diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,38791 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 64453,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999997030225484e-05,
+      "loss": 3.0012,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998812090263e-05,
+      "loss": 2.9619,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997327203357e-05,
+      "loss": 2.8142,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995248362182e-05,
+      "loss": 2.7896,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992575567232e-05,
+      "loss": 2.8117,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989308819143e-05,
+      "loss": 2.7935,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998544811869e-05,
+      "loss": 2.7057,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999809934667904e-05,
+      "loss": 2.5933,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999759448645024e-05,
+      "loss": 2.6133,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999703023130264e-05,
+      "loss": 2.6661,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964065813702e-05,
+      "loss": 2.7384,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999572353680116e-05,
+      "loss": 2.6686,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999949810977577e-05,
+      "loss": 2.5632,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999417926441636e-05,
+      "loss": 2.592,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999933180369675e-05,
+      "loss": 2.8662,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999239741561584e-05,
+      "loss": 2.6443,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999141740058e-05,
+      "loss": 2.3841,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999037799209286e-05,
+      "loss": 2.3181,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999892791904014e-05,
+      "loss": 2.8112,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998812099576664e-05,
+      "loss": 2.5552,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998690340846375e-05,
+      "loss": 2.6409,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99985626428782e-05,
+      "loss": 2.8176,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998429005702475e-05,
+      "loss": 2.5206,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999828942935097e-05,
+      "loss": 2.3381,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999814391385681e-05,
+      "loss": 2.3726,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999799245925459e-05,
+      "loss": 2.5409,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999783506558029e-05,
+      "loss": 2.5615,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99976717328713e-05,
+      "loss": 2.4968,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999750246116643e-05,
+      "loss": 2.4802,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99973272505059e-05,
+      "loss": 2.459,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999714610093132e-05,
+      "loss": 2.4231,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999695901248575e-05,
+      "loss": 2.3061,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999676598521362e-05,
+      "loss": 2.5796,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99965670191608e-05,
+      "loss": 2.52,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999636211437455e-05,
+      "loss": 2.4434,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996151270903566e-05,
+      "loss": 2.8005,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999593448879793e-05,
+      "loss": 2.59,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999571176810916e-05,
+      "loss": 2.6115,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999548310889015e-05,
+      "loss": 2.5592,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995248511195236e-05,
+      "loss": 2.5598,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995007975080155e-05,
+      "loss": 2.459,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999476150060205e-05,
+      "loss": 2.436,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999450908781949e-05,
+      "loss": 2.4784,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999425073679243e-05,
+      "loss": 2.4681,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999398644758225e-05,
+      "loss": 2.6846,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993716220251754e-05,
+      "loss": 2.5075,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999344005486513e-05,
+      "loss": 2.4964,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999315795148799e-05,
+      "loss": 2.4126,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999286991018736e-05,
+      "loss": 2.4928,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992575931031685e-05,
+      "loss": 2.416,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992276014090786e-05,
+      "loss": 2.345,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999197015943594e-05,
+      "loss": 2.2783,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999165836713979e-05,
+      "loss": 2.2773,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999134063727643e-05,
+      "loss": 2.4833,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999101696992135e-05,
+      "loss": 2.3003,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999068736515143e-05,
+      "loss": 2.5178,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990351823044995e-05,
+      "loss": 2.5985,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999001034368175e-05,
+      "loss": 2.6049,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998966292714283e-05,
+      "loss": 2.5506,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998930957351078e-05,
+      "loss": 2.3595,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998895028286955e-05,
+      "loss": 2.232,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988585055304485e-05,
+      "loss": 2.2808,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998821389090238e-05,
+      "loss": 2.2583,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99878367897514e-05,
+      "loss": 2.4466,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998745375194114e-05,
+      "loss": 2.6608,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998706477756261e-05,
+      "loss": 2.3876,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998666986670821e-05,
+      "loss": 2.3225,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986269019471784e-05,
+      "loss": 2.4232,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998586223594855e-05,
+      "loss": 2.2644,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998544951623516e-05,
+      "loss": 2.4051,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998503086042966e-05,
+      "loss": 2.3918,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984606268631524e-05,
+      "loss": 2.3766,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998417574094162e-05,
+      "loss": 2.3092,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998373927746225e-05,
+      "loss": 2.4057,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983296878297085e-05,
+      "loss": 2.3432,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998284854355125e-05,
+      "loss": 2.4986,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998239427333126e-05,
+      "loss": 2.5343,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998193406774502e-05,
+      "loss": 2.5031,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99814679269019e-05,
+      "loss": 2.494,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980995850912623e-05,
+      "loss": 2.2148,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998051783988935e-05,
+      "loss": 2.3924,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998003389394565e-05,
+      "loss": 2.417,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997954401319651e-05,
+      "loss": 2.4294,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979048197758295e-05,
+      "loss": 2.5631,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997854644774882e-05,
+      "loss": 2.3411,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997803876328728e-05,
+      "loss": 2.3587,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977525144494296e-05,
+      "loss": 2.3098,
+      "step": 870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99770055914919e-05,
+      "loss": 2.2527,
+      "step": 880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976480104403536e-05,
+      "loss": 2.5115,
+      "step": 890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975948683354023e-05,
+      "loss": 2.3659,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997541132846963e-05,
+      "loss": 2.2096,
+      "step": 910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974868039878034e-05,
+      "loss": 2.3797,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99743188177083e-05,
+      "loss": 2.3605,
+      "step": 930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997376366209091e-05,
+      "loss": 2.2011,
+      "step": 940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973202573157776e-05,
+      "loss": 2.2481,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997263555104218e-05,
+      "loss": 2.2433,
+      "step": 960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997206259587884e-05,
+      "loss": 2.3716,
+      "step": 970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99714837078039e-05,
+      "loss": 2.4106,
+      "step": 980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997089888695488e-05,
+      "loss": 2.2783,
+      "step": 990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997030813347072e-05,
+      "loss": 2.2129,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9969711447491776e-05,
+      "loss": 2.2076,
+      "step": 1010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9969108829159804e-05,
+      "loss": 2.134,
+      "step": 1020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9968500278617984e-05,
+      "loss": 2.1275,
+      "step": 1030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996788579601089e-05,
+      "loss": 2.1536,
+      "step": 1040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996726538148452e-05,
+      "loss": 2.1009,
+      "step": 1050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996663903518627e-05,
+      "loss": 2.3487,
+      "step": 1060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9966006757264936e-05,
+      "loss": 2.28,
+      "step": 1070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996536854787076e-05,
+      "loss": 2.299,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996472440715535e-05,
+      "loss": 2.379,
+      "step": 1090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996407433527174e-05,
+      "loss": 2.3946,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996341833237439e-05,
+      "loss": 2.3273,
+      "step": 1110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9962756398619155e-05,
+      "loss": 2.2376,
+      "step": 1120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9962088534163284e-05,
+      "loss": 2.3319,
+      "step": 1130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996141473916546e-05,
+      "loss": 2.3277,
+      "step": 1140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996073501378575e-05,
+      "loss": 2.2619,
+      "step": 1150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9960049358185666e-05,
+      "loss": 2.4775,
+      "step": 1160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9959357772528096e-05,
+      "loss": 2.4691,
+      "step": 1170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995866025697735e-05,
+      "loss": 2.4217,
+      "step": 1180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995795681169915e-05,
+      "loss": 2.2668,
+      "step": 1190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99572474368606e-05,
+      "loss": 2.6242,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995653213263027e-05,
+      "loss": 2.3795,
+      "step": 1210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995581089917807e-05,
+      "loss": 2.3136,
+      "step": 1220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995508373667538e-05,
+      "loss": 2.1655,
+      "step": 1230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9954350645294935e-05,
+      "loss": 2.4984,
+      "step": 1240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995361162521092e-05,
+      "loss": 2.0179,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9952866676598915e-05,
+      "loss": 2.3977,
+      "step": 1260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99521157996359e-05,
+      "loss": 2.3967,
+      "step": 1270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995135899450026e-05,
+      "loss": 2.2054,
+      "step": 1280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995059626137182e-05,
+      "loss": 2.0958,
+      "step": 1290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9949827600431774e-05,
+      "loss": 2.3841,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9949053011862755e-05,
+      "loss": 2.3579,
+      "step": 1310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994827249584878e-05,
+      "loss": 2.1933,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994748605257529e-05,
+      "loss": 2.1585,
+      "step": 1330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9946693682229136e-05,
+      "loss": 2.3349,
+      "step": 1340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994589538499856e-05,
+      "loss": 2.3072,
+      "step": 1350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994509116107323e-05,
+      "loss": 2.4915,
+      "step": 1360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994428101064421e-05,
+      "loss": 2.3594,
+      "step": 1370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9943464933903984e-05,
+      "loss": 2.2165,
+      "step": 1380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9942642931046425e-05,
+      "loss": 2.1827,
+      "step": 1390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994181500226685e-05,
+      "loss": 2.6124,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9940981147761926e-05,
+      "loss": 2.5578,
+      "step": 1410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9940141367729785e-05,
+      "loss": 2.5208,
+      "step": 1420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9939295662369944e-05,
+      "loss": 2.528,
+      "step": 1430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9938444031883326e-05,
+      "loss": 2.4929,
+      "step": 1440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993758647647225e-05,
+      "loss": 2.4645,
+      "step": 1450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993672299634047e-05,
+      "loss": 2.4204,
+      "step": 1460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9935853591693126e-05,
+      "loss": 2.4796,
+      "step": 1470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9934978262736774e-05,
+      "loss": 2.2953,
+      "step": 1480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993409700967938e-05,
+      "loss": 2.2583,
+      "step": 1490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993320983273031e-05,
+      "loss": 2.3071,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9932316732100334e-05,
+      "loss": 2.2646,
+      "step": 1510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993141770800166e-05,
+      "loss": 2.39,
+      "step": 1520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993051276064785e-05,
+      "loss": 2.3467,
+      "step": 1530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992960189025393e-05,
+      "loss": 2.0758,
+      "step": 1540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9928685097036285e-05,
+      "loss": 2.3439,
+      "step": 1550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992776238121275e-05,
+      "loss": 2.2491,
+      "step": 1560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9926833743002524e-05,
+      "loss": 2.5075,
+      "step": 1570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9925899182626246e-05,
+      "loss": 2.2464,
+      "step": 1580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992495870030595e-05,
+      "loss": 2.2552,
+      "step": 1590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992401229626508e-05,
+      "loss": 2.2311,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992305997072847e-05,
+      "loss": 2.0316,
+      "step": 1610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99221017239224e-05,
+      "loss": 2.3166,
+      "step": 1620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992113755607451e-05,
+      "loss": 2.2292,
+      "step": 1630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992016746741388e-05,
+      "loss": 2.2514,
+      "step": 1640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991919145817098e-05,
+      "loss": 2.19,
+      "step": 1650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9918209528577694e-05,
+      "loss": 2.3848,
+      "step": 1660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991722167886732e-05,
+      "loss": 2.1526,
+      "step": 1670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991622790927454e-05,
+      "loss": 2.3185,
+      "step": 1680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9915228220035466e-05,
+      "loss": 2.2688,
+      "step": 1690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991422261138759e-05,
+      "loss": 2.5844,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9913211083569855e-05,
+      "loss": 2.2667,
+      "step": 1710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991219363682255e-05,
+      "loss": 2.3673,
+      "step": 1720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991117027138742e-05,
+      "loss": 2.1373,
+      "step": 1730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99101409875076e-05,
+      "loss": 2.1753,
+      "step": 1740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990910578542762e-05,
+      "loss": 2.0088,
+      "step": 1750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9908064665393426e-05,
+      "loss": 2.2182,
+      "step": 1760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990701762765238e-05,
+      "loss": 2.0533,
+      "step": 1770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990596467245322e-05,
+      "loss": 2.3013,
+      "step": 1780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990490580004613e-05,
+      "loss": 2.3737,
+      "step": 1790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9903841010682665e-05,
+      "loss": 2.3559,
+      "step": 1800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9902770304615807e-05,
+      "loss": 2.1684,
+      "step": 1810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9901693682099936e-05,
+      "loss": 2.287,
+      "step": 1820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990061114339084e-05,
+      "loss": 2.2476,
+      "step": 1830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989952268874569e-05,
+      "loss": 2.3043,
+      "step": 1840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989842831842312e-05,
+      "loss": 2.3861,
+      "step": 1850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98973280326831e-05,
+      "loss": 2.4816,
+      "step": 1860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989622183178706e-05,
+      "loss": 2.2807,
+      "step": 1870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98951097159978e-05,
+      "loss": 2.2469,
+      "step": 1880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989399168557954e-05,
+      "loss": 2.1102,
+      "step": 1890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9892867740797913e-05,
+      "loss": 2.0544,
+      "step": 1900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9891737881919934e-05,
+      "loss": 2.0653,
+      "step": 1910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989060210921405e-05,
+      "loss": 2.0007,
+      "step": 1920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988946042295009e-05,
+      "loss": 2.2082,
+      "step": 1930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9888312823399306e-05,
+      "loss": 2.1967,
+      "step": 1940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988715931083434e-05,
+      "loss": 2.3142,
+      "step": 1950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988599988552925e-05,
+      "loss": 2.286,
+      "step": 1960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9884834547759504e-05,
+      "loss": 2.2712,
+      "step": 1970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9883663297801955e-05,
+      "loss": 2.1449,
+      "step": 1980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988248613593486e-05,
+      "loss": 2.054,
+      "step": 1990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988130306243791e-05,
+      "loss": 1.896,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988011407759217e-05,
+      "loss": 2.2043,
+      "step": 2010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987891918168013e-05,
+      "loss": 2.0708,
+      "step": 2020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9877718374985675e-05,
+      "loss": 2.1179,
+      "step": 2030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987651165779408e-05,
+      "loss": 1.9875,
+      "step": 2040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987529903039206e-05,
+      "loss": 2.2379,
+      "step": 2050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987408049306771e-05,
+      "loss": 2.1113,
+      "step": 2060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987285604611051e-05,
+      "loss": 2.0418,
+      "step": 2070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9871625689811395e-05,
+      "loss": 2.1074,
+      "step": 2080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987038942446266e-05,
+      "loss": 2.0016,
+      "step": 2090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9869147250358025e-05,
+      "loss": 2.2086,
+      "step": 2100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986789916779261e-05,
+      "loss": 2.3601,
+      "step": 2110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9866645177062925e-05,
+      "loss": 2.2846,
+      "step": 2120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986538527846691e-05,
+      "loss": 2.2031,
+      "step": 2130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9864119472303885e-05,
+      "loss": 2.2138,
+      "step": 2140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9862847758874586e-05,
+      "loss": 2.3275,
+      "step": 2150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9861570138481154e-05,
+      "loss": 2.2675,
+      "step": 2160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986028661142712e-05,
+      "loss": 1.941,
+      "step": 2170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985899717801743e-05,
+      "loss": 2.0147,
+      "step": 2180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985770183855842e-05,
+      "loss": 2.0947,
+      "step": 2190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985640059335787e-05,
+      "loss": 2.2822,
+      "step": 2200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9855093442724895e-05,
+      "loss": 2.064,
+      "step": 2210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985378038697007e-05,
+      "loss": 2.1018,
+      "step": 2220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9852461426405355e-05,
+      "loss": 2.0853,
+      "step": 2230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985113656134411e-05,
+      "loss": 2.3735,
+      "step": 2240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984980579210109e-05,
+      "loss": 2.195,
+      "step": 2250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9848469118992467e-05,
+      "loss": 2.1305,
+      "step": 2260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984712654233582e-05,
+      "loss": 2.3085,
+      "step": 2270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984577806245011e-05,
+      "loss": 2.2605,
+      "step": 2280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984442367965571e-05,
+      "loss": 2.2594,
+      "step": 2290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9843063394274405e-05,
+      "loss": 2.226,
+      "step": 2300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984169720662937e-05,
+      "loss": 2.1836,
+      "step": 2310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984032511704518e-05,
+      "loss": 2.4017,
+      "step": 2320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983894712584782e-05,
+      "loss": 2.3924,
+      "step": 2330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983756323336469e-05,
+      "loss": 2.2412,
+      "step": 2340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9836173439924574e-05,
+      "loss": 2.024,
+      "step": 2350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983477774585765e-05,
+      "loss": 2.2287,
+      "step": 2360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983337615149552e-05,
+      "loss": 2.1847,
+      "step": 2370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9831968657171175e-05,
+      "loss": 2.4511,
+      "step": 2380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9830555263219006e-05,
+      "loss": 2.1062,
+      "step": 2390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9829135969974815e-05,
+      "loss": 1.9827,
+      "step": 2400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982771077777579e-05,
+      "loss": 1.8538,
+      "step": 2410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9826279686960556e-05,
+      "loss": 1.9839,
+      "step": 2420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982484269786909e-05,
+      "loss": 2.116,
+      "step": 2430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98233998108428e-05,
+      "loss": 2.2749,
+      "step": 2440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9821951026224504e-05,
+      "loss": 2.3435,
+      "step": 2450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982049634435838e-05,
+      "loss": 2.2564,
+      "step": 2460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981903576559006e-05,
+      "loss": 2.0415,
+      "step": 2470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981756929026653e-05,
+      "loss": 2.1574,
+      "step": 2480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981609691873622e-05,
+      "loss": 2.3005,
+      "step": 2490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9814618651348934e-05,
+      "loss": 2.2791,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981313448845586e-05,
+      "loss": 2.2492,
+      "step": 2510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981164443040963e-05,
+      "loss": 2.2691,
+      "step": 2520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9810148477564254e-05,
+      "loss": 2.2665,
+      "step": 2530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980864663027514e-05,
+      "loss": 2.3859,
+      "step": 2540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980713888889909e-05,
+      "loss": 2.2164,
+      "step": 2550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980562525379433e-05,
+      "loss": 2.1845,
+      "step": 2560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980410572532046e-05,
+      "loss": 2.0928,
+      "step": 2570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9802580303838506e-05,
+      "loss": 2.1488,
+      "step": 2580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9801048989710874e-05,
+      "loss": 2.1667,
+      "step": 2590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9799511783301375e-05,
+      "loss": 2.199,
+      "step": 2600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979796868497523e-05,
+      "loss": 2.181,
+      "step": 2610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979641969509903e-05,
+      "loss": 2.2468,
+      "step": 2620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979486481404081e-05,
+      "loss": 2.1072,
+      "step": 2630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9793304042169976e-05,
+      "loss": 2.0543,
+      "step": 2640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979173737985733e-05,
+      "loss": 2.2469,
+      "step": 2650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97901648274751e-05,
+      "loss": 2.1893,
+      "step": 2660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978858638539687e-05,
+      "loss": 2.1738,
+      "step": 2670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9787002053997674e-05,
+      "loss": 2.4275,
+      "step": 2680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97854118336539e-05,
+      "loss": 2.3653,
+      "step": 2690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978381572474338e-05,
+      "loss": 2.254,
+      "step": 2700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978221372764531e-05,
+      "loss": 2.2414,
+      "step": 2710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978060584274028e-05,
+      "loss": 2.1424,
+      "step": 2720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9778992070410315e-05,
+      "loss": 2.0868,
+      "step": 2730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9777372411038817e-05,
+      "loss": 2.1442,
+      "step": 2740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977574686501057e-05,
+      "loss": 2.0401,
+      "step": 2750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9774115432711785e-05,
+      "loss": 2.1412,
+      "step": 2760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9772478114530074e-05,
+      "loss": 2.1729,
+      "step": 2770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9770834910854414e-05,
+      "loss": 2.1737,
+      "step": 2780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976918582207521e-05,
+      "loss": 1.953,
+      "step": 2790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9767530848584254e-05,
+      "loss": 2.2031,
+      "step": 2800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976586999077474e-05,
+      "loss": 2.2072,
+      "step": 2810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976420324904126e-05,
+      "loss": 2.1298,
+      "step": 2820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976253062377979e-05,
+      "loss": 2.1546,
+      "step": 2830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976085211538773e-05,
+      "loss": 2.256,
+      "step": 2840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9759167724263856e-05,
+      "loss": 2.2208,
+      "step": 2850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9757477450808346e-05,
+      "loss": 2.1131,
+      "step": 2860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975578129542279e-05,
+      "loss": 2.1572,
+      "step": 2870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975407925851015e-05,
+      "loss": 2.0763,
+      "step": 2880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97523713404748e-05,
+      "loss": 2.339,
+      "step": 2890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9750657541722524e-05,
+      "loss": 2.155,
+      "step": 2900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9748937862660483e-05,
+      "loss": 2.0969,
+      "step": 2910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9747212303697236e-05,
+      "loss": 2.1013,
+      "step": 2920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974548086524275e-05,
+      "loss": 2.0357,
+      "step": 2930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9743743547708384e-05,
+      "loss": 2.0447,
+      "step": 2940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974200035150689e-05,
+      "loss": 2.1169,
+      "step": 2950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9740251277052424e-05,
+      "loss": 2.0658,
+      "step": 2960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973849632476053e-05,
+      "loss": 1.9411,
+      "step": 2970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973673549504816e-05,
+      "loss": 2.2571,
+      "step": 2980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973496878833365e-05,
+      "loss": 2.2832,
+      "step": 2990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9733196205036735e-05,
+      "loss": 1.9626,
+      "step": 3000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973141774557855e-05,
+      "loss": 2.3252,
+      "step": 3010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972963341038163e-05,
+      "loss": 2.7221,
+      "step": 3020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9727843199869905e-05,
+      "loss": 2.6555,
+      "step": 3030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9726047114468676e-05,
+      "loss": 2.287,
+      "step": 3040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9724245154604686e-05,
+      "loss": 2.1365,
+      "step": 3050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972243732070603e-05,
+      "loss": 2.0094,
+      "step": 3060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9720623613202234e-05,
+      "loss": 1.867,
+      "step": 3070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971880403252418e-05,
+      "loss": 2.3603,
+      "step": 3080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971697857910418e-05,
+      "loss": 2.1237,
+      "step": 3090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971514725337593e-05,
+      "loss": 2.0849,
+      "step": 3100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971331005577452e-05,
+      "loss": 1.7417,
+      "step": 3110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971146698673643e-05,
+      "loss": 2.2803,
+      "step": 3120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9709618046699534e-05,
+      "loss": 2.15,
+      "step": 3130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9707763236103133e-05,
+      "loss": 2.2166,
+      "step": 3140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970590255538787e-05,
+      "loss": 2.0576,
+      "step": 3150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9704036004995816e-05,
+      "loss": 2.1542,
+      "step": 3160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970216358537043e-05,
+      "loss": 2.1302,
+      "step": 3170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970028529695657e-05,
+      "loss": 1.9554,
+      "step": 3180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969840114020048e-05,
+      "loss": 2.1815,
+      "step": 3190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9696511115549806e-05,
+      "loss": 2.3238,
+      "step": 3200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969461522345358e-05,
+      "loss": 2.1987,
+      "step": 3210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969271346436223e-05,
+      "loss": 2.2443,
+      "step": 3220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9690805838727575e-05,
+      "loss": 2.642,
+      "step": 3230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9688892347002845e-05,
+      "loss": 2.6266,
+      "step": 3240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9686972989642644e-05,
+      "loss": 2.5613,
+      "step": 3250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968504776710298e-05,
+      "loss": 2.5365,
+      "step": 3260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968311667984124e-05,
+      "loss": 2.3755,
+      "step": 3270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9681179728316235e-05,
+      "loss": 2.1146,
+      "step": 3280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967923691298814e-05,
+      "loss": 2.1764,
+      "step": 3290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967728823431852e-05,
+      "loss": 2.2956,
+      "step": 3300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9675333692770364e-05,
+      "loss": 2.4662,
+      "step": 3310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967337328880803e-05,
+      "loss": 2.2265,
+      "step": 3320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967140702289727e-05,
+      "loss": 2.1081,
+      "step": 3330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966943489550524e-05,
+      "loss": 2.0846,
+      "step": 3340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966745690710048e-05,
+      "loss": 1.9382,
+      "step": 3350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9665473058152925e-05,
+      "loss": 2.0287,
+      "step": 3360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96634833491339e-05,
+      "loss": 1.9763,
+      "step": 3370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9661487780516124e-05,
+      "loss": 2.1072,
+      "step": 3380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96594863527737e-05,
+      "loss": 2.0486,
+      "step": 3390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965747906638215e-05,
+      "loss": 2.0953,
+      "step": 3400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965546592181836e-05,
+      "loss": 2.1719,
+      "step": 3410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965344691956061e-05,
+      "loss": 2.0439,
+      "step": 3420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965142206008858e-05,
+      "loss": 2.0581,
+      "step": 3430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.964939134388336e-05,
+      "loss": 2.1861,
+      "step": 3440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.964735477142738e-05,
+      "loss": 2.099,
+      "step": 3450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9645312343204506e-05,
+      "loss": 2.018,
+      "step": 3460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96432640597e-05,
+      "loss": 2.3352,
+      "step": 3470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9641209921400475e-05,
+      "loss": 2.1058,
+      "step": 3480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963914992879396e-05,
+      "loss": 1.9687,
+      "step": 3490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963708408236988e-05,
+      "loss": 2.0786,
+      "step": 3500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963501238261904e-05,
+      "loss": 2.1054,
+      "step": 3510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963293483003363e-05,
+      "loss": 2.1543,
+      "step": 3520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963085142510725e-05,
+      "loss": 1.8358,
+      "step": 3530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.962876216833488e-05,
+      "loss": 1.865,
+      "step": 3540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962666706021288e-05,
+      "loss": 2.0973,
+      "step": 3550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9624566101239024e-05,
+      "loss": 2.1343,
+      "step": 3560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9622459291912445e-05,
+      "loss": 2.1087,
+      "step": 3570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9620346632733695e-05,
+      "loss": 2.1601,
+      "step": 3580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9618228124204705e-05,
+      "loss": 2.0268,
+      "step": 3590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961610376682878e-05,
+      "loss": 1.8539,
+      "step": 3600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961397356111064e-05,
+      "loss": 1.8472,
+      "step": 3610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961183750755639e-05,
+      "loss": 1.8152,
+      "step": 3620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.96096956066735e-05,
+      "loss": 1.831,
+      "step": 3630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960754785897086e-05,
+      "loss": 2.162,
+      "step": 3640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960539426495874e-05,
+      "loss": 2.1966,
+      "step": 3650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960323482514878e-05,
+      "loss": 1.8904,
+      "step": 3660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9601069540054035e-05,
+      "loss": 1.9847,
+      "step": 3670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959889841018893e-05,
+      "loss": 1.8438,
+      "step": 3680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9596721436069295e-05,
+      "loss": 2.0892,
+      "step": 3690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9594538618212334e-05,
+      "loss": 2.0328,
+      "step": 3700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959234995713665e-05,
+      "loss": 2.0356,
+      "step": 3710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959015545336222e-05,
+      "loss": 2.1961,
+      "step": 3720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958795510741043e-05,
+      "loss": 2.314,
+      "step": 3730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9585748919804035e-05,
+      "loss": 2.2465,
+      "step": 3740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9583536891067184e-05,
+      "loss": 2.2194,
+      "step": 3750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9581319021725424e-05,
+      "loss": 2.1749,
+      "step": 3760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9579095312305677e-05,
+      "loss": 2.1942,
+      "step": 3770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957686576333625e-05,
+      "loss": 2.0216,
+      "step": 3780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9574630375346855e-05,
+      "loss": 1.7777,
+      "step": 3790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9572389148868564e-05,
+      "loss": 1.9338,
+      "step": 3800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957014208443387e-05,
+      "loss": 2.1368,
+      "step": 3810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.956788918257662e-05,
+      "loss": 1.9417,
+      "step": 3820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9565630443832065e-05,
+      "loss": 1.8576,
+      "step": 3830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.956336586873686e-05,
+      "loss": 1.8562,
+      "step": 3840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9561095457828994e-05,
+      "loss": 2.0311,
+      "step": 3850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95588192116479e-05,
+      "loss": 2.2187,
+      "step": 3860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.955653713073437e-05,
+      "loss": 2.555,
+      "step": 3870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9554249215630575e-05,
+      "loss": 2.5267,
+      "step": 3880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9551955466880094e-05,
+      "loss": 2.4391,
+      "step": 3890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.954965588502788e-05,
+      "loss": 2.4525,
+      "step": 3900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9547350470620256e-05,
+      "loss": 2.3782,
+      "step": 3910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.954503922420497e-05,
+      "loss": 2.2834,
+      "step": 3920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.954272214633111e-05,
+      "loss": 2.0641,
+      "step": 3930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.954039923754919e-05,
+      "loss": 1.9868,
+      "step": 3940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.953807049841108e-05,
+      "loss": 2.0768,
+      "step": 3950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.953573592947006e-05,
+      "loss": 2.2646,
+      "step": 3960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9533395531280764e-05,
+      "loss": 2.0901,
+      "step": 3970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.953104930439924e-05,
+      "loss": 2.161,
+      "step": 3980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9528697249382896e-05,
+      "loss": 2.6214,
+      "step": 3990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9526339366790556e-05,
+      "loss": 2.5649,
+      "step": 4000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95239756571824e-05,
+      "loss": 2.4146,
+      "step": 4010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.952160612112e-05,
+      "loss": 2.002,
+      "step": 4020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9519230759166324e-05,
+      "loss": 2.2773,
+      "step": 4030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.951684957188571e-05,
+      "loss": 2.2472,
+      "step": 4040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.951446255984389e-05,
+      "loss": 1.9906,
+      "step": 4050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9512069723607967e-05,
+      "loss": 1.8697,
+      "step": 4060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950967106374644e-05,
+      "loss": 2.014,
+      "step": 4070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950726658082918e-05,
+      "loss": 2.2205,
+      "step": 4080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950485627542747e-05,
+      "loss": 2.291,
+      "step": 4090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950244014811393e-05,
+      "loss": 2.061,
+      "step": 4100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95000181994626e-05,
+      "loss": 2.0847,
+      "step": 4110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949759043004889e-05,
+      "loss": 1.981,
+      "step": 4120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.94951568404496e-05,
+      "loss": 1.9461,
+      "step": 4130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.94927174312429e-05,
+      "loss": 1.9566,
+      "step": 4140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9490272203008345e-05,
+      "loss": 2.0425,
+      "step": 4150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9487821156326884e-05,
+      "loss": 2.048,
+      "step": 4160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.948536429178084e-05,
+      "loss": 1.8071,
+      "step": 4170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9482901609953924e-05,
+      "loss": 1.9987,
+      "step": 4180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.948043311143122e-05,
+      "loss": 1.9567,
+      "step": 4190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9477958796799206e-05,
+      "loss": 1.875,
+      "step": 4200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9475478666645725e-05,
+      "loss": 1.9359,
+      "step": 4210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947299272156001e-05,
+      "loss": 2.1412,
+      "step": 4220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9470500962132683e-05,
+      "loss": 2.1799,
+      "step": 4230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9468003388955743e-05,
+      "loss": 1.9198,
+      "step": 4240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946550000262256e-05,
+      "loss": 1.967,
+      "step": 4250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94629908037279e-05,
+      "loss": 1.9976,
+      "step": 4260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9460475792867904e-05,
+      "loss": 2.13,
+      "step": 4270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9457954970640086e-05,
+      "loss": 1.8709,
+      "step": 4280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945542833764336e-05,
+      "loss": 2.0669,
+      "step": 4290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9452895894477995e-05,
+      "loss": 2.0228,
+      "step": 4300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945035764174566e-05,
+      "loss": 2.277,
+      "step": 4310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94478135800494e-05,
+      "loss": 1.9109,
+      "step": 4320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944526370999364e-05,
+      "loss": 1.9506,
+      "step": 4330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944270803218417e-05,
+      "loss": 1.9117,
+      "step": 4340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944014654722819e-05,
+      "loss": 1.9772,
+      "step": 4350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943757925573425e-05,
+      "loss": 1.9477,
+      "step": 4360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94350061583123e-05,
+      "loss": 1.913,
+      "step": 4370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943242725557366e-05,
+      "loss": 2.0901,
+      "step": 4380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942984254813102e-05,
+      "loss": 2.021,
+      "step": 4390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942725203659848e-05,
+      "loss": 1.7747,
+      "step": 4400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9424655721591486e-05,
+      "loss": 2.07,
+      "step": 4410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942205360372687e-05,
+      "loss": 2.0515,
+      "step": 4420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941944568362287e-05,
+      "loss": 1.9969,
+      "step": 4430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9416831961899054e-05,
+      "loss": 1.9436,
+      "step": 4440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941421243917641e-05,
+      "loss": 1.8913,
+      "step": 4450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9411587116077284e-05,
+      "loss": 1.9566,
+      "step": 4460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940895599322542e-05,
+      "loss": 2.058,
+      "step": 4470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9406319071245907e-05,
+      "loss": 2.1318,
+      "step": 4480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940367635076524e-05,
+      "loss": 1.9649,
+      "step": 4490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940102783241127e-05,
+      "loss": 2.0137,
+      "step": 4500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9398373516813254e-05,
+      "loss": 1.9616,
+      "step": 4510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9395713404601795e-05,
+      "loss": 2.5022,
+      "step": 4520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.93930474964089e-05,
+      "loss": 2.3886,
+      "step": 4530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9390375792867934e-05,
+      "loss": 2.354,
+      "step": 4540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938769829461364e-05,
+      "loss": 2.1582,
+      "step": 4550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9385015002282154e-05,
+      "loss": 1.9769,
+      "step": 4560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938232591651097e-05,
+      "loss": 2.4004,
+      "step": 4570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937963103793898e-05,
+      "loss": 2.1883,
+      "step": 4580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937693036720642e-05,
+      "loss": 2.2007,
+      "step": 4590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937422390495493e-05,
+      "loss": 1.9345,
+      "step": 4600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937151165182752e-05,
+      "loss": 2.0859,
+      "step": 4610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936879360846856e-05,
+      "loss": 2.2311,
+      "step": 4620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936606977552382e-05,
+      "loss": 2.0618,
+      "step": 4630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936334015364043e-05,
+      "loss": 2.1401,
+      "step": 4640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.93606047434669e-05,
+      "loss": 2.1817,
+      "step": 4650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935786354565311e-05,
+      "loss": 2.2033,
+      "step": 4660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9355116560850325e-05,
+      "loss": 2.0494,
+      "step": 4670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935236378971117e-05,
+      "loss": 1.98,
+      "step": 4680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934960523288966e-05,
+      "loss": 1.9362,
+      "step": 4690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934684089104118e-05,
+      "loss": 2.1071,
+      "step": 4700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934407076482249e-05,
+      "loss": 2.1135,
+      "step": 4710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934129485489171e-05,
+      "loss": 1.8801,
+      "step": 4720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.933851316190836e-05,
+      "loss": 1.7928,
+      "step": 4730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9335725686533313e-05,
+      "loss": 1.7444,
+      "step": 4740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.933293242942882e-05,
+      "loss": 1.7374,
+      "step": 4750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.933013339125851e-05,
+      "loss": 1.7355,
+      "step": 4760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.932732857268739e-05,
+      "loss": 1.7329,
+      "step": 4770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9324517974381835e-05,
+      "loss": 1.9436,
+      "step": 4780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9321701597009586e-05,
+      "loss": 1.8888,
+      "step": 4790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.931887944123976e-05,
+      "loss": 1.8875,
+      "step": 4800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.931605150774287e-05,
+      "loss": 1.833,
+      "step": 4810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9313217797190744e-05,
+      "loss": 2.4164,
+      "step": 4820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9310378310256664e-05,
+      "loss": 1.9184,
+      "step": 4830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930753304761522e-05,
+      "loss": 1.809,
+      "step": 4840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9304682009942384e-05,
+      "loss": 1.7764,
+      "step": 4850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930182519791553e-05,
+      "loss": 1.7491,
+      "step": 4860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929896261221338e-05,
+      "loss": 1.7499,
+      "step": 4870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929609425351602e-05,
+      "loss": 1.8303,
+      "step": 4880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9293220122504944e-05,
+      "loss": 1.7255,
+      "step": 4890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929034021986299e-05,
+      "loss": 1.693,
+      "step": 4900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9287454546274336e-05,
+      "loss": 1.884,
+      "step": 4910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928456310242461e-05,
+      "loss": 2.0695,
+      "step": 4920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9281665889000743e-05,
+      "loss": 2.0197,
+      "step": 4930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9278762906691066e-05,
+      "loss": 1.9041,
+      "step": 4940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927585415618528e-05,
+      "loss": 2.0866,
+      "step": 4950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9272939638174444e-05,
+      "loss": 2.1303,
+      "step": 4960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9270019353351e-05,
+      "loss": 1.9051,
+      "step": 4970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926709330240875e-05,
+      "loss": 1.9725,
+      "step": 4980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926416148604288e-05,
+      "loss": 2.0612,
+      "step": 4990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926122390494993e-05,
+      "loss": 2.0576,
+      "step": 5000
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 1.898725152015686,
+      "eval_runtime": 82.1036,
+      "eval_samples_per_second": 36.539,
+      "eval_steps_per_second": 4.567,
+      "step": 5000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9258280559827806e-05,
+      "loss": 2.02,
+      "step": 5010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925533145137582e-05,
+      "loss": 1.8561,
+      "step": 5020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92523765802946e-05,
+      "loss": 1.8566,
+      "step": 5030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9249415947286185e-05,
+      "loss": 1.7604,
+      "step": 5040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924644955305397e-05,
+      "loss": 2.1573,
+      "step": 5050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924347739830271e-05,
+      "loss": 2.0955,
+      "step": 5060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924049948373853e-05,
+      "loss": 2.2568,
+      "step": 5070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923751581006894e-05,
