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.875571820677035e-07,
+      "loss": 0.1534,
+      "step": 43240
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.806953339432754e-07,
+      "loss": 0.1892,
+      "step": 43250
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.738334858188472e-07,
+      "loss": 0.1494,
+      "step": 43260
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.669716376944191e-07,
+      "loss": 0.1294,
+      "step": 43270
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.601097895699909e-07,
+      "loss": 0.1417,
+      "step": 43280
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.532479414455626e-07,
+      "loss": 0.1477,
+      "step": 43290
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.463860933211346e-07,
+      "loss": 0.1646,
+      "step": 43300
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.395242451967063e-07,
+      "loss": 0.1222,
+      "step": 43310
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.326623970722781e-07,
+      "loss": 0.204,
+      "step": 43320
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.2580054894785e-07,
+      "loss": 0.1464,
+      "step": 43330
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.189387008234218e-07,
+      "loss": 0.1419,
+      "step": 43340
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.120768526989936e-07,
+      "loss": 0.1127,
+      "step": 43350
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.052150045745654e-07,
+      "loss": 0.2284,
+      "step": 43360
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.983531564501372e-07,
+      "loss": 0.1026,
+      "step": 43370
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.914913083257091e-07,
+      "loss": 0.0951,
+      "step": 43380
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.846294602012809e-07,
+      "loss": 0.1732,
+      "step": 43390
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.777676120768527e-07,
+      "loss": 0.1664,
+      "step": 43400
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.709057639524244e-07,
+      "loss": 0.1925,
+      "step": 43410
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.640439158279964e-07,
+      "loss": 0.1328,
+      "step": 43420
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.571820677035682e-07,
+      "loss": 0.095,
+      "step": 43430
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.5032021957914e-07,
+      "loss": 0.2191,
+      "step": 43440
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.434583714547118e-07,
+      "loss": 0.1986,
+      "step": 43450
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.365965233302837e-07,
+      "loss": 0.1481,
+      "step": 43460
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.297346752058555e-07,
+      "loss": 0.2082,
+      "step": 43470
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.228728270814273e-07,
+      "loss": 0.6052,
+      "step": 43480
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.16010978956999e-07,
+      "loss": 0.1735,
+      "step": 43490
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.09149130832571e-07,
+      "loss": 0.1351,
+      "step": 43500
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.022872827081428e-07,
+      "loss": 0.2337,
+      "step": 43510
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.954254345837145e-07,
+      "loss": 0.1253,
+      "step": 43520
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.885635864592864e-07,
+      "loss": 0.1308,
+      "step": 43530
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.817017383348582e-07,
+      "loss": 0.1409,
+      "step": 43540
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.7483989021043004e-07,
+      "loss": 0.1485,
+      "step": 43550
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.679780420860018e-07,
+      "loss": 0.1389,
+      "step": 43560
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.6111619396157366e-07,
+      "loss": 0.1509,
+      "step": 43570
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.542543458371455e-07,
+      "loss": 0.1476,
+      "step": 43580
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.4739249771271733e-07,
+      "loss": 0.1736,
+      "step": 43590
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.405306495882891e-07,
+      "loss": 0.165,
+      "step": 43600
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.3366880146386095e-07,
+      "loss": 0.1212,
+      "step": 43610
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.2680695333943274e-07,
+      "loss": 0.1983,
+      "step": 43620
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.199451052150046e-07,
+      "loss": 0.1114,
+      "step": 43630
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.130832570905764e-07,
+      "loss": 0.1245,
+      "step": 43640
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.0622140896614825e-07,
+      "loss": 0.1718,
+      "step": 43650
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.9935956084172003e-07,
+      "loss": 0.1805,
+      "step": 43660
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.9249771271729186e-07,
+      "loss": 0.1285,
+      "step": 43670
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.856358645928637e-07,
+      "loss": 0.1561,
+      "step": 43680
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.7877401646843554e-07,
+      "loss": 0.1798,
+      "step": 43690
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.719121683440073e-07,
+      "loss": 0.1445,
+      "step": 43700
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.6505032021957916e-07,
+      "loss": 0.2283,
+      "step": 43710
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.5818847209515094e-07,
+      "loss": 0.1231,
+      "step": 43720
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.5132662397072283e-07,
+      "loss": 0.1546,
+      "step": 43730
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.444647758462946e-07,
+      "loss": 0.1513,
+      "step": 43740
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.3760292772186645e-07,
+      "loss": 0.1895,
+      "step": 43750
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.3074107959743823e-07,
+      "loss": 0.1283,
+      "step": 43760
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.238792314730101e-07,
+      "loss": 0.3066,
+      "step": 43770
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.170173833485819e-07,
+      "loss": 0.128,
+      "step": 43780
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.1015553522415374e-07,
+      "loss": 0.1246,
+      "step": 43790
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.032936870997255e-07,
+      "loss": 0.1677,
+      "step": 43800
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.9643183897529736e-07,
+      "loss": 0.1229,
+      "step": 43810
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.895699908508692e-07,
+      "loss": 0.1555,
+      "step": 43820
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.8270814272644103e-07,
+      "loss": 0.1382,
+      "step": 43830
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.758462946020128e-07,
+      "loss": 0.147,
+      "step": 43840
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.6898444647758465e-07,
+      "loss": 0.1826,
+      "step": 43850
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.6212259835315643e-07,
+      "loss": 0.1767,
+      "step": 43860
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.552607502287283e-07,
+      "loss": 0.161,
+      "step": 43870
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.483989021043001e-07,
+      "loss": 0.1212,
+      "step": 43880
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.4153705397987194e-07,
+      "loss": 0.172,
+      "step": 43890
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.3467520585544372e-07,
+      "loss": 0.1464,
+      "step": 43900
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.2781335773101556e-07,
+      "loss": 0.1519,
+      "step": 43910
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.2095150960658737e-07,
+      "loss": 0.1657,
+      "step": 43920
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.140896614821592e-07,
+      "loss": 0.1528,
+      "step": 43930
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.0722781335773102e-07,
+      "loss": 0.1718,
+      "step": 43940
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.0036596523330285e-07,
+      "loss": 0.131,
+      "step": 43950
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.9350411710887466e-07,
+      "loss": 0.149,
+      "step": 43960
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.8664226898444647e-07,
+      "loss": 0.0924,
+      "step": 43970
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.797804208600183e-07,
+      "loss": 0.0975,
+      "step": 43980
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.7291857273559012e-07,
+      "loss": 0.1688,
+      "step": 43990
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6605672461116195e-07,
+      "loss": 0.2282,
+      "step": 44000
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.5919487648673376e-07,
+      "loss": 0.1203,
+      "step": 44010
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.5233302836230557e-07,
+      "loss": 0.1407,
+      "step": 44020
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.454711802378774e-07,
+      "loss": 0.1465,
+      "step": 44030
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.3860933211344922e-07,
+      "loss": 0.1194,
+      "step": 44040
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.3174748398902105e-07,
+      "loss": 0.1385,
+      "step": 44050
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.2488563586459286e-07,
+      "loss": 0.1606,
+      "step": 44060
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.180237877401647e-07,
+      "loss": 0.1766,
+      "step": 44070
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.1116193961573651e-07,
+      "loss": 0.1534,
+      "step": 44080
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.0430009149130833e-07,
+      "loss": 0.1617,
+      "step": 44090
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 9.743824336688016e-08,
+      "loss": 0.1464,
+      "step": 44100
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 9.057639524245198e-08,
+      "loss": 0.1745,
+      "step": 44110
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.371454711802379e-08,
+      "loss": 0.1772,
+      "step": 44120
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.685269899359561e-08,
+      "loss": 0.1255,
+      "step": 44130
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.999085086916743e-08,
+      "loss": 0.1977,
+      "step": 44140
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.312900274473926e-08,
+      "loss": 0.1553,
+      "step": 44150
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.6267154620311066e-08,
+      "loss": 0.1111,
+      "step": 44160
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.940530649588289e-08,
+      "loss": 0.1373,
+      "step": 44170
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.254345837145471e-08,
+      "loss": 0.1667,
+      "step": 44180
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.568161024702653e-08,
+      "loss": 0.1136,
+      "step": 44190
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.8819762122598354e-08,
+      "loss": 0.1942,
+      "step": 44200
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.1957913998170177e-08,
+      "loss": 0.1626,
+      "step": 44210
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.5096065873741996e-08,
+      "loss": 0.1564,
+      "step": 44220
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 0.47174692153930664,
+      "eval_runtime": 304.7093,
+      "eval_samples_per_second": 6.449,
+      "eval_steps_per_second": 1.615,
+      "eval_wer": 0.18347606432136035,
+      "step": 44220
+    },
+    {
+      "epoch": 10.0,
+      "step": 44220,
+      "total_flos": 5.359967055632466e+19,
+      "train_loss": 0.5920302843168708,
+      "train_runtime": 56436.9449,
+      "train_samples_per_second": 3.134,
+      "train_steps_per_second": 0.784
+    }
+  ],
+  "max_steps": 44220,
+  "num_train_epochs": 10,
+  "total_flos": 5.359967055632466e+19,
+  "trial_name": null,
+  "trial_params": null
+}