+      "loss": 2.1872,
+      "step": 5080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923452637800281e-05,
+      "loss": 2.0216,
+      "step": 5090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923153118825036e-05,
+      "loss": 1.9783,
+      "step": 5100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9228530241523205e-05,
+      "loss": 1.9228,
+      "step": 5110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9225523538534306e-05,
+      "loss": 1.9523,
+      "step": 5120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9222511079998015e-05,
+      "loss": 2.2417,
+      "step": 5130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.921949286663003e-05,
+      "loss": 2.2712,
+      "step": 5140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.921646889914742e-05,
+      "loss": 2.5226,
+      "step": 5150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9213439178268626e-05,
+      "loss": 2.4187,
+      "step": 5160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.921040370471346e-05,
+      "loss": 2.2626,
+      "step": 5170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92073624792031e-05,
+      "loss": 2.057,
+      "step": 5180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.920431550246007e-05,
+      "loss": 2.1018,
+      "step": 5190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.920126277520829e-05,
+      "loss": 2.1591,
+      "step": 5200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9198204298173036e-05,
+      "loss": 2.2859,
+      "step": 5210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9195140072080934e-05,
+      "loss": 2.1389,
+      "step": 5220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.919207009766e-05,
+      "loss": 1.9886,
+      "step": 5230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.91889943756396e-05,
+      "loss": 2.0741,
+      "step": 5240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9185912906750477e-05,
+      "loss": 1.7405,
+      "step": 5250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9182825691724714e-05,
+      "loss": 1.65,
+      "step": 5260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.917973273129579e-05,
+      "loss": 1.5947,
+      "step": 5270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9176634026198545e-05,
+      "loss": 1.6044,
+      "step": 5280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9173529577169164e-05,
+      "loss": 1.8158,
+      "step": 5290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.917041938494521e-05,
+      "loss": 2.0296,
+      "step": 5300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.91673034502656e-05,
+      "loss": 1.9597,
+      "step": 5310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9164181773870647e-05,
+      "loss": 2.1001,
+      "step": 5320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.916105435650198e-05,
+      "loss": 1.8555,
+      "step": 5330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.915792119890263e-05,
+      "loss": 2.0666,
+      "step": 5340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.915478230181698e-05,
+      "loss": 2.0482,
+      "step": 5350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9151637665990775e-05,
+      "loss": 1.9815,
+      "step": 5360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9148487292171116e-05,
+      "loss": 1.9244,
+      "step": 5370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.914533118110648e-05,
+      "loss": 1.9903,
+      "step": 5380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.914216933354671e-05,
+      "loss": 1.841,
+      "step": 5390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.913900175024298e-05,
+      "loss": 2.0886,
+      "step": 5400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9135828431947875e-05,
+      "loss": 2.0558,
+      "step": 5410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.913264937941531e-05,
+      "loss": 2.1695,
+      "step": 5420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9129464593400565e-05,
+      "loss": 2.3233,
+      "step": 5430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.91262740746603e-05,
+      "loss": 2.1296,
+      "step": 5440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9123077823952513e-05,
+      "loss": 2.1855,
+      "step": 5450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9119875842036587e-05,
+      "loss": 2.0441,
+      "step": 5460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.911666812967324e-05,
+      "loss": 2.0958,
+      "step": 5470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.911345468762457e-05,
+      "loss": 2.2783,
+      "step": 5480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.911023551665405e-05,
+      "loss": 2.2121,
+      "step": 5490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910701061752648e-05,
+      "loss": 2.148,
+      "step": 5500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910377999100804e-05,
+      "loss": 2.0967,
+      "step": 5510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910054363786628e-05,
+      "loss": 2.1708,
+      "step": 5520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909730155887008e-05,
+      "loss": 2.0055,
+      "step": 5530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909405375478971e-05,
+      "loss": 2.1761,
+      "step": 5540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90908002263968e-05,
+      "loss": 2.0528,
+      "step": 5550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908754097446431e-05,
+      "loss": 1.8527,
+      "step": 5560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908427599976658e-05,
+      "loss": 1.783,
+      "step": 5570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908100530307933e-05,
+      "loss": 1.7497,
+      "step": 5580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90777288851796e-05,
+      "loss": 1.9056,
+      "step": 5590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.907444674684582e-05,
+      "loss": 2.0527,
+      "step": 5600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.907115888885776e-05,
+      "loss": 2.064,
+      "step": 5610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.906786531199655e-05,
+      "loss": 1.9632,
+      "step": 5620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90645660170447e-05,
+      "loss": 1.8798,
+      "step": 5630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9061261004786045e-05,
+      "loss": 2.0581,
+      "step": 5640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.905795027600581e-05,
+      "loss": 2.0661,
+      "step": 5650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9054633831490557e-05,
+      "loss": 2.2821,
+      "step": 5660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.905131167202822e-05,
+      "loss": 2.1977,
+      "step": 5670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904798379840808e-05,
+      "loss": 2.3421,
+      "step": 5680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904465021142077e-05,
+      "loss": 2.18,
+      "step": 5690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904131091185831e-05,
+      "loss": 2.1462,
+      "step": 5700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.903796590051405e-05,
+      "loss": 2.1342,
+      "step": 5710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90346151781827e-05,
+      "loss": 2.2187,
+      "step": 5720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9031258745660326e-05,
+      "loss": 2.2763,
+      "step": 5730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9027896603744364e-05,
+      "loss": 1.851,
+      "step": 5740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90245287532336e-05,
+      "loss": 1.7995,
+      "step": 5750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.902115519492817e-05,
+      "loss": 1.7303,
+      "step": 5760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.901777592962957e-05,
+      "loss": 2.0054,
+      "step": 5770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.901439095814067e-05,
+      "loss": 1.9593,
+      "step": 5780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.901100028126565e-05,
+      "loss": 2.0311,
+      "step": 5790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.900760389981008e-05,
+      "loss": 2.089,
+      "step": 5800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9004201814580896e-05,
+      "loss": 1.9953,
+      "step": 5810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9000794026386354e-05,
+      "loss": 2.1757,
+      "step": 5820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.89973805360361e-05,
+      "loss": 2.2484,
+      "step": 5830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.89939613443411e-05,
+      "loss": 2.277,
+      "step": 5840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.89905364521137e-05,
+      "loss": 2.2497,
+      "step": 5850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.898710586016761e-05,
+      "loss": 2.2037,
+      "step": 5860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.898366956931784e-05,
+      "loss": 2.2424,
+      "step": 5870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.898022758038081e-05,
+      "loss": 2.0302,
+      "step": 5880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8976779894174286e-05,
+      "loss": 2.0582,
+      "step": 5890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.897332651151736e-05,
+      "loss": 1.838,
+      "step": 5900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.89698674332305e-05,
+      "loss": 1.7248,
+      "step": 5910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.896640266013552e-05,
+      "loss": 1.7454,
+      "step": 5920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8962932193055586e-05,
+      "loss": 1.855,
+      "step": 5930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.895945603281522e-05,
+      "loss": 1.9816,
+      "step": 5940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8955974180240296e-05,
+      "loss": 2.0163,
+      "step": 5950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.895248663615803e-05,
+      "loss": 2.1321,
+      "step": 5960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.894899340139702e-05,
+      "loss": 1.9038,
+      "step": 5970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8945494476787166e-05,
+      "loss": 1.8843,
+      "step": 5980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8941989863159773e-05,
+      "loss": 1.7639,
+      "step": 5990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.893847956134747e-05,
+      "loss": 2.1,
+      "step": 6000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.893496357218423e-05,
+      "loss": 2.3216,
+      "step": 6010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.893144189650541e-05,
+      "loss": 2.1858,
+      "step": 6020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.892791453514768e-05,
+      "loss": 2.1404,
+      "step": 6030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8924381488949074e-05,
+      "loss": 2.0588,
+      "step": 6040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.892084275874899e-05,
+      "loss": 2.0701,
+      "step": 6050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.891729834538816e-05,
+      "loss": 1.9458,
+      "step": 6060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8913748249708684e-05,
+      "loss": 2.0036,
+      "step": 6070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.891019247255399e-05,
+      "loss": 1.8968,
+      "step": 6080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.890663101476887e-05,
+      "loss": 1.7224,
+      "step": 6090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8903063877199465e-05,
+      "loss": 1.658,
+      "step": 6100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.889949106069325e-05,
+      "loss": 1.6651,
+      "step": 6110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8895912566099075e-05,
+      "loss": 2.1595,
+      "step": 6120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8892328394267114e-05,
+      "loss": 2.3534,
+      "step": 6130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8888738546048926e-05,
+      "loss": 2.2488,
+      "step": 6140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8885143022297365e-05,
+      "loss": 2.2491,
+      "step": 6150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.888154182386668e-05,
+      "loss": 1.7979,
+      "step": 6160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.887793495161244e-05,
+      "loss": 2.0072,
+      "step": 6170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.887432240639158e-05,
+      "loss": 2.0917,
+      "step": 6180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.887070418906238e-05,
+      "loss": 1.7792,
+      "step": 6190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8867080300484454e-05,
+      "loss": 2.0356,
+      "step": 6200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.886345074151877e-05,
+      "loss": 2.025,
+      "step": 6210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885981551302766e-05,
+      "loss": 1.9747,
+      "step": 6220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885617461587478e-05,
+      "loss": 1.9287,
+      "step": 6230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885252805092514e-05,
+      "loss": 1.8151,
+      "step": 6240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.884887581904509e-05,
+      "loss": 1.8781,
+      "step": 6250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.884521792110236e-05,
+      "loss": 2.2645,
+      "step": 6260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.884155435796598e-05,
+      "loss": 1.9585,
+      "step": 6270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8837885130506345e-05,
+      "loss": 1.9634,
+      "step": 6280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.883421023959521e-05,
+      "loss": 1.9319,
+      "step": 6290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.883052968610565e-05,
+      "loss": 1.8935,
+      "step": 6300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.882684347091211e-05,
+      "loss": 1.9621,
+      "step": 6310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.882315159489036e-05,
+      "loss": 1.8953,
+      "step": 6320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.881945405891752e-05,
+      "loss": 1.9326,
+      "step": 6330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8815750863872075e-05,
+      "loss": 2.0082,
+      "step": 6340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8812042010633815e-05,
+      "loss": 1.824,
+      "step": 6350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8808327500083914e-05,
+      "loss": 2.051,
+      "step": 6360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880460733310487e-05,
+      "loss": 1.4587,
+      "step": 6370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880088151058052e-05,
+      "loss": 1.9454,
+      "step": 6380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8797150033396054e-05,
+      "loss": 2.0016,
+      "step": 6390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8793412902438015e-05,
+      "loss": 1.9901,
+      "step": 6400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.878967011859426e-05,
+      "loss": 1.9554,
+      "step": 6410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.878592168275402e-05,
+      "loss": 2.0197,
+      "step": 6420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8782167595807846e-05,
+      "loss": 1.7733,
+      "step": 6430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8778407858647646e-05,
+      "loss": 1.7212,
+      "step": 6440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.877464247216668e-05,
+      "loss": 2.0501,
+      "step": 6450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.87708714372595e-05,
+      "loss": 1.9117,
+      "step": 6460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8767094754822074e-05,
+      "loss": 1.944,
+      "step": 6470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.876331242575166e-05,
+      "loss": 1.8063,
+      "step": 6480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.875952445094686e-05,
+      "loss": 1.932,
+      "step": 6490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8755730831307645e-05,
+      "loss": 2.0349,
+      "step": 6500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.875193156773529e-05,
+      "loss": 1.8732,
+      "step": 6510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.874812666113245e-05,
+      "loss": 2.0624,
+      "step": 6520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.874431611240311e-05,
+      "loss": 1.8679,
+      "step": 6530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.874049992245257e-05,
+      "loss": 2.325,
+      "step": 6540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8736678092187485e-05,
+      "loss": 2.1108,
+      "step": 6550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.873285062251586e-05,
+      "loss": 2.0823,
+      "step": 6560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.872901751434703e-05,
+      "loss": 1.9649,
+      "step": 6570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8725178768591685e-05,
+      "loss": 1.9148,
+      "step": 6580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.872133438616182e-05,
+      "loss": 1.8602,
+      "step": 6590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.871748436797081e-05,
+      "loss": 1.9709,
+      "step": 6600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8713628714933345e-05,
+      "loss": 1.7993,
+      "step": 6610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.870976742796546e-05,
+      "loss": 1.749,
+      "step": 6620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8705900507984514e-05,
+      "loss": 1.6584,
+      "step": 6630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.870202795590924e-05,
+      "loss": 1.9275,
+      "step": 6640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8698149772659654e-05,
+      "loss": 2.0005,
+      "step": 6650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8694265959157175e-05,
+      "loss": 2.0807,
+      "step": 6660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.869037651632451e-05,
+      "loss": 1.9795,
+      "step": 6670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.868648144508573e-05,
+      "loss": 1.9871,
+      "step": 6680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8682580746366225e-05,
+      "loss": 2.03,
+      "step": 6690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.867867442109273e-05,
+      "loss": 2.1176,
+      "step": 6700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.867476247019332e-05,
+      "loss": 2.1169,
+      "step": 6710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8670844894597414e-05,
+      "loss": 2.1753,
+      "step": 6720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.866692169523574e-05,
+      "loss": 2.2104,
+      "step": 6730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.86629928730404e-05,
+      "loss": 2.1714,
+      "step": 6740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8659058428944784e-05,
+      "loss": 2.2279,
+      "step": 6750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.865511836388367e-05,
+      "loss": 2.2077,
+      "step": 6760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.865117267879313e-05,
+      "loss": 2.1795,
+      "step": 6770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8647221374610595e-05,
+      "loss": 2.1064,
+      "step": 6780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8643264452274825e-05,
+      "loss": 2.0315,
+      "step": 6790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.863930191272591e-05,
+      "loss": 1.9821,
+      "step": 6800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.863533375690529e-05,
+      "loss": 1.8336,
+      "step": 6810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.86313599857557e-05,
+      "loss": 1.8944,
+      "step": 6820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862738060022126e-05,
+      "loss": 1.7907,
+      "step": 6830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862339560124739e-05,
+      "loss": 1.6787,
+      "step": 6840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.861940498978086e-05,
+      "loss": 2.1433,
+      "step": 6850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.861540876676976e-05,
+      "loss": 1.9924,
+      "step": 6860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.861140693316353e-05,
+      "loss": 2.0064,
+      "step": 6870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.860739948991292e-05,
+      "loss": 2.0399,
+      "step": 6880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.860338643797004e-05,
+      "loss": 2.0269,
+      "step": 6890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8599367778288306e-05,
+      "loss": 1.985,
+      "step": 6900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8595343511822494e-05,
+      "loss": 2.0193,
+      "step": 6910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8591313639528685e-05,
+      "loss": 1.9701,
+      "step": 6920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.858727816236431e-05,
+      "loss": 1.9802,
+      "step": 6930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8583237081288126e-05,
+      "loss": 1.9126,
+      "step": 6940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.857919039726021e-05,
+      "loss": 2.1847,
+      "step": 6950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8575138111241995e-05,
+      "loss": 1.8313,
+      "step": 6960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.857108022419622e-05,
+      "loss": 1.7952,
+      "step": 6970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.856701673708698e-05,
+      "loss": 1.7582,
+      "step": 6980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.856294765087968e-05,
+      "loss": 1.8362,
+      "step": 6990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.855887296654105e-05,
+      "loss": 1.8562,
+      "step": 7000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.855479268503917e-05,
+      "loss": 1.9698,
+      "step": 7010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.855070680734345e-05,
+      "loss": 1.8776,
+      "step": 7020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.854661533442461e-05,
+      "loss": 2.0073,
+      "step": 7030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.854251826725471e-05,
+      "loss": 2.0788,
+      "step": 7040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.853841560680715e-05,
+      "loss": 1.9275,
+      "step": 7050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.853430735405663e-05,
+      "loss": 1.7879,
+      "step": 7060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.853019350997922e-05,
+      "loss": 1.6989,
+      "step": 7070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.852607407555227e-05,
+      "loss": 1.9847,
+      "step": 7080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.85219490517545e-05,
+      "loss": 2.168,
+      "step": 7090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.851781843956594e-05,
+      "loss": 2.0781,
+      "step": 7100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.851368223996794e-05,
+      "loss": 2.0415,
+      "step": 7110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8509540453943194e-05,
+      "loss": 1.9486,
+      "step": 7120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.850539308247571e-05,
+      "loss": 1.9106,
+      "step": 7130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.850124012655084e-05,
+      "loss": 2.2245,
+      "step": 7140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8497081587155245e-05,
+      "loss": 1.9552,
+      "step": 7150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.849291746527691e-05,
+      "loss": 1.871,
+      "step": 7160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.848874776190517e-05,
+      "loss": 2.0217,
+      "step": 7170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.848457247803066e-05,
+      "loss": 2.1396,
+      "step": 7180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.848039161464536e-05,
+      "loss": 2.0519,
+      "step": 7190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.847620517274256e-05,
+      "loss": 1.9063,
+      "step": 7200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.847201315331689e-05,
+      "loss": 1.9367,
+      "step": 7210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.846781555736429e-05,
+      "loss": 2.0058,
+      "step": 7220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.846361238588205e-05,
+      "loss": 1.9975,
+      "step": 7230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.845940363986875e-05,
+      "loss": 1.9838,
+      "step": 7240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.845518932032431e-05,
+      "loss": 1.9991,
+      "step": 7250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.845096942825e-05,
+      "loss": 1.8271,
+      "step": 7260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.844674396464837e-05,
+      "loss": 1.9611,
+      "step": 7270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8442512930523306e-05,
+      "loss": 2.1237,
+      "step": 7280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.843827632688006e-05,
+      "loss": 2.4828,
+      "step": 7290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.843403415472514e-05,
+      "loss": 2.3333,
+      "step": 7300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8429786415066415e-05,
+      "loss": 2.3092,
+      "step": 7310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8425533108913085e-05,
+      "loss": 1.853,
+      "step": 7320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.842127423727565e-05,
+      "loss": 1.6972,
+      "step": 7330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8417009801165936e-05,
+      "loss": 2.0058,
+      "step": 7340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.841273980159711e-05,
+      "loss": 2.0079,
+      "step": 7350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8408464239583626e-05,
+      "loss": 1.9739,
+      "step": 7360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.84041831161413e-05,
+      "loss": 2.0129,
+      "step": 7370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.839989643228725e-05,
+      "loss": 2.0464,
+      "step": 7380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.83956041890399e-05,
+      "loss": 2.0549,
+      "step": 7390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.839130638741902e-05,
+      "loss": 1.9158,
+      "step": 7400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.838700302844569e-05,
+      "loss": 1.726,
+      "step": 7410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83826941131423e-05,
+      "loss": 1.8976,
+      "step": 7420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.837837964253258e-05,
+      "loss": 1.8251,
+      "step": 7430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8374059617641574e-05,
+      "loss": 1.9301,
+      "step": 7440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.836973403949563e-05,
+      "loss": 1.8708,
+      "step": 7450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8365402909122436e-05,
+      "loss": 1.9437,
+      "step": 7460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8361066227551e-05,
+      "loss": 1.8773,
+      "step": 7470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8356723995811615e-05,
+      "loss": 1.7485,
+      "step": 7480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.835237621493592e-05,
+      "loss": 1.9634,
+      "step": 7490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83480228859569e-05,
+      "loss": 1.8195,
+      "step": 7500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83436640099088e-05,
+      "loss": 1.9039,
+      "step": 7510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8339299587827215e-05,
+      "loss": 1.9509,
+      "step": 7520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.833492962074905e-05,
+      "loss": 1.9431,
+      "step": 7530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.833055410971254e-05,
+      "loss": 2.0907,
+      "step": 7540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8326173055757215e-05,
+      "loss": 2.0375,
+      "step": 7550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.832178645992396e-05,
+      "loss": 2.1631,
+      "step": 7560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8317394323254914e-05,
+      "loss": 1.798,
+      "step": 7570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83129966467936e-05,
+      "loss": 2.0854,
+      "step": 7580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8308593431584816e-05,
+      "loss": 2.0271,
+      "step": 7590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.830418467867467e-05,
+      "loss": 2.0026,
+      "step": 7600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.829977038911063e-05,
+      "loss": 2.3122,
+      "step": 7610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8295350563941446e-05,
+      "loss": 1.9115,
+      "step": 7620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.829092520421717e-05,
+      "loss": 1.7976,
+      "step": 7630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8286494310989207e-05,
+      "loss": 1.7556,
+      "step": 7640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8282057885310236e-05,
+      "loss": 1.873,
+      "step": 7650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.82776159282343e-05,
+      "loss": 1.85,
+      "step": 7660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.827316844081671e-05,
+      "loss": 1.8485,
+      "step": 7670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8268715424114115e-05,
+      "loss": 1.8614,
+      "step": 7680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8264256879184466e-05,
+      "loss": 1.7338,
+      "step": 7690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8259792807087036e-05,
+      "loss": 1.6865,
+      "step": 7700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8255323208882406e-05,
+      "loss": 1.8391,
+      "step": 7710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.825084808563247e-05,
+      "loss": 1.8365,
+      "step": 7720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8246367438400444e-05,
+      "loss": 1.7852,
+      "step": 7730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8241881268250846e-05,
+      "loss": 2.1699,
+      "step": 7740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8237389576249514e-05,
+      "loss": 1.9428,
+      "step": 7750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.823289236346358e-05,
+      "loss": 1.912,
+      "step": 7760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.822838963096151e-05,
+      "loss": 2.0396,
+      "step": 7770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.822388137981306e-05,
+      "loss": 1.759,
+      "step": 7780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8219367611089336e-05,
+      "loss": 1.9741,
+      "step": 7790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8214848325862705e-05,
+      "loss": 2.0953,
+      "step": 7800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.821032352520687e-05,
+      "loss": 2.017,
+      "step": 7810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.820579321019685e-05,
+      "loss": 1.9514,
+      "step": 7820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.820125738190895e-05,
+      "loss": 1.9039,
+      "step": 7830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.819671604142082e-05,
+      "loss": 1.9749,
+      "step": 7840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.81921691898114e-05,
+      "loss": 1.9322,
+      "step": 7850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.818761682816093e-05,
+      "loss": 2.1528,
+      "step": 7860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.818305895755096e-05,
+      "loss": 2.0001,
+      "step": 7870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8178495579064384e-05,
+      "loss": 1.8025,
+      "step": 7880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.817392669378536e-05,
+      "loss": 2.1475,
+      "step": 7890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8169352302799386e-05,
+      "loss": 2.0314,
+      "step": 7900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8164772407193234e-05,
+      "loss": 2.0755,
+      "step": 7910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8160187008055024e-05,
+      "loss": 1.9103,
+      "step": 7920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8155596106474155e-05,
+      "loss": 1.9736,
+      "step": 7930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.815099970354134e-05,
+      "loss": 1.8619,
+      "step": 7940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8146397800348607e-05,
+      "loss": 2.0516,
+      "step": 7950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8141790397989286e-05,
+      "loss": 1.6771,
+      "step": 7960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.813717749755801e-05,
+      "loss": 1.6609,
+      "step": 7970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.813255910015072e-05,
+      "loss": 1.9101,
+      "step": 7980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.812793520686467e-05,
+      "loss": 2.0758,
+      "step": 7990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8123305818798406e-05,
+      "loss": 1.9979,
+      "step": 8000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811867093705178e-05,
+      "loss": 2.1423,
+      "step": 8010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811403056272598e-05,
+      "loss": 2.1008,
+      "step": 8020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810938469692346e-05,
+      "loss": 1.7519,
+      "step": 8030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8104733340747996e-05,
+      "loss": 1.7586,
+      "step": 8040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8100076495304655e-05,
+      "loss": 1.9796,
+      "step": 8050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8095414161699835e-05,
+      "loss": 1.8869,
+      "step": 8060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8090746341041224e-05,
+      "loss": 1.7599,
+      "step": 8070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8086073034437805e-05,
+      "loss": 1.803,
+      "step": 8080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.808139424299987e-05,
+      "loss": 1.7705,
+      "step": 8090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.807670996783902e-05,
+      "loss": 2.0492,
+      "step": 8100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.807202021006814e-05,
+      "loss": 2.061,
+      "step": 8110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8067324970801454e-05,
+      "loss": 2.0152,
+      "step": 8120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.806262425115445e-05,
+      "loss": 1.9784,
+      "step": 8130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8057918052243944e-05,
+      "loss": 2.0255,
+      "step": 8140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8053206375188046e-05,
+      "loss": 1.9468,
+      "step": 8150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.804848922110615e-05,
+      "loss": 1.9584,
+      "step": 8160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.804376659111898e-05,
+      "loss": 1.9546,
+      "step": 8170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.803903848634854e-05,
+      "loss": 1.9174,
+      "step": 8180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.803430490791815e-05,
+      "loss": 2.0649,
+      "step": 8190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.802956585695241e-05,
+      "loss": 2.1623,
+      "step": 8200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8024821334577254e-05,
+      "loss": 2.1094,
+      "step": 8210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8020071341919884e-05,
+      "loss": 2.0405,
+      "step": 8220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.80153158801088e-05,
+      "loss": 2.1514,
+      "step": 8230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8010554950273834e-05,
+      "loss": 2.1036,
+      "step": 8240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800578855354608e-05,
+      "loss": 2.2558,
+      "step": 8250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800101669105796e-05,
+      "loss": 2.4583,
+      "step": 8260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799623936394318e-05,
+      "loss": 2.1735,
+      "step": 8270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799145657333675e-05,
+      "loss": 1.833,
+      "step": 8280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.798666832037497e-05,
+      "loss": 1.9369,
+      "step": 8290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7981874606195444e-05,
+      "loss": 1.9243,
+      "step": 8300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797707543193706e-05,
+      "loss": 1.9941,
+      "step": 8310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797227079874004e-05,
+      "loss": 2.0482,
+      "step": 8320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7967460707745855e-05,
+      "loss": 2.0051,
+      "step": 8330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796264516009731e-05,
+      "loss": 1.9693,
+      "step": 8340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.795782415693848e-05,
+      "loss": 1.9817,
+      "step": 8350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.795299769941476e-05,
+      "loss": 2.0272,
+      "step": 8360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7948165788672825e-05,
+      "loss": 2.2728,
+      "step": 8370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7943328425860656e-05,
+      "loss": 2.3472,
+      "step": 8380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793848561212751e-05,
+      "loss": 2.1766,
+      "step": 8390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793363734862396e-05,
+      "loss": 1.9,
+      "step": 8400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7928783636501866e-05,
+      "loss": 1.742,
+      "step": 8410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792392447691438e-05,
+      "loss": 2.033,
+      "step": 8420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791905987101596e-05,
+      "loss": 1.9269,
+      "step": 8430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791418981996233e-05,
+      "loss": 1.7271,
+      "step": 8440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7909314324910546e-05,
+      "loss": 1.6478,
+      "step": 8450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790443338701893e-05,
+      "loss": 1.8186,
+      "step": 8460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789954700744709e-05,
+      "loss": 1.9366,
+      "step": 8470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789465518735597e-05,
+      "loss": 1.9426,
+      "step": 8480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7889757927907754e-05,
+      "loss": 1.8036,
+      "step": 8490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788485523026596e-05,
+      "loss": 1.7854,
+      "step": 8500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787994709559537e-05,
+      "loss": 1.8721,
+      "step": 8510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7875033525062065e-05,
+      "loss": 2.0534,
+      "step": 8520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787011451983343e-05,
+      "loss": 2.1017,
+      "step": 8530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786519008107813e-05,
+      "loss": 2.1193,
+      "step": 8540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786026020996612e-05,
+      "loss": 1.6915,
+      "step": 8550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7855324907668656e-05,
+      "loss": 2.2558,
+      "step": 8560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7850384175358264e-05,
+      "loss": 2.15,
+      "step": 8570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784543801420878e-05,
+      "loss": 2.1195,
+      "step": 8580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784048642539533e-05,
+      "loss": 1.9804,
+      "step": 8590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78355294100943e-05,
+      "loss": 2.1879,
+      "step": 8600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78305669694834e-05,
+      "loss": 2.1071,
+      "step": 8610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782559910474163e-05,
+      "loss": 2.0897,
+      "step": 8620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782062581704925e-05,
+      "loss": 1.9715,
+      "step": 8630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7815647107587826e-05,
+      "loss": 2.0952,
+      "step": 8640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781066297754021e-05,
+      "loss": 2.0414,
+      "step": 8650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780567342809055e-05,
+      "loss": 1.9536,
+      "step": 8660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780067846042425e-05,
+      "loss": 1.882,
+      "step": 8670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7795678075728046e-05,
+      "loss": 2.4317,
+      "step": 8680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779067227518993e-05,
+      "loss": 2.2772,
+      "step": 8690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7785661059999185e-05,
+      "loss": 2.1378,
+      "step": 8700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.778064443134641e-05,
+      "loss": 2.3588,
+      "step": 8710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7775622390423436e-05,
+      "loss": 2.0895,
+      "step": 8720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777059493842342e-05,
+      "loss": 2.2796,
+      "step": 8730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7765562076540796e-05,
+      "loss": 2.4353,
+      "step": 8740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776052380597128e-05,
+      "loss": 2.3649,
+      "step": 8750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775548012791188e-05,
+      "loss": 2.2979,
+      "step": 8760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7750431043560866e-05,
+      "loss": 2.2781,
+      "step": 8770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774537655411783e-05,
+      "loss": 2.2893,
+      "step": 8780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7740316660783614e-05,
+      "loss": 2.1151,
+      "step": 8790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7735251364760356e-05,
+      "loss": 2.0411,
+      "step": 8800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7730180667251496e-05,
+      "loss": 2.2134,
+      "step": 8810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772510456946172e-05,
+      "loss": 2.2249,
+      "step": 8820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772002307259703e-05,
+      "loss": 2.0907,
+      "step": 8830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771493617786469e-05,
+      "loss": 2.058,
+      "step": 8840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770984388647327e-05,
+      "loss": 2.1777,
+      "step": 8850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770474619963258e-05,
+      "loss": 2.1117,
+      "step": 8860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769964311855376e-05,
+      "loss": 2.1006,
+      "step": 8870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7694534644449204e-05,
+      "loss": 2.0163,
+      "step": 8880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768942077853259e-05,
+      "loss": 1.9393,
+      "step": 8890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768430152201888e-05,
+      "loss": 1.6398,
+      "step": 8900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767917687612433e-05,
+      "loss": 2.2575,
+      "step": 8910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7674046842066445e-05,
+      "loss": 2.3521,
+      "step": 8920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7668911421064046e-05,
+      "loss": 2.3196,
+      "step": 8930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76637706143372e-05,
+      "loss": 2.3459,
+      "step": 8940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765862442310728e-05,
+      "loss": 1.855,
+      "step": 8950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765347284859692e-05,
+      "loss": 1.7455,
+      "step": 8960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764831589203005e-05,
+      "loss": 1.9444,
+      "step": 8970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764315355463187e-05,
+      "loss": 1.9757,
+      "step": 8980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763798583762885e-05,
+      "loss": 1.8883,
+      "step": 8990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763281274224875e-05,
+      "loss": 1.8203,
+      "step": 9000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762763426972061e-05,
+      "loss": 1.7857,
+      "step": 9010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762245042127473e-05,
+      "loss": 1.8144,
+      "step": 9020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761726119814271e-05,
+      "loss": 1.762,
+      "step": 9030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761206660155742e-05,
+      "loss": 1.7896,
+      "step": 9040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7606866632752985e-05,
+      "loss": 1.7461,
+      "step": 9050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7601661292964826e-05,
+      "loss": 1.8114,
+      "step": 9060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.759645058342966e-05,
+      "loss": 2.1743,
+      "step": 9070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.759123450538543e-05,
+      "loss": 2.0353,
+      "step": 9080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.75860130600714e-05,
+      "loss": 2.1254,
+      "step": 9090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7580786248728086e-05,
+      "loss": 2.0014,
+      "step": 9100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.757555407259728e-05,
+      "loss": 1.885,
+      "step": 9110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7570316532922066e-05,
+      "loss": 1.8056,
+      "step": 9120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.756507363094677e-05,
+      "loss": 1.7643,
+      "step": 9130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.755982536791702e-05,
+      "loss": 1.9196,
+      "step": 9140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.755457174507971e-05,
+      "loss": 2.0162,
+      "step": 9150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.754931276368301e-05,
+      "loss": 1.953,
+      "step": 9160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.754404842497635e-05,
+      "loss": 2.2563,
+      "step": 9170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.753877873021044e-05,
+      "loss": 1.899,
+      "step": 9180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.753350368063728e-05,
+      "loss": 1.8397,
+      "step": 9190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.752822327751011e-05,
+      "loss": 2.0235,
+      "step": 9200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.752293752208347e-05,
+      "loss": 1.8929,
+      "step": 9210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.751764641561315e-05,
+      "loss": 1.9323,
+      "step": 9220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.751234995935623e-05,
+      "loss": 1.8821,
+      "step": 9230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7507048154571054e-05,
+      "loss": 1.9615,
+      "step": 9240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7501741002517224e-05,
+      "loss": 1.9782,
+      "step": 9250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.749642850445564e-05,
+      "loss": 2.0528,
+      "step": 9260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.749111066164843e-05,
+      "loss": 1.9991,
+      "step": 9270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.748578747535905e-05,
+      "loss": 2.023,
+      "step": 9280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.748045894685217e-05,
+      "loss": 1.9597,
+      "step": 9290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.747512507739376e-05,
+      "loss": 1.6098,
+      "step": 9300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.746978586825104e-05,
+      "loss": 1.7034,
+      "step": 9310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.746444132069253e-05,
+      "loss": 2.0285,
+      "step": 9320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7459091435987987e-05,
+      "loss": 1.9661,
+      "step": 9330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.745373621540844e-05,
+      "loss": 1.6788,
+      "step": 9340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744837566022621e-05,
+      "loss": 1.6256,
+      "step": 9350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744300977171485e-05,
+      "loss": 1.5972,
+      "step": 9360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.743763855114922e-05,
+      "loss": 2.0121,
+      "step": 9370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7432261999805405e-05,
+      "loss": 2.1146,
+      "step": 9380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.742688011896078e-05,
+      "loss": 1.9229,
+      "step": 9390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.742149290989399e-05,
+      "loss": 1.9889,
+      "step": 9400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.741610037388494e-05,
+      "loss": 1.8988,
+      "step": 9410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.741070251221479e-05,
+      "loss": 1.9786,
+      "step": 9420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.740529932616598e-05,
+      "loss": 2.0236,
+      "step": 9430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.739989081702222e-05,
+      "loss": 1.9261,
+      "step": 9440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.739447698606845e-05,
+      "loss": 1.9716,
+      "step": 9450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.738905783459091e-05,
+      "loss": 1.9437,
+      "step": 9460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.73836333638771e-05,
+      "loss": 2.1043,
+      "step": 9470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.737820357521577e-05,
+      "loss": 1.7465,
+      "step": 9480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.737276846989694e-05,
+      "loss": 2.0127,
+      "step": 9490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7367328049211896e-05,
+      "loss": 1.6638,
+      "step": 9500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.736188231445318e-05,
+      "loss": 1.9884,
+      "step": 9510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.73564312669146e-05,
+      "loss": 1.9697,
+      "step": 9520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.735097490789123e-05,
+      "loss": 2.0905,
+      "step": 9530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7345513238679395e-05,
+      "loss": 2.0008,
+      "step": 9540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7340046260576706e-05,
+      "loss": 1.8835,
+      "step": 9550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7334573974882005e-05,
+      "loss": 1.8408,
+      "step": 9560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.73290963828954e-05,
+      "loss": 1.8761,
+      "step": 9570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.732361348591828e-05,
+      "loss": 2.0504,
+      "step": 9580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.731812528525329e-05,
+      "loss": 1.8004,
+      "step": 9590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.731263178220432e-05,
+      "loss": 1.8499,
+      "step": 9600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.730713297807652e-05,
+      "loss": 1.9923,
+      "step": 9610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7301628874176306e-05,
+      "loss": 2.0158,
+      "step": 9620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.729611947181137e-05,
+      "loss": 1.9316,
+      "step": 9630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.729060477229064e-05,
+      "loss": 1.9267,
+      "step": 9640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7285084776924296e-05,
+      "loss": 1.8112,
+      "step": 9650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.727955948702381e-05,
+      "loss": 1.7576,
+      "step": 9660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.727402890390187e-05,
+      "loss": 1.9649,
+      "step": 9670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.726849302887245e-05,
+      "loss": 2.0302,
+      "step": 9680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7262951863250786e-05,
+      "loss": 2.0225,
+      "step": 9690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.725740540835335e-05,
+      "loss": 2.0515,
+      "step": 9700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.725185366549788e-05,
+      "loss": 1.8163,
+      "step": 9710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7246296636003366e-05,
+      "loss": 2.0532,
+      "step": 9720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7240734321190065e-05,
+      "loss": 2.0369,
+      "step": 9730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.723516672237948e-05,
+      "loss": 2.0018,
+      "step": 9740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.722959384089437e-05,
+      "loss": 1.9775,
+      "step": 9750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.722401567805875e-05,
+      "loss": 1.8971,
+      "step": 9760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7218432235197904e-05,
+      "loss": 1.8902,
+      "step": 9770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.721284351363834e-05,
+      "loss": 1.868,
+      "step": 9780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.720724951470784e-05,
+      "loss": 1.8857,
+      "step": 9790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.720165023973545e-05,
+      "loss": 1.8486,
+      "step": 9800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.719604569005144e-05,
+      "loss": 1.7548,
+      "step": 9810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.719043586698736e-05,
+      "loss": 1.9658,
+      "step": 9820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7184820771876e-05,
+      "loss": 1.9502,
+      "step": 9830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.717920040605142e-05,
+      "loss": 1.9131,
+      "step": 9840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.717357477084889e-05,
+      "loss": 1.9672,
+      "step": 9850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.716794386760498e-05,
+      "loss": 1.7682,
+      "step": 9860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7162307697657474e-05,
+      "loss": 1.7251,
+      "step": 9870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7156666262345446e-05,
+      "loss": 1.9932,
+      "step": 9880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.715101956300918e-05,
+      "loss": 1.9698,
+      "step": 9890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.714536760099024e-05,
+      "loss": 1.935,
+      "step": 9900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.713971037763143e-05,
+      "loss": 1.8971,
+      "step": 9910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.71340478942768e-05,
+      "loss": 1.8972,
+      "step": 9920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.712838015227166e-05,
+      "loss": 2.0097,
+      "step": 9930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.712270715296256e-05,
+      "loss": 1.7283,
+      "step": 9940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7117028897697294e-05,
+      "loss": 2.2552,
+      "step": 9950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.711134538782493e-05,
+      "loss": 2.3688,
+      "step": 9960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.710565662469575e-05,
+      "loss": 2.5769,
+      "step": 9970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.709996260966131e-05,
+      "loss": 2.6396,
+      "step": 9980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.709426334407441e-05,
+      "loss": 2.5805,
+      "step": 9990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7088558829289084e-05,
+      "loss": 2.621,
+      "step": 10000
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 1.8739286661148071,
+      "eval_runtime": 82.2522,
+      "eval_samples_per_second": 36.473,
+      "eval_steps_per_second": 4.559,
+      "step": 10000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.708284906666063e-05,
+      "loss": 2.1815,
+      "step": 10010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.707713405754558e-05,
+      "loss": 1.765,
+      "step": 10020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7071413803301714e-05,
+      "loss": 1.9038,
+      "step": 10030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.706568830528806e-05,
+      "loss": 1.8567,
+      "step": 10040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.70599575648649e-05,
+      "loss": 1.857,
+      "step": 10050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.705422158339376e-05,
+      "loss": 1.8576,
+      "step": 10060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.704848036223739e-05,
+      "loss": 1.7714,
+      "step": 10070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.704273390275981e-05,
+      "loss": 1.7014,
+      "step": 10080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.703698220632628e-05,
+      "loss": 1.8278,
+      "step": 10090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.703122527430328e-05,
+      "loss": 2.0879,
+      "step": 10100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.702546310805857e-05,
+      "loss": 1.6694,
+      "step": 10110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.701969570896113e-05,
+      "loss": 1.739,
+      "step": 10120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.70139230783812e-05,
+      "loss": 1.7439,
+      "step": 10130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.700814521769025e-05,
+      "loss": 1.6742,
+      "step": 10140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.700236212826098e-05,
+      "loss": 1.699,
+      "step": 10150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.699657381146736e-05,
+      "loss": 1.9494,
+      "step": 10160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6990780268684586e-05,
+      "loss": 1.9078,
+      "step": 10170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.698498150128911e-05,
+      "loss": 1.9222,
+      "step": 10180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.69791775106586e-05,
+      "loss": 1.9113,
+      "step": 10190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.697336829817199e-05,
+      "loss": 1.9318,
+      "step": 10200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.696755386520943e-05,
+      "loss": 1.9121,
+      "step": 10210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6961734213152346e-05,
+      "loss": 1.9519,
+      "step": 10220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.695590934338336e-05,
+      "loss": 1.8991,
+      "step": 10230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6950079257286385e-05,
+      "loss": 1.9348,
+      "step": 10240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.694424395624651e-05,
+      "loss": 1.9006,
+      "step": 10250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.693840344165011e-05,
+      "loss": 1.9004,
+      "step": 10260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6932557714884804e-05,
+      "loss": 1.9497,
+      "step": 10270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6926706777339404e-05,
+      "loss": 1.9068,
+      "step": 10280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6920850630404e-05,
+      "loss": 1.8316,
+      "step": 10290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.691498927546991e-05,
+      "loss": 1.966,
+      "step": 10300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.690912271392969e-05,
+      "loss": 1.9009,
+      "step": 10310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.690325094717711e-05,
+      "loss": 2.0095,
+      "step": 10320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6897373976607205e-05,
+      "loss": 1.9975,
+      "step": 10330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.689149180361625e-05,
+      "loss": 1.9553,
+      "step": 10340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.688560442960173e-05,
+      "loss": 1.9482,
+      "step": 10350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6879711855962377e-05,
+      "loss": 1.8805,
+      "step": 10360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6873814084098164e-05,
+      "loss": 1.8808,
+      "step": 10370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.68679111154103e-05,
+      "loss": 1.8206,
+      "step": 10380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6862002951301224e-05,
+      "loss": 1.9451,
+      "step": 10390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.68560895931746e-05,
+      "loss": 1.9995,
+      "step": 10400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.685017104243534e-05,
+      "loss": 1.7784,
+      "step": 10410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.684424730048959e-05,
+      "loss": 1.8627,
+      "step": 10420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.683831836874472e-05,
+      "loss": 1.9029,
+      "step": 10430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.683238424860933e-05,
+      "loss": 1.6917,
+      "step": 10440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.682644494149327e-05,
+      "loss": 1.6733,
+      "step": 10450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6820500448807614e-05,
+      "loss": 1.7877,
+      "step": 10460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.681455077196466e-05,
+      "loss": 1.7785,
+      "step": 10470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.680859591237794e-05,
+      "loss": 1.6942,
+      "step": 10480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.680263587146223e-05,
+      "loss": 1.6603,
+      "step": 10490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.679667065063352e-05,
+      "loss": 1.8127,
+      "step": 10500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.679070025130905e-05,
+      "loss": 1.711,
+      "step": 10510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6784724674907265e-05,
+      "loss": 1.7378,
+      "step": 10520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.677874392284787e-05,
+      "loss": 1.8696,
+      "step": 10530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.677275799655177e-05,
+      "loss": 1.7392,
+      "step": 10540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.676676689744112e-05,
+      "loss": 1.6893,
+      "step": 10550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6760770626939296e-05,
+      "loss": 1.8008,
+      "step": 10560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6754769186470905e-05,
+      "loss": 1.8705,
+      "step": 10570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.674876257746178e-05,
+      "loss": 1.8482,
+      "step": 10580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6742750801338975e-05,
+      "loss": 1.8464,
+      "step": 10590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6736733859530796e-05,
+      "loss": 1.7576,
+      "step": 10600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.673071175346675e-05,
+      "loss": 1.6226,
+      "step": 10610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6724684484577574e-05,
+      "loss": 1.5831,
+      "step": 10620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.671865205429525e-05,
+      "loss": 1.8734,
+      "step": 10630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.671261446405297e-05,
+      "loss": 1.8517,
+      "step": 10640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.670657171528516e-05,
+      "loss": 2.0043,
+      "step": 10650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6700523809427463e-05,
+      "loss": 1.9505,
+      "step": 10660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.669447074791676e-05,
+      "loss": 1.8915,
+      "step": 10670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.668841253219114e-05,
+      "loss": 1.8408,
+      "step": 10680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.668234916368992e-05,
+      "loss": 1.9041,
+      "step": 10690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.667628064385367e-05,
+      "loss": 1.8687,
+      "step": 10700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6670206974124145e-05,
+      "loss": 1.9502,
+      "step": 10710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.666412815594434e-05,
+      "loss": 1.7661,
+      "step": 10720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6658044190758474e-05,
+      "loss": 1.6825,
+      "step": 10730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.665195508001198e-05,
+      "loss": 1.9132,
+      "step": 10740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6645860825151536e-05,
+      "loss": 1.9045,
+      "step": 10750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6639761427625024e-05,
+      "loss": 1.8062,
+      "step": 10760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.663365688888154e-05,
+      "loss": 1.9197,
+      "step": 10770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.662754721037142e-05,
+      "loss": 1.9311,
+      "step": 10780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.662143239354621e-05,
+      "loss": 2.028,
+      "step": 10790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.661531243985869e-05,
+      "loss": 1.8718,
+      "step": 10800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.660918735076284e-05,
+      "loss": 1.8887,
+      "step": 10810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.660305712771387e-05,
+      "loss": 1.866,
+      "step": 10820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.659692177216821e-05,
+      "loss": 1.9326,
+      "step": 10830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.659078128558352e-05,
+      "loss": 1.8323,
+      "step": 10840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6584635669418666e-05,
+      "loss": 1.7588,
+      "step": 10850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6578484925133726e-05,
+      "loss": 1.7223,
+      "step": 10860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.657232905419001e-05,
+      "loss": 1.8745,
+      "step": 10870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6566168058050055e-05,
+      "loss": 1.8508,
+      "step": 10880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.656000193817758e-05,
+      "loss": 1.9472,
+      "step": 10890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.655383069603756e-05,
+      "loss": 1.6377,
+      "step": 10900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.654765433309616e-05,
+      "loss": 1.5515,
+      "step": 10910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.654147285082079e-05,
+      "loss": 1.5038,
+      "step": 10920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.653528625068004e-05,
+      "loss": 1.5037,
+      "step": 10930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6529094534143745e-05,
+      "loss": 1.8102,
+      "step": 10940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.652289770268294e-05,
+      "loss": 1.8538,
+      "step": 10950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.651669575776988e-05,
+      "loss": 1.6151,
+      "step": 10960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6510488700878044e-05,
+      "loss": 1.5582,
+      "step": 10970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.650427653348211e-05,
+      "loss": 1.6568,
+      "step": 10980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.649805925705797e-05,
+      "loss": 1.8329,
+      "step": 10990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.649183687308275e-05,
+      "loss": 1.7992,
+      "step": 11000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.648560938303477e-05,
+      "loss": 1.8048,
+      "step": 11010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.647937678839356e-05,
+      "loss": 1.8764,
+      "step": 11020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.647313909063989e-05,
+      "loss": 1.9041,
+      "step": 11030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6466896291255724e-05,
+      "loss": 1.8926,
+      "step": 11040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6460648391724215e-05,
+      "loss": 1.7009,
+      "step": 11050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.645439539352978e-05,
+      "loss": 1.8641,
+      "step": 11060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6448137298158e-05,
+      "loss": 1.8535,
+      "step": 11070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6441874107095684e-05,
+      "loss": 1.7434,
+      "step": 11080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.643560582183087e-05,
+      "loss": 1.9244,
+      "step": 11090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.642933244385277e-05,
+      "loss": 1.9989,
+      "step": 11100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.642305397465185e-05,
+      "loss": 1.9856,
+      "step": 11110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6416770415719736e-05,
+      "loss": 1.8266,
+      "step": 11120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6410481768549294e-05,
+      "loss": 2.0761,
+      "step": 11130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.640418803463461e-05,
+      "loss": 2.0146,
+      "step": 11140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.639788921547093e-05,
+      "loss": 1.9131,
+      "step": 11150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.639158531255478e-05,
+      "loss": 1.9705,
+      "step": 11160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.638527632738382e-05,
+      "loss": 1.9331,
+      "step": 11170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6378962261456974e-05,
+      "loss": 1.8699,
+      "step": 11180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6372643116274335e-05,
+      "loss": 1.8694,
+      "step": 11190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6366318893337216e-05,
+      "loss": 1.8108,
+      "step": 11200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.635998959414816e-05,
+      "loss": 1.6571,
+      "step": 11210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.635365522021086e-05,
+      "loss": 1.5681,
+      "step": 11220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.634731577303028e-05,
+      "loss": 1.7951,
+      "step": 11230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6340971254112544e-05,
+      "loss": 1.8145,
+      "step": 11240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6334621664965e-05,
+      "loss": 1.6982,
+      "step": 11250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6328267007096196e-05,
+      "loss": 1.7583,
+      "step": 11260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.632190728201587e-05,
+      "loss": 1.9065,
+      "step": 11270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6315542491235005e-05,
+      "loss": 1.9994,
+      "step": 11280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.630917263626574e-05,
+      "loss": 1.8391,
+      "step": 11290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6302797718621436e-05,
+      "loss": 2.2777,
+      "step": 11300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6296417739816675e-05,
+      "loss": 2.056,
+      "step": 11310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.629003270136721e-05,
+      "loss": 1.6578,
+      "step": 11320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6283642604790015e-05,
+      "loss": 1.9703,
+      "step": 11330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6277247451603266e-05,
+      "loss": 1.9719,
+      "step": 11340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6270847243326334e-05,
+      "loss": 1.9471,
+      "step": 11350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6264441981479786e-05,
+      "loss": 2.0095,
+      "step": 11360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.625803166758541e-05,
+      "loss": 1.9687,
+      "step": 11370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.625161630316617e-05,
+      "loss": 1.9939,
+      "step": 11380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.624519588974624e-05,
+      "loss": 1.9446,
+      "step": 11390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6238770428851e-05,
+      "loss": 1.9293,
+      "step": 11400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6232339922007026e-05,
+      "loss": 1.8977,
+      "step": 11410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.622590437074209e-05,
+      "loss": 1.8549,
+      "step": 11420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6219463776585155e-05,
+      "loss": 1.8216,
+      "step": 11430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6213018141066394e-05,
+      "loss": 1.8008,
+      "step": 11440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6206567465717175e-05,
+      "loss": 1.9247,
+      "step": 11450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6200111752070066e-05,
+      "loss": 2.4714,
+      "step": 11460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.619365100165882e-05,
+      "loss": 1.939,
+      "step": 11470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.618718521601839e-05,
+      "loss": 1.9736,
+      "step": 11480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.618071439668495e-05,
+      "loss": 1.9322,
+      "step": 11490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.617423854519583e-05,
+      "loss": 1.7523,
+      "step": 11500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6167757663089596e-05,
+      "loss": 1.7063,
+      "step": 11510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.616127175190598e-05,
+      "loss": 1.847,
+      "step": 11520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6154780813185896e-05,
+      "loss": 1.7941,
+      "step": 11530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6148284848471503e-05,
+      "loss": 1.7386,
+      "step": 11540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.614178385930611e-05,
+      "loss": 1.8555,
+      "step": 11550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.613527784723424e-05,
+      "loss": 1.9049,
+      "step": 11560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.612876681380161e-05,
+      "loss": 1.9095,
+      "step": 11570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6122250760555105e-05,
+      "loss": 1.9012,
+      "step": 11580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.611572968904284e-05,
+      "loss": 1.9786,
+      "step": 11590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6109203600814094e-05,
+      "loss": 1.8501,
+      "step": 11600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.610267249741935e-05,
+      "loss": 2.006,
+      "step": 11610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6096136380410295e-05,
+      "loss": 1.8907,
+      "step": 11620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.608959525133977e-05,
+      "loss": 1.7982,
+      "step": 11630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6083049111761837e-05,
+      "loss": 1.7995,
+      "step": 11640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.607649796323175e-05,
+      "loss": 1.8037,
+      "step": 11650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6069941807305926e-05,
+      "loss": 2.2565,
+      "step": 11660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6063380645542005e-05,
+      "loss": 1.8256,
+      "step": 11670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6056814479498786e-05,
+      "loss": 1.8434,
+      "step": 11680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.60502433107363e-05,
+      "loss": 1.8675,
+      "step": 11690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.604366714081571e-05,
+      "loss": 1.8115,
+      "step": 11700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.60370859712994e-05,
+      "loss": 1.7835,
+      "step": 11710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6030499803750936e-05,
+      "loss": 1.7173,
+      "step": 11720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.602390863973509e-05,
+      "loss": 1.8121,
+      "step": 11730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.601731248081779e-05,
+      "loss": 1.9854,
+      "step": 11740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.601071132856616e-05,
+      "loss": 1.8921,
+      "step": 11750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.600410518454853e-05,
+      "loss": 1.8271,
+      "step": 11760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.599749405033438e-05,
+      "loss": 2.0631,
+      "step": 11770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5990877927494414e-05,
+      "loss": 1.9179,
+      "step": 11780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5984256817600496e-05,
+      "loss": 1.9427,
+      "step": 11790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5977630722225685e-05,
+      "loss": 1.8756,
+      "step": 11800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.597099964294422e-05,
+      "loss": 1.8167,
+      "step": 11810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.596436358133153e-05,
+      "loss": 1.7602,
+      "step": 11820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.595772253896421e-05,
+      "loss": 1.7515,
+      "step": 11830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5951076517420066e-05,
+      "loss": 1.6922,
+      "step": 11840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.594442551827807e-05,
+      "loss": 1.6731,
+      "step": 11850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5937769543118374e-05,
+      "loss": 1.7259,
+      "step": 11860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.593110859352232e-05,
+      "loss": 1.6148,
+      "step": 11870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5924442671072435e-05,
+      "loss": 1.6675,
+      "step": 11880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.591777177735242e-05,
+      "loss": 1.8021,
+      "step": 11890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5911095913947155e-05,
+      "loss": 1.8148,
+      "step": 11900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.59044150824427e-05,
+      "loss": 1.6678,
+      "step": 11910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.589772928442632e-05,
+      "loss": 1.6347,
+      "step": 11920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.589103852148641e-05,
+      "loss": 1.9682,
+      "step": 11930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5884342795212595e-05,
+      "loss": 1.764,
+      "step": 11940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.587764210719566e-05,
+      "loss": 1.7597,
+      "step": 11950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5870936459027564e-05,
+      "loss": 1.7164,
+      "step": 11960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.586422585230145e-05,
+      "loss": 2.3256,
+      "step": 11970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.585751028861162e-05,
+      "loss": 2.0235,
+      "step": 11980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.585078976955359e-05,
+      "loss": 1.9836,
+      "step": 11990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.584406429672403e-05,
+      "loss": 1.941,
+      "step": 12000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5837333871720786e-05,
+      "loss": 1.9123,
+      "step": 12010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.58305984961429e-05,
+      "loss": 1.8873,
+      "step": 12020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.582385817159055e-05,
+      "loss": 1.8653,
+      "step": 12030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.581711289966514e-05,
+      "loss": 1.7987,
+      "step": 12040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.581036268196921e-05,
+      "loss": 1.7142,
+      "step": 12050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.580360752010651e-05,
+      "loss": 1.6927,
+      "step": 12060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.579684741568191e-05,
+      "loss": 1.683,
+      "step": 12070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.579008237030152e-05,
+      "loss": 1.6997,
+      "step": 12080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5783312385572574e-05,
+      "loss": 1.84,
+      "step": 12090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5776537463103506e-05,
+      "loss": 1.9425,
+      "step": 12100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.576975760450393e-05,
+      "loss": 1.8408,
+      "step": 12110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.576297281138459e-05,
+      "loss": 1.7698,
+      "step": 12120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.575618308535744e-05,
+      "loss": 1.9723,
+      "step": 12130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5749388428035604e-05,
+      "loss": 1.8772,
+      "step": 12140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.574258884103337e-05,
+      "loss": 1.8779,
+      "step": 12150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.573578432596619e-05,
+      "loss": 2.0864,
+      "step": 12160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.57289748844507e-05,
+      "loss": 2.0642,
+      "step": 12170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5722160518104695e-05,
+      "loss": 1.7313,
+      "step": 12180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5715341228547146e-05,
+      "loss": 1.6198,
+      "step": 12190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5708517017398203e-05,
+      "loss": 1.5874,
+      "step": 12200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.570168788627916e-05,
+      "loss": 1.9228,
+      "step": 12210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.569485383681251e-05,
+      "loss": 1.9219,
+      "step": 12220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.568801487062189e-05,
+      "loss": 1.8953,
+      "step": 12230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.568117098933212e-05,
+      "loss": 1.8927,
+      "step": 12240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.567432219456918e-05,
+      "loss": 1.8993,
+      "step": 12250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.566746848796022e-05,
+      "loss": 1.8572,
+      "step": 12260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.566060987113356e-05,
+      "loss": 1.8601,
+      "step": 12270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.565374634571868e-05,
+      "loss": 1.9086,
+      "step": 12280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.564687791334623e-05,
+      "loss": 1.9078,
+      "step": 12290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.564000457564803e-05,
+      "loss": 1.9682,
+      "step": 12300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5633126334257056e-05,
+      "loss": 1.9442,
+      "step": 12310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.562624319080746e-05,
+      "loss": 1.7455,
+      "step": 12320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.561935514693455e-05,
+      "loss": 1.5605,
+      "step": 12330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5612462204274796e-05,
+      "loss": 1.9678,
+      "step": 12340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.560556436446584e-05,
+      "loss": 1.4714,
+      "step": 12350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.559866162914649e-05,
+      "loss": 1.4454,
+      "step": 12360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.55917539999567e-05,
+      "loss": 1.9687,
+      "step": 12370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.558484147853761e-05,
+      "loss": 1.7806,
+      "step": 12380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5577924066531506e-05,
+      "loss": 1.9462,
+      "step": 12390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5571001765581836e-05,
+      "loss": 2.0398,
+      "step": 12400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.556407457733322e-05,
+      "loss": 1.9248,
+      "step": 12410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.555714250343143e-05,
+      "loss": 1.9582,
+      "step": 12420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.55502055455234e-05,
+      "loss": 1.9771,
+      "step": 12430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.554326370525723e-05,
+      "loss": 1.9089,
+      "step": 12440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5536316984282165e-05,
+      "loss": 1.7116,
+      "step": 12450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5529365384248645e-05,
+      "loss": 1.7041,
+      "step": 12460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.552240890680822e-05,
+      "loss": 1.804,
+      "step": 12470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.551544755361363e-05,
+      "loss": 2.0004,
+      "step": 12480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.550848132631877e-05,
+      "loss": 1.9521,
+      "step": 12490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.55015102265787e-05,
+      "loss": 1.8869,
+      "step": 12500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.549453425604961e-05,
+      "loss": 1.9398,
+      "step": 12510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.548755341638888e-05,
+      "loss": 1.9281,
+      "step": 12520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.548056770925502e-05,
+      "loss": 1.932,
+      "step": 12530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.54735771363077e-05,
+      "loss": 1.8425,
+      "step": 12540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.546658169920778e-05,
+      "loss": 1.7888,
+      "step": 12550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.545958139961724e-05,
+      "loss": 1.6667,
+      "step": 12560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5452576239199204e-05,
+      "loss": 1.6906,
+      "step": 12570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.544556621961799e-05,
+      "loss": 1.8205,
+      "step": 12580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.543855134253906e-05,
+      "loss": 1.8937,
+      "step": 12590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.543153160962901e-05,
+      "loss": 1.8233,
+      "step": 12600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.54245070225556e-05,
+      "loss": 1.913,
+      "step": 12610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5417477582987747e-05,
+      "loss": 1.7161,
+      "step": 12620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.541044329259553e-05,
+      "loss": 1.6269,
+      "step": 12630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.540340415305016e-05,
+      "loss": 1.8336,
+      "step": 12640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5396360166024e-05,
+      "loss": 1.9987,
+      "step": 12650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.53893113331906e-05,
+      "loss": 1.9546,
+      "step": 12660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.538225765622462e-05,
+      "loss": 1.9335,
+      "step": 12670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.537519913680187e-05,
+      "loss": 1.8625,
+      "step": 12680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5368135776599365e-05,
+      "loss": 1.8887,
+      "step": 12690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5361067577295195e-05,
+      "loss": 1.9487,
+      "step": 12700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.535399454056866e-05,
+      "loss": 1.9788,
+      "step": 12710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.534691666810017e-05,
+      "loss": 2.1135,
+      "step": 12720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5339833961571324e-05,
+      "loss": 2.1068,
+      "step": 12730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.533274642266482e-05,
+      "loss": 2.1422,
+      "step": 12740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.532565405306454e-05,
+      "loss": 1.9906,
+      "step": 12750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.531855685445549e-05,
+      "loss": 2.0665,
+      "step": 12760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.531145482852387e-05,
+      "loss": 1.5896,
+      "step": 12770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5304347976956954e-05,
+      "loss": 1.4532,
+      "step": 12780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.529723630144323e-05,
+      "loss": 1.7132,
+      "step": 12790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5290119803672295e-05,
+      "loss": 1.5862,
+      "step": 12800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.528299848533489e-05,
+      "loss": 1.8516,
+      "step": 12810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.527587234812292e-05,
+      "loss": 2.0298,
+      "step": 12820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.526874139372943e-05,
+      "loss": 1.9277,
+      "step": 12830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.526160562384861e-05,
+      "loss": 1.9302,
+      "step": 12840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.525446504017578e-05,
+      "loss": 2.0459,
+      "step": 12850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5247319644407415e-05,
+      "loss": 1.8374,
+      "step": 12860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.524016943824113e-05,
+      "loss": 1.8198,
+      "step": 12870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5233014423375694e-05,
+      "loss": 1.9469,
+      "step": 12880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5225854601511006e-05,
+      "loss": 1.874,
+      "step": 12890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.521868997434811e-05,
+      "loss": 1.8618,
+      "step": 12900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5211520543589195e-05,
+      "loss": 1.9154,
+      "step": 12910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.520434631093757e-05,
+      "loss": 1.9372,
+      "step": 12920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.519716727809773e-05,
+      "loss": 1.7493,
+      "step": 12930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.518998344677527e-05,
+      "loss": 1.8592,
+      "step": 12940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.518279481867695e-05,
+      "loss": 1.9,
+      "step": 12950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5175601395510635e-05,
+      "loss": 1.7847,
+      "step": 12960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5168403178985374e-05,
+      "loss": 1.6724,
+      "step": 12970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5161200170811326e-05,
+      "loss": 1.6806,
+      "step": 12980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5153992372699796e-05,
+      "loss": 1.9181,
+      "step": 12990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5146779786363225e-05,
+      "loss": 2.0027,
+      "step": 13000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.513956241351519e-05,
+      "loss": 1.9227,
+      "step": 13010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.513234025587042e-05,
+      "loss": 1.8939,
+      "step": 13020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.512511331514477e-05,
+      "loss": 1.8229,
+      "step": 13030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.511788159305521e-05,
+      "loss": 1.7994,
+      "step": 13040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5110645091319894e-05,
+      "loss": 1.7008,
+      "step": 13050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.510340381165806e-05,
+      "loss": 1.9179,
+      "step": 13060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.509615775579012e-05,
+      "loss": 1.6109,
+      "step": 13070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.50889069254376e-05,
+      "loss": 1.6107,
+      "step": 13080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5081651322323174e-05,
+      "loss": 1.9068,
+      "step": 13090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5074390948170635e-05,
+      "loss": 1.8588,
+      "step": 13100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.506712580470492e-05,
+      "loss": 1.7813,
+      "step": 13110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.505985589365209e-05,
+      "loss": 1.7971,
+      "step": 13120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.505258121673936e-05,
+      "loss": 1.8641,
+      "step": 13130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.504530177569505e-05,
+      "loss": 1.9082,
+      "step": 13140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.503801757224861e-05,
+      "loss": 1.9987,
+      "step": 13150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.503072860813067e-05,
+      "loss": 1.9675,
+      "step": 13160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.502343488507293e-05,
+      "loss": 1.9649,
+      "step": 13170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.501613640480826e-05,
+      "loss": 1.9364,
+      "step": 13180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.500883316907064e-05,
+      "loss": 1.9237,
+      "step": 13190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.500152517959519e-05,
+      "loss": 1.8864,
+      "step": 13200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.499421243811815e-05,
+      "loss": 1.9767,
+      "step": 13210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.49868949463769e-05,
+      "loss": 2.03,
+      "step": 13220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.497957270610995e-05,
+      "loss": 2.008,
+      "step": 13230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.497224571905693e-05,
+      "loss": 1.8984,
+      "step": 13240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.49649139869586e-05,
+      "loss": 1.6864,
+      "step": 13250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4957577511556836e-05,
+      "loss": 1.826,
+      "step": 13260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.495023629459466e-05,
+      "loss": 1.9153,
+      "step": 13270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.494289033781622e-05,
+      "loss": 1.802,
+      "step": 13280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4935539642966776e-05,
+      "loss": 1.8751,
+      "step": 13290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.492818421179271e-05,
+      "loss": 1.8342,
+      "step": 13300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.492082404604156e-05,
+      "loss": 1.9485,
+      "step": 13310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4913459147461955e-05,
+      "loss": 1.5999,
+      "step": 13320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.490608951780367e-05,
+      "loss": 1.8489,
+      "step": 13330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.489871515881759e-05,
+      "loss": 1.8404,
+      "step": 13340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.489133607225573e-05,
+      "loss": 1.5456,
+      "step": 13350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4883952259871234e-05,
+      "loss": 1.8751,
+      "step": 13360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.487656372341836e-05,
+      "loss": 1.9009,
+      "step": 13370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4869170464652485e-05,
+      "loss": 1.8085,
+      "step": 13380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4861772485330114e-05,
+      "loss": 1.919,
+      "step": 13390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.485436978720889e-05,
+      "loss": 1.6947,
+      "step": 13400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.484696237204754e-05,
+      "loss": 1.7379,
+      "step": 13410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.483955024160594e-05,
+      "loss": 1.8707,
+      "step": 13420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.483213339764508e-05,
+      "loss": 1.9474,
+      "step": 13430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4824711841927075e-05,
+      "loss": 1.8756,
+      "step": 13440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.481728557621514e-05,
+      "loss": 1.8177,
+      "step": 13450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.480985460227362e-05,
+      "loss": 1.8495,
+      "step": 13460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4802418921867996e-05,
+      "loss": 1.8704,
+      "step": 13470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.479497853676484e-05,
+      "loss": 1.8,
+      "step": 13480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.478753344873185e-05,
+      "loss": 1.7696,
+      "step": 13490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.478008365953786e-05,
+      "loss": 1.9174,
+      "step": 13500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4772629170952795e-05,
+      "loss": 1.9657,
+      "step": 13510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.476516998474771e-05,
+      "loss": 1.7958,
+      "step": 13520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4757706102694766e-05,
+      "loss": 2.0145,
+      "step": 13530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.475023752656725e-05,
+      "loss": 1.7189,
+      "step": 13540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.474276425813957e-05,
+      "loss": 1.6287,
+      "step": 13550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.473528629918723e-05,
+      "loss": 1.8224,
+      "step": 13560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.472780365148686e-05,
+      "loss": 1.6843,
+      "step": 13570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4720316316816204e-05,
+      "loss": 1.782,
+      "step": 13580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4712824296954114e-05,
+      "loss": 1.801,
+      "step": 13590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.470532759368056e-05,
+      "loss": 1.8339,
+      "step": 13600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.469782620877663e-05,
+      "loss": 1.7849,
+      "step": 13610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.469032014402451e-05,
+      "loss": 1.9943,
+      "step": 13620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.468280940120751e-05,
+      "loss": 2.0836,
+      "step": 13630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.467529398211005e-05,
+      "loss": 2.1812,
+      "step": 13640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4667773888517645e-05,
+      "loss": 1.9783,
+      "step": 13650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4660249122216944e-05,
+      "loss": 1.9577,
+      "step": 13660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.46527196849957e-05,
+      "loss": 1.7278,
+      "step": 13670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4645185578642755e-05,
+      "loss": 1.6278,
+      "step": 13680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4637646804948095e-05,
+      "loss": 1.866,
+      "step": 13690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.463010336570278e-05,
+      "loss": 1.8716,
+      "step": 13700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4622555262699014e-05,
+      "loss": 1.8191,
+      "step": 13710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4615002497730076e-05,
+      "loss": 1.9734,
+      "step": 13720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.460744507259037e-05,
+      "loss": 1.7782,
+      "step": 13730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.45998829890754e-05,
+      "loss": 1.8092,
+      "step": 13740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.459231624898179e-05,
+      "loss": 1.6374,
+      "step": 13750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.458474485410725e-05,
+      "loss": 1.4119,
+      "step": 13760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4577168806250625e-05,
+      "loss": 1.8767,
+      "step": 13770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.456958810721182e-05,
+      "loss": 1.8548,
+      "step": 13780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.45620027587919e-05,
+      "loss": 1.7237,
+      "step": 13790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.455441276279299e-05,
+      "loss": 1.7079,
+      "step": 13800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.454681812101834e-05,
+      "loss": 1.7156,
+      "step": 13810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.45392188352723e-05,
+      "loss": 1.6823,
+      "step": 13820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4531614907360325e-05,
+      "loss": 1.9576,
+      "step": 13830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.452400633908897e-05,
+      "loss": 2.1093,
+      "step": 13840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.451639313226589e-05,
+      "loss": 2.0721,
+      "step": 13850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.450877528869985e-05,
+      "loss": 2.033,
+      "step": 13860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.450115281020071e-05,
+      "loss": 1.8167,
+      "step": 13870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4493525698579444e-05,
+      "loss": 2.1634,
+      "step": 13880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.448589395564809e-05,
+      "loss": 2.4067,
+      "step": 13890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4478257583219846e-05,
+      "loss": 2.0891,
+      "step": 13900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.447061658310895e-05,
+      "loss": 1.6136,
+      "step": 13910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.446297095713078e-05,
+      "loss": 1.5512,
+      "step": 13920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4455320707101785e-05,
+      "loss": 2.0977,
+      "step": 13930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4447665834839544e-05,
+      "loss": 1.8516,
+      "step": 13940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.444000634216271e-05,
+      "loss": 1.776,
+      "step": 13950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.443234223089103e-05,
+      "loss": 1.6728,
+      "step": 13960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.442467350284537e-05,
+      "loss": 1.68,
+      "step": 13970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.441700015984768e-05,
+      "loss": 1.9841,
+      "step": 13980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.440932220372101e-05,
+      "loss": 2.0927,
+      "step": 13990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4401639636289495e-05,
+      "loss": 1.8751,
+      "step": 14000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.439395245937838e-05,
+      "loss": 1.8214,
+      "step": 14010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4386260674813994e-05,
+      "loss": 1.7763,
+      "step": 14020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.437856428442378e-05,
+      "loss": 1.792,
+      "step": 14030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.437086329003625e-05,
+      "loss": 1.7321,
+      "step": 14040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.436315769348103e-05,
+      "loss": 1.8559,
+      "step": 14050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4355447496588814e-05,
+      "loss": 2.0059,
+      "step": 14060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.434773270119143e-05,
+      "loss": 2.0193,
+      "step": 14070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.434001330912174e-05,
+      "loss": 2.0029,
+      "step": 14080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.433228932221377e-05,
+      "loss": 1.9328,
+      "step": 14090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.432456074230258e-05,
+      "loss": 1.6896,
+      "step": 14100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.431682757122434e-05,
+      "loss": 1.7283,
+      "step": 14110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.430908981081633e-05,
+      "loss": 1.9206,
+      "step": 14120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.430134746291687e-05,
+      "loss": 1.8314,
+      "step": 14130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.429360052936543e-05,
+      "loss": 1.8634,
+      "step": 14140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4285849012002534e-05,
+      "loss": 1.794,
+      "step": 14150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.42780929126698e-05,
+      "loss": 1.8407,
+      "step": 14160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.427033223320994e-05,
+      "loss": 1.8993,
+      "step": 14170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.426256697546674e-05,
+      "loss": 1.8699,
+      "step": 14180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.425479714128511e-05,
+      "loss": 1.792,
+      "step": 14190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4247022732511e-05,
+      "loss": 1.747,
+      "step": 14200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.423924375099148e-05,
+      "loss": 1.7641,
+      "step": 14210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.423146019857469e-05,
+      "loss": 1.535,
+      "step": 14220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4223672077109866e-05,
+      "loss": 1.6136,
+      "step": 14230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4215879388447325e-05,
+      "loss": 1.5902,
+      "step": 14240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.420808213443845e-05,
+      "loss": 1.8859,
+      "step": 14250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4200280316935764e-05,
+      "loss": 1.7579,
+      "step": 14260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.419247393779282e-05,
+      "loss": 1.704,
+      "step": 14270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.418466299886427e-05,
+      "loss": 1.766,
+      "step": 14280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.417684750200585e-05,
+      "loss": 1.8919,
+      "step": 14290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.416902744907438e-05,
+      "loss": 1.8188,
+      "step": 14300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4161202841927784e-05,
+      "loss": 1.7794,
+      "step": 14310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4153373682425026e-05,
+      "loss": 1.7279,
+      "step": 14320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4145539972426186e-05,
+      "loss": 1.6114,
+      "step": 14330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.413770171379239e-05,
+      "loss": 1.8225,
+      "step": 14340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.41298589083859e-05,
+      "loss": 1.6789,
+      "step": 14350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.412201155807e-05,
+      "loss": 1.5272,
+      "step": 14360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4114159664709085e-05,
+      "loss": 1.8824,
+      "step": 14370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.410630323016863e-05,
+      "loss": 1.9019,
+      "step": 14380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.409844225631518e-05,
+      "loss": 2.1993,
+      "step": 14390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4090576745016355e-05,
+      "loss": 1.8192,
+      "step": 14400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.408270669814086e-05,
+      "loss": 1.6379,
+      "step": 14410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.407483211755847e-05,
+      "loss": 1.6886,
+      "step": 14420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.406695300514007e-05,
+      "loss": 1.7652,
+      "step": 14430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4059069362757555e-05,
+      "loss": 1.8505,
+      "step": 14440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.405118119228397e-05,
+      "loss": 1.7787,
+      "step": 14450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.404328849559338e-05,
+      "loss": 1.8214,
+      "step": 14460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.403539127456096e-05,
+      "loss": 1.9068,
+      "step": 14470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4027489531062954e-05,
+      "loss": 1.6333,
+      "step": 14480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4019583266976655e-05,
+      "loss": 1.6862,
+      "step": 14490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.401167248418045e-05,
+      "loss": 1.8079,
+      "step": 14500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.400375718455382e-05,
+      "loss": 1.9234,
+      "step": 14510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.399583736997728e-05,
+      "loss": 1.972,
+      "step": 14520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.398791304233243e-05,
+      "loss": 1.9224,
+      "step": 14530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3979984203501955e-05,
+      "loss": 1.7961,
+      "step": 14540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.397205085536961e-05,
+      "loss": 1.7981,
+      "step": 14550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.396411299982021e-05,
+      "loss": 1.728,
+      "step": 14560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.395617063873965e-05,
+      "loss": 2.0786,
+      "step": 14570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3948223774014875e-05,
+      "loss": 2.108,
+      "step": 14580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.394027240753393e-05,
+      "loss": 2.1114,
+      "step": 14590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.393231654118593e-05,
+      "loss": 1.9395,
+      "step": 14600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3924356176861024e-05,
+      "loss": 1.758,
+      "step": 14610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.391639131645045e-05,
+      "loss": 1.8279,
+      "step": 14620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3908421961846534e-05,
+      "loss": 1.8606,
+      "step": 14630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.390044811494264e-05,
+      "loss": 1.8434,
+      "step": 14640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.38924697776332e-05,
+      "loss": 1.7888,
+      "step": 14650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3884486951813747e-05,
+      "loss": 1.7293,
+      "step": 14660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.387649963938084e-05,
+      "loss": 1.7429,
+      "step": 14670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3868507842232115e-05,
+      "loss": 1.8577,
+      "step": 14680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.386051156226629e-05,
+      "loss": 1.9004,
+      "step": 14690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.385251080138314e-05,
+      "loss": 1.9588,
+      "step": 14700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.384450556148349e-05,
+      "loss": 1.8877,
+      "step": 14710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.383649584446924e-05,
+      "loss": 2.0125,
+      "step": 14720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.382848165224337e-05,
+      "loss": 1.7729,
+      "step": 14730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3820462986709894e-05,
+      "loss": 1.7276,
+      "step": 14740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.38124398497739e-05,
+      "loss": 1.7206,
+      "step": 14750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.380441224334156e-05,
+      "loss": 1.7593,
+      "step": 14760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.379638016932006e-05,
+      "loss": 1.8081,
+      "step": 14770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.37883436296177e-05,
+      "loss": 1.7556,
+      "step": 14780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.378030262614381e-05,
+      "loss": 1.9479,
+      "step": 14790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.377225716080878e-05,
+      "loss": 1.7268,
+      "step": 14800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.376420723552407e-05,
+      "loss": 1.792,
+      "step": 14810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.37561528522022e-05,
+      "loss": 1.777,
+      "step": 14820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3748094012756756e-05,
+      "loss": 1.9572,
+      "step": 14830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3740030719102356e-05,
+      "loss": 1.9225,
+      "step": 14840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3731962973154706e-05,
+      "loss": 1.8801,
+      "step": 14850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.372389077683055e-05,
+      "loss": 1.8461,
+      "step": 14860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3715814132047684e-05,
+      "loss": 1.7351,
+      "step": 14870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3707733040725e-05,
+      "loss": 1.6492,
+      "step": 14880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.369964750478239e-05,
+      "loss": 1.6887,
+      "step": 14890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.369155752614086e-05,
+      "loss": 1.8136,
+      "step": 14900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.368346310672242e-05,
+      "loss": 1.7833,
+      "step": 14910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.367536424845018e-05,
+      "loss": 1.7457,
+      "step": 14920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.366726095324827e-05,
+      "loss": 1.8059,
+      "step": 14930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.365915322304188e-05,
+      "loss": 1.7849,
+      "step": 14940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.365104105975727e-05,
+      "loss": 1.827,
+      "step": 14950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3642924465321745e-05,
+      "loss": 1.8097,
+      "step": 14960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.363480344166365e-05,
+      "loss": 1.7577,
+      "step": 14970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.362667799071241e-05,
+      "loss": 1.7518,
+      "step": 14980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3618548114398475e-05,
+      "loss": 1.8154,
+      "step": 14990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.361041381465336e-05,
+      "loss": 1.7627,
+      "step": 15000
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 1.7180181741714478,
+      "eval_runtime": 82.2878,
+      "eval_samples_per_second": 36.457,
+      "eval_steps_per_second": 4.557,
+      "step": 15000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.360227509340963e-05,
+      "loss": 1.7868,
+      "step": 15010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3594131952600894e-05,
+      "loss": 1.7538,
+      "step": 15020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.358598439416182e-05,
+      "loss": 1.8729,
+      "step": 15030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.357783242002811e-05,
+      "loss": 1.7501,
+      "step": 15040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.356967603213654e-05,
+      "loss": 1.6776,
+      "step": 15050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.356151523242491e-05,
+      "loss": 2.0497,
+      "step": 15060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3553350022832083e-05,
+      "loss": 1.9301,
+      "step": 15070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.354518040529797e-05,
+      "loss": 1.8151,
+      "step": 15080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.353700638176351e-05,
+      "loss": 2.0247,
+      "step": 15090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3528827954170726e-05,
+      "loss": 1.8791,
+      "step": 15100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3520645124462656e-05,
+      "loss": 1.7957,
+      "step": 15110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.351245789458338e-05,
+      "loss": 1.812,
+      "step": 15120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.350426626647805e-05,
+      "loss": 1.9326,
+      "step": 15130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.349607024209285e-05,
+      "loss": 2.0055,
+      "step": 15140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3487869823374995e-05,
+      "loss": 2.093,
+      "step": 15150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.347966501227276e-05,
+      "loss": 2.1203,
+      "step": 15160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3471455810735474e-05,
+      "loss": 1.9363,
+      "step": 15170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.346324222071348e-05,
+      "loss": 1.9597,
+      "step": 15180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3455024244158184e-05,
+      "loss": 2.0135,
+      "step": 15190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.344680188302203e-05,
+      "loss": 2.1012,
+      "step": 15200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.343857513925851e-05,
+      "loss": 1.9789,
+      "step": 15210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3430344014822125e-05,
+      "loss": 1.7305,
+      "step": 15220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.342210851166847e-05,
+      "loss": 1.7928,
+      "step": 15230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.341386863175414e-05,
+      "loss": 1.7754,
+      "step": 15240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.340562437703678e-05,
+      "loss": 1.9918,
+      "step": 15250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.339737574947508e-05,
+      "loss": 1.8024,
+      "step": 15260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3389122751028764e-05,
+      "loss": 1.9443,
+      "step": 15270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3380865383658594e-05,
+      "loss": 1.8665,
+      "step": 15280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.337260364932637e-05,
+      "loss": 1.9303,
+      "step": 15290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.336433754999494e-05,
+      "loss": 1.8062,
+      "step": 15300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3356067087628174e-05,
+      "loss": 1.7333,
+      "step": 15310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3347792264190976e-05,
+      "loss": 1.5874,
+      "step": 15320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.333951308164931e-05,
+      "loss": 1.5032,
+      "step": 15330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.333122954197015e-05,
+      "loss": 1.7122,
+      "step": 15340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.332294164712152e-05,
+      "loss": 1.6085,
+      "step": 15350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.331464939907248e-05,
+      "loss": 1.7119,
+      "step": 15360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3306352799793105e-05,
+      "loss": 1.7717,
+      "step": 15370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.329805185125453e-05,
+      "loss": 1.8143,
+      "step": 15380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.32897465554289e-05,
+      "loss": 1.7925,
+      "step": 15390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.328143691428941e-05,
+      "loss": 1.7802,
+      "step": 15400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.327312292981028e-05,
+      "loss": 1.6896,
+      "step": 15410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.326480460396677e-05,
+      "loss": 1.6639,
+      "step": 15420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.325648193873514e-05,
+      "loss": 1.6593,
+      "step": 15430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3248154936092734e-05,
+      "loss": 1.7572,
+      "step": 15440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.323982359801788e-05,
+      "loss": 1.911,
+      "step": 15450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.323148792648996e-05,
+      "loss": 2.201,
+      "step": 15460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3223147923489385e-05,
+      "loss": 1.9133,
+      "step": 15470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3214803590997574e-05,
+      "loss": 1.4673,
+      "step": 15480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3206454930997e-05,
+      "loss": 1.5952,
+      "step": 15490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.319810194547115e-05,
+      "loss": 1.8063,
+      "step": 15500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.318974463640455e-05,
+      "loss": 1.7932,
+      "step": 15510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.318138300578274e-05,
+      "loss": 1.7659,
+      "step": 15520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.317301705559229e-05,
+      "loss": 1.6929,
+      "step": 15530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.31646467878208e-05,
+      "loss": 1.6017,
+      "step": 15540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.31562722044569e-05,
+      "loss": 1.7831,
+      "step": 15550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.314789330749024e-05,
+      "loss": 1.6174,
+      "step": 15560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.313951009891148e-05,
+      "loss": 1.4159,
+      "step": 15570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.313112258071235e-05,
+      "loss": 1.4068,
+      "step": 15580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.312273075488554e-05,
+      "loss": 1.6638,
+      "step": 15590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3114334623424816e-05,
+      "loss": 1.8426,
+      "step": 15600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.310593418832494e-05,
+      "loss": 1.7097,
+      "step": 15610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3097529451581695e-05,
+      "loss": 1.8638,
+      "step": 15620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.308912041519192e-05,
+      "loss": 2.0008,
+      "step": 15630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.308070708115343e-05,
+      "loss": 1.9124,
+      "step": 15640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3072289451465086e-05,
+      "loss": 1.8525,
+      "step": 15650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3063867528126766e-05,
+      "loss": 1.9728,
+      "step": 15660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3055441313139366e-05,
+      "loss": 1.8224,
+      "step": 15670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.304701080850481e-05,
+      "loss": 1.8074,
+      "step": 15680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.303857601622602e-05,
+      "loss": 1.8022,
+      "step": 15690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3030136938306966e-05,
+      "loss": 1.7989,
+      "step": 15700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.302169357675261e-05,
+      "loss": 1.7999,
+      "step": 15710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3013245933568937e-05,
+      "loss": 1.8495,
+      "step": 15720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.300479401076297e-05,
+      "loss": 1.676,
+      "step": 15730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.299633781034272e-05,
+      "loss": 1.8275,
+      "step": 15740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.298787733431724e-05,
+      "loss": 1.852,
+      "step": 15750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.297941258469659e-05,
+      "loss": 1.7432,
+      "step": 15760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.2970943563491816e-05,
+      "loss": 1.6872,
+      "step": 15770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.296247027271503e-05,
+      "loss": 1.6908,
+      "step": 15780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.2953992714379314e-05,
+      "loss": 1.9134,
+      "step": 15790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.29455108904988e-05,
+      "loss": 1.784,
+      "step": 15800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.293702480308861e-05,
+      "loss": 1.8022,
+      "step": 15810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2928534454164884e-05,
+      "loss": 1.8676,
+      "step": 15820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2920039845744775e-05,
+      "loss": 1.7982,
+      "step": 15830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.291154097984644e-05,
+      "loss": 1.7909,
+      "step": 15840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.290303785848908e-05,
+      "loss": 1.6917,
+      "step": 15850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.289453048369287e-05,
+      "loss": 1.817,
+      "step": 15860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2886018857479e-05,
+      "loss": 1.845,
+      "step": 15870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2877502981869686e-05,
+      "loss": 1.7028,
+      "step": 15880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2868982858888166e-05,
+      "loss": 1.8574,
+      "step": 15890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.286045849055863e-05,
+      "loss": 1.605,
+      "step": 15900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.285192987890634e-05,
+      "loss": 1.6146,
+      "step": 15910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2843397025957523e-05,
+      "loss": 1.5675,
+      "step": 15920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2834859933739455e-05,
+      "loss": 1.5633,
+      "step": 15930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2826318604280375e-05,
+      "loss": 1.4985,
+      "step": 15940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.281777303960955e-05,
+      "loss": 1.4398,
+      "step": 15950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.280922324175727e-05,
+      "loss": 1.453,
+      "step": 15960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2800669212754795e-05,
+      "loss": 1.6388,
+      "step": 15970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.279211095463441e-05,
+      "loss": 1.8499,
+      "step": 15980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.278354846942941e-05,
+      "loss": 1.8666,
+      "step": 15990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.277498175917408e-05,
+      "loss": 1.7608,
+      "step": 16000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2766410825903716e-05,
+      "loss": 1.6489,
+      "step": 16010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.275783567165462e-05,
+      "loss": 1.9551,
+      "step": 16020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.27492562984641e-05,
+      "loss": 1.8426,
+      "step": 16030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2740672708370454e-05,
+      "loss": 2.0718,
+      "step": 16040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2732084903412985e-05,
+      "loss": 1.847,
+      "step": 16050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.272349288563201e-05,
+      "loss": 1.8333,
+      "step": 16060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.271489665706882e-05,
+      "loss": 1.7727,
+      "step": 16070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.270629621976574e-05,
+      "loss": 1.7242,
+      "step": 16080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.269769157576608e-05,
+      "loss": 1.5498,
+      "step": 16090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.268908272711413e-05,
+      "loss": 2.0275,
+      "step": 16100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.268046967585522e-05,
+      "loss": 1.8603,
+      "step": 16110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.267185242403564e-05,
+      "loss": 1.721,
+      "step": 16120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.26632309737027e-05,
+      "loss": 1.7883,
+      "step": 16130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.265460532690469e-05,
+      "loss": 1.7635,
+      "step": 16140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.264597548569093e-05,
+      "loss": 1.8881,
+      "step": 16150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.26373414521117e-05,
+      "loss": 1.9935,
+      "step": 16160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.262870322821829e-05,
+      "loss": 1.7706,
+      "step": 16170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.262006081606299e-05,
+      "loss": 1.7359,
+      "step": 16180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.261141421769908e-05,
+      "loss": 1.7273,
+      "step": 16190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.260276343518084e-05,
+      "loss": 1.5001,
+      "step": 16200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2594108470563524e-05,
+      "loss": 1.8077,
+      "step": 16210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2585449325903416e-05,
+      "loss": 1.7671,
+      "step": 16220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.257678600325777e-05,
+      "loss": 1.566,
+      "step": 16230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2568118504684814e-05,
+      "loss": 1.9281,
+      "step": 16240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.255944683224382e-05,
+      "loss": 1.9293,
+      "step": 16250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2550770987994995e-05,
+      "loss": 1.6836,
+      "step": 16260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2542090973999574e-05,
+      "loss": 1.6259,
+      "step": 16270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2533406792319774e-05,
+      "loss": 1.5957,
+      "step": 16280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.25247184450188e-05,
+      "loss": 1.5949,
+      "step": 16290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.251602593416084e-05,
+      "loss": 1.683,
+      "step": 16300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.250732926181108e-05,
+      "loss": 1.9353,
+      "step": 16310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.249862843003569e-05,
+      "loss": 1.9002,
+      "step": 16320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.248992344090185e-05,
+      "loss": 1.8219,
+      "step": 16330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.248121429647768e-05,
+      "loss": 1.7036,
+      "step": 16340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2472500998832334e-05,
+      "loss": 1.6086,
+      "step": 16350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.246378355003593e-05,
+      "loss": 1.693,
+      "step": 16360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.245506195215957e-05,
+      "loss": 1.7098,
+      "step": 16370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.244633620727535e-05,
+      "loss": 1.8342,
+      "step": 16380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2437606317456366e-05,
+      "loss": 1.8023,
+      "step": 16390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.242887228477666e-05,
+      "loss": 1.8904,
+      "step": 16400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2420134111311293e-05,
+      "loss": 1.7619,
+      "step": 16410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2411391799136304e-05,
+      "loss": 2.0016,
+      "step": 16420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.240264535032868e-05,
+      "loss": 1.5288,
+      "step": 16430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.239389476696646e-05,
+      "loss": 1.4601,
+      "step": 16440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2385140051128594e-05,
+      "loss": 1.5915,
+      "step": 16450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.237638120489505e-05,
+      "loss": 1.675,
+      "step": 16460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.236761823034678e-05,
+      "loss": 1.7,
+      "step": 16470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2358851129565715e-05,
+      "loss": 1.8357,
+      "step": 16480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.235007990463474e-05,
+      "loss": 1.8126,
+      "step": 16490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.234130455763775e-05,
+      "loss": 1.8795,
+      "step": 16500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.233252509065961e-05,
+      "loss": 1.7031,
+      "step": 16510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2323741505786156e-05,
+      "loss": 1.8439,
+      "step": 16520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.231495380510422e-05,
+      "loss": 1.7563,
+      "step": 16530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.23061619907016e-05,
+      "loss": 1.7746,
+      "step": 16540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.229736606466707e-05,
+      "loss": 1.7549,
+      "step": 16550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.228856602909037e-05,
+      "loss": 1.793,
+      "step": 16560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2279761886062255e-05,
+      "loss": 1.7028,
+      "step": 16570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.227095363767441e-05,
+      "loss": 2.0031,
+      "step": 16580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.226214128601953e-05,
+      "loss": 1.8168,
+      "step": 16590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.225332483319125e-05,
+      "loss": 1.7478,
+      "step": 16600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.224450428128423e-05,
+      "loss": 1.8152,
+      "step": 16610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2235679632394055e-05,
+      "loss": 1.7568,
+      "step": 16620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.22268508886173e-05,
+      "loss": 1.7691,
+      "step": 16630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.221801805205152e-05,
+      "loss": 1.7557,
+      "step": 16640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.220918112479525e-05,
+      "loss": 1.786,
+      "step": 16650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2200340108947965e-05,
+      "loss": 1.7086,
+      "step": 16660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.219149500661014e-05,
+      "loss": 1.7002,
+      "step": 16670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.218264581988321e-05,
+      "loss": 1.6567,
+      "step": 16680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.217379255086958e-05,
+      "loss": 1.6914,
+      "step": 16690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.216493520167264e-05,
+      "loss": 1.6897,
+      "step": 16700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.215607377439672e-05,
+      "loss": 2.014,
+      "step": 16710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.214720827114714e-05,
+      "loss": 1.9544,
+      "step": 16720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.213833869403019e-05,
+      "loss": 1.928,
+      "step": 16730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.212946504515311e-05,
+      "loss": 1.938,
+      "step": 16740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2120587326624126e-05,
+      "loss": 2.0092,
+      "step": 16750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.211170554055244e-05,
+      "loss": 1.9373,
+      "step": 16760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.210281968904817e-05,
+      "loss": 1.7966,
+      "step": 16770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.209392977422245e-05,
+      "loss": 2.1905,
+      "step": 16780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.208503579818738e-05,
+      "loss": 1.8587,
+      "step": 16790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.207613776305598e-05,
+      "loss": 2.0832,
+      "step": 16800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2067235670942284e-05,
+      "loss": 1.9328,
+      "step": 16810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.205832952396126e-05,
+      "loss": 1.8872,
+      "step": 16820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.204941932422885e-05,
+      "loss": 1.7503,
+      "step": 16830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.204050507386195e-05,
+      "loss": 1.7135,
+      "step": 16840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2031586774978424e-05,
+      "loss": 1.6992,
+      "step": 16850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.202266442969712e-05,
+      "loss": 1.7776,
+      "step": 16860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.20137380401378e-05,
+      "loss": 1.6132,
+      "step": 16870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2004807608421225e-05,
+      "loss": 1.8119,
+      "step": 16880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.199587313666911e-05,
+      "loss": 1.7936,
+      "step": 16890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1986934627004114e-05,
+      "loss": 1.8504,
+      "step": 16900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1977992081549874e-05,
+      "loss": 1.7269,
+      "step": 16910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.196904550243097e-05,
+      "loss": 1.7408,
+      "step": 16920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.196009489177296e-05,
+      "loss": 1.9128,
+      "step": 16930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.195114025170233e-05,
+      "loss": 1.9862,
+      "step": 16940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.194218158434655e-05,
+      "loss": 1.9998,
+      "step": 16950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1933218891834034e-05,
+      "loss": 1.9382,
+      "step": 16960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.192425217629417e-05,
+      "loss": 1.7846,
+      "step": 16970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.191528143985727e-05,
+      "loss": 1.5135,
+      "step": 16980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1906306684654636e-05,
+      "loss": 1.5438,
+      "step": 16990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.189732791281849e-05,
+      "loss": 2.0039,
+      "step": 17000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1888345126482036e-05,
+      "loss": 1.9236,
+      "step": 17010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1879358327779426e-05,
+      "loss": 1.8032,
+      "step": 17020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.187036751884576e-05,
+      "loss": 1.6551,
+      "step": 17030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.186137270181707e-05,
+      "loss": 1.5531,
+      "step": 17040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.18523738788304e-05,
+      "loss": 1.4822,
+      "step": 17050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.184337105202367e-05,
+      "loss": 1.421,
+      "step": 17060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.183436422353582e-05,
+      "loss": 1.4202,
+      "step": 17070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.182535339550669e-05,
+      "loss": 1.4155,
+      "step": 17080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.181633857007711e-05,
+      "loss": 1.5707,
+      "step": 17090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1807319749388826e-05,
+      "loss": 1.8303,
+      "step": 17100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.179829693558454e-05,
+      "loss": 1.7824,
+      "step": 17110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.178927013080792e-05,
+      "loss": 1.7659,
+      "step": 17120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1780239337203576e-05,
+      "loss": 1.7013,
+      "step": 17130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1771204556917056e-05,
+      "loss": 1.7361,
+      "step": 17140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.176216579209487e-05,
+      "loss": 1.6975,
+      "step": 17150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.175312304488443e-05,
+      "loss": 1.7941,
+      "step": 17160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1744076317434176e-05,
+      "loss": 1.7507,
+      "step": 17170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.173502561189343e-05,
+      "loss": 1.6078,
+      "step": 17180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.172597093041247e-05,
+      "loss": 1.5817,
+      "step": 17190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1716912275142525e-05,
+      "loss": 1.4377,
+      "step": 17200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.170784964823577e-05,
+      "loss": 1.9275,
+      "step": 17210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.169878305184533e-05,
+      "loss": 1.7549,
+      "step": 17220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1689712488125255e-05,
+      "loss": 1.6684,
+      "step": 17230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.168063795923055e-05,
+      "loss": 1.7631,
+      "step": 17240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.167155946731717e-05,
+      "loss": 1.7593,
+      "step": 17250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.166247701454198e-05,
+      "loss": 1.875,
+      "step": 17260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.165339060306282e-05,
+      "loss": 1.8556,
+      "step": 17270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.164430023503846e-05,
+      "loss": 1.8407,
+      "step": 17280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.16352059126286e-05,
+      "loss": 1.8785,
+      "step": 17290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1626107637993886e-05,
+      "loss": 1.8172,
+      "step": 17300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1617005413295915e-05,
+      "loss": 1.7869,
+      "step": 17310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.16078992406972e-05,
+      "loss": 1.7307,
+      "step": 17320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.159878912236121e-05,
+      "loss": 1.8682,
+      "step": 17330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.158967506045234e-05,
+      "loss": 1.9418,
+      "step": 17340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.158055705713593e-05,
+      "loss": 1.9209,
+      "step": 17350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.157143511457825e-05,
+      "loss": 1.8189,
+      "step": 17360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1562309234946514e-05,
+      "loss": 1.8056,
+      "step": 17370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.155317942040887e-05,
+      "loss": 1.7732,
+      "step": 17380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1544045673134377e-05,
+      "loss": 1.7047,
+      "step": 17390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.153490799529307e-05,
+      "loss": 1.956,
+      "step": 17400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1525766389055895e-05,
+      "loss": 1.5792,
+      "step": 17410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.151662085659471e-05,
+      "loss": 1.5705,
+      "step": 17420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.150747140008236e-05,
+      "loss": 1.8104,
+      "step": 17430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.149831802169257e-05,
+      "loss": 1.8144,
+      "step": 17440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1489160723600026e-05,
+      "loss": 1.8224,
+      "step": 17450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.147999950798033e-05,
+      "loss": 1.9291,
+      "step": 17460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1470834377010024e-05,
+      "loss": 1.9241,
+      "step": 17470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1461665332866595e-05,
+      "loss": 1.9035,
+      "step": 17480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1452492377728415e-05,
+      "loss": 1.7754,
+      "step": 17490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1443315513774825e-05,
+      "loss": 1.926,
+      "step": 17500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.143413474318608e-05,
+      "loss": 1.8511,
+      "step": 17510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1424950068143374e-05,
+      "loss": 1.9259,
+      "step": 17520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.141576149082881e-05,
+      "loss": 1.9261,
+      "step": 17530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.140656901342543e-05,
+      "loss": 1.7308,
+      "step": 17540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.139737263811722e-05,
+      "loss": 1.9385,
+      "step": 17550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.138817236708904e-05,
+      "loss": 1.8203,
+      "step": 17560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.137896820252672e-05,
+      "loss": 1.7017,
+      "step": 17570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1369760146617014e-05,
+      "loss": 1.7135,
+      "step": 17580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.136054820154759e-05,
+      "loss": 1.5351,
+      "step": 17590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.135133236950702e-05,
+      "loss": 1.9286,
+      "step": 17600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.134211265268484e-05,
+      "loss": 1.8042,
+      "step": 17610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.133288905327148e-05,
+      "loss": 1.5988,
+      "step": 17620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1323661573458306e-05,
+      "loss": 1.7819,
+      "step": 17630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.13144302154376e-05,
+      "loss": 1.968,
+      "step": 17640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.130519498140256e-05,
+      "loss": 1.6891,
+      "step": 17650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.129595587354731e-05,
+      "loss": 1.8239,
+      "step": 17660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.12867128940669e-05,
+      "loss": 1.7686,
+      "step": 17670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.12774660451573e-05,
+      "loss": 1.8351,
+      "step": 17680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.126821532901538e-05,
+      "loss": 1.8614,
+      "step": 17690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.125896074783896e-05,
+      "loss": 1.818,
+      "step": 17700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.124970230382676e-05,
+      "loss": 1.8144,
+      "step": 17710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.124043999917841e-05,
+      "loss": 1.7685,
+      "step": 17720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1231173836094464e-05,
+      "loss": 1.843,
+      "step": 17730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.122190381677641e-05,
+      "loss": 1.8512,
+      "step": 17740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1212629943426616e-05,
+      "loss": 1.7045,
+      "step": 17750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.12033522182484e-05,
+      "loss": 1.6487,
+      "step": 17760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1194070643445994e-05,
+      "loss": 1.7144,
+      "step": 17770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.118478522122451e-05,
+      "loss": 1.8606,
+      "step": 17780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.117549595379001e-05,
+      "loss": 1.9938,
+      "step": 17790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1166202843349456e-05,
+      "loss": 1.7362,
+      "step": 17800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1156905892110725e-05,
+      "loss": 1.799,
+      "step": 17810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1147605102282595e-05,
+      "loss": 1.8124,
+      "step": 17820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.113830047607478e-05,
+      "loss": 1.6871,
+      "step": 17830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.112899201569787e-05,
+      "loss": 1.7202,
+      "step": 17840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.11196797233634e-05,
+      "loss": 1.6538,
+      "step": 17850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.111036360128381e-05,
+      "loss": 1.6437,
+      "step": 17860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1101043651672444e-05,
+      "loss": 1.6758,
+      "step": 17870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.109171987674353e-05,
+      "loss": 1.6477,
+      "step": 17880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.108239227871226e-05,
+      "loss": 1.7014,
+      "step": 17890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.107306085979467e-05,
+      "loss": 1.8359,
+      "step": 17900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.106372562220777e-05,
+      "loss": 1.7704,
+      "step": 17910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.105438656816942e-05,
+      "loss": 1.7824,
+      "step": 17920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1045043699898413e-05,
+      "loss": 1.8951,
+      "step": 17930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.103569701961447e-05,
+      "loss": 1.7652,
+      "step": 17940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.102634652953816e-05,
+      "loss": 1.7841,
+      "step": 17950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.101699223189101e-05,
+      "loss": 1.7326,
+      "step": 17960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.100763412889543e-05,
+      "loss": 1.8165,
+      "step": 17970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0998272222774734e-05,
+      "loss": 1.8275,
+      "step": 17980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0988906515753143e-05,
+      "loss": 1.9799,
+      "step": 17990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.097953701005578e-05,
+      "loss": 1.605,
+      "step": 18000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.097016370790868e-05,
+      "loss": 1.7013,
+      "step": 18010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.096078661153875e-05,
+      "loss": 1.6276,
+      "step": 18020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0951405723173836e-05,
+      "loss": 1.5157,
+      "step": 18030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0942021045042664e-05,
+      "loss": 1.7399,
+      "step": 18040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0932632579374855e-05,
+      "loss": 1.7422,
+      "step": 18050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0923240328400946e-05,
+      "loss": 1.7179,
+      "step": 18060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0913844294352374e-05,
+      "loss": 1.8205,
+      "step": 18070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0904444479461456e-05,
+      "loss": 1.6875,
+      "step": 18080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0895040885961426e-05,
+      "loss": 1.7905,
+      "step": 18090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.088563351608641e-05,
+      "loss": 1.6654,
+      "step": 18100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.087622237207141e-05,
+      "loss": 1.9952,
+      "step": 18110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.086680745615237e-05,
+      "loss": 1.8413,
+      "step": 18120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0857388770566086e-05,
+      "loss": 1.6973,
+      "step": 18130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.084796631755028e-05,
+      "loss": 1.8969,
+      "step": 18140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.083854009934354e-05,
+      "loss": 1.9469,
+      "step": 18150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.082911011818539e-05,
+      "loss": 1.7181,
+      "step": 18160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.08196763763162e-05,
+      "loss": 1.837,
+      "step": 18170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.081023887597727e-05,
+      "loss": 1.7704,
+      "step": 18180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0800797619410775e-05,
+      "loss": 1.8593,
+      "step": 18190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.079135260885979e-05,
+      "loss": 1.8582,
+      "step": 18200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0781903846568283e-05,
+      "loss": 1.694,
+      "step": 18210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0772451334781094e-05,
+      "loss": 1.8066,
+      "step": 18220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0762995075743986e-05,
+      "loss": 1.7904,
+      "step": 18230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.075353507170359e-05,
+      "loss": 1.5556,
+      "step": 18240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.074407132490743e-05,
+      "loss": 2.0854,
+      "step": 18250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.073460383760392e-05,
+      "loss": 2.2883,
+      "step": 18260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0725132612042374e-05,
+      "loss": 1.9893,
+      "step": 18270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.071565765047298e-05,
+      "loss": 1.6285,
+      "step": 18280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.070617895514681e-05,
+      "loss": 1.6268,
+      "step": 18290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.069669652831584e-05,
+      "loss": 1.7356,
+      "step": 18300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.068721037223292e-05,
+      "loss": 1.8117,
+      "step": 18310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.067772048915178e-05,
+      "loss": 1.9067,
+      "step": 18320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0668226881327064e-05,
+      "loss": 2.0273,
+      "step": 18330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0658729551014275e-05,
+      "loss": 1.9985,
+      "step": 18340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.06492285004698e-05,
+      "loss": 1.8458,
+      "step": 18350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.063972373195092e-05,
+      "loss": 2.0248,
+      "step": 18360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0630215247715806e-05,
+      "loss": 1.8124,
+      "step": 18370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.062070305002349e-05,
+      "loss": 1.6876,
+      "step": 18380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.061118714113391e-05,
+      "loss": 1.8265,
+      "step": 18390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.060166752330786e-05,
+      "loss": 1.7734,
+      "step": 18400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.059214419880705e-05,
+      "loss": 1.8269,
+      "step": 18410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.058261716989402e-05,
+      "loss": 1.8366,
+      "step": 18420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0573086438832254e-05,
+      "loss": 1.8681,
+      "step": 18430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.056355200788606e-05,
+      "loss": 1.7621,
+      "step": 18440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.055401387932065e-05,
+      "loss": 1.6917,
+      "step": 18450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0544472055402116e-05,
+      "loss": 1.7434,
+      "step": 18460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0534926538397426e-05,
+      "loss": 1.9047,
+      "step": 18470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.052537733057441e-05,
+      "loss": 1.7002,
+      "step": 18480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0515824434201796e-05,
+      "loss": 1.782,
+      "step": 18490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.050626785154918e-05,
+      "loss": 1.9203,
+      "step": 18500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.049670758488704e-05,
+      "loss": 1.7146,
+      "step": 18510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.048714363648671e-05,
+      "loss": 1.4322,
+      "step": 18520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.047757600862042e-05,
+      "loss": 1.7889,
+      "step": 18530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0468004703561264e-05,
+      "loss": 1.9318,
+      "step": 18540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.04584297235832e-05,
+      "loss": 1.7388,
+      "step": 18550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.044885107096109e-05,
+      "loss": 1.8296,
+      "step": 18560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0439268747970645e-05,
+      "loss": 2.0499,
+      "step": 18570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0429682756888436e-05,
+      "loss": 1.9024,
+      "step": 18580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.042009309999194e-05,
+      "loss": 1.8188,
+      "step": 18590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.041049977955948e-05,
+      "loss": 1.7791,
+      "step": 18600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.040090279787025e-05,
+      "loss": 1.7076,
+      "step": 18610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.039130215720433e-05,
+      "loss": 1.6423,
+      "step": 18620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.038169785984265e-05,
+      "loss": 1.6914,
+      "step": 18630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.037208990806702e-05,
+      "loss": 1.7897,
+      "step": 18640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0362478304160124e-05,
+      "loss": 1.7449,
+      "step": 18650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.03528630504055e-05,
+      "loss": 1.7032,
+      "step": 18660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0343244149087554e-05,
+      "loss": 1.6831,
+      "step": 18670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.033362160249157e-05,
+      "loss": 1.6692,
+      "step": 18680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.032399541290368e-05,
+      "loss": 1.7421,
+      "step": 18690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.031436558261091e-05,
+      "loss": 2.0957,
+      "step": 18700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.030473211390113e-05,
+      "loss": 1.8474,
+      "step": 18710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0295095009063076e-05,
+      "loss": 1.7627,
+      "step": 18720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.028545427038635e-05,
+      "loss": 1.6315,
+      "step": 18730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0275809900161404e-05,
+      "loss": 1.9766,
+      "step": 18740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.026616190067958e-05,
+      "loss": 2.0223,
+      "step": 18750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.025651027423308e-05,
+      "loss": 1.735,
+      "step": 18760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.024685502311494e-05,
+      "loss": 1.875,
+      "step": 18770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.023719614961907e-05,
+      "loss": 1.73,
+      "step": 18780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.022753365604026e-05,
+      "loss": 1.8447,
+      "step": 18790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.021786754467413e-05,
+      "loss": 1.7906,
+      "step": 18800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0208197817817186e-05,
+      "loss": 1.6478,
+      "step": 18810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.019852447776676e-05,
+      "loss": 1.7192,
+      "step": 18820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.018884752682109e-05,
+      "loss": 1.6251,
+      "step": 18830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.017916696727924e-05,
+      "loss": 1.7682,
+      "step": 18840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0169482801441114e-05,
+      "loss": 1.7846,
+      "step": 18850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0159795031607516e-05,
+      "loss": 1.7793,
+      "step": 18860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.015010366008007e-05,
+      "loss": 1.7393,
+      "step": 18870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.014040868916129e-05,
+      "loss": 1.8395,
+      "step": 18880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0130710121154517e-05,
+      "loss": 1.7264,
+      "step": 18890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0121007958363946e-05,
+      "loss": 1.8341,
+      "step": 18900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.011130220309465e-05,
+      "loss": 1.78,
+      "step": 18910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.010159285765253e-05,
+      "loss": 2.0015,
+      "step": 18920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0091879924344366e-05,
+      "loss": 1.785,
+      "step": 18930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.008216340547777e-05,
+      "loss": 1.7484,
+      "step": 18940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.00724433033612e-05,
+      "loss": 1.7765,
+      "step": 18950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.006271962030399e-05,
+      "loss": 1.7668,
+      "step": 18960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.005299235861631e-05,
+      "loss": 1.8168,
+      "step": 18970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0043261520609174e-05,
+      "loss": 1.7325,
+      "step": 18980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0033527108594465e-05,
+      "loss": 1.7765,
+      "step": 18990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.002378912488489e-05,
+      "loss": 1.8701,
+      "step": 19000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.001404757179403e-05,
+      "loss": 1.7496,
+      "step": 19010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.00043024516363e-05,
+      "loss": 1.793,
+      "step": 19020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999455376672697e-05,
+      "loss": 1.7605,
+      "step": 19030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.998480151938214e-05,
+      "loss": 1.8325,
+      "step": 19040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.997504571191877e-05,
+      "loss": 1.8125,
+      "step": 19050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.996528634665466e-05,
+      "loss": 1.7646,
+      "step": 19060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.995552342590848e-05,
+      "loss": 1.6872,
+      "step": 19070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9945756951999695e-05,
+      "loss": 1.8314,
+      "step": 19080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.993598692724867e-05,
+      "loss": 1.9722,
+      "step": 19090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.992621335397657e-05,
+      "loss": 1.7721,
+      "step": 19100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.991643623450542e-05,
+      "loss": 1.7816,
+      "step": 19110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.99066555711581e-05,
+      "loss": 1.8161,
+      "step": 19120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.98968713662583e-05,
+      "loss": 1.7745,
+      "step": 19130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.988708362213059e-05,
+      "loss": 1.7645,
+      "step": 19140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.987729234110035e-05,
+      "loss": 1.731,
+      "step": 19150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9867497525493815e-05,
+      "loss": 1.7704,
+      "step": 19160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9857699177638056e-05,
+      "loss": 1.7141,
+      "step": 19170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.984789729986098e-05,
+      "loss": 1.7778,
+      "step": 19180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9838091894491345e-05,
+      "loss": 1.7421,
+      "step": 19190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.982828296385872e-05,
+      "loss": 1.7052,
+      "step": 19200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.981847051029356e-05,
+      "loss": 1.7307,
+      "step": 19210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.980865453612711e-05,
+      "loss": 1.8335,
+      "step": 19220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.979883504369146e-05,
+      "loss": 1.8264,
+      "step": 19230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9789012035319554e-05,
+      "loss": 1.7544,
+      "step": 19240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9779185513345164e-05,
+      "loss": 1.7298,
+      "step": 19250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9769355480102885e-05,
+      "loss": 1.7222,
+      "step": 19260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.975952193792817e-05,
+      "loss": 1.6603,
+      "step": 19270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.974968488915728e-05,
+      "loss": 1.7057,
+      "step": 19280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.973984433612732e-05,
+      "loss": 1.7883,
+      "step": 19290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9730000281176226e-05,
+      "loss": 1.7152,
+      "step": 19300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.972015272664278e-05,
+      "loss": 1.7797,
+      "step": 19310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9710301674866576e-05,
+      "loss": 1.8524,
+      "step": 19320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.970044712818805e-05,
+      "loss": 1.7798,
+      "step": 19330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9690589088948453e-05,
+      "loss": 1.8383,
+      "step": 19340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9680727559489895e-05,
+      "loss": 1.735,
+      "step": 19350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.967086254215527e-05,
+      "loss": 1.7415,
+      "step": 19360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9660994039288365e-05,
+      "loss": 1.8081,
+      "step": 19370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.965112205323373e-05,
+      "loss": 1.8115,
+      "step": 19380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.964124658633678e-05,
+      "loss": 1.7512,
+      "step": 19390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.963136764094375e-05,
+      "loss": 1.775,
+      "step": 19400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.962148521940169e-05,
+      "loss": 1.9015,
+      "step": 19410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9611599324058505e-05,
+      "loss": 1.8619,
+      "step": 19420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.960170995726289e-05,
+      "loss": 1.8775,
+      "step": 19430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9591817121364374e-05,
+      "loss": 1.9008,
+      "step": 19440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9581920818713325e-05,
+      "loss": 1.9972,
+      "step": 19450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9572021051660937e-05,
+      "loss": 1.8781,
+      "step": 19460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.956211782255919e-05,
+      "loss": 1.9165,
+      "step": 19470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.955221113376094e-05,
+      "loss": 1.6503,
+      "step": 19480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.954230098761982e-05,
+      "loss": 1.7148,
+      "step": 19490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.953238738649031e-05,
+      "loss": 1.9949,
+      "step": 19500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.95224703327277e-05,
+      "loss": 1.7523,
+      "step": 19510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9512549828688096e-05,
+      "loss": 1.4507,
+      "step": 19520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9502625876728444e-05,
+      "loss": 1.7698,
+      "step": 19530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.949269847920649e-05,
+      "loss": 1.7696,
+      "step": 19540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.94827676384808e-05,
+      "loss": 1.7032,
+      "step": 19550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.947283335691078e-05,
+      "loss": 1.7362,
+      "step": 19560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9462895636856606e-05,
+      "loss": 1.6577,
+      "step": 19570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.945295448067933e-05,
+      "loss": 1.5756,
+      "step": 19580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.944300989074077e-05,
+      "loss": 2.3291,
+      "step": 19590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9433061869403596e-05,
+      "loss": 1.8899,
+      "step": 19600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.942311041903127e-05,
+      "loss": 1.9932,
+      "step": 19610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9413155541988086e-05,
+      "loss": 1.924,
+      "step": 19620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9403197240639134e-05,
+      "loss": 1.8888,
+      "step": 19630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.939323551735033e-05,
+      "loss": 1.7904,
+      "step": 19640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.93832703744884e-05,
+      "loss": 1.8913,
+      "step": 19650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9373301814420885e-05,
+      "loss": 1.8015,
+      "step": 19660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.936332983951613e-05,
+      "loss": 1.8726,
+      "step": 19670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9353354452143296e-05,
+      "loss": 2.3297,
+      "step": 19680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.934337565467236e-05,
+      "loss": 2.3726,
+      "step": 19690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9333393449474096e-05,
+      "loss": 2.2656,
+      "step": 19700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.932340783892011e-05,
+      "loss": 1.8524,
+      "step": 19710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9313418825382797e-05,
+      "loss": 1.8907,
+      "step": 19720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.930342641123536e-05,
+      "loss": 1.6221,
+      "step": 19730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.929343059885182e-05,
+      "loss": 1.6913,
+      "step": 19740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9283431390607e-05,
+      "loss": 1.7511,
+      "step": 19750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9273428788876545e-05,
+      "loss": 1.775,
+      "step": 19760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.926342279603686e-05,
+      "loss": 1.7368,
+      "step": 19770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.925341341446524e-05,
+      "loss": 1.8073,
+      "step": 19780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.924340064653968e-05,
+      "loss": 1.683,
+      "step": 19790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.923338449463906e-05,
+      "loss": 2.0044,
+      "step": 19800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.922336496114304e-05,
+      "loss": 1.8457,
+      "step": 19810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9213342048432064e-05,
+      "loss": 1.9592,
+      "step": 19820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.920331575888742e-05,
+      "loss": 2.2334,
+      "step": 19830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.919328609489115e-05,
+      "loss": 2.165,
+      "step": 19840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9183253058826126e-05,
+      "loss": 1.883,
+      "step": 19850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.917321665307602e-05,
+      "loss": 1.8251,
+      "step": 19860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9163176880025306e-05,
+      "loss": 1.6437,
+      "step": 19870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.915313374205925e-05,
+      "loss": 1.6483,
+      "step": 19880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.914308724156392e-05,
+      "loss": 1.6747,
+      "step": 19890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9133037380926174e-05,
+      "loss": 1.7958,
+      "step": 19900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.912298416253369e-05,
+      "loss": 1.8711,
+      "step": 19910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.911292758877493e-05,
+      "loss": 1.7244,
+      "step": 19920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.910286766203915e-05,
+      "loss": 1.3913,
+      "step": 19930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9092804384716405e-05,
+      "loss": 1.2789,
+      "step": 19940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.908273775919756e-05,
+      "loss": 1.267,
+      "step": 19950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.907266778787425e-05,
+      "loss": 1.305,
+      "step": 19960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9062594473138914e-05,
+      "loss": 1.6949,
+      "step": 19970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9052517817384814e-05,
+      "loss": 1.726,
+      "step": 19980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.904243782300596e-05,
+      "loss": 1.6375,
+      "step": 19990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.903235449239719e-05,
+      "loss": 1.7818,
+      "step": 20000
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.6719199419021606,
+      "eval_runtime": 82.1614,
+      "eval_samples_per_second": 36.513,
+      "eval_steps_per_second": 4.564,
+      "step": 20000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.902226782795411e-05,
+      "loss": 1.7389,
+      "step": 20010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9012177832073135e-05,
+      "loss": 1.7549,
+      "step": 20020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.900208450715147e-05,
+      "loss": 2.4886,
+      "step": 20030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8991987855587106e-05,
+      "loss": 1.7799,
+      "step": 20040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8981887879778826e-05,
+      "loss": 1.8575,
+      "step": 20050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.897178458212618e-05,
+      "loss": 1.8324,
+      "step": 20060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8961677965029564e-05,
+      "loss": 1.8776,
+      "step": 20070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8951568030890106e-05,
+      "loss": 1.8885,
+      "step": 20080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.894145478210975e-05,
+      "loss": 1.6727,
+      "step": 20090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.893133822109122e-05,
+      "loss": 1.5491,
+      "step": 20100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.892121835023803e-05,
+      "loss": 1.6234,
+      "step": 20110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.891109517195448e-05,
+      "loss": 2.0649,
+      "step": 20120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.890096868864566e-05,
+      "loss": 1.7587,
+      "step": 20130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8890838902717416e-05,
+      "loss": 1.7782,
+      "step": 20140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.888070581657644e-05,
+      "loss": 1.7231,
+      "step": 20150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.887056943263013e-05,
+      "loss": 1.661,
+      "step": 20160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.886042975328674e-05,
+      "loss": 1.6638,
+      "step": 20170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.885028678095526e-05,
+      "loss": 1.578,
+      "step": 20180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8840140518045476e-05,
+      "loss": 1.7251,
+      "step": 20190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.882999096696795e-05,
+      "loss": 2.0096,
+      "step": 20200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.881983813013407e-05,
+      "loss": 1.6559,
+      "step": 20210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.880968200995592e-05,
+      "loss": 1.4021,
+      "step": 20220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.879952260884644e-05,
+      "loss": 1.3743,
+      "step": 20230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.878935992921931e-05,
+      "loss": 1.9068,
+      "step": 20240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8779193973488994e-05,
+      "loss": 1.9849,
+      "step": 20250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8769024744070746e-05,
+      "loss": 1.7681,
+      "step": 20260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.875885224338061e-05,
+      "loss": 2.0144,
+      "step": 20270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.874867647383535e-05,
+      "loss": 1.7185,
+      "step": 20280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8738497437852574e-05,
+      "loss": 1.5899,
+      "step": 20290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.872831513785062e-05,
+      "loss": 1.5627,
+      "step": 20300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.871812957624864e-05,
+      "loss": 1.5517,
+      "step": 20310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.870794075546652e-05,
+      "loss": 1.543,
+      "step": 20320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.869774867792495e-05,
+      "loss": 1.5452,
+      "step": 20330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8687553346045394e-05,
+      "loss": 1.5281,
+      "step": 20340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.867735476225005e-05,
+      "loss": 1.5573,
+      "step": 20350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8667152928961935e-05,
+      "loss": 1.5516,
+      "step": 20360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.865694784860483e-05,
+      "loss": 1.7044,
+      "step": 20370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.864673952360326e-05,
+      "loss": 1.899,
+      "step": 20380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8636527956382545e-05,
+      "loss": 1.9532,
+      "step": 20390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8626313149368774e-05,
+      "loss": 1.7651,
+      "step": 20400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8616095104988796e-05,
+      "loss": 1.7377,
+      "step": 20410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8605873825670236e-05,
+      "loss": 1.849,
+      "step": 20420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.859564931384149e-05,
+      "loss": 2.0254,
+      "step": 20430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8585421571931705e-05,
+      "loss": 1.7248,
+      "step": 20440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.857519060237082e-05,
+      "loss": 1.7746,
+      "step": 20450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.856495640758952e-05,
+      "loss": 1.7915,
+      "step": 20460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.855471899001927e-05,
+      "loss": 1.8315,
+      "step": 20470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.85444783520923e-05,
+      "loss": 1.6788,
+      "step": 20480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.853423449624159e-05,
+      "loss": 1.8222,
+      "step": 20490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.852398742490091e-05,
+      "loss": 1.9023,
+      "step": 20500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8513737140504766e-05,
+      "loss": 1.7445,
+      "step": 20510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.850348364548844e-05,
+      "loss": 1.5802,
+      "step": 20520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8493226942287986e-05,
+      "loss": 1.9335,
+      "step": 20530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.848296703334022e-05,
+      "loss": 1.8343,
+      "step": 20540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8472703921082685e-05,
+      "loss": 1.8327,
+      "step": 20550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.846243760795373e-05,
+      "loss": 1.8993,
+      "step": 20560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.845216809639245e-05,
+      "loss": 2.1533,
+      "step": 20570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8441895388838675e-05,
+      "loss": 2.0013,
+      "step": 20580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.843161948773304e-05,
+      "loss": 2.0934,
+      "step": 20590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.84213403955169e-05,
+      "loss": 2.1578,
+      "step": 20600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.841105811463238e-05,
+      "loss": 2.1593,
+      "step": 20610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.840077264752236e-05,
+      "loss": 1.7487,
+      "step": 20620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.83904839966305e-05,
+      "loss": 1.8957,
+      "step": 20630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.838019216440119e-05,
+      "loss": 2.0142,
+      "step": 20640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.836989715327957e-05,
+      "loss": 1.4616,
+      "step": 20650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8359598965711566e-05,
+      "loss": 1.6836,
+      "step": 20660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8349297604143824e-05,
+      "loss": 1.7139,
+      "step": 20670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.833899307102378e-05,
+      "loss": 1.6656,
+      "step": 20680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.83286853687996e-05,
+      "loss": 1.6214,
+      "step": 20690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.831837449992021e-05,
+      "loss": 1.6208,
+      "step": 20700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.830806046683528e-05,
+      "loss": 1.6895,
+      "step": 20710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.829774327199524e-05,
+      "loss": 1.6621,
+      "step": 20720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.828742291785127e-05,
+      "loss": 1.7969,
+      "step": 20730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8277099406855305e-05,
+      "loss": 2.0026,
+      "step": 20740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8266772741460015e-05,
+      "loss": 1.8381,
+      "step": 20750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.825644292411884e-05,
+      "loss": 1.7816,
+      "step": 20760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8246109957285945e-05,
+      "loss": 1.7813,
+      "step": 20770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8235773843416276e-05,
+      "loss": 1.8329,
+      "step": 20780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.822543458496548e-05,
+      "loss": 1.7267,
+      "step": 20790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.821509218439001e-05,
+      "loss": 1.8711,
+      "step": 20800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.820474664414701e-05,
+      "loss": 1.7582,
+      "step": 20810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.81943979666944e-05,
+      "loss": 1.6506,
+      "step": 20820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8184046154490846e-05,
+      "loss": 1.8114,
+      "step": 20830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.817369120999574e-05,
+      "loss": 1.7804,
+      "step": 20840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8163333135669245e-05,
+      "loss": 1.7462,
+      "step": 20850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.815297193397224e-05,
+      "loss": 1.7284,
+      "step": 20860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.814260760736636e-05,
+      "loss": 1.7066,
+      "step": 20870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.813224015831399e-05,
+      "loss": 1.7458,
+      "step": 20880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.812186958927824e-05,
+      "loss": 1.7124,
+      "step": 20890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.811149590272298e-05,
+      "loss": 1.8492,
+      "step": 20900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.810111910111281e-05,
+      "loss": 1.7383,
+      "step": 20910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.809073918691306e-05,
+      "loss": 1.7421,
+      "step": 20920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.808035616258982e-05,
+      "loss": 1.7694,
+      "step": 20930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8069970030609905e-05,
+      "loss": 1.775,
+      "step": 20940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.805958079344088e-05,
+      "loss": 1.6878,
+      "step": 20950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.804918845355102e-05,
+      "loss": 1.5963,
+      "step": 20960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.803879301340938e-05,
+      "loss": 1.903,
+      "step": 20970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.802839447548572e-05,
+      "loss": 1.7145,
+      "step": 20980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8017992842250544e-05,
+      "loss": 1.8377,
+      "step": 20990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8007588116175094e-05,
+      "loss": 1.8695,
+      "step": 21000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.799718029973135e-05,
+      "loss": 1.7653,
+      "step": 21010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.798676939539201e-05,
+      "loss": 1.801,
+      "step": 21020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7976355405630534e-05,
+      "loss": 1.6953,
+      "step": 21030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.796593833292108e-05,
+      "loss": 1.8006,
+      "step": 21040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7955518179738564e-05,
+      "loss": 1.7951,
+      "step": 21050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7945094948558635e-05,
+      "loss": 1.7566,
+      "step": 21060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7934668641857654e-05,
+      "loss": 1.7827,
+      "step": 21070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.792423926211272e-05,
+      "loss": 1.8567,
+      "step": 21080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7913806811801675e-05,
+      "loss": 1.7523,
+      "step": 21090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.790337129340308e-05,
+      "loss": 1.8241,
+      "step": 21100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7892932709396224e-05,
+      "loss": 1.8981,
+      "step": 21110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.788249106226111e-05,
+      "loss": 1.8243,
+      "step": 21120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.787204635447853e-05,
+      "loss": 1.7446,
+      "step": 21130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7861598588529915e-05,
+      "loss": 1.9259,
+      "step": 21140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7851147766897474e-05,
+      "loss": 1.8173,
+      "step": 21150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.784069389206414e-05,
+      "loss": 1.7805,
+      "step": 21160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7830236966513566e-05,
+      "loss": 1.7785,
+      "step": 21170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.781977699273013e-05,
+      "loss": 1.7508,
+      "step": 21180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.780931397319893e-05,
+      "loss": 1.7411,
+      "step": 21190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.77988479104058e-05,
+      "loss": 1.7592,
+      "step": 21200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.778837880683727e-05,
+      "loss": 1.5076,
+      "step": 21210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.777790666498061e-05,
+      "loss": 1.5261,
+      "step": 21220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.776743148732384e-05,
+      "loss": 1.8198,
+      "step": 21230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.775695327635564e-05,
+      "loss": 1.6624,
+      "step": 21240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.774647203456547e-05,
+      "loss": 1.8671,
+      "step": 21250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.773598776444347e-05,
+      "loss": 1.8776,
+      "step": 21260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.772550046848051e-05,
+      "loss": 1.9879,
+      "step": 21270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.771501014916818e-05,
+      "loss": 1.7667,
+      "step": 21280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.770451680899881e-05,
+      "loss": 1.786,
+      "step": 21290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.769402045046542e-05,
+      "loss": 1.6326,
+      "step": 21300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7683521076061746e-05,
+      "loss": 2.1336,
+      "step": 21310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7673018688282255e-05,
+      "loss": 1.8962,
+      "step": 21320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7662513289622136e-05,
+      "loss": 1.9161,
+      "step": 21330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7652004882577264e-05,
+      "loss": 1.837,
+      "step": 21340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.764149346964426e-05,
+      "loss": 2.0038,
+      "step": 21350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7630979053320435e-05,
+      "loss": 1.6784,
+      "step": 21360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.762046163610384e-05,
+      "loss": 1.6955,
+      "step": 21370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7609941220493206e-05,
+      "loss": 1.9777,
+      "step": 21380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.759941780898801e-05,
+      "loss": 2.3311,
+      "step": 21390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7588891404088415e-05,
+      "loss": 1.7348,
+      "step": 21400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.757836200829531e-05,
+      "loss": 1.7724,
+      "step": 21410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.756782962411028e-05,
+      "loss": 1.8227,
+      "step": 21420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.755729425403565e-05,
+      "loss": 1.7858,
+      "step": 21430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.754675590057441e-05,
+      "loss": 1.7256,
+      "step": 21440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.753621456623029e-05,
+      "loss": 1.8612,
+      "step": 21450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7525670253507735e-05,
+      "loss": 2.4958,
+      "step": 21460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.751512296491186e-05,
+      "loss": 1.9053,
+      "step": 21470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.750457270294854e-05,
+      "loss": 1.7845,
+      "step": 21480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7494019470124295e-05,
+      "loss": 1.7628,
+      "step": 21490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.74834632689464e-05,
+      "loss": 1.725,
+      "step": 21500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.747290410192282e-05,
+      "loss": 1.7116,
+      "step": 21510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7462341971562214e-05,
+      "loss": 1.7632,
+      "step": 21520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.745177688037396e-05,
+      "loss": 1.993,
+      "step": 21530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.744120883086813e-05,
+      "loss": 1.6328,
+      "step": 21540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.74306378255555e-05,
+      "loss": 1.5776,
+      "step": 21550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.742006386694756e-05,
+      "loss": 1.7845,
+      "step": 21560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7409486957556486e-05,
+      "loss": 1.889,
+      "step": 21570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7398907099895154e-05,
+      "loss": 1.7302,
+      "step": 21580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.738832429647716e-05,
+      "loss": 1.8648,
+      "step": 21590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7377738549816774e-05,
+      "loss": 1.7418,
+      "step": 21600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.736714986242899e-05,
+      "loss": 1.6086,
+      "step": 21610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7356558236829476e-05,
+      "loss": 1.7707,
+      "step": 21620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7345963675534623e-05,
+      "loss": 1.8387,
+      "step": 21630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.733536618106151e-05,
+      "loss": 1.7437,
+      "step": 21640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7324765755927896e-05,
+      "loss": 1.7369,
+      "step": 21650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.731416240265226e-05,
+      "loss": 1.7419,
+      "step": 21660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.730355612375377e-05,
+      "loss": 1.717,
+      "step": 21670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.729294692175228e-05,
+      "loss": 1.808,
+      "step": 21680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.728233479916835e-05,
+      "loss": 1.8349,
+      "step": 21690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7271719758523225e-05,
+      "loss": 1.6731,
+      "step": 21700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.726110180233885e-05,
+      "loss": 1.7641,
+      "step": 21710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.725048093313787e-05,
+      "loss": 1.8213,
+      "step": 21720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.723985715344359e-05,
+      "loss": 1.8582,
+      "step": 21730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.722923046578004e-05,
+      "loss": 1.8905,
+      "step": 21740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7218600872671926e-05,
+      "loss": 1.7755,
+      "step": 21750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.720796837664466e-05,
+      "loss": 1.749,
+      "step": 21760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.719733298022431e-05,
+      "loss": 1.7706,
+      "step": 21770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.718669468593767e-05,
+      "loss": 1.8133,
+      "step": 21780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.71760534963122e-05,
+      "loss": 1.7395,
+      "step": 21790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.716540941387606e-05,
+      "loss": 1.8009,
+      "step": 21800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.71547624411581e-05,
+      "loss": 1.9699,
+      "step": 21810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.714411258068783e-05,
+      "loss": 1.9111,
+      "step": 21820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7133459834995465e-05,
+      "loss": 1.763,
+      "step": 21830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.712280420661192e-05,
+      "loss": 1.6418,
+      "step": 21840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7112145698068776e-05,
+      "loss": 1.6618,
+      "step": 21850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.71014843118983e-05,
+      "loss": 1.5545,
+      "step": 21860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7090820050633436e-05,
+      "loss": 1.5446,
+      "step": 21870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.708015291680783e-05,
+      "loss": 1.6402,
+      "step": 21880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.706948291295581e-05,
+      "loss": 1.5775,
+      "step": 21890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7058810041612356e-05,
+      "loss": 1.8741,
+      "step": 21900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.704813430531316e-05,
+      "loss": 1.8697,
+      "step": 21910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7037455706594596e-05,
+      "loss": 1.8899,
+      "step": 21920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.702677424799368e-05,
+      "loss": 1.6944,
+      "step": 21930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.701608993204815e-05,
+      "loss": 1.7497,
+      "step": 21940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.70054027612964e-05,
+      "loss": 1.7852,
+      "step": 21950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.699471273827753e-05,
+      "loss": 1.734,
+      "step": 21960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.698401986553127e-05,
+      "loss": 1.6974,
+      "step": 21970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6973324145598065e-05,
+      "loss": 1.7039,
+      "step": 21980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6962625581019024e-05,
+      "loss": 1.7583,
+      "step": 21990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.695192417433593e-05,
+      "loss": 1.8593,
+      "step": 22000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.694121992809125e-05,
+      "loss": 1.9412,
+      "step": 22010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.693051284482812e-05,
+      "loss": 1.9218,
+      "step": 22020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6919802927090345e-05,
+      "loss": 1.8648,
+      "step": 22030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.690909017742241e-05,
+      "loss": 1.5874,
+      "step": 22040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.689837459836948e-05,
+      "loss": 1.7104,
+      "step": 22050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.688765619247736e-05,
+      "loss": 1.8052,
+      "step": 22060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6876934962292575e-05,
+      "loss": 1.8381,
+      "step": 22070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6866210910362284e-05,
+      "loss": 1.8112,
+      "step": 22080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6855484039234325e-05,
+      "loss": 1.7958,
+      "step": 22090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.684475435145722e-05,
+      "loss": 1.9857,
+      "step": 22100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6834021849580135e-05,
+      "loss": 1.7658,
+      "step": 22110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.682328653615293e-05,
+      "loss": 1.7657,
+      "step": 22120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.681254841372612e-05,
+      "loss": 1.5535,
+      "step": 22130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6801807484850885e-05,
+      "loss": 1.7207,
+      "step": 22140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.679106375207908e-05,
+      "loss": 1.7034,
+      "step": 22150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.678031721796323e-05,
+      "loss": 2.0085,
+      "step": 22160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.67695678850565e-05,
+      "loss": 1.7749,
+      "step": 22170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.675881575591275e-05,
+      "loss": 1.8751,
+      "step": 22180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.674806083308648e-05,
+      "loss": 1.8139,
+      "step": 22190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.673730311913288e-05,
+      "loss": 1.5534,
+      "step": 22200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.672654261660779e-05,
+      "loss": 1.6991,
+      "step": 22210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.671577932806769e-05,
+      "loss": 2.0929,
+      "step": 22220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6705013256069773e-05,
+      "loss": 1.6748,
+      "step": 22230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.669424440317183e-05,
+      "loss": 1.7667,
+      "step": 22240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.668347277193238e-05,
+      "loss": 1.8185,
+      "step": 22250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.667269836491054e-05,
+      "loss": 1.7565,
+      "step": 22260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.666192118466614e-05,
+      "loss": 1.7857,
+      "step": 22270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.665114123375961e-05,
+      "loss": 1.6989,
+      "step": 22280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.664035851475211e-05,
+      "loss": 1.7107,
+      "step": 22290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.66295730302054e-05,
+      "loss": 2.0024,
+      "step": 22300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6618784782681915e-05,
+      "loss": 2.0356,
+      "step": 22310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.660799377474475e-05,
+      "loss": 1.9373,
+      "step": 22320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6597200008957654e-05,
+      "loss": 1.7353,
+      "step": 22330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6586403487885036e-05,
+      "loss": 1.7009,
+      "step": 22340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6575604214091955e-05,
+      "loss": 1.7529,
+      "step": 22350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.656480219014411e-05,
+      "loss": 1.6625,
+      "step": 22360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.655399741860788e-05,
+      "loss": 1.6758,
+      "step": 22370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.654318990205029e-05,
+      "loss": 1.8066,
+      "step": 22380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6532379643038986e-05,
+      "loss": 1.6777,
+      "step": 22390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.652156664414231e-05,
+      "loss": 1.7312,
+      "step": 22400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.651075090792924e-05,
+      "loss": 2.0819,
+      "step": 22410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.649993243696938e-05,
+      "loss": 2.0563,
+      "step": 22420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6489111233833026e-05,
+      "loss": 1.9119,
+      "step": 22430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6478287301091085e-05,
+      "loss": 1.9691,
+      "step": 22440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.646746064131513e-05,
+      "loss": 1.925,
+      "step": 22450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6456631257077383e-05,
+      "loss": 1.7243,
+      "step": 22460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.644579915095071e-05,
+      "loss": 1.736,
+      "step": 22470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.643496432550864e-05,
+      "loss": 1.8087,
+      "step": 22480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6424126783325296e-05,
+      "loss": 1.9584,
+      "step": 22490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.641328652697551e-05,
+      "loss": 1.9199,
+      "step": 22500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6402443559034725e-05,
+      "loss": 1.9966,
+      "step": 22510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6391597882079035e-05,
+      "loss": 1.8249,
+      "step": 22520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6380749498685175e-05,
+      "loss": 1.8854,
+      "step": 22530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6369898411430526e-05,
+      "loss": 1.9154,
+      "step": 22540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6359044622893104e-05,
+      "loss": 1.7194,
+      "step": 22550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.634818813565159e-05,
+      "loss": 1.6552,
+      "step": 22560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.633732895228527e-05,
+      "loss": 1.7559,
+      "step": 22570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.632646707537411e-05,
+      "loss": 2.0456,
+      "step": 22580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.631560250749868e-05,
+      "loss": 2.4504,
+      "step": 22590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.630473525124021e-05,
+      "loss": 2.4843,
+      "step": 22600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.629386530918057e-05,
+      "loss": 2.4479,
+      "step": 22610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.628299268390225e-05,
+      "loss": 1.8605,
+      "step": 22620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.62721173779884e-05,
+      "loss": 1.7632,
+      "step": 22630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.62612393940228e-05,
+      "loss": 1.7765,
+      "step": 22640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.625035873458985e-05,
+      "loss": 1.6999,
+      "step": 22650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6239475402274604e-05,
+      "loss": 1.7993,
+      "step": 22660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.622858939966275e-05,
+      "loss": 1.8779,
+      "step": 22670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.62177007293406e-05,
+      "loss": 1.8765,
+      "step": 22680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6206809393895106e-05,
+      "loss": 1.7534,
+      "step": 22690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6195915395913856e-05,
+      "loss": 1.8675,
+      "step": 22700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6185018737985074e-05,
+      "loss": 1.7767,
+      "step": 22710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.617411942269759e-05,
+      "loss": 1.6999,
+      "step": 22720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6163217452640895e-05,
+      "loss": 1.661,
+      "step": 22730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.61523128304051e-05,
+      "loss": 1.67,
+      "step": 22740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.614140555858096e-05,
+      "loss": 1.7231,
+      "step": 22750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6130495639759827e-05,
+      "loss": 1.9564,
+      "step": 22760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.611958307653371e-05,
+      "loss": 1.7446,
+      "step": 22770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.610866787149523e-05,
+      "loss": 1.8313,
+      "step": 22780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6097750027237654e-05,
+      "loss": 1.6877,
+      "step": 22790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.608682954635485e-05,
+      "loss": 1.6592,
+      "step": 22800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.607590643144135e-05,
+      "loss": 1.5947,
+      "step": 22810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.606498068509226e-05,
+      "loss": 1.7021,
+      "step": 22820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.605405230990336e-05,
+      "loss": 1.4892,
+      "step": 22830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.604312130847104e-05,
+      "loss": 1.8037,
+      "step": 22840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.603218768339229e-05,
+      "loss": 1.6658,
+      "step": 22850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6021251437264746e-05,
+      "loss": 1.8919,
+      "step": 22860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6010312572686676e-05,
+      "loss": 1.5664,
+      "step": 22870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.599937109225695e-05,
+      "loss": 1.6721,
+      "step": 22880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5988426998575057e-05,
+      "loss": 1.7573,
+      "step": 22890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.597748029424113e-05,
+      "loss": 1.7993,
+      "step": 22900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.59665309818559e-05,
+      "loss": 1.7229,
+      "step": 22910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.595557906402073e-05,
+      "loss": 1.7968,
+      "step": 22920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.594462454333759e-05,
+      "loss": 1.917,
+      "step": 22930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.59336674224091e-05,
+      "loss": 1.7526,
+      "step": 22940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.592270770383844e-05,
+      "loss": 1.6933,
+      "step": 22950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.591174539022947e-05,
+      "loss": 1.7998,
+      "step": 22960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.590078048418663e-05,
+      "loss": 1.7173,
+      "step": 22970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5889812988314976e-05,
+      "loss": 1.7017,
+      "step": 22980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.587884290522019e-05,
+      "loss": 1.784,
+      "step": 22990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5867870237508574e-05,
+      "loss": 1.8082,
+      "step": 23000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5856894987787023e-05,
+      "loss": 1.5939,
+      "step": 23010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.584591715866307e-05,
+      "loss": 1.7126,
+      "step": 23020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.583493675274484e-05,
+      "loss": 1.7157,
+      "step": 23030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.582395377264109e-05,
+      "loss": 1.7234,
+      "step": 23040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5812968220961165e-05,
+      "loss": 1.6123,
+      "step": 23050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5801980100315046e-05,
+      "loss": 1.7661,
+      "step": 23060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5790989413313304e-05,
+      "loss": 1.9309,
+      "step": 23070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.577999616256713e-05,
+      "loss": 1.8323,
+      "step": 23080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.576900035068833e-05,
+      "loss": 1.8078,
+      "step": 23090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5758001980289304e-05,
+      "loss": 1.6526,
+      "step": 23100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.574700105398306e-05,
+      "loss": 1.7426,
+      "step": 23110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.573599757438322e-05,
+      "loss": 1.7144,
+      "step": 23120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.572499154410403e-05,
+      "loss": 1.5783,
+      "step": 23130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.571398296576031e-05,
+      "loss": 1.7074,
+      "step": 23140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5702971841967495e-05,
+      "loss": 1.6907,
+      "step": 23150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.569195817534164e-05,
+      "loss": 1.6853,
+      "step": 23160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.56809419684994e-05,
+      "loss": 1.7841,
+      "step": 23170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.566992322405801e-05,
+      "loss": 1.6522,
+      "step": 23180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.565890194463533e-05,
+      "loss": 1.6243,
+      "step": 23190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5647878132849814e-05,
+      "loss": 1.5946,
+      "step": 23200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.563685179132053e-05,
+      "loss": 1.5563,
+      "step": 23210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.562582292266713e-05,
+      "loss": 1.5263,
+      "step": 23220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5614791529509886e-05,
+      "loss": 1.5209,
+      "step": 23230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.560375761446964e-05,
+      "loss": 1.8463,
+      "step": 23240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5592721180167865e-05,
+      "loss": 1.7597,
+      "step": 23250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5581682229226613e-05,
+      "loss": 1.709,
+      "step": 23260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.557064076426855e-05,
+      "loss": 1.82,
+      "step": 23270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.555959678791692e-05,
+      "loss": 1.9364,
+      "step": 23280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.554855030279556e-05,
+      "loss": 1.9167,
+      "step": 23290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.553750131152894e-05,
+      "loss": 1.912,
+      "step": 23300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.552644981674209e-05,
+      "loss": 1.9452,
+      "step": 23310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5515395821060635e-05,
+      "loss": 1.9852,
+      "step": 23320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.550433932711083e-05,
+      "loss": 1.8639,
+      "step": 23330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.549328033751947e-05,
+      "loss": 2.0317,
+      "step": 23340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.548221885491399e-05,
+      "loss": 1.7329,
+      "step": 23350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.547115488192239e-05,
+      "loss": 1.7058,
+      "step": 23360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.546008842117328e-05,
+      "loss": 1.8143,
+      "step": 23370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5449019475295846e-05,
+      "loss": 1.7216,
+      "step": 23380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5437948046919865e-05,
+      "loss": 1.6946,
+      "step": 23390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.542687413867572e-05,
+      "loss": 1.6823,
+      "step": 23400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.541579775319436e-05,
+      "loss": 1.6412,
+      "step": 23410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.540471889310734e-05,
+      "loss": 1.7579,
+      "step": 23420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.53936375610468e-05,
+      "loss": 1.7311,
+      "step": 23430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.538255375964546e-05,
+      "loss": 1.719,
+      "step": 23440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.537146749153663e-05,
+      "loss": 1.6885,
+      "step": 23450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5360378759354216e-05,
+      "loss": 1.6851,
+      "step": 23460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.534928756573269e-05,
+      "loss": 1.6414,
+      "step": 23470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.533819391330713e-05,
+      "loss": 1.7166,
+      "step": 23480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.532709780471317e-05,
+      "loss": 1.6955,
+      "step": 23490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5315999242587074e-05,
+      "loss": 1.7456,
+      "step": 23500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.530489822956563e-05,
+      "loss": 1.8366,
+      "step": 23510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5293794768286254e-05,
+      "loss": 1.7837,
+      "step": 23520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.528268886138693e-05,
+      "loss": 1.9097,
+      "step": 23530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.527158051150622e-05,
+      "loss": 1.6959,
+      "step": 23540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.526046972128326e-05,
+      "loss": 1.759,
+      "step": 23550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.524935649335777e-05,
+      "loss": 1.7727,
+      "step": 23560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5238240830370075e-05,
+      "loss": 1.8515,
+      "step": 23570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.522712273496103e-05,
+      "loss": 1.8419,
+      "step": 23580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.521600220977211e-05,
+      "loss": 1.7575,
+      "step": 23590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.520487925744535e-05,
+      "loss": 1.8707,
+      "step": 23600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.519375388062336e-05,
+      "loss": 1.7468,
+      "step": 23610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.518262608194932e-05,
+      "loss": 1.6861,
+      "step": 23620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.517149586406702e-05,
+      "loss": 1.5096,
+      "step": 23630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.516036322962076e-05,
+      "loss": 1.7559,
+      "step": 23640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5149228181255485e-05,
+      "loss": 1.9187,
+      "step": 23650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5138090721616676e-05,
+      "loss": 1.8076,
+      "step": 23660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5126950853350384e-05,
+      "loss": 1.8971,
+      "step": 23670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5115808579103235e-05,
+      "loss": 1.5526,
+      "step": 23680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5104663901522456e-05,
+      "loss": 1.3989,
+      "step": 23690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.50935168232558e-05,
+      "loss": 1.397,
+      "step": 23700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5082367346951625e-05,
+      "loss": 1.8248,
+      "step": 23710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.507121547525883e-05,
+      "loss": 1.82,
+      "step": 23720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5060061210826914e-05,
+      "loss": 1.6933,
+      "step": 23730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.504890455630592e-05,
+      "loss": 1.5885,
+      "step": 23740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.503774551434647e-05,
+      "loss": 1.5817,
+      "step": 23750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5026584087599765e-05,
+      "loss": 1.7256,
+      "step": 23760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.501542027871753e-05,
+      "loss": 1.9109,
+      "step": 23770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5004254090352104e-05,
+      "loss": 1.768,
+      "step": 23780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4993085525156376e-05,
+      "loss": 1.7429,
+      "step": 23790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4981914585783784e-05,
+      "loss": 1.7856,
+      "step": 23800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.497074127488834e-05,
+      "loss": 1.7591,
+      "step": 23810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4959565595124624e-05,
+      "loss": 1.8351,
+      "step": 23820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4948387549147794e-05,
+      "loss": 1.8248,
+      "step": 23830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.493720713961353e-05,
+      "loss": 1.767,
+      "step": 23840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.492602436917809e-05,
+      "loss": 1.7004,
+      "step": 23850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.491483924049833e-05,
+      "loss": 1.7291,
+      "step": 23860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.49036517562316e-05,
+      "loss": 1.802,
+      "step": 23870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.489246191903587e-05,
+      "loss": 1.8571,
+      "step": 23880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4881269731569634e-05,
+      "loss": 1.9238,
+      "step": 23890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.487007519649195e-05,
+      "loss": 1.8023,
+      "step": 23900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.485887831646243e-05,
+      "loss": 1.7083,
+      "step": 23910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.484767909414128e-05,
+      "loss": 1.7932,
+      "step": 23920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.48364775321892e-05,
+      "loss": 1.774,
+      "step": 23930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4825273633267506e-05,
+      "loss": 1.6841,
+      "step": 23940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.481406740003803e-05,
+      "loss": 1.6688,
+      "step": 23950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4802858835163166e-05,
+      "loss": 1.6681,
+      "step": 23960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4791647941305874e-05,
+      "loss": 1.7979,
+      "step": 23970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.478043472112965e-05,
+      "loss": 1.7689,
+      "step": 23980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4769219177298576e-05,
+      "loss": 1.8383,
+      "step": 23990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4758001312477235e-05,
+      "loss": 1.867,
+      "step": 24000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.474678112933081e-05,
+      "loss": 1.8459,
+      "step": 24010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4735558630525e-05,
+      "loss": 1.719,
+      "step": 24020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4724333818726074e-05,
+      "loss": 2.2192,
+      "step": 24030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.471310669660085e-05,
+      "loss": 1.9411,
+      "step": 24040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4701877266816684e-05,
+      "loss": 1.7642,
+      "step": 24050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4690645532041485e-05,
+      "loss": 1.791,
+      "step": 24060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.467941149494372e-05,
+      "loss": 1.7735,
+      "step": 24070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.466817515819238e-05,
+      "loss": 1.7356,
+      "step": 24080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4656936524457026e-05,
+      "loss": 1.7402,
+      "step": 24090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.464569559640775e-05,
+      "loss": 1.819,
+      "step": 24100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.463445237671519e-05,
+      "loss": 1.7783,
+      "step": 24110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.462320686805054e-05,
+      "loss": 1.8609,
+      "step": 24120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.461195907308553e-05,
+      "loss": 1.973,
+      "step": 24130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.460070899449242e-05,
+      "loss": 1.7322,
+      "step": 24140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.458945663494405e-05,
+      "loss": 1.7004,
+      "step": 24150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.457820199711375e-05,
+      "loss": 1.8108,
+      "step": 24160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4566945083675436e-05,
+      "loss": 1.7443,
+      "step": 24170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4555685897303544e-05,
+      "loss": 1.573,
+      "step": 24180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.454442444067305e-05,
+      "loss": 1.4901,
+      "step": 24190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4533160716459485e-05,
+      "loss": 1.5956,
+      "step": 24200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.452189472733889e-05,
+      "loss": 2.021,
+      "step": 24210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.451062647598788e-05,
+      "loss": 1.9465,
+      "step": 24220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4499355965083564e-05,
+      "loss": 1.8425,
+      "step": 24230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.448808319730363e-05,
+      "loss": 1.6531,
+      "step": 24240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.447680817532629e-05,
+      "loss": 1.952,
+      "step": 24250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.446553090183026e-05,
+      "loss": 1.7268,
+      "step": 24260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4454251379494835e-05,
+      "loss": 1.7894,
+      "step": 24270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.444296961099983e-05,
+      "loss": 1.7273,
+      "step": 24280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.443168559902557e-05,
+      "loss": 1.7656,
+      "step": 24290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4420399346252965e-05,
+      "loss": 1.7506,
+      "step": 24300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4409110855363384e-05,
+      "loss": 1.8547,
+      "step": 24310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.439782012903881e-05,
+      "loss": 1.9149,
+      "step": 24320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.438652716996169e-05,
+      "loss": 1.8917,
+      "step": 24330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4375231980815026e-05,
+      "loss": 1.702,
+      "step": 24340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.436393456428236e-05,
+      "loss": 1.7413,
+      "step": 24350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.435263492304776e-05,
+      "loss": 1.6912,
+      "step": 24360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.43413330597958e-05,
+      "loss": 1.798,
+      "step": 24370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.433002897721162e-05,
+      "loss": 1.8972,
+      "step": 24380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.431872267798085e-05,
+      "loss": 1.9135,
+      "step": 24390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4307414164789665e-05,
+      "loss": 1.7388,
+      "step": 24400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.429610344032477e-05,
+      "loss": 1.7196,
+      "step": 24410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.428479050727339e-05,
+      "loss": 1.6962,
+      "step": 24420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4273475368323266e-05,
+      "loss": 1.7335,
+      "step": 24430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.426215802616267e-05,
+      "loss": 1.7125,
+      "step": 24440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4250838483480414e-05,
+      "loss": 1.6688,
+      "step": 24450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4239516742965795e-05,
+      "loss": 1.6918,
+      "step": 24460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.422819280730867e-05,
+      "loss": 1.751,
+      "step": 24470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.42168666791994e-05,
+      "loss": 1.7352,
+      "step": 24480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.420553836132886e-05,
+      "loss": 1.6766,
+      "step": 24490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.419420785638846e-05,
+      "loss": 1.7238,
+      "step": 24500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.418287516707012e-05,
+      "loss": 1.705,
+      "step": 24510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.417154029606629e-05,
+      "loss": 1.7676,
+      "step": 24520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.416020324606993e-05,
+      "loss": 1.5816,
+      "step": 24530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.414886401977451e-05,
+      "loss": 1.7767,
+      "step": 24540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4137522619874035e-05,
+      "loss": 1.6206,
+      "step": 24550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.412617904906301e-05,
+      "loss": 1.6197,
+      "step": 24560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4114833310036465e-05,
+      "loss": 1.9195,
+      "step": 24570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4103485405489954e-05,
+      "loss": 1.8067,
+      "step": 24580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4092135338119514e-05,
+      "loss": 1.8727,
+      "step": 24590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4080783110621725e-05,
+      "loss": 1.7131,
+      "step": 24600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.406942872569368e-05,
+      "loss": 1.5751,
+      "step": 24610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.405807218603297e-05,
+      "loss": 1.6031,
+      "step": 24620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4046713494337694e-05,
+      "loss": 1.8843,
+      "step": 24630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.403535265330648e-05,
+      "loss": 1.8602,
+      "step": 24640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.402398966563848e-05,
+      "loss": 1.8609,
+      "step": 24650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.40126245340333e-05,
+      "loss": 1.7298,
+      "step": 24660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4001257261191105e-05,
+      "loss": 1.6763,
+      "step": 24670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.398988784981255e-05,
+      "loss": 1.6175,
+      "step": 24680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.397851630259882e-05,
+      "loss": 1.5832,
+      "step": 24690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3967142622251555e-05,
+      "loss": 1.6236,
+      "step": 24700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.395576681147298e-05,
+      "loss": 1.7545,
+      "step": 24710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.394438887296575e-05,
+      "loss": 1.751,
+      "step": 24720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3933008809433064e-05,
+      "loss": 1.7365,
+      "step": 24730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3921626623578624e-05,
+      "loss": 1.8056,
+      "step": 24740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.391024231810664e-05,
+      "loss": 1.7138,
+      "step": 24750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3898855895721795e-05,
+      "loss": 1.6259,
+      "step": 24760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.388746735912932e-05,
+      "loss": 1.7193,
+      "step": 24770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.387607671103492e-05,
+      "loss": 1.8801,
+      "step": 24780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3864683954144805e-05,
+      "loss": 1.8366,
+      "step": 24790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.385328909116568e-05,
+      "loss": 2.0822,
+      "step": 24800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.384189212480477e-05,
+      "loss": 2.0006,
+      "step": 24810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.38304930577698e-05,
+      "loss": 1.7844,
+      "step": 24820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.381909189276895e-05,
+      "loss": 1.6923,
+      "step": 24830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3807688632510956e-05,
+      "loss": 1.6599,
+      "step": 24840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.379628327970502e-05,
+      "loss": 1.8322,
+      "step": 24850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.378487583706085e-05,
+      "loss": 1.9181,
+      "step": 24860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3773466307288644e-05,
+      "loss": 1.7382,
+      "step": 24870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.376205469309911e-05,
+      "loss": 1.5476,
+      "step": 24880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.375064099720343e-05,
+      "loss": 1.7766,
+      "step": 24890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3739225222313296e-05,
+      "loss": 1.7224,
+      "step": 24900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.372780737114091e-05,
+      "loss": 1.664,
+      "step": 24910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.371638744639891e-05,
+      "loss": 1.7273,
+      "step": 24920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.370496545080048e-05,
+      "loss": 1.697,
+      "step": 24930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.36935413870593e-05,
+      "loss": 1.6109,
+      "step": 24940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.36821152578895e-05,
+      "loss": 1.4285,
+      "step": 24950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.367068706600572e-05,
+      "loss": 1.7077,
+      "step": 24960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3659256814123106e-05,
+      "loss": 1.6574,
+      "step": 24970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3647824504957274e-05,
+      "loss": 1.6948,
+      "step": 24980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3636390141224334e-05,
+      "loss": 1.6229,
+      "step": 24990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3624953725640883e-05,
+      "loss": 1.7728,
+      "step": 25000
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 1.6268333196640015,
+      "eval_runtime": 81.9489,
+      "eval_samples_per_second": 36.608,
+      "eval_steps_per_second": 4.576,
+      "step": 25000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3613515260924014e-05,
+      "loss": 2.0887,
+      "step": 25010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3602074749791285e-05,
+      "loss": 1.7694,
+      "step": 25020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3590632194960764e-05,
+      "loss": 1.8469,
+      "step": 25030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.357918759915101e-05,
+      "loss": 1.5551,
+      "step": 25040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.356774096508102e-05,
+      "loss": 1.6253,
+      "step": 25050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3556292295470334e-05,
+      "loss": 1.7699,
+      "step": 25060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.354484159303894e-05,
+      "loss": 1.6644,
+      "step": 25070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3533388860507324e-05,
+      "loss": 1.584,
+      "step": 25080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.352193410059644e-05,
+      "loss": 1.4726,
+      "step": 25090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3510477316027736e-05,
+      "loss": 1.5807,
+      "step": 25100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.349901850952315e-05,
+      "loss": 1.5795,
+      "step": 25110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.348755768380507e-05,
+      "loss": 1.5597,
+      "step": 25120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.347609484159638e-05,
+      "loss": 1.5436,
+      "step": 25130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.346462998562046e-05,
+      "loss": 1.5582,
+      "step": 25140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.345316311860113e-05,
+      "loss": 1.5671,
+      "step": 25150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3441694243262745e-05,
+      "loss": 1.5275,
+      "step": 25160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.343022336233008e-05,
+      "loss": 1.5793,
+      "step": 25170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3418750478528407e-05,
+      "loss": 1.6574,
+      "step": 25180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3407275594583484e-05,
+      "loss": 1.6223,
+      "step": 25190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3395798713221526e-05,
+      "loss": 1.5103,
+      "step": 25200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.338431983716926e-05,
+      "loss": 1.4893,
+      "step": 25210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.337283896915383e-05,
+      "loss": 1.5049,
+      "step": 25220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.33613561119029e-05,
+      "loss": 1.5761,
+      "step": 25230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.334987126814458e-05,
+      "loss": 1.5808,
+      "step": 25240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3338384440607466e-05,
+      "loss": 1.567,
+      "step": 25250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.332689563202062e-05,
+      "loss": 1.4998,
+      "step": 25260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3315404845113585e-05,
+      "loss": 1.7106,
+      "step": 25270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.330391208261635e-05,
+      "loss": 1.7445,
+      "step": 25280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.329241734725939e-05,
+      "loss": 1.7429,
+      "step": 25290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.328092064177366e-05,
+      "loss": 1.6719,
+      "step": 25300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.326942196889056e-05,
+      "loss": 1.654,
+      "step": 25310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.325792133134196e-05,
+      "loss": 1.6559,
+      "step": 25320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.324641873186022e-05,
+      "loss": 1.7359,
+      "step": 25330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3234914173178143e-05,
+      "loss": 1.7824,
+      "step": 25340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.322340765802901e-05,
+      "loss": 1.6409,
+      "step": 25350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.321189918914654e-05,
+      "loss": 1.5096,
+      "step": 25360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.320038876926497e-05,
+      "loss": 2.0661,
+      "step": 25370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3188876401118935e-05,
+      "loss": 2.0201,
+      "step": 25380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.317736208744359e-05,
+      "loss": 1.8802,
+      "step": 25390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.316584583097453e-05,
+      "loss": 1.7726,
+      "step": 25400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3154327634447785e-05,
+      "loss": 1.6791,
+      "step": 25410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.314280750059989e-05,
+      "loss": 1.6414,
+      "step": 25420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.313128543216782e-05,
+      "loss": 1.8079,
+      "step": 25430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3119761431889016e-05,
+      "loss": 1.9467,
+      "step": 25440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.310823550250135e-05,
+      "loss": 1.7744,
+      "step": 25450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3096707646743195e-05,
+      "loss": 1.5397,
+      "step": 25460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.308517786735336e-05,
+      "loss": 1.5905,
+      "step": 25470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.30736461670711e-05,
+      "loss": 1.5812,
+      "step": 25480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.306211254863615e-05,
+      "loss": 1.6715,
+      "step": 25490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3050577014788696e-05,
+      "loss": 1.6572,
+      "step": 25500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3039039568269356e-05,
+      "loss": 1.7116,
+      "step": 25510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.302750021181923e-05,
+      "loss": 1.7914,
+      "step": 25520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.301595894817985e-05,
+      "loss": 1.8371,
+      "step": 25530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3004415780093234e-05,
+      "loss": 1.5683,
+      "step": 25540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2992870710301804e-05,
+      "loss": 1.6656,
+      "step": 25550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2981323741548484e-05,
+      "loss": 1.895,
+      "step": 25560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.296977487657661e-05,
+      "loss": 1.6954,
+      "step": 25570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.295822411812999e-05,
+      "loss": 1.6994,
+      "step": 25580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.294667146895287e-05,
+      "loss": 1.8332,
+      "step": 25590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.293511693178996e-05,
+      "loss": 1.7575,
+      "step": 25600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.29235605093864e-05,
+      "loss": 1.7178,
+      "step": 25610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.291200220448779e-05,
+      "loss": 1.7724,
+      "step": 25620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.290044201984018e-05,
+      "loss": 1.7646,
+      "step": 25630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2888879958190055e-05,
+      "loss": 1.714,
+      "step": 25640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.287731602228436e-05,
+      "loss": 1.7595,
+      "step": 25650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.286575021487047e-05,
+      "loss": 1.6687,
+      "step": 25660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.285418253869623e-05,
+      "loss": 1.6742,
+      "step": 25670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2842612996509884e-05,
+      "loss": 1.6377,
+      "step": 25680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.283104159106015e-05,
+      "loss": 1.629,
+      "step": 25690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.281946832509621e-05,
+      "loss": 1.7158,
+      "step": 25700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2807893201367645e-05,
+      "loss": 1.6402,
+      "step": 25710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.27963162226245e-05,
+      "loss": 1.6443,
+      "step": 25720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.278473739161725e-05,
+      "loss": 1.95,
+      "step": 25730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2773156711096824e-05,
+      "loss": 1.8658,
+      "step": 25740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2761574183814576e-05,
+      "loss": 1.685,
+      "step": 25750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.274998981252231e-05,
+      "loss": 1.8152,
+      "step": 25760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.273840359997227e-05,
+      "loss": 1.6196,
+      "step": 25770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2726815548917114e-05,
+      "loss": 1.5927,
+      "step": 25780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.271522566210997e-05,
+      "loss": 1.7696,
+      "step": 25790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.270363394230438e-05,
+      "loss": 1.6634,
+      "step": 25800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.269204039225432e-05,
+      "loss": 1.8589,
+      "step": 25810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2680445014714226e-05,
+      "loss": 1.784,
+      "step": 25820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.266884781243894e-05,
+      "loss": 1.6899,
+      "step": 25830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.265724878818374e-05,
+      "loss": 1.808,
+      "step": 25840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2645647944704364e-05,
+      "loss": 1.734,
+      "step": 25850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.263404528475695e-05,
+      "loss": 1.5994,
+      "step": 25860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.262244081109807e-05,
+      "loss": 1.6928,
+      "step": 25870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.261083452648477e-05,
+      "loss": 1.5715,
+      "step": 25880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.259922643367446e-05,
+      "loss": 1.5805,
+      "step": 25890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.258761653542504e-05,
+      "loss": 1.6229,
+      "step": 25900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.257600483449479e-05,
+      "loss": 1.7378,
+      "step": 25910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.256439133364246e-05,
+      "loss": 1.714,
+      "step": 25920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2552776035627194e-05,
+      "loss": 1.7747,
+      "step": 25930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.254115894320859e-05,
+      "loss": 1.9148,
+      "step": 25940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.252954005914665e-05,
+      "loss": 1.671,
+      "step": 25950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.251791938620181e-05,
+      "loss": 1.8409,
+      "step": 25960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2506296927134935e-05,
+      "loss": 1.7661,
+      "step": 25970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.249467268470731e-05,
+      "loss": 1.707,
+      "step": 25980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.248304666168066e-05,
+      "loss": 1.7039,
+      "step": 25990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.24714188608171e-05,
+      "loss": 1.9488,
+      "step": 26000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.245978928487919e-05,
+      "loss": 1.8123,
+      "step": 26010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2448157936629905e-05,
+      "loss": 1.406,
+      "step": 26020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2436524818832665e-05,
+      "loss": 1.6617,
+      "step": 26030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.242488993425126e-05,
+      "loss": 1.9082,
+      "step": 26040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2413253285649944e-05,
+      "loss": 1.8644,
+      "step": 26050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.240161487579337e-05,
+      "loss": 1.8957,
+      "step": 26060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2389974707446616e-05,
+      "loss": 1.8585,
+      "step": 26070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.237833278337518e-05,
+      "loss": 1.9341,
+      "step": 26080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.236668910634496e-05,
+      "loss": 1.8685,
+      "step": 26090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2355043679122295e-05,
+      "loss": 1.8628,
+      "step": 26100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2343396504473926e-05,
+      "loss": 1.8625,
+      "step": 26110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2331747585167016e-05,
+      "loss": 1.8669,
+      "step": 26120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.232009692396913e-05,
+      "loss": 1.9103,
+      "step": 26130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.230844452364826e-05,
+      "loss": 1.7012,
+      "step": 26140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.229679038697281e-05,
+      "loss": 1.6406,
+      "step": 26150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2285134516711585e-05,
+      "loss": 1.587,
+      "step": 26160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.22734769156338e-05,
+      "loss": 1.5711,
+      "step": 26170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2261817586509116e-05,
+      "loss": 1.5205,
+      "step": 26180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2250156532107566e-05,
+      "loss": 1.7986,
+      "step": 26190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.223849375519961e-05,
+      "loss": 1.6426,
+      "step": 26200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2226829258556096e-05,
+      "loss": 1.5944,
+      "step": 26210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.221516304494831e-05,
+      "loss": 1.5593,
+      "step": 26220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2203495117147954e-05,
+      "loss": 1.5733,
+      "step": 26230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2191825477927086e-05,
+      "loss": 1.5301,
+      "step": 26240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.218015413005822e-05,
+      "loss": 1.538,
+      "step": 26250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.216848107631424e-05,
+      "loss": 1.5263,
+      "step": 26260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.215680631946848e-05,
+      "loss": 1.515,
+      "step": 26270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.214512986229463e-05,
+      "loss": 1.5197,
+      "step": 26280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.213345170756682e-05,
+      "loss": 1.6252,
+      "step": 26290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.212177185805954e-05,
+      "loss": 1.8574,
+      "step": 26300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2110090316547754e-05,
+      "loss": 1.5719,
+      "step": 26310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.209840708580676e-05,
+      "loss": 1.4922,
+      "step": 26320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.208672216861228e-05,
+      "loss": 1.708,
+      "step": 26330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.207503556774045e-05,
+      "loss": 1.5512,
+      "step": 26340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.20633472859678e-05,
+      "loss": 1.7736,
+      "step": 26350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.205165732607124e-05,
+      "loss": 2.1982,
+      "step": 26360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.203996569082811e-05,
+      "loss": 1.6502,
+      "step": 26370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2028272383016105e-05,
+      "loss": 1.6477,
+      "step": 26380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.201657740541338e-05,
+      "loss": 1.6217,
+      "step": 26390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2004880760798415e-05,
+      "loss": 1.6244,
+      "step": 26400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.199318245195015e-05,
+      "loss": 1.7033,
+      "step": 26410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.198148248164788e-05,
+      "loss": 1.7792,
+      "step": 26420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1969780852671305e-05,
+      "loss": 1.5474,
+      "step": 26430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1958077567800525e-05,
+      "loss": 1.5157,
+      "step": 26440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.194637262981603e-05,
+      "loss": 1.6204,
+      "step": 26450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1934666041498704e-05,
+      "loss": 1.5332,
+      "step": 26460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.192295780562981e-05,
+      "loss": 1.66,
+      "step": 26470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.191124792499102e-05,
+      "loss": 1.5516,
+      "step": 26480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.189953640236441e-05,
+      "loss": 1.7165,
+      "step": 26490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1887823240532386e-05,
+      "loss": 1.5814,
+      "step": 26500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.187610844227782e-05,
+      "loss": 1.6904,
+      "step": 26510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.186439201038392e-05,
+      "loss": 1.6722,
+      "step": 26520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.18526739476343e-05,
+      "loss": 1.6511,
+      "step": 26530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.184095425681296e-05,
+      "loss": 1.6741,
+      "step": 26540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.182923294070429e-05,
+      "loss": 1.6868,
+      "step": 26550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.181751000209307e-05,
+      "loss": 1.8265,
+      "step": 26560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.180578544376445e-05,
+      "loss": 1.713,
+      "step": 26570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1794059268503965e-05,
+      "loss": 1.5994,
+      "step": 26580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.178233147909756e-05,
+      "loss": 1.7666,
+      "step": 26590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1770602078331536e-05,
+      "loss": 1.6172,
+      "step": 26600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.175887106899259e-05,
+      "loss": 1.7012,
+      "step": 26610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.17471384538678e-05,
+      "loss": 1.7924,
+      "step": 26620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.173540423574462e-05,
+      "loss": 1.6811,
+      "step": 26630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.172366841741089e-05,
+      "loss": 1.6598,
+      "step": 26640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.171193100165484e-05,
+      "loss": 1.6031,
+      "step": 26650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.170019199126505e-05,
+      "loss": 1.5951,
+      "step": 26660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.16884513890305e-05,
+      "loss": 1.6063,
+      "step": 26670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.167670919774055e-05,
+      "loss": 1.6082,
+      "step": 26680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.166496542018494e-05,
+      "loss": 1.4703,
+      "step": 26690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1653220059153764e-05,
+      "loss": 1.3815,
+      "step": 26700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.164147311743752e-05,
+      "loss": 1.3212,
+      "step": 26710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.162972459782707e-05,
+      "loss": 1.3003,
+      "step": 26720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.161797450311365e-05,
+      "loss": 1.262,
+      "step": 26730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.160622283608885e-05,
+      "loss": 1.2724,
+      "step": 26740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.159446959954468e-05,
+      "loss": 1.744,
+      "step": 26750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1582714796273485e-05,
+      "loss": 1.9589,
+      "step": 26760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1570958429068e-05,
+      "loss": 1.5921,
+      "step": 26770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.155920050072132e-05,
+      "loss": 1.7693,
+      "step": 26780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.154744101402692e-05,
+      "loss": 1.6997,
+      "step": 26790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.153567997177864e-05,
+      "loss": 1.5777,
+      "step": 26800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.15239173767707e-05,
+      "loss": 1.685,
+      "step": 26810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.151215323179767e-05,
+      "loss": 1.7791,
+      "step": 26820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.15003875396545e-05,
+      "loss": 1.5098,
+      "step": 26830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.148862030313651e-05,
+      "loss": 1.6355,
+      "step": 26840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.147685152503938e-05,
+      "loss": 1.8056,
+      "step": 26850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1465081208159166e-05,
+      "loss": 1.7474,
+      "step": 26860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.145330935529226e-05,
+      "loss": 1.6095,
+      "step": 26870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.144153596923548e-05,
+      "loss": 1.5368,
+      "step": 26880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1429761052785936e-05,
+      "loss": 1.631,
+      "step": 26890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.141798460874116e-05,
+      "loss": 1.6739,
+      "step": 26900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1406206639899e-05,
+      "loss": 1.7434,
+      "step": 26910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1394427149057706e-05,
+      "loss": 1.7871,
+      "step": 26920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.138264613901587e-05,
+      "loss": 1.7276,
+      "step": 26930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1370863612572446e-05,
+      "loss": 1.8821,
+      "step": 26940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1359079572526745e-05,
+      "loss": 1.8255,
+      "step": 26950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.134729402167845e-05,
+      "loss": 1.7216,
+      "step": 26960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1335506962827584e-05,
+      "loss": 1.4447,
+      "step": 26970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1323718398774555e-05,
+      "loss": 1.3712,
+      "step": 26980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.13119283323201e-05,
+      "loss": 1.3425,
+      "step": 26990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1300136766265334e-05,
+      "loss": 1.4896,
+      "step": 27000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1288343703411715e-05,
+      "loss": 1.4382,
+      "step": 27010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1276549146561066e-05,
+      "loss": 1.4238,
+      "step": 27020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.126475309851556e-05,
+      "loss": 1.5625,
+      "step": 27030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1252955562077725e-05,
+      "loss": 1.5133,
+      "step": 27040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.124115654005044e-05,
+      "loss": 1.4896,
+      "step": 27050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1229356035236946e-05,
+      "loss": 1.7696,
+      "step": 27060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1217554050440824e-05,
+      "loss": 1.456,
+      "step": 27070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.120575058846602e-05,
+      "loss": 1.8534,
+      "step": 27080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1193945652116805e-05,
+      "loss": 1.7251,
+      "step": 27090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1182139244197845e-05,
+      "loss": 1.7076,
+      "step": 27100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1170331367514125e-05,
+      "loss": 1.6781,
+      "step": 27110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1158522024870964e-05,
+      "loss": 1.5175,
+      "step": 27120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1146711219074055e-05,
+      "loss": 1.6627,
+      "step": 27130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.113489895292945e-05,
+      "loss": 1.6183,
+      "step": 27140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.112308522924352e-05,
+      "loss": 1.7912,
+      "step": 27150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.111127005082298e-05,
+      "loss": 1.7367,
+      "step": 27160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.109945342047491e-05,
+      "loss": 1.7409,
+      "step": 27170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.108763534100675e-05,
+      "loss": 1.8691,
+      "step": 27180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1075815815226236e-05,
+      "loss": 1.8603,
+      "step": 27190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.106399484594148e-05,
+      "loss": 1.6118,
+      "step": 27200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.105217243596094e-05,
+      "loss": 1.5487,
+      "step": 27210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1040348588093396e-05,
+      "loss": 1.6202,
+      "step": 27220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.102852330514799e-05,
+      "loss": 1.8411,
+      "step": 27230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.101669658993419e-05,
+      "loss": 1.8134,
+      "step": 27240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.100486844526182e-05,
+      "loss": 1.6982,
+      "step": 27250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.099303887394102e-05,
+      "loss": 1.7489,
+      "step": 27260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.098120787878229e-05,
+      "loss": 1.5734,
+      "step": 27270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.096937546259646e-05,
+      "loss": 1.5495,
+      "step": 27280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.09575416281947e-05,
+      "loss": 1.5152,
+      "step": 27290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.094570637838851e-05,
+      "loss": 1.9858,
+      "step": 27300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.093386971598975e-05,
+      "loss": 1.749,
+      "step": 27310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.092203164381057e-05,
+      "loss": 1.8611,
+      "step": 27320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0910192164663495e-05,
+      "loss": 1.838,
+      "step": 27330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.089835128136139e-05,
+      "loss": 1.9634,
+      "step": 27340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.08865089967174e-05,
+      "loss": 1.7006,
+      "step": 27350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.087466531354506e-05,
+      "loss": 1.8916,
+      "step": 27360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.086282023465822e-05,
+      "loss": 1.8243,
+      "step": 27370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.085097376287104e-05,
+      "loss": 1.5791,
+      "step": 27380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.083912590099804e-05,
+      "loss": 1.8089,
+      "step": 27390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0827276651854055e-05,
+      "loss": 1.7644,
+      "step": 27400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.081542601825426e-05,
+      "loss": 1.7128,
+      "step": 27410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0803574003014136e-05,
+      "loss": 1.3885,
+      "step": 27420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0791720608949516e-05,
+      "loss": 1.2528,
+      "step": 27430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0779865838876566e-05,
+      "loss": 1.3976,
+      "step": 27440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.076800969561174e-05,
+      "loss": 1.7092,
+      "step": 27450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.075615218197186e-05,
+      "loss": 1.6607,
+      "step": 27460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.074429330077405e-05,
+      "loss": 1.6261,
+      "step": 27470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.073243305483578e-05,
+      "loss": 1.6382,
+      "step": 27480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.072057144697481e-05,
+      "loss": 1.5335,
+      "step": 27490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.070870848000927e-05,
+      "loss": 1.5949,
+      "step": 27500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.069684415675756e-05,
+      "loss": 1.6318,
+      "step": 27510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.068497848003845e-05,
+      "loss": 1.5606,
+      "step": 27520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.067311145267099e-05,
+      "loss": 1.6279,
+      "step": 27530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.066124307747459e-05,
+      "loss": 1.8066,
+      "step": 27540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0649373357268954e-05,
+      "loss": 1.7916,
+      "step": 27550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0637502294874115e-05,
+      "loss": 1.7638,
+      "step": 27560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.062562989311043e-05,
+      "loss": 1.9256,
+      "step": 27570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.061375615479855e-05,
+      "loss": 1.7197,
+      "step": 27580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.060188108275947e-05,
+      "loss": 1.6927,
+      "step": 27590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0590004679814496e-05,
+      "loss": 1.634,
+      "step": 27600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0578126948785256e-05,
+      "loss": 1.5652,
+      "step": 27610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0566247892493674e-05,
+      "loss": 1.763,
+      "step": 27620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.055436751376199e-05,
+      "loss": 1.6584,
+      "step": 27630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.054248581541278e-05,
+      "loss": 1.8066,
+      "step": 27640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.053060280026892e-05,
+      "loss": 1.7909,
+      "step": 27650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.05187184711536e-05,
+      "loss": 1.8564,
+      "step": 27660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.050683283089032e-05,
+      "loss": 1.8838,
+      "step": 27670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0494945882302896e-05,
+      "loss": 1.8863,
+      "step": 27680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0483057628215445e-05,
+      "loss": 1.8339,
+      "step": 27690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.047116807145241e-05,
+      "loss": 1.7681,
+      "step": 27700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0459277214838532e-05,
+      "loss": 1.7846,
+      "step": 27710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0447385061198867e-05,
+      "loss": 1.7298,
+      "step": 27720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0435491613358775e-05,
+      "loss": 1.6944,
+      "step": 27730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0423596874143928e-05,
+      "loss": 1.7725,
+      "step": 27740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0411700846380294e-05,
+      "loss": 1.5585,
+      "step": 27750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0399803532894155e-05,
+      "loss": 1.8569,
+      "step": 27760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0387904936512107e-05,
+      "loss": 1.7329,
+      "step": 27770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.037600506006103e-05,
+      "loss": 1.6108,
+      "step": 27780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.036410390636813e-05,
+      "loss": 1.6965,
+      "step": 27790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0352201478260904e-05,
+      "loss": 1.8549,
+      "step": 27800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0340297778567152e-05,
+      "loss": 1.6424,
+      "step": 27810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.032839281011497e-05,
+      "loss": 1.5738,
+      "step": 27820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0316486575732782e-05,
+      "loss": 1.5422,
+      "step": 27830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0304579078249284e-05,
+      "loss": 1.5878,
+      "step": 27840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0292670320493483e-05,
+      "loss": 1.5613,
+      "step": 27850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0280760305294686e-05,
+      "loss": 1.5252,
+      "step": 27860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0268849035482493e-05,
+      "loss": 1.6483,
+      "step": 27870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0256936513886812e-05,
+      "loss": 1.6335,
+      "step": 27880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.024502274333785e-05,
+      "loss": 1.7424,
+      "step": 27890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0233107726666092e-05,
+      "loss": 1.8983,
+      "step": 27900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0221191466702337e-05,
+      "loss": 1.688,
+      "step": 27910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0209273966277675e-05,
+      "loss": 1.7073,
+      "step": 27920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.019735522822348e-05,
+      "loss": 1.7371,
+      "step": 27930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.018543525537144e-05,
+      "loss": 1.7472,
+      "step": 27940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0173514050553518e-05,
+      "loss": 1.698,
+      "step": 27950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0161591616601982e-05,
+      "loss": 1.6788,
+      "step": 27960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0149667956349393e-05,
+      "loss": 1.5568,
+      "step": 27970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0137743072628578e-05,
+      "loss": 1.4975,
+      "step": 27980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0125816968272692e-05,
+      "loss": 1.4826,
+      "step": 27990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0113889646115156e-05,
+      "loss": 1.6935,
+      "step": 28000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0101961108989685e-05,
+      "loss": 1.8728,
+      "step": 28010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0090031359730288e-05,
+      "loss": 2.0207,
+      "step": 28020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0078100401171256e-05,
+      "loss": 1.8829,
+      "step": 28030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0066168236147168e-05,
+      "loss": 1.6192,
+      "step": 28040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0054234867492893e-05,
+      "loss": 1.4405,
+      "step": 28050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0042300298043584e-05,
+      "loss": 1.4945,
+      "step": 28060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.003036453063469e-05,
+      "loss": 1.7648,
+      "step": 28070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0018427568101916e-05,
+      "loss": 1.7953,
+      "step": 28080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0006489413281274e-05,
+      "loss": 1.6217,
+      "step": 28090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.999455006900907e-05,
+      "loss": 1.7079,
+      "step": 28100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9982609538121852e-05,
+      "loss": 1.7493,
+      "step": 28110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9970667823456487e-05,
+      "loss": 1.738,
+      "step": 28120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9958724927850116e-05,
+      "loss": 1.7836,
+      "step": 28130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9946780854140148e-05,
+      "loss": 1.7041,
+      "step": 28140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.993483560516428e-05,
+      "loss": 1.7077,
+      "step": 28150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9922889183760495e-05,
+      "loss": 1.6833,
+      "step": 28160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.991094159276704e-05,
+      "loss": 1.7331,
+      "step": 28170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9898992835022445e-05,
+      "loss": 1.6459,
+      "step": 28180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.988704291336552e-05,
+      "loss": 1.5785,
+      "step": 28190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.987509183063536e-05,
+      "loss": 1.6738,
+      "step": 28200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.986313958967132e-05,
+      "loss": 1.4948,
+      "step": 28210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.985118619331303e-05,
+      "loss": 1.6859,
+      "step": 28220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9839231644400422e-05,
+      "loss": 1.5797,
+      "step": 28230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9827275945773658e-05,
+      "loss": 1.428,
+      "step": 28240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9815319100273208e-05,
+      "loss": 1.3978,
+      "step": 28250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9803361110739807e-05,
+      "loss": 1.405,
+      "step": 28260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.979140198001445e-05,
+      "loss": 1.4238,
+      "step": 28270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9779441710938406e-05,
+      "loss": 1.4123,
+      "step": 28280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.976748030635323e-05,
+      "loss": 1.4245,
+      "step": 28290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9755517769100732e-05,
+      "loss": 1.4344,
+      "step": 28300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9743554102022993e-05,
+      "loss": 1.4337,
+      "step": 28310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9731589307962364e-05,
+      "loss": 1.4068,
+      "step": 28320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.971962338976147e-05,
+      "loss": 1.5998,
+      "step": 28330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.970765635026319e-05,
+      "loss": 1.7379,
+      "step": 28340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9695688192310684e-05,
+      "loss": 1.7453,
+      "step": 28350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9683718918747365e-05,
+      "loss": 1.6703,
+      "step": 28360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9671748532416922e-05,
+      "loss": 1.761,
+      "step": 28370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9659777036163293e-05,
+      "loss": 1.7083,
+      "step": 28380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.96478044328307e-05,
+      "loss": 1.7116,
+      "step": 28390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9635830725263613e-05,
+      "loss": 1.6249,
+      "step": 28400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.962385591630677e-05,
+      "loss": 1.5438,
+      "step": 28410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9611880008805164e-05,
+      "loss": 1.5816,
+      "step": 28420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9599903005604067e-05,
+      "loss": 1.7611,
+      "step": 28430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9587924909548987e-05,
+      "loss": 1.719,
+      "step": 28440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.957594572348571e-05,
+      "loss": 1.687,
+      "step": 28450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9563965450260273e-05,
+      "loss": 1.643,
+      "step": 28460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.955198409271897e-05,
+      "loss": 1.6359,
+      "step": 28470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9540001653708354e-05,
+      "loss": 1.6959,
+      "step": 28480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.952801813607524e-05,
+      "loss": 1.8389,
+      "step": 28490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9516033542666705e-05,
+      "loss": 1.6751,
+      "step": 28500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9504047876330054e-05,
+      "loss": 1.5968,
+      "step": 28510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9492061139912873e-05,
+      "loss": 1.5431,
+      "step": 28520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.948007333626301e-05,
+      "loss": 1.9211,
+      "step": 28530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9468084468228524e-05,
+      "loss": 1.8268,
+      "step": 28540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9456094538657763e-05,
+      "loss": 1.6954,
+      "step": 28550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9444103550399332e-05,
+      "loss": 1.7823,
+      "step": 28560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9432111506302058e-05,
+      "loss": 1.9149,
+      "step": 28570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9420118409215036e-05,
+      "loss": 1.8603,
+      "step": 28580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.940812426198762e-05,
+      "loss": 1.5809,
+      "step": 28590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9396129067469387e-05,
+      "loss": 1.5678,
+      "step": 28600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9384132828510186e-05,
+      "loss": 1.7516,
+      "step": 28610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.937213554796011e-05,
+      "loss": 1.7715,
+      "step": 28620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9360137228669505e-05,
+      "loss": 1.7714,
+      "step": 28630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9348137873488936e-05,
+      "loss": 1.6796,
+      "step": 28640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9336137485269238e-05,
+      "loss": 1.6385,
+      "step": 28650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9324136066861496e-05,
+      "loss": 1.648,
+      "step": 28660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9312133621117023e-05,
+      "loss": 1.6922,
+      "step": 28670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.930013015088739e-05,
+      "loss": 1.7492,
+      "step": 28680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.92881256590244e-05,
+      "loss": 1.8912,
+      "step": 28690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9276120148380105e-05,
+      "loss": 1.6817,
+      "step": 28700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.92641136218068e-05,
+      "loss": 1.6823,
+      "step": 28710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9252106082157016e-05,
+      "loss": 1.5804,
+      "step": 28720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9240097532283527e-05,
+      "loss": 1.602,
+      "step": 28730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9228087975039348e-05,
+      "loss": 1.5708,
+      "step": 28740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9216077413277736e-05,
+      "loss": 1.52,
+      "step": 28750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9204065849852186e-05,
+      "loss": 1.4574,
+      "step": 28760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9192053287616423e-05,
+      "loss": 1.494,
+      "step": 28770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9180039729424413e-05,
+      "loss": 1.3531,
+      "step": 28780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9168025178130376e-05,
+      "loss": 1.316,
+      "step": 28790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.915600963658873e-05,
+      "loss": 1.295,
+      "step": 28800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9143993107654166e-05,
+      "loss": 1.3081,
+      "step": 28810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9131975594181595e-05,
+      "loss": 1.8561,
+      "step": 28820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9119957099026152e-05,
+      "loss": 1.6407,
+      "step": 28830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9107937625043217e-05,
+      "loss": 1.6726,
+      "step": 28840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9095917175088412e-05,
+      "loss": 1.9138,
+      "step": 28850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.908389575201757e-05,
+      "loss": 1.5129,
+      "step": 28860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9071873358686762e-05,
+      "loss": 1.4201,
+      "step": 28870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9059849997952294e-05,
+      "loss": 1.3606,
+      "step": 28880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.904782567267071e-05,
+      "loss": 1.3055,
+      "step": 28890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9035800385698754e-05,
+      "loss": 1.2525,
+      "step": 28900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9023774139893424e-05,
+      "loss": 1.263,
+      "step": 28910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9011746938111945e-05,
+      "loss": 1.2766,
+      "step": 28920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.899971878321176e-05,
+      "loss": 1.7647,
+      "step": 28930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.898768967805054e-05,
+      "loss": 1.7972,
+      "step": 28940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.897565962548618e-05,
+      "loss": 1.8002,
+      "step": 28950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8963628628376815e-05,
+      "loss": 1.7473,
+      "step": 28960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8951596689580786e-05,
+      "loss": 1.7605,
+      "step": 28970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.893956381195666e-05,
+      "loss": 1.742,
+      "step": 28980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.892752999836324e-05,
+      "loss": 1.7703,
+      "step": 28990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8915495251659537e-05,
+      "loss": 1.7393,
+      "step": 29000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.890345957470479e-05,
+      "loss": 1.6704,
+      "step": 29010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8891422970358467e-05,
+      "loss": 1.6492,
+      "step": 29020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8879385441480232e-05,
+      "loss": 1.578,
+      "step": 29030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8867346990930004e-05,
+      "loss": 1.593,
+      "step": 29040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8855307621567895e-05,
+      "loss": 1.6041,
+      "step": 29050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8843267336254233e-05,
+      "loss": 1.6087,
+      "step": 29060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8831226137849578e-05,
+      "loss": 1.7016,
+      "step": 29070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.88191840292147e-05,
+      "loss": 1.6694,
+      "step": 29080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8807141013210596e-05,
+      "loss": 1.8301,
+      "step": 29090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.879509709269846e-05,
+      "loss": 2.0191,
+      "step": 29100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8783052270539712e-05,
+      "loss": 1.6776,
+      "step": 29110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8771006549595985e-05,
+      "loss": 1.7866,
+      "step": 29120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8758959932729125e-05,
+      "loss": 1.9466,
+      "step": 29130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8746912422801192e-05,
+      "loss": 1.7838,
+      "step": 29140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8734864022674452e-05,
+      "loss": 1.7466,
+      "step": 29150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8722814735211395e-05,
+      "loss": 1.6114,
+      "step": 29160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8710764563274712e-05,
+      "loss": 1.4811,
+      "step": 29170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8698713509727303e-05,
+      "loss": 1.5204,
+      "step": 29180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.868666157743229e-05,
+      "loss": 1.7502,
+      "step": 29190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.867460876925298e-05,
+      "loss": 1.3763,
+      "step": 29200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8662555088052913e-05,
+      "loss": 1.3838,
+      "step": 29210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8650500536695833e-05,
+      "loss": 1.3707,
+      "step": 29220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8638445118045666e-05,
+      "loss": 1.6074,
+      "step": 29230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.862638883496657e-05,
+      "loss": 1.8818,
+      "step": 29240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.861433169032291e-05,
+      "loss": 1.5339,
+      "step": 29250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.860227368697923e-05,
+      "loss": 1.4017,
+      "step": 29260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8590214827800303e-05,
+      "loss": 1.3372,
+      "step": 29270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.85781551156511e-05,
+      "loss": 1.4301,
+      "step": 29280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8566094553396783e-05,
+      "loss": 1.9031,
+      "step": 29290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8554033143902724e-05,
+      "loss": 1.6892,
+      "step": 29300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8541970890034503e-05,
+      "loss": 1.4832,
+      "step": 29310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8529907794657883e-05,
+      "loss": 1.3524,
+      "step": 29320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8517843860638855e-05,
+      "loss": 1.3083,
+      "step": 29330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8505779090843575e-05,
+      "loss": 1.2614,
+      "step": 29340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8493713488138433e-05,
+      "loss": 1.2801,
+      "step": 29350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8481647055389982e-05,
+      "loss": 1.6889,
+      "step": 29360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.846957979546499e-05,
+      "loss": 1.7342,
+      "step": 29370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8457511711230435e-05,
+      "loss": 1.7363,
+      "step": 29380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8445442805553458e-05,
+      "loss": 1.8042,
+      "step": 29390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8433373081301418e-05,
+      "loss": 1.6996,
+      "step": 29400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.842130254134187e-05,
+      "loss": 1.7642,
+      "step": 29410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8409231188542562e-05,
+      "loss": 1.7229,
+      "step": 29420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8397159025771412e-05,
+      "loss": 1.8722,
+      "step": 29430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.838508605589656e-05,
+      "loss": 1.8483,
+      "step": 29440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.837301228178632e-05,
+      "loss": 1.8254,
+      "step": 29450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8360937706309203e-05,
+      "loss": 1.8034,
+      "step": 29460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.834886233233392e-05,
+      "loss": 1.887,
+      "step": 29470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8336786162729355e-05,
+      "loss": 1.9504,
+      "step": 29480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.832470920036459e-05,
+      "loss": 1.8102,
+      "step": 29490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8312631448108888e-05,
+      "loss": 1.6279,
+      "step": 29500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.830055290883172e-05,
+      "loss": 1.8089,
+      "step": 29510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8288473585402703e-05,
+      "loss": 1.6192,
+      "step": 29520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.827639348069169e-05,
+      "loss": 1.5336,
+      "step": 29530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8264312597568675e-05,
+      "loss": 1.8823,
+      "step": 29540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.825223093890389e-05,
+      "loss": 1.748,
+      "step": 29550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8240148507567687e-05,
+      "loss": 1.7695,
+      "step": 29560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8228065306430644e-05,
+      "loss": 1.8945,
+      "step": 29570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8215981338363523e-05,
+      "loss": 1.889,
+      "step": 29580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.820389660623724e-05,
+      "loss": 1.8328,
+      "step": 29590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8191811112922916e-05,
+      "loss": 1.7398,
+      "step": 29600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8179724861291852e-05,
+      "loss": 1.7906,
+      "step": 29610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8167637854215518e-05,
+      "loss": 1.5268,
+      "step": 29620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8155550094565562e-05,
+      "loss": 1.7252,
+      "step": 29630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8143461585213832e-05,
+      "loss": 1.7338,
+      "step": 29640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8131372329032333e-05,
+      "loss": 1.707,
+      "step": 29650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8119282328893255e-05,
+      "loss": 1.6499,
+      "step": 29660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.810719158766896e-05,
+      "loss": 1.7669,
+      "step": 29670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8095100108231998e-05,
+      "loss": 1.5605,
+      "step": 29680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8083007893455077e-05,
+      "loss": 1.5086,
+      "step": 29690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.807091494621109e-05,
+      "loss": 1.4872,
+      "step": 29700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.805882126937312e-05,
+      "loss": 1.46,
+      "step": 29710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8046726865814383e-05,
+      "loss": 1.4358,
+      "step": 29720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8034631738408296e-05,
+      "loss": 1.5663,
+      "step": 29730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8022535890028462e-05,
+      "loss": 1.8098,
+      "step": 29740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.801043932354861e-05,
+      "loss": 1.8617,
+      "step": 29750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7998342041842683e-05,
+      "loss": 1.836,
+      "step": 29760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.798624404778477e-05,
+      "loss": 1.7425,
+      "step": 29770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7974145344249142e-05,
+      "loss": 1.7579,
+      "step": 29780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7962045934110225e-05,
+      "loss": 1.7358,
+      "step": 29790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7949945820242628e-05,
+      "loss": 1.89,
+      "step": 29800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7937845005521125e-05,
+      "loss": 1.8189,
+      "step": 29810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7925743492820632e-05,
+      "loss": 1.6827,
+      "step": 29820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.791364128501627e-05,
+      "loss": 1.807,
+      "step": 29830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7901538384983296e-05,
+      "loss": 1.6965,
+      "step": 29840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.788943479559713e-05,
+      "loss": 2.0146,
+      "step": 29850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7877330519733386e-05,
+      "loss": 1.9342,
+      "step": 29860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.786522556026781e-05,
+      "loss": 1.8879,
+      "step": 29870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.785311992007633e-05,
+      "loss": 1.926,
+      "step": 29880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7841013602035022e-05,
+      "loss": 1.7062,
+      "step": 29890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7828906609020128e-05,
+      "loss": 1.4731,
+      "step": 29900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7816798943908057e-05,
+      "loss": 1.3973,
+      "step": 29910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7804690609575362e-05,
+      "loss": 1.6483,
+      "step": 29920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7792581608898772e-05,
+      "loss": 1.5568,
+      "step": 29930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7780471944755166e-05,
+      "loss": 1.7068,
+      "step": 29940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7768361620021578e-05,
+      "loss": 1.7387,
+      "step": 29950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.775625063757521e-05,
+      "loss": 1.6133,
+      "step": 29960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7744139000293413e-05,
+      "loss": 1.5842,
+      "step": 29970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.773202671105369e-05,
+      "loss": 1.5144,
+      "step": 29980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7719913772733698e-05,
+      "loss": 1.7253,
+      "step": 29990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7707800188211252e-05,
+      "loss": 2.3502,
+      "step": 30000
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 1.6172226667404175,
+      "eval_runtime": 82.1504,
+      "eval_samples_per_second": 36.518,
+      "eval_steps_per_second": 4.565,
+      "step": 30000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7695685960364338e-05,
+      "loss": 1.8308,
+      "step": 30010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.768357109207106e-05,
+      "loss": 1.8016,
+      "step": 30020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7671455586209695e-05,
+      "loss": 1.7465,
+      "step": 30030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.765933944565868e-05,
+      "loss": 1.6972,
+      "step": 30040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7647222673296575e-05,
+      "loss": 1.5448,
+      "step": 30050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7635105272002116e-05,
+      "loss": 1.8143,
+      "step": 30060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7622987244654185e-05,
+      "loss": 1.8477,
+      "step": 30070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.761086859413179e-05,
+      "loss": 1.8495,
+      "step": 30080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7598749323314117e-05,
+      "loss": 1.9997,
+      "step": 30090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.758662943508048e-05,
+      "loss": 1.6225,
+      "step": 30100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.757450893231035e-05,
+      "loss": 1.7279,
+      "step": 30110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.756238781788334e-05,
+      "loss": 1.6533,
+      "step": 30120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7550266094679205e-05,
+      "loss": 1.5769,
+      "step": 30130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7538143765577852e-05,
+      "loss": 1.6187,
+      "step": 30140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.752602083345932e-05,
+      "loss": 1.6829,
+      "step": 30150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7513897301203808e-05,
+      "loss": 1.8075,
+      "step": 30160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7501773171691648e-05,
+      "loss": 1.8943,
+      "step": 30170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.74896484478033e-05,
+      "loss": 1.5768,
+      "step": 30180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.747752313241939e-05,
+      "loss": 1.4907,
+      "step": 30190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7465397228420688e-05,
+      "loss": 1.5801,
+      "step": 30200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.745327073868807e-05,
+      "loss": 1.7734,
+      "step": 30210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.744114366610257e-05,
+      "loss": 1.6839,
+      "step": 30220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7429016013545376e-05,
+      "loss": 1.6181,
+      "step": 30230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7416887783897798e-05,
+      "loss": 1.6177,
+      "step": 30240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.740475898004127e-05,
+      "loss": 1.6521,
+      "step": 30250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7392629604857393e-05,
+      "loss": 1.7239,
+      "step": 30260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7380499661227882e-05,
+      "loss": 1.6157,
+      "step": 30270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7368369152034583e-05,
+      "loss": 1.6048,
+      "step": 30280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7356238080159503e-05,
+      "loss": 1.7573,
+      "step": 30290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7344106448484763e-05,
+      "loss": 1.6857,
+      "step": 30300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7331974259892606e-05,
+      "loss": 1.5751,
+      "step": 30310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7319841517265438e-05,
+      "loss": 1.7957,
+      "step": 30320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7307708223485762e-05,
+      "loss": 1.7801,
+      "step": 30330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7295574381436252e-05,
+      "loss": 1.7531,
+      "step": 30340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7283439993999673e-05,
+      "loss": 1.6598,
+      "step": 30350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7271305064058938e-05,
+      "loss": 1.6426,
+      "step": 30360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7259169594497098e-05,
+      "loss": 1.5347,
+      "step": 30370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7247033588197306e-05,
+      "loss": 1.7294,
+      "step": 30380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.723489704804287e-05,
+      "loss": 1.7253,
+      "step": 30390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.722275997691721e-05,
+      "loss": 1.6307,
+      "step": 30400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7210622377703877e-05,
+      "loss": 1.6531,
+      "step": 30410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7198484253286538e-05,
+      "loss": 1.5858,
+      "step": 30420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7186345606548995e-05,
+      "loss": 1.5903,
+      "step": 30430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.717420644037518e-05,
+      "loss": 1.5994,
+      "step": 30440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7162066757649134e-05,
+      "loss": 1.5711,
+      "step": 30450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.714992656125503e-05,
+      "loss": 1.557,
+      "step": 30460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7137785854077153e-05,
+      "loss": 1.5776,
+      "step": 30470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.712564463899992e-05,
+      "loss": 1.5899,
+      "step": 30480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7113502918907863e-05,
+      "loss": 1.7311,
+      "step": 30490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7101360696685644e-05,
+      "loss": 1.6193,
+      "step": 30500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7089217975218022e-05,
+      "loss": 1.5175,
+      "step": 30510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7077074757389896e-05,
+      "loss": 1.496,
+      "step": 30520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7064931046086277e-05,
+      "loss": 1.5115,
+      "step": 30530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.705278684419229e-05,
+      "loss": 1.4888,
+      "step": 30540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7040642154593177e-05,
+      "loss": 1.7007,
+      "step": 30550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7028496980174295e-05,
+      "loss": 1.7672,
+      "step": 30560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7016351323821126e-05,
+      "loss": 1.6953,
+      "step": 30570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7004205188419252e-05,
+      "loss": 1.4908,
+      "step": 30580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6992058576854378e-05,
+      "loss": 1.4833,
+      "step": 30590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.697991149201232e-05,
+      "loss": 1.4356,
+      "step": 30600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.696776393677901e-05,
+      "loss": 1.4056,
+      "step": 30610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6955615914040485e-05,
+      "loss": 1.3645,
+      "step": 30620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6943467426682893e-05,
+      "loss": 1.3449,
+      "step": 30630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6931318477592494e-05,
+      "loss": 1.5329,
+      "step": 30640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6919169069655663e-05,
+      "loss": 2.0459,
+      "step": 30650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.690701920575889e-05,
+      "loss": 1.6377,
+      "step": 30660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6894868888788742e-05,
+      "loss": 1.729,
+      "step": 30670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6882718121631934e-05,
+      "loss": 1.7079,
+      "step": 30680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6870566907175255e-05,
+      "loss": 1.7092,
+      "step": 30690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.685841524830563e-05,
+      "loss": 1.7776,
+      "step": 30700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6846263147910055e-05,
+      "loss": 1.8686,
+      "step": 30710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6834110608875667e-05,
+      "loss": 1.7686,
+      "step": 30720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6821957634089678e-05,
+      "loss": 1.6347,
+      "step": 30730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.680980422643942e-05,
+      "loss": 1.5857,
+      "step": 30740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.679765038881233e-05,
+      "loss": 1.6671,
+      "step": 30750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.678549612409594e-05,
+      "loss": 1.6739,
+      "step": 30760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6773341435177866e-05,
+      "loss": 1.5915,
+      "step": 30770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.676118632494586e-05,
+      "loss": 1.5745,
+      "step": 30780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.674903079628775e-05,
+      "loss": 1.7472,
+      "step": 30790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6736874852091483e-05,
+      "loss": 1.8191,
+      "step": 30800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.672471849524507e-05,
+      "loss": 1.7328,
+      "step": 30810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.671256172863666e-05,
+      "loss": 1.7156,
+      "step": 30820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6700404555154474e-05,
+      "loss": 1.8492,
+      "step": 30830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6688246977686836e-05,
+      "loss": 1.8925,
+      "step": 30840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6676088999122173e-05,
+      "loss": 1.5997,
+      "step": 30850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6663930622348997e-05,
+      "loss": 1.58,
+      "step": 30860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.665177185025592e-05,
+      "loss": 1.7138,
+      "step": 30870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6639612685731648e-05,
+      "loss": 1.7152,
+      "step": 30880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6627453131664977e-05,
+      "loss": 1.7096,
+      "step": 30890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6615293190944794e-05,
+      "loss": 1.6762,
+      "step": 30900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.660313286646009e-05,
+      "loss": 1.9552,
+      "step": 30910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6590972161099937e-05,
+      "loss": 2.2446,
+      "step": 30920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6578811077753495e-05,
+      "loss": 1.4388,
+      "step": 30930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6566649619310025e-05,
+      "loss": 1.5694,
+      "step": 30940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.655448778865886e-05,
+      "loss": 1.7624,
+      "step": 30950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6542325588689443e-05,
+      "loss": 1.5322,
+      "step": 30960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6530163022291282e-05,
+      "loss": 1.5382,
+      "step": 30970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.651800009235399e-05,
+      "loss": 1.5375,
+      "step": 30980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6505836801767264e-05,
+      "loss": 1.5104,
+      "step": 30990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6493673153420868e-05,
+      "loss": 1.5047,
+      "step": 31000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6481509150204676e-05,
+      "loss": 1.5781,
+      "step": 31010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.646934479500864e-05,
+      "loss": 1.7457,
+      "step": 31020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6457180090722782e-05,
+      "loss": 1.6599,
+      "step": 31030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.644501504023722e-05,
+      "loss": 1.5761,
+      "step": 31040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6432849646442148e-05,
+      "loss": 2.4015,
+      "step": 31050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.642068391222785e-05,
+      "loss": 2.4452,
+      "step": 31060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6408517840484674e-05,
+      "loss": 1.6371,
+      "step": 31070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6396351434103077e-05,
+      "loss": 1.6392,
+      "step": 31080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.638418469597357e-05,
+      "loss": 1.6121,
+      "step": 31090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6372017628986735e-05,
+      "loss": 1.5775,
+      "step": 31100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6359850236033267e-05,
+      "loss": 1.6328,
+      "step": 31110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6347682520003914e-05,
+      "loss": 1.6165,
+      "step": 31120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6335514483789502e-05,
+      "loss": 1.5576,
+      "step": 31130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.632334613028094e-05,
+      "loss": 1.9131,
+      "step": 31140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6311177462369208e-05,
+      "loss": 1.7581,
+      "step": 31150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.629900848294537e-05,
+      "loss": 1.6805,
+      "step": 31160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6286839194900544e-05,
+      "loss": 1.6619,
+      "step": 31170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6274669601125933e-05,
+      "loss": 1.7099,
+      "step": 31180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6262499704512822e-05,
+      "loss": 1.6765,
+      "step": 31190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6250329507952554e-05,
+      "loss": 1.7046,
+      "step": 31200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.623815901433655e-05,
+      "loss": 1.6142,
+      "step": 31210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.62259882265563e-05,
+      "loss": 1.6231,
+      "step": 31220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6213817147503357e-05,
+      "loss": 1.5973,
+      "step": 31230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6201645780069355e-05,
+      "loss": 1.51,
+      "step": 31240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6189474127145998e-05,
+      "loss": 1.4731,
+      "step": 31250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6177302191625043e-05,
+      "loss": 1.5109,
+      "step": 31260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.616512997639832e-05,
+      "loss": 1.5098,
+      "step": 31270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6152957484357733e-05,
+      "loss": 1.6014,
+      "step": 31280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.614078471839525e-05,
+      "loss": 1.6762,
+      "step": 31290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6128611681402894e-05,
+      "loss": 1.5151,
+      "step": 31300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.611643837627276e-05,
+      "loss": 1.792,
+      "step": 31310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.610426480589701e-05,
+      "loss": 1.8181,
+      "step": 31320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.609209097316786e-05,
+      "loss": 1.8565,
+      "step": 31330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.607991688097759e-05,
+      "loss": 1.7751,
+      "step": 31340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6067742532218552e-05,
+      "loss": 1.597,
+      "step": 31350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6055567929783155e-05,
+      "loss": 1.7352,
+      "step": 31360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.604339307656385e-05,
+      "loss": 1.6425,
+      "step": 31370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.603121797545317e-05,
+      "loss": 1.5603,
+      "step": 31380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6019042629343705e-05,
+      "loss": 1.9553,
+      "step": 31390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.600686704112809e-05,
+      "loss": 1.7989,
+      "step": 31400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.599469121369903e-05,
+      "loss": 1.7747,
+      "step": 31410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.598251514994929e-05,
+      "loss": 1.6601,
+      "step": 31420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5970338852771665e-05,
+      "loss": 1.6134,
+      "step": 31430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5958162325059025e-05,
+      "loss": 1.621,
+      "step": 31440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.594598556970432e-05,
+      "loss": 1.705,
+      "step": 31450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.593380858960049e-05,
+      "loss": 1.7126,
+      "step": 31460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5921631387640587e-05,
+      "loss": 1.6731,
+      "step": 31470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5909453966717694e-05,
+      "loss": 1.6747,
+      "step": 31480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5897276329724947e-05,
+      "loss": 1.6363,
+      "step": 31490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5885098479555525e-05,
+      "loss": 1.6252,
+      "step": 31500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.587292041910267e-05,
+      "loss": 1.6745,
+      "step": 31510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5860742151259677e-05,
+      "loss": 1.6305,
+      "step": 31520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5848563678919862e-05,
+      "loss": 1.5926,
+      "step": 31530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.583638500497663e-05,
+      "loss": 1.6396,
+      "step": 31540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5824206132323408e-05,
+      "loss": 1.6165,
+      "step": 31550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5812027063853677e-05,
+      "loss": 1.6309,
+      "step": 31560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5799847802460964e-05,
+      "loss": 1.6165,
+      "step": 31570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5787668351038835e-05,
+      "loss": 1.5766,
+      "step": 31580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5775488712480928e-05,
+      "loss": 1.608,
+      "step": 31590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5763308889680883e-05,
+      "loss": 1.6832,
+      "step": 31600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5751128885532412e-05,
+      "loss": 1.6309,
+      "step": 31610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.573894870292927e-05,
+      "loss": 1.6565,
+      "step": 31620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.572676834476525e-05,
+      "loss": 1.8253,
+      "step": 31630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5714587813934177e-05,
+      "loss": 1.822,
+      "step": 31640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5702407113329934e-05,
+      "loss": 1.738,
+      "step": 31650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5690226245846422e-05,
+      "loss": 1.806,
+      "step": 31660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5678045214377612e-05,
+      "loss": 1.7034,
+      "step": 31670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5665864021817488e-05,
+      "loss": 1.615,
+      "step": 31680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.565368267106008e-05,
+      "loss": 1.751,
+      "step": 31690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.564150116499946e-05,
+      "loss": 1.5873,
+      "step": 31700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5629319506529742e-05,
+      "loss": 1.6959,
+      "step": 31710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5617137698545063e-05,
+      "loss": 1.8869,
+      "step": 31720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.560495574393959e-05,
+      "loss": 1.7533,
+      "step": 31730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.559277364560755e-05,
+      "loss": 1.8836,
+      "step": 31740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.558059140644319e-05,
+      "loss": 1.6901,
+      "step": 31750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5568409029340773e-05,
+      "loss": 1.4849,
+      "step": 31760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.555622651719463e-05,
+      "loss": 1.662,
+      "step": 31770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5544043872899103e-05,
+      "loss": 1.9107,
+      "step": 31780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5531861099348557e-05,
+      "loss": 1.6115,
+      "step": 31790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5519678199437413e-05,
+      "loss": 1.6896,
+      "step": 31800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5507495176060103e-05,
+      "loss": 1.7752,
+      "step": 31810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5495312032111096e-05,
+      "loss": 1.7192,
+      "step": 31820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5483128770484882e-05,
+      "loss": 1.918,
+      "step": 31830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.547094539407599e-05,
+      "loss": 1.8316,
+      "step": 31840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5458761905778962e-05,
+      "loss": 1.6121,
+      "step": 31850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5446578308488384e-05,
+      "loss": 1.7559,
+      "step": 31860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.543439460509885e-05,
+      "loss": 1.7316,
+      "step": 31870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5422210798505003e-05,
+      "loss": 1.5897,
+      "step": 31880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5410026891601487e-05,
+      "loss": 1.5622,
+      "step": 31890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5397842887282968e-05,
+      "loss": 1.7648,
+      "step": 31900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5385658788444168e-05,
+      "loss": 1.7416,
+      "step": 31910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5373474597979785e-05,
+      "loss": 1.6322,
+      "step": 31920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5361290318784577e-05,
+      "loss": 2.0612,
+      "step": 31930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5349105953753306e-05,
+      "loss": 2.408,
+      "step": 31940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5336921505780763e-05,
+      "loss": 2.0981,
+      "step": 31950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5324736977761742e-05,
+      "loss": 1.9849,
+      "step": 31960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5312552372591075e-05,
+      "loss": 1.7358,
+      "step": 31970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5300367693163607e-05,
+      "loss": 1.7846,
+      "step": 31980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.528818294237419e-05,
+      "loss": 1.5975,
+      "step": 31990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.52759981231177e-05,
+      "loss": 1.6959,
+      "step": 32000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5263813238289036e-05,
+      "loss": 1.6827,
+      "step": 32010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5251628290783108e-05,
+      "loss": 1.633,
+      "step": 32020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5239443283494835e-05,
+      "loss": 1.7342,
+      "step": 32030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5227258219319154e-05,
+      "loss": 1.7013,
+      "step": 32040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5215073101151033e-05,
+      "loss": 1.6876,
+      "step": 32050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.520288793188541e-05,
+      "loss": 1.7396,
+      "step": 32060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5190702714417276e-05,
+      "loss": 1.8015,
+      "step": 32070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.517851745164162e-05,
+      "loss": 1.7347,
+      "step": 32080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5166332146453437e-05,
+      "loss": 1.6833,
+      "step": 32090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5154146801747736e-05,
+      "loss": 1.84,
+      "step": 32100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5141961420419542e-05,
+      "loss": 1.5983,
+      "step": 32110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5129776005363865e-05,
+      "loss": 1.7959,
+      "step": 32120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5117590559475752e-05,
+      "loss": 2.1036,
+      "step": 32130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5105405085650246e-05,
+      "loss": 1.8333,
+      "step": 32140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.509321958678239e-05,
+      "loss": 1.7488,
+      "step": 32150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5081034065767245e-05,
+      "loss": 1.5693,
+      "step": 32160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5068848525499865e-05,
+      "loss": 1.7141,
+      "step": 32170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5056662968875316e-05,
+      "loss": 1.8203,
+      "step": 32180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5044477398788672e-05,
+      "loss": 1.8285,
+      "step": 32190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5032291818134995e-05,
+      "loss": 1.6109,
+      "step": 32200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5020106229809377e-05,
+      "loss": 1.7752,
+      "step": 32210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5007920636706867e-05,
+      "loss": 1.5972,
+      "step": 32220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4995735041722567e-05,
+      "loss": 1.6089,
+      "step": 32230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4983549447751533e-05,
+      "loss": 1.7108,
+      "step": 32240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4971363857688864e-05,
+      "loss": 1.7389,
+      "step": 32250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4959178274429625e-05,
+      "loss": 1.6557,
+      "step": 32260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4946992700868886e-05,
+      "loss": 1.6699,
+      "step": 32270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4934807139901728e-05,
+      "loss": 1.536,
+      "step": 32280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4922621594423224e-05,
+      "loss": 1.7225,
+      "step": 32290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.491043606732842e-05,
+      "loss": 1.4861,
+      "step": 32300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.48982505615124e-05,
+      "loss": 1.4572,
+      "step": 32310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4886065079870202e-05,
+      "loss": 1.4441,
+      "step": 32320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4873879625296894e-05,
+      "loss": 1.4619,
+      "step": 32330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.486169420068751e-05,
+      "loss": 1.6902,
+      "step": 32340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.484950880893709e-05,
+      "loss": 1.9447,
+      "step": 32350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.483732345294066e-05,
+      "loss": 1.6986,
+      "step": 32360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.482513813559325e-05,
+      "loss": 1.7045,
+      "step": 32370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.481295285978985e-05,
+      "loss": 1.6438,
+      "step": 32380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.480076762842549e-05,
+      "loss": 1.5444,
+      "step": 32390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.478858244439515e-05,
+      "loss": 1.5298,
+      "step": 32400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4776397310593792e-05,
+      "loss": 1.5227,
+      "step": 32410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4764212229916413e-05,
+      "loss": 1.5614,
+      "step": 32420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4752027205257957e-05,
+      "loss": 1.6088,
+      "step": 32430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.473984223951335e-05,
+      "loss": 1.6516,
+      "step": 32440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4727657335577546e-05,
+      "loss": 1.7938,
+      "step": 32450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4715472496345445e-05,
+      "loss": 1.7171,
+      "step": 32460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.470328772471194e-05,
+      "loss": 1.6899,
+      "step": 32470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4691103023571927e-05,
+      "loss": 1.6393,
+      "step": 32480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4678918395820256e-05,
+      "loss": 1.5695,
+      "step": 32490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4666733844351786e-05,
+      "loss": 1.6906,
+      "step": 32500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.465454937206134e-05,
+      "loss": 1.767,
+      "step": 32510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4642364981843738e-05,
+      "loss": 1.608,
+      "step": 32520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4630180676593756e-05,
+      "loss": 1.605,
+      "step": 32530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4617996459206183e-05,
+      "loss": 1.7836,
+      "step": 32540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.460581233257575e-05,
+      "loss": 1.6599,
+      "step": 32550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4593628299597208e-05,
+      "loss": 2.0553,
+      "step": 32560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4581444363165253e-05,
+      "loss": 1.7034,
+      "step": 32570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4569260526174552e-05,
+      "loss": 1.7329,
+      "step": 32580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4557076791519794e-05,
+      "loss": 1.5378,
+      "step": 32590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.45448931620956e-05,
+      "loss": 1.6747,
+      "step": 32600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4532709640796574e-05,
+      "loss": 1.7547,
+      "step": 32610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4520526230517314e-05,
+      "loss": 1.5435,
+      "step": 32620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4508342934152368e-05,
+      "loss": 1.9551,
+      "step": 32630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4496159754596274e-05,
+      "loss": 1.7664,
+      "step": 32640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4483976694743537e-05,
+      "loss": 1.5342,
+      "step": 32650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4471793757488625e-05,
+      "loss": 1.7877,
+      "step": 32660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4459610945725987e-05,
+      "loss": 1.835,
+      "step": 32670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4447428262350043e-05,
+      "loss": 1.587,
+      "step": 32680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4435245710255177e-05,
+      "loss": 1.9906,
+      "step": 32690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4423063292335734e-05,
+      "loss": 1.7502,
+      "step": 32700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.441088101148606e-05,
+      "loss": 1.6386,
+      "step": 32710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4398698870600428e-05,
+      "loss": 1.5704,
+      "step": 32720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4386516872573086e-05,
+      "loss": 1.5277,
+      "step": 32730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.437433502029828e-05,
+      "loss": 1.5251,
+      "step": 32740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.436215331667019e-05,
+      "loss": 1.5114,
+      "step": 32750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4349971764582953e-05,
+      "loss": 1.4764,
+      "step": 32760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.433779036693071e-05,
+      "loss": 1.6542,
+      "step": 32770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4325609126607517e-05,
+      "loss": 1.719,
+      "step": 32780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.431342804650745e-05,
+      "loss": 1.8886,
+      "step": 32790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.430124712952448e-05,
+      "loss": 1.7042,
+      "step": 32800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.428906637855259e-05,
+      "loss": 1.6884,
+      "step": 32810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4276885796485704e-05,
+      "loss": 1.7696,
+      "step": 32820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.426470538621772e-05,
+      "loss": 1.7448,
+      "step": 32830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4252525150642452e-05,
+      "loss": 1.7439,
+      "step": 32840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4240345092653745e-05,
+      "loss": 1.6832,
+      "step": 32850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4228165215145336e-05,
+      "loss": 1.5707,
+      "step": 32860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4215985521010937e-05,
+      "loss": 1.5439,
+      "step": 32870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4203806013144247e-05,
+      "loss": 1.5626,
+      "step": 32880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.419162669443889e-05,
+      "loss": 1.96,
+      "step": 32890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4179447567788434e-05,
+      "loss": 1.6574,
+      "step": 32900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4167268636086446e-05,
+      "loss": 1.739,
+      "step": 32910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4155089902226412e-05,
+      "loss": 1.5862,
+      "step": 32920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.414291136910176e-05,
+      "loss": 1.418,
+      "step": 32930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4130733039605922e-05,
+      "loss": 1.7893,
+      "step": 32940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4118554916632226e-05,
+      "loss": 1.7871,
+      "step": 32950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4106377003073978e-05,
+      "loss": 1.8599,
+      "step": 32960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4094199301824437e-05,
+      "loss": 1.5863,
+      "step": 32970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4082021815776813e-05,
+      "loss": 1.4855,
+      "step": 32980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.406984454782423e-05,
+      "loss": 1.5832,
+      "step": 32990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4057667500859816e-05,
+      "loss": 1.8194,
+      "step": 33000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4045490677776596e-05,
+      "loss": 1.7327,
+      "step": 33010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4033314081467582e-05,
+      "loss": 2.0048,
+      "step": 33020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4021137714825708e-05,
+      "loss": 1.7481,
+      "step": 33030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4008961580743837e-05,
+      "loss": 1.7322,
+      "step": 33040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3996785682114836e-05,
+      "loss": 1.7494,
+      "step": 33050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.398461002183146e-05,
+      "loss": 1.7358,
+      "step": 33060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.397243460278641e-05,
+      "loss": 1.6446,
+      "step": 33070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.396025942787238e-05,
+      "loss": 1.3793,
+      "step": 33080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.394808449998195e-05,
+      "loss": 1.5024,
+      "step": 33090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.393590982200766e-05,
+      "loss": 1.7198,
+      "step": 33100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3923735396842012e-05,
+      "loss": 1.6659,
+      "step": 33110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3911561227377412e-05,
+      "loss": 1.5172,
+      "step": 33120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3899387316506232e-05,
+      "loss": 1.469,
+      "step": 33130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3887213667120774e-05,
+      "loss": 1.7945,
+      "step": 33140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3875040282113278e-05,
+      "loss": 1.8832,
+      "step": 33150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.386286716437591e-05,
+      "loss": 1.7948,
+      "step": 33160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.38506943168008e-05,
+      "loss": 1.5751,
+      "step": 33170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.383852174227999e-05,
+      "loss": 1.5009,
+      "step": 33180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.382634944370545e-05,
+      "loss": 1.4573,
+      "step": 33190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3814177423969125e-05,
+      "loss": 1.5138,
+      "step": 33200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3802005685962853e-05,
+      "loss": 1.5747,
+      "step": 33210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3789834232578405e-05,
+      "loss": 1.5056,
+      "step": 33220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3777663066707527e-05,
+      "loss": 1.7035,
+      "step": 33230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3765492191241848e-05,
+      "loss": 1.6348,
+      "step": 33240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3753321609072962e-05,
+      "loss": 1.8221,
+      "step": 33250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3741151323092374e-05,
+      "loss": 1.6239,
+      "step": 33260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3728981336191518e-05,
+      "loss": 1.6051,
+      "step": 33270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3716811651261776e-05,
+      "loss": 1.6145,
+      "step": 33280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3704642271194432e-05,
+      "loss": 1.6199,
+      "step": 33290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3692473198880717e-05,
+      "loss": 1.5711,
+      "step": 33300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.368030443721179e-05,
+      "loss": 1.603,
+      "step": 33310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3668135989078723e-05,
+      "loss": 1.4979,
+      "step": 33320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3655967857372514e-05,
+      "loss": 1.6174,
+      "step": 33330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3643800044984106e-05,
+      "loss": 1.8854,
+      "step": 33340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.363163255480434e-05,
+      "loss": 1.6554,
+      "step": 33350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.361946538972399e-05,
+      "loss": 1.8907,
+      "step": 33360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3607298552633773e-05,
+      "loss": 1.9422,
+      "step": 33370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.359513204642429e-05,
+      "loss": 1.8492,
+      "step": 33380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3582965873986086e-05,
+      "loss": 1.7607,
+      "step": 33390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.357080003820964e-05,
+      "loss": 1.7617,
+      "step": 33400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3558634541985322e-05,
+      "loss": 1.6094,
+      "step": 33410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3546469388203427e-05,
+      "loss": 1.5686,
+      "step": 33420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3534304579754195e-05,
+      "loss": 1.6858,
+      "step": 33430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3522140119527752e-05,
+      "loss": 1.6803,
+      "step": 33440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.350997601041416e-05,
+      "loss": 1.7798,
+      "step": 33450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3497812255303388e-05,
+      "loss": 1.7511,
+      "step": 33460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3485648857085323e-05,
+      "loss": 1.9284,
+      "step": 33470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3473485818649778e-05,
+      "loss": 1.689,
+      "step": 33480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.346132314288647e-05,
+      "loss": 1.6359,
+      "step": 33490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.344916083268502e-05,
+      "loss": 1.6527,
+      "step": 33500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3436998890934987e-05,
+      "loss": 1.542,
+      "step": 33510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3424837320525827e-05,
+      "loss": 1.3756,
+      "step": 33520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3412676124346894e-05,
+      "loss": 1.3429,
+      "step": 33530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3400515305287495e-05,
+      "loss": 1.3121,
+      "step": 33540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.33883548662368e-05,
+      "loss": 1.3005,
+      "step": 33550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.337619481008391e-05,
+      "loss": 1.6357,
+      "step": 33560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3364035139717855e-05,
+      "loss": 1.6519,
+      "step": 33570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3351875858027537e-05,
+      "loss": 1.7905,
+      "step": 33580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3339716967901777e-05,
+      "loss": 1.6664,
+      "step": 33590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3327558472229323e-05,
+      "loss": 1.6835,
+      "step": 33600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.331540037389881e-05,
+      "loss": 1.5074,
+      "step": 33610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.330324267579877e-05,
+      "loss": 1.5436,
+      "step": 33620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3291085380817672e-05,
+      "loss": 1.5462,
+      "step": 33630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3278928491843862e-05,
+      "loss": 1.5847,
+      "step": 33640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3266772011765588e-05,
+      "loss": 1.7635,
+      "step": 33650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.325461594347103e-05,
+      "loss": 1.7619,
+      "step": 33660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3242460289848238e-05,
+      "loss": 1.6929,
+      "step": 33670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.323030505378517e-05,
+      "loss": 1.6625,
+      "step": 33680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3218150238169707e-05,
+      "loss": 1.6484,
+      "step": 33690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3205995845889596e-05,
+      "loss": 1.835,
+      "step": 33700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.319384187983252e-05,
+      "loss": 1.6508,
+      "step": 33710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.318168834288603e-05,
+      "loss": 1.8109,
+      "step": 33720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3169535237937584e-05,
+      "loss": 1.6658,
+      "step": 33730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3157382567874558e-05,
+      "loss": 1.8626,
+      "step": 33740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3145230335584186e-05,
+      "loss": 1.8833,
+      "step": 33750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3133078543953632e-05,
+      "loss": 1.6479,
+      "step": 33760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3120927195869947e-05,
+      "loss": 2.0983,
+      "step": 33770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3108776294220054e-05,
+      "loss": 1.7872,
+      "step": 33780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.30966258418908e-05,
+      "loss": 1.6232,
+      "step": 33790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3084475841768913e-05,
+      "loss": 1.6766,
+      "step": 33800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.307232629674102e-05,
+      "loss": 1.7139,
+      "step": 33810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.306017720969361e-05,
+      "loss": 1.6988,
+      "step": 33820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.304802858351312e-05,
+      "loss": 1.4856,
+      "step": 33830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3035880421085827e-05,
+      "loss": 1.4142,
+      "step": 33840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3023732725297903e-05,
+      "loss": 1.6597,
+      "step": 33850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3011585499035445e-05,
+      "loss": 1.7434,
+      "step": 33860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2999438745184407e-05,
+      "loss": 1.6781,
+      "step": 33870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.298729246663062e-05,
+      "loss": 1.7279,
+      "step": 33880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2975146666259855e-05,
+      "loss": 1.7297,
+      "step": 33890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.296300134695771e-05,
+      "loss": 1.6467,
+      "step": 33900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2950856511609696e-05,
+      "loss": 1.7753,
+      "step": 33910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2938712163101212e-05,
+      "loss": 1.7859,
+      "step": 33920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2926568304317536e-05,
+      "loss": 1.7883,
+      "step": 33930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.291442493814383e-05,
+      "loss": 1.902,
+      "step": 33940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2902282067465144e-05,
+      "loss": 1.8431,
+      "step": 33950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2890139695166387e-05,
+      "loss": 1.739,
+      "step": 33960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.287799782413239e-05,
+      "loss": 1.836,
+      "step": 33970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2865856457247833e-05,
+      "loss": 1.7779,
+      "step": 33980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2853715597397276e-05,
+      "loss": 1.713,
+      "step": 33990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2841575247465185e-05,
+      "loss": 1.7279,
+      "step": 34000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2829435410335883e-05,
+      "loss": 1.8186,
+      "step": 34010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.281729608889356e-05,
+      "loss": 1.9258,
+      "step": 34020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2805157286022327e-05,
+      "loss": 1.9581,
+      "step": 34030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.279301900460613e-05,
+      "loss": 1.6842,
+      "step": 34040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2780881247528796e-05,
+      "loss": 1.6601,
+      "step": 34050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.276874401767406e-05,
+      "loss": 1.5923,
+      "step": 34060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.275660731792549e-05,
+      "loss": 1.6098,
+      "step": 34070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.274447115116655e-05,
+      "loss": 1.773,
+      "step": 34080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2732335520280584e-05,
+      "loss": 1.9136,
+      "step": 34090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2720200428150797e-05,
+      "loss": 1.7324,
+      "step": 34100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2708065877660254e-05,
+      "loss": 1.7334,
+      "step": 34110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.269593187169192e-05,
+      "loss": 1.6025,
+      "step": 34120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.268379841312862e-05,
+      "loss": 1.7218,
+      "step": 34130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2671665504853022e-05,
+      "loss": 1.894,
+      "step": 34140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.265953314974771e-05,
+      "loss": 1.7248,
+      "step": 34150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.264740135069509e-05,
+      "loss": 1.646,
+      "step": 34160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2635270110577486e-05,
+      "loss": 1.8224,
+      "step": 34170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2623139432277048e-05,
+      "loss": 1.6072,
+      "step": 34180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2611009318675793e-05,
+      "loss": 1.883,
+      "step": 34190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.259887977265564e-05,
+      "loss": 1.6763,
+      "step": 34200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2586750797098344e-05,
+      "loss": 1.6696,
+      "step": 34210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2574622394885512e-05,
+      "loss": 1.6135,
+      "step": 34220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.256249456889866e-05,
+      "loss": 1.4837,
+      "step": 34230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2550367322019128e-05,
+      "loss": 1.4442,
+      "step": 34240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.253824065712813e-05,
+      "loss": 1.4963,
+      "step": 34250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2526114577106745e-05,
+      "loss": 1.4023,
+      "step": 34260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2513989084835918e-05,
+      "loss": 1.7389,
+      "step": 34270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.250186418319643e-05,
+      "loss": 1.7651,
+      "step": 34280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2489739875068954e-05,
+      "loss": 1.8217,
+      "step": 34290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2477616163334007e-05,
+      "loss": 1.7917,
+      "step": 34300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.246549305087195e-05,
+      "loss": 1.9634,
+      "step": 34310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2453370540563036e-05,
+      "loss": 1.6658,
+      "step": 34320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2441248635287344e-05,
+      "loss": 1.6517,
+      "step": 34330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2429127337924806e-05,
+      "loss": 1.8958,
+      "step": 34340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.241700665135525e-05,
+      "loss": 1.733,
+      "step": 34350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2404886578458323e-05,
+      "loss": 1.5721,
+      "step": 34360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.239276712211352e-05,
+      "loss": 1.6829,
+      "step": 34370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2380648285200227e-05,
+      "loss": 1.7942,
+      "step": 34380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2368530070597645e-05,
+      "loss": 1.6126,
+      "step": 34390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.235641248118486e-05,
+      "loss": 1.638,
+      "step": 34400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2344295519840778e-05,
+      "loss": 1.6918,
+      "step": 34410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2332179189444176e-05,
+      "loss": 1.7117,
+      "step": 34420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.232006349287367e-05,
+      "loss": 1.7946,
+      "step": 34430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2307948433007747e-05,
+      "loss": 1.8358,
+      "step": 34440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2295834012724702e-05,
+      "loss": 1.835,
+      "step": 34450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2283720234902727e-05,
+      "loss": 2.0003,
+      "step": 34460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2271607102419827e-05,
+      "loss": 1.7104,
+      "step": 34470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2259494618153853e-05,
+      "loss": 1.6024,
+      "step": 34480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2247382784982533e-05,
+      "loss": 1.7935,
+      "step": 34490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2235271605783413e-05,
+      "loss": 1.6697,
+      "step": 34500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2223161083433874e-05,
+      "loss": 1.5126,
+      "step": 34510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2211051220811187e-05,
+      "loss": 1.4102,
+      "step": 34520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2198942020792416e-05,
+      "loss": 1.6864,
+      "step": 34530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2186833486254482e-05,
+      "loss": 1.809,
+      "step": 34540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2174725620074184e-05,
+      "loss": 2.4907,
+      "step": 34550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2162618425128102e-05,
+      "loss": 1.6648,
+      "step": 34560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2150511904292704e-05,
+      "loss": 1.4899,
+      "step": 34570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2138406060444276e-05,
+      "loss": 1.5963,
+      "step": 34580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2126300896458955e-05,
+      "loss": 1.7502,
+      "step": 34590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2114196415212693e-05,
+      "loss": 1.7971,
+      "step": 34600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2102092619581315e-05,
+      "loss": 1.7811,
+      "step": 34610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.208998951244045e-05,
+      "loss": 1.8287,
+      "step": 34620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.207788709666559e-05,
+      "loss": 1.807,
+      "step": 34630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.206578537513205e-05,
+      "loss": 1.7596,
+      "step": 34640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2053684350714966e-05,
+      "loss": 1.773,
+      "step": 34650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.204158402628934e-05,
+      "loss": 1.7895,
+      "step": 34660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2029484404729987e-05,
+      "loss": 1.7783,
+      "step": 34670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2017385488911543e-05,
+      "loss": 1.7443,
+      "step": 34680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2005287281708513e-05,
+      "loss": 1.7495,
+      "step": 34690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1993189785995212e-05,
+      "loss": 1.7505,
+      "step": 34700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1981093004645762e-05,
+      "loss": 1.751,
+      "step": 34710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.196899694053417e-05,
+      "loss": 1.7445,
+      "step": 34720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1956901596534228e-05,
+      "loss": 1.7619,
+      "step": 34730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1944806975519568e-05,
+      "loss": 1.7533,
+      "step": 34740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1932713080363664e-05,
+      "loss": 1.7821,
+      "step": 34750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1920619913939805e-05,
+      "loss": 1.7179,
+      "step": 34760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1908527479121096e-05,
+      "loss": 1.7433,
+      "step": 34770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1896435778780505e-05,
+      "loss": 1.7481,
+      "step": 34780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.188434481579079e-05,
+      "loss": 1.7492,
+      "step": 34790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1872254593024532e-05,
+      "loss": 1.722,
+      "step": 34800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1860165113354176e-05,
+      "loss": 1.7646,
+      "step": 34810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.184807637965195e-05,
+      "loss": 1.7601,
+      "step": 34820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1835988394789906e-05,
+      "loss": 1.718,
+      "step": 34830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.182390116163996e-05,
+      "loss": 1.703,
+      "step": 34840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1811814683073794e-05,
+      "loss": 1.6917,
+      "step": 34850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1799728961962962e-05,
+      "loss": 1.7516,
+      "step": 34860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.178764400117879e-05,
+      "loss": 1.7078,
+      "step": 34870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.177555980359246e-05,
+      "loss": 1.7169,
+      "step": 34880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.176347637207496e-05,
+      "loss": 1.7344,
+      "step": 34890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.175139370949709e-05,
+      "loss": 1.7129,
+      "step": 34900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1739311818729474e-05,
+      "loss": 1.717,
+      "step": 34910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.172723070264255e-05,
+      "loss": 1.7029,
+      "step": 34920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.171515036410658e-05,
+      "loss": 1.6984,
+      "step": 34930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.170307080599162e-05,
+      "loss": 1.711,
+      "step": 34940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1690992031167578e-05,
+      "loss": 1.7022,
+      "step": 34950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1678914042504135e-05,
+      "loss": 1.6856,
+      "step": 34960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.16668368428708e-05,
+      "loss": 1.7337,
+      "step": 34970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1654760435136917e-05,
+      "loss": 1.6995,
+      "step": 34980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.164268482217161e-05,
+      "loss": 1.7,
+      "step": 34990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.163061000684382e-05,
+      "loss": 1.6957,
+      "step": 35000
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 1.5867847204208374,
+      "eval_runtime": 82.4161,
+      "eval_samples_per_second": 36.401,
+      "eval_steps_per_second": 4.55,
+      "step": 35000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.161853599202232e-05,
+      "loss": 1.7199,
+      "step": 35010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1606462780575667e-05,
+      "loss": 1.6812,
+      "step": 35020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1594390375372246e-05,
+      "loss": 1.7047,
+      "step": 35030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1582318779280236e-05,
+      "loss": 1.7232,
+      "step": 35040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1570247995167643e-05,
+      "loss": 1.7013,
+      "step": 35050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1558178025902247e-05,
+      "loss": 1.7285,
+      "step": 35060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1546108874351673e-05,
+      "loss": 1.7174,
+      "step": 35070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.153404054338332e-05,
+      "loss": 1.6271,
+      "step": 35080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1521973035864426e-05,
+      "loss": 1.7001,
+      "step": 35090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1509906354661997e-05,
+      "loss": 1.6902,
+      "step": 35100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.149784050264285e-05,
+      "loss": 1.7759,
+      "step": 35110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1485775482673643e-05,
+      "loss": 1.5853,
+      "step": 35120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.147371129762078e-05,
+      "loss": 1.6965,
+      "step": 35130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1461647950350502e-05,
+      "loss": 1.6932,
+      "step": 35140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.144958544372885e-05,
+      "loss": 1.6664,
+      "step": 35150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1437523780621653e-05,
+      "loss": 1.7326,
+      "step": 35160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1425462963894535e-05,
+      "loss": 1.7017,
+      "step": 35170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.141340299641295e-05,
+      "loss": 1.5896,
+      "step": 35180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.140134388104211e-05,
+      "loss": 1.6619,
+      "step": 35190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1389285620647042e-05,
+      "loss": 1.6838,
+      "step": 35200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1377228218092586e-05,
+      "loss": 1.6775,
+      "step": 35210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1365171676243357e-05,
+      "loss": 1.5946,
+      "step": 35220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1353115997963764e-05,
+      "loss": 1.6967,
+      "step": 35230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1341061186118032e-05,
+      "loss": 1.6656,
+      "step": 35240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1329007243570163e-05,
+      "loss": 1.6311,
+      "step": 35250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1316954173183943e-05,
+      "loss": 1.7566,
+      "step": 35260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1304901977822982e-05,
+      "loss": 1.6327,
+      "step": 35270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.129285066035066e-05,
+      "loss": 1.7071,
+      "step": 35280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.128080022363014e-05,
+      "loss": 1.674,
+      "step": 35290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1268750670524405e-05,
+      "loss": 1.6822,
+      "step": 35300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1256702003896197e-05,
+      "loss": 1.6378,
+      "step": 35310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.124465422660808e-05,
+      "loss": 1.6938,
+      "step": 35320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1232607341522376e-05,
+      "loss": 1.6866,
+      "step": 35330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1220561351501202e-05,
+      "loss": 1.69,
+      "step": 35340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.120851625940648e-05,
+      "loss": 1.692,
+      "step": 35350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.11964720680999e-05,
+      "loss": 1.6697,
+      "step": 35360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1184428780442947e-05,
+      "loss": 1.6843,
+      "step": 35370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1172386399296887e-05,
+      "loss": 1.7021,
+      "step": 35380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.116034492752278e-05,
+      "loss": 1.6651,
+      "step": 35390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1148304367981446e-05,
+      "loss": 1.6863,
+      "step": 35400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.113626472353353e-05,
+      "loss": 1.6608,
+      "step": 35410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.112422599703941e-05,
+      "loss": 1.6148,
+      "step": 35420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.111218819135928e-05,
+      "loss": 1.6601,
+      "step": 35430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1100151309353106e-05,
+      "loss": 1.6247,
+      "step": 35440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1088115353880638e-05,
+      "loss": 1.6718,
+      "step": 35450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1076080327801386e-05,
+      "loss": 1.6574,
+      "step": 35460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.106404623397468e-05,
+      "loss": 1.6764,
+      "step": 35470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1052013075259585e-05,
+      "loss": 1.6576,
+      "step": 35480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.103998085451496e-05,
+      "loss": 1.6744,
+      "step": 35490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1027949574599463e-05,
+      "loss": 1.6743,
+      "step": 35500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.101591923837149e-05,
+      "loss": 1.6804,
+      "step": 35510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1003889848689236e-05,
+      "loss": 1.646,
+      "step": 35520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0991861408410672e-05,
+      "loss": 1.6693,
+      "step": 35530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0979833920393538e-05,
+      "loss": 1.6902,
+      "step": 35540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0967807387495346e-05,
+      "loss": 1.6448,
+      "step": 35550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.095578181257339e-05,
+      "loss": 1.6497,
+      "step": 35560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.094375719848471e-05,
+      "loss": 1.6736,
+      "step": 35570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0931733548086162e-05,
+      "loss": 1.6736,
+      "step": 35580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.091971086423434e-05,
+      "loss": 1.5904,
+      "step": 35590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0907689149785603e-05,
+      "loss": 1.6969,
+      "step": 35600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0895668407596113e-05,
+      "loss": 1.6721,
+      "step": 35610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0883648640521777e-05,
+      "loss": 1.7031,
+      "step": 35620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0871629851418255e-05,
+      "loss": 1.6977,
+      "step": 35630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0859612043141027e-05,
+      "loss": 1.6856,
+      "step": 35640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.084759521854529e-05,
+      "loss": 1.6906,
+      "step": 35650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.083557938048601e-05,
+      "loss": 1.675,
+      "step": 35660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0823564531817967e-05,
+      "loss": 1.6969,
+      "step": 35670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0811550675395645e-05,
+      "loss": 1.6719,
+      "step": 35680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0799537814073327e-05,
+      "loss": 1.7009,
+      "step": 35690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.078752595070506e-05,
+      "loss": 1.7159,
+      "step": 35700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0775515088144643e-05,
+      "loss": 1.6867,
+      "step": 35710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0763505229245624e-05,
+      "loss": 1.6888,
+      "step": 35720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0751496376861355e-05,
+      "loss": 1.6791,
+      "step": 35730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.073948853384491e-05,
+      "loss": 1.6872,
+      "step": 35740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.072748170304913e-05,
+      "loss": 1.6545,
+      "step": 35750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.071547588732663e-05,
+      "loss": 1.7282,
+      "step": 35760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0703471089529766e-05,
+      "loss": 1.7023,
+      "step": 35770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0691467312510678e-05,
+      "loss": 1.6865,
+      "step": 35780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.067946455912124e-05,
+      "loss": 1.7061,
+      "step": 35790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0667462832213073e-05,
+      "loss": 1.7203,
+      "step": 35800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0655462134637593e-05,
+      "loss": 1.7236,
+      "step": 35810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0643462469245945e-05,
+      "loss": 1.7272,
+      "step": 35820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0631463838889016e-05,
+      "loss": 1.7049,
+      "step": 35830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.061946624641749e-05,
+      "loss": 1.6904,
+      "step": 35840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0607469694681757e-05,
+      "loss": 1.6967,
+      "step": 35850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0595474186531988e-05,
+      "loss": 1.6654,
+      "step": 35860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0583479724818107e-05,
+      "loss": 1.6826,
+      "step": 35870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0571486312389776e-05,
+      "loss": 1.698,
+      "step": 35880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0559493952096403e-05,
+      "loss": 1.6943,
+      "step": 35890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.054750264678718e-05,
+      "loss": 1.6718,
+      "step": 35900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.053551239931101e-05,
+      "loss": 1.6603,
+      "step": 35910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0523523212516552e-05,
+      "loss": 1.6593,
+      "step": 35920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.051153508925224e-05,
+      "loss": 1.669,
+      "step": 35930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0499548032366224e-05,
+      "loss": 1.6858,
+      "step": 35940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0487562044706402e-05,
+      "loss": 1.679,
+      "step": 35950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0475577129120454e-05,
+      "loss": 1.6793,
+      "step": 35960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0463593288455758e-05,
+      "loss": 1.6693,
+      "step": 35970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.045161052555947e-05,
+      "loss": 1.6794,
+      "step": 35980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0439628843278474e-05,
+      "loss": 1.6656,
+      "step": 35990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0427648244459403e-05,
+      "loss": 1.6536,
+      "step": 36000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0415668731948625e-05,
+      "loss": 1.6534,
+      "step": 36010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.040369030859227e-05,
+      "loss": 1.6886,
+      "step": 36020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0391712977236185e-05,
+      "loss": 1.6546,
+      "step": 36030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0379736740725968e-05,
+      "loss": 1.6652,
+      "step": 36040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0367761601906964e-05,
+      "loss": 1.6507,
+      "step": 36050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0355787563624234e-05,
+      "loss": 1.6413,
+      "step": 36060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0343814628722614e-05,
+      "loss": 1.6671,
+      "step": 36070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0331842800046646e-05,
+      "loss": 1.6714,
+      "step": 36080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0319872080440617e-05,
+      "loss": 1.6567,
+      "step": 36090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0307902472748564e-05,
+      "loss": 1.6193,
+      "step": 36100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0295933979814246e-05,
+      "loss": 1.6734,
+      "step": 36110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.028396660448115e-05,
+      "loss": 1.6715,
+      "step": 36120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.027200034959253e-05,
+      "loss": 1.6476,
+      "step": 36130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0260035217991337e-05,
+      "loss": 1.6556,
+      "step": 36140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.024807121252027e-05,
+      "loss": 1.6485,
+      "step": 36150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0236108336021764e-05,
+      "loss": 1.6567,
+      "step": 36160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0224146591337994e-05,
+      "loss": 1.657,
+      "step": 36170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.021218598131083e-05,
+      "loss": 1.6314,
+      "step": 36180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0200226508781925e-05,
+      "loss": 1.6215,
+      "step": 36190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.018826817659262e-05,
+      "loss": 1.6383,
+      "step": 36200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0176310987583986e-05,
+      "loss": 1.6625,
+      "step": 36210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0164354944596862e-05,
+      "loss": 1.6495,
+      "step": 36220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0152400050471766e-05,
+      "loss": 1.6553,
+      "step": 36230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0140446308048987e-05,
+      "loss": 1.6519,
+      "step": 36240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0128493720168503e-05,
+      "loss": 1.6591,
+      "step": 36250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0116542289670023e-05,
+      "loss": 1.6122,
+      "step": 36260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0104592019393018e-05,
+      "loss": 1.6488,
+      "step": 36270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.009264291217664e-05,
+      "loss": 1.6568,
+      "step": 36280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.008069497085978e-05,
+      "loss": 1.6478,
+      "step": 36290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0068748198281066e-05,
+      "loss": 1.6827,
+      "step": 36300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.005680259727882e-05,
+      "loss": 1.6815,
+      "step": 36310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0044858170691112e-05,
+      "loss": 1.676,
+      "step": 36320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0032914921355715e-05,
+      "loss": 1.7173,
+      "step": 36330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.002097285211014e-05,
+      "loss": 1.6834,
+      "step": 36340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0009031965791594e-05,
+      "loss": 1.691,
+      "step": 36350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9997092265237034e-05,
+      "loss": 1.672,
+      "step": 36360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9985153753283105e-05,
+      "loss": 1.6638,
+      "step": 36370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.997321643276617e-05,
+      "loss": 1.6726,
+      "step": 36380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.996128030652235e-05,
+      "loss": 1.6455,
+      "step": 36390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9949345377387434e-05,
+      "loss": 1.662,
+      "step": 36400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.993741164819694e-05,
+      "loss": 1.6576,
+      "step": 36410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.992547912178613e-05,
+      "loss": 1.6539,
+      "step": 36420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.991354780098993e-05,
+      "loss": 1.6428,
+      "step": 36430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9901617688643033e-05,
+      "loss": 1.7203,
+      "step": 36440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9889688787579807e-05,
+      "loss": 1.7004,
+      "step": 36450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.987776110063433e-05,
+      "loss": 1.6572,
+      "step": 36460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.986583463064043e-05,
+      "loss": 1.6404,
+      "step": 36470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9853909380431602e-05,
+      "loss": 1.6314,
+      "step": 36480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9841985352841078e-05,
+      "loss": 1.6172,
+      "step": 36490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.983006255070179e-05,
+      "loss": 1.6235,
+      "step": 36500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.981814097684639e-05,
+      "loss": 1.6229,
+      "step": 36510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9806220634107206e-05,
+      "loss": 1.6283,
+      "step": 36520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9794301525316323e-05,
+      "loss": 1.6487,
+      "step": 36530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9782383653305496e-05,
+      "loss": 1.6501,
+      "step": 36540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9770467020906182e-05,
+      "loss": 1.6227,
+      "step": 36550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9758551630949584e-05,
+      "loss": 1.6341,
+      "step": 36560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.974663748626657e-05,
+      "loss": 1.661,
+      "step": 36570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9734724589687714e-05,
+      "loss": 1.6329,
+      "step": 36580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.972281294404333e-05,
+      "loss": 1.608,
+      "step": 36590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9710902552163397e-05,
+      "loss": 1.6393,
+      "step": 36600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.96989934168776e-05,
+      "loss": 1.635,
+      "step": 36610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9687085541015355e-05,
+      "loss": 1.633,
+      "step": 36620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.967517892740574e-05,
+      "loss": 1.6157,
+      "step": 36630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9663273578877564e-05,
+      "loss": 1.6495,
+      "step": 36640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9651369498259316e-05,
+      "loss": 1.6712,
+      "step": 36650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9639466688379195e-05,
+      "loss": 1.6215,
+      "step": 36660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9627565152065093e-05,
+      "loss": 1.644,
+      "step": 36670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9615664892144608e-05,
+      "loss": 1.6315,
+      "step": 36680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9603765911445003e-05,
+      "loss": 1.6257,
+      "step": 36690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.959186821279329e-05,
+      "loss": 1.6185,
+      "step": 36700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.957997179901614e-05,
+      "loss": 1.6329,
+      "step": 36710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.95680766729399e-05,
+      "loss": 1.6276,
+      "step": 36720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9556182837390678e-05,
+      "loss": 1.6251,
+      "step": 36730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9544290295194216e-05,
+      "loss": 1.6479,
+      "step": 36740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.953239904917595e-05,
+      "loss": 1.6345,
+      "step": 36750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9520509102161056e-05,
+      "loss": 1.6258,
+      "step": 36760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.950862045697436e-05,
+      "loss": 1.6358,
+      "step": 36770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9496733116440372e-05,
+      "loss": 1.649,
+      "step": 36780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.948484708338334e-05,
+      "loss": 1.6902,
+      "step": 36790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.947296236062714e-05,
+      "loss": 1.6752,
+      "step": 36800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.946107895099539e-05,
+      "loss": 1.662,
+      "step": 36810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9449196857311373e-05,
+      "loss": 1.6664,
+      "step": 36820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9437316082398052e-05,
+      "loss": 1.6244,
+      "step": 36830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9425436629078082e-05,
+      "loss": 1.6154,
+      "step": 36840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9413558500173818e-05,
+      "loss": 1.6371,
+      "step": 36850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9401681698507284e-05,
+      "loss": 1.643,
+      "step": 36860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9389806226900182e-05,
+      "loss": 1.6153,
+      "step": 36870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9377932088173935e-05,
+      "loss": 1.6275,
+      "step": 36880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9366059285149595e-05,
+      "loss": 1.6104,
+      "step": 36890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9354187820647955e-05,
+      "loss": 1.6675,
+      "step": 36900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9342317697489446e-05,
+      "loss": 1.6076,
+      "step": 36910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.933044891849418e-05,
+      "loss": 1.6242,
+      "step": 36920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9318581486482e-05,
+      "loss": 1.619,
+      "step": 36930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.930671540427237e-05,
+      "loss": 1.6285,
+      "step": 36940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9294850674684447e-05,
+      "loss": 1.6157,
+      "step": 36950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9282987300537104e-05,
+      "loss": 1.6029,
+      "step": 36960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9271125284648848e-05,
+      "loss": 1.6473,
+      "step": 36970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9259264629837886e-05,
+      "loss": 1.6199,
+      "step": 36980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9247405338922092e-05,
+      "loss": 1.6086,
+      "step": 36990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9235547414719024e-05,
+      "loss": 1.6335,
+      "step": 37000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.92236908600459e-05,
+      "loss": 1.6311,
+      "step": 37010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.921183567771964e-05,
+      "loss": 1.618,
+      "step": 37020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9199981870556815e-05,
+      "loss": 1.6124,
+      "step": 37030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9188129441373662e-05,
+      "loss": 1.6852,
+      "step": 37040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9176278392986124e-05,
+      "loss": 1.6135,
+      "step": 37050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9164428728209792e-05,
+      "loss": 1.6176,
+      "step": 37060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.915258044985992e-05,
+      "loss": 1.6052,
+      "step": 37070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9140733560751462e-05,
+      "loss": 1.6169,
+      "step": 37080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9128888063699012e-05,
+      "loss": 1.5959,
+      "step": 37090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.911704396151685e-05,
+      "loss": 1.6304,
+      "step": 37100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9105201257018928e-05,
+      "loss": 1.6042,
+      "step": 37110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.909335995301885e-05,
+      "loss": 1.6218,
+      "step": 37120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.90815200523299e-05,
+      "loss": 1.6279,
+      "step": 37130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9069681557765032e-05,
+      "loss": 1.623,
+      "step": 37140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9057844472136845e-05,
+      "loss": 1.5995,
+      "step": 37150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9046008798257624e-05,
+      "loss": 1.6197,
+      "step": 37160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9034174538939316e-05,
+      "loss": 1.6644,
+      "step": 37170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9022341696993513e-05,
+      "loss": 1.6634,
+      "step": 37180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.901051027523151e-05,
+      "loss": 1.6782,
+      "step": 37190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8998680276464215e-05,
+      "loss": 1.6734,
+      "step": 37200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8986851703502227e-05,
+      "loss": 1.6527,
+      "step": 37210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8975024559155813e-05,
+      "loss": 1.6456,
+      "step": 37220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8963198846234882e-05,
+      "loss": 1.6312,
+      "step": 37230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8951374567549e-05,
+      "loss": 1.6193,
+      "step": 37240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8939551725907426e-05,
+      "loss": 1.625,
+      "step": 37250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.892773032411903e-05,
+      "loss": 1.6118,
+      "step": 37260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.891591036499238e-05,
+      "loss": 1.6162,
+      "step": 37270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8904091851335674e-05,
+      "loss": 1.632,
+      "step": 37280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.889227478595679e-05,
+      "loss": 1.6279,
+      "step": 37290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8880459171663235e-05,
+      "loss": 1.6107,
+      "step": 37300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8868645011262198e-05,
+      "loss": 1.6299,
+      "step": 37310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.885683230756051e-05,
+      "loss": 1.6124,
+      "step": 37320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8845021063364638e-05,
+      "loss": 1.612,
+      "step": 37330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.883321128148075e-05,
+      "loss": 1.6124,
+      "step": 37340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8821402964714607e-05,
+      "loss": 1.6117,
+      "step": 37350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8809596115871677e-05,
+      "loss": 1.5816,
+      "step": 37360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8797790737757045e-05,
+      "loss": 1.6274,
+      "step": 37370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8785986833175445e-05,
+      "loss": 1.6172,
+      "step": 37380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8774184404931294e-05,
+      "loss": 1.621,
+      "step": 37390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.876238345582862e-05,
+      "loss": 1.6078,
+      "step": 37400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.875058398867111e-05,
+      "loss": 1.6112,
+      "step": 37410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8738786006262127e-05,
+      "loss": 1.6331,
+      "step": 37420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.872698951140464e-05,
+      "loss": 1.6103,
+      "step": 37430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8715194506901287e-05,
+      "loss": 1.6117,
+      "step": 37440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.870340099555435e-05,
+      "loss": 1.614,
+      "step": 37450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8691608980165765e-05,
+      "loss": 1.592,
+      "step": 37460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8679818463537076e-05,
+      "loss": 1.602,
+      "step": 37470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8668029448469526e-05,
+      "loss": 1.6005,
+      "step": 37480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8656241937763964e-05,
+      "loss": 1.5889,
+      "step": 37490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864445593422087e-05,
+      "loss": 1.6192,
+      "step": 37500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863267144064041e-05,
+      "loss": 1.6034,
+      "step": 37510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862088845982236e-05,
+      "loss": 1.6003,
+      "step": 37520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8609106994566135e-05,
+      "loss": 1.6116,
+      "step": 37530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8597327047670814e-05,
+      "loss": 1.6345,
+      "step": 37540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8585548621935088e-05,
+      "loss": 1.6368,
+      "step": 37550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8573771720157297e-05,
+      "loss": 1.6586,
+      "step": 37560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8561996345135435e-05,
+      "loss": 1.6127,
+      "step": 37570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8550222499667098e-05,
+      "loss": 1.6258,
+      "step": 37580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.853845018654956e-05,
+      "loss": 1.6294,
+      "step": 37590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8526679408579697e-05,
+      "loss": 1.6193,
+      "step": 37600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.851491016855404e-05,
+      "loss": 1.6417,
+      "step": 37610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.850314246926874e-05,
+      "loss": 1.5921,
+      "step": 37620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8491376313519604e-05,
+      "loss": 1.6261,
+      "step": 37630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8479611704102038e-05,
+      "loss": 1.635,
+      "step": 37640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8467848643811123e-05,
+      "loss": 1.6165,
+      "step": 37650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.845608713544154e-05,
+      "loss": 1.6248,
+      "step": 37660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8444327181787602e-05,
+      "loss": 1.6395,
+      "step": 37670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.843256878564328e-05,
+      "loss": 1.6479,
+      "step": 37680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.842081194980215e-05,
+      "loss": 1.687,
+      "step": 37690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.840905667705741e-05,
+      "loss": 1.6954,
+      "step": 37700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.839730297020192e-05,
+      "loss": 1.7003,
+      "step": 37710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8385550832028144e-05,
+      "loss": 1.6894,
+      "step": 37720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.837380026532816e-05,
+      "loss": 1.7122,
+      "step": 37730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8362051272893725e-05,
+      "loss": 1.6924,
+      "step": 37740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.835030385751616e-05,
+      "loss": 1.7223,
+      "step": 37750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8338558021986447e-05,
+      "loss": 1.7023,
+      "step": 37760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8326813769095187e-05,
+      "loss": 1.7259,
+      "step": 37770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8315071101632604e-05,
+      "loss": 1.698,
+      "step": 37780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8303330022388528e-05,
+      "loss": 1.7212,
+      "step": 37790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8291590534152457e-05,
+      "loss": 1.6951,
+      "step": 37800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.827985263971345e-05,
+      "loss": 1.6868,
+      "step": 37810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8268116341860247e-05,
+      "loss": 1.6604,
+      "step": 37820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8256381643381166e-05,
+      "loss": 1.6979,
+      "step": 37830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8244648547064153e-05,
+      "loss": 1.6589,
+      "step": 37840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.82329170556968e-05,
+      "loss": 1.6791,
+      "step": 37850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8221187172066285e-05,
+      "loss": 1.6811,
+      "step": 37860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.820945889895941e-05,
+      "loss": 1.6398,
+      "step": 37870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8197732239162613e-05,
+      "loss": 1.5977,
+      "step": 37880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8186007195461937e-05,
+      "loss": 1.6212,
+      "step": 37890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8174283770643027e-05,
+      "loss": 1.5862,
+      "step": 37900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8162561967491178e-05,
+      "loss": 1.5805,
+      "step": 37910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.815084178879126e-05,
+      "loss": 1.6151,
+      "step": 37920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8139123237327787e-05,
+      "loss": 1.7078,
+      "step": 37930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8127406315884873e-05,
+      "loss": 1.6221,
+      "step": 37940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.811569102724625e-05,
+      "loss": 1.6135,
+      "step": 37950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8103977374195248e-05,
+      "loss": 1.6406,
+      "step": 37960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8092265359514837e-05,
+      "loss": 1.6536,
+      "step": 37970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.808055498598757e-05,
+      "loss": 1.614,
+      "step": 37980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.806884625639561e-05,
+      "loss": 1.6013,
+      "step": 37990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8057139173520764e-05,
+      "loss": 1.6079,
+      "step": 38000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.804543374014441e-05,
+      "loss": 1.6078,
+      "step": 38010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8033729959047545e-05,
+      "loss": 1.5776,
+      "step": 38020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8022027833010786e-05,
+      "loss": 1.6088,
+      "step": 38030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.801032736481433e-05,
+      "loss": 1.6384,
+      "step": 38040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7998628557238023e-05,
+      "loss": 1.6569,
+      "step": 38050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.798693141306127e-05,
+      "loss": 1.624,
+      "step": 38060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7975235935063102e-05,
+      "loss": 1.6564,
+      "step": 38070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.796354212602217e-05,
+      "loss": 1.585,
+      "step": 38080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7951849988716694e-05,
+      "loss": 1.5716,
+      "step": 38090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7940159525924525e-05,
+      "loss": 1.6015,
+      "step": 38100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.79284707404231e-05,
+      "loss": 1.5994,
+      "step": 38110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7916783634989475e-05,
+      "loss": 1.5848,
+      "step": 38120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7905098212400274e-05,
+      "loss": 1.5871,
+      "step": 38130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.789341447543177e-05,
+      "loss": 1.5862,
+      "step": 38140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7881732426859793e-05,
+      "loss": 1.588,
+      "step": 38150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7870052069459777e-05,
+      "loss": 1.5747,
+      "step": 38160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7858373406006786e-05,
+      "loss": 1.6165,
+      "step": 38170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7846696439275453e-05,
+      "loss": 1.5906,
+      "step": 38180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7835021172039997e-05,
+      "loss": 1.5845,
+      "step": 38190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7823347607074274e-05,
+      "loss": 1.6178,
+      "step": 38200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7811675747151702e-05,
+      "loss": 1.5876,
+      "step": 38210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.78000055950453e-05,
+      "loss": 1.6362,
+      "step": 38220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7788337153527695e-05,
+      "loss": 1.6062,
+      "step": 38230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.77766704253711e-05,
+      "loss": 1.636,
+      "step": 38240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.77650054133473e-05,
+      "loss": 1.6348,
+      "step": 38250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.775334212022772e-05,
+      "loss": 1.6208,
+      "step": 38260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7741680548783328e-05,
+      "loss": 1.623,
+      "step": 38270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7730020701784706e-05,
+      "loss": 1.5966,
+      "step": 38280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.771836258200203e-05,
+      "loss": 1.5977,
+      "step": 38290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7706706192205044e-05,
+      "loss": 1.5994,
+      "step": 38300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7695051535163116e-05,
+      "loss": 1.5909,
+      "step": 38310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7683398613645174e-05,
+      "loss": 1.5842,
+      "step": 38320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7671747430419732e-05,
+      "loss": 1.5904,
+      "step": 38330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.766009798825492e-05,
+      "loss": 1.5989,
+      "step": 38340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.764845028991842e-05,
+      "loss": 1.6016,
+      "step": 38350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7636804338177516e-05,
+      "loss": 1.577,
+      "step": 38360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7625160135799086e-05,
+      "loss": 1.5831,
+      "step": 38370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.761351768554957e-05,
+      "loss": 1.5808,
+      "step": 38380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7601876990195016e-05,
+      "loss": 1.6154,
+      "step": 38390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7590238052501034e-05,
+      "loss": 1.5626,
+      "step": 38400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7578600875232837e-05,
+      "loss": 1.5827,
+      "step": 38410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.756696546115519e-05,
+      "loss": 1.5814,
+      "step": 38420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7555331813032478e-05,
+      "loss": 1.5946,
+      "step": 38430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7543699933628632e-05,
+      "loss": 1.5841,
+      "step": 38440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7532069825707175e-05,
+      "loss": 1.6331,
+      "step": 38450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.752044149203123e-05,
+      "loss": 1.6097,
+      "step": 38460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.750881493536346e-05,
+      "loss": 1.6138,
+      "step": 38470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7497190158466127e-05,
+      "loss": 1.588,
+      "step": 38480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.748556716410108e-05,
+      "loss": 1.5934,
+      "step": 38490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7473945955029714e-05,
+      "loss": 1.5615,
+      "step": 38500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.746232653401304e-05,
+      "loss": 1.5744,
+      "step": 38510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7450708903811616e-05,
+      "loss": 1.6236,
+      "step": 38520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.743909306718557e-05,
+      "loss": 1.5934,
+      "step": 38530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.742747902689463e-05,
+      "loss": 1.5735,
+      "step": 38540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.741586678569807e-05,
+      "loss": 1.5778,
+      "step": 38550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7404256346354755e-05,
+      "loss": 1.5866,
+      "step": 38560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7392647711623118e-05,
+      "loss": 1.598,
+      "step": 38570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.738104088426116e-05,
+      "loss": 1.6246,
+      "step": 38580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.736943586702645e-05,
+      "loss": 1.6181,
+      "step": 38590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.735783266267614e-05,
+      "loss": 1.6335,
+      "step": 38600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7346231273966937e-05,
+      "loss": 1.6548,
+      "step": 38610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.733463170365511e-05,
+      "loss": 1.6111,
+      "step": 38620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.732303395449653e-05,
+      "loss": 1.6206,
+      "step": 38630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7311438029246596e-05,
+      "loss": 1.5805,
+      "step": 38640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.729984393066029e-05,
+      "loss": 1.5742,
+      "step": 38650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7288251661492177e-05,
+      "loss": 1.568,
+      "step": 38660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.727666122449636e-05,
+      "loss": 1.569,
+      "step": 38670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7265072622426507e-05,
+      "loss": 1.5819,
+      "step": 38680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7253485858035884e-05,
+      "loss": 1.6272,
+      "step": 38690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7241900934077277e-05,
+      "loss": 1.5797,
+      "step": 38700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7230317853303064e-05,
+      "loss": 1.5887,
+      "step": 38710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.721873661846518e-05,
+      "loss": 1.5757,
+      "step": 38720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.720715723231511e-05,
+      "loss": 1.5795,
+      "step": 38730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.719557969760391e-05,
+      "loss": 1.5713,
+      "step": 38740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7184004017082196e-05,
+      "loss": 1.6052,
+      "step": 38750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.717243019350013e-05,
+      "loss": 1.557,
+      "step": 38760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7160858229607463e-05,
+      "loss": 1.5745,
+      "step": 38770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7149288128153472e-05,
+      "loss": 1.5799,
+      "step": 38780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7137719891887e-05,
+      "loss": 1.5643,
+      "step": 38790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7126153523556466e-05,
+      "loss": 1.6055,
+      "step": 38800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7114589025909823e-05,
+      "loss": 1.558,
+      "step": 38810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7103026401694574e-05,
+      "loss": 1.5983,
+      "step": 38820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7091465653657814e-05,
+      "loss": 1.5699,
+      "step": 38830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7079906784546156e-05,
+      "loss": 1.5673,
+      "step": 38840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.706834979710577e-05,
+      "loss": 1.572,
+      "step": 38850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7056794694082405e-05,
+      "loss": 1.6072,
+      "step": 38860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.704524147822133e-05,
+      "loss": 1.5643,
+      "step": 38870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.703369015226739e-05,
+      "loss": 1.5748,
+      "step": 38880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.702214071896497e-05,
+      "loss": 1.5863,
+      "step": 38890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7010593181058012e-05,
+      "loss": 1.5839,
+      "step": 38900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6999047541289987e-05,
+      "loss": 1.5741,
+      "step": 38910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6987503802403948e-05,
+      "loss": 1.5978,
+      "step": 38920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.697596196714247e-05,
+      "loss": 1.6085,
+      "step": 38930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6964422038247685e-05,
+      "loss": 1.632,
+      "step": 38940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6952884018461277e-05,
+      "loss": 1.6179,
+      "step": 38950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.694134791052446e-05,
+      "loss": 1.6036,
+      "step": 38960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6929813717178023e-05,
+      "loss": 1.5891,
+      "step": 38970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.691828144116227e-05,
+      "loss": 1.588,
+      "step": 38980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.690675108521706e-05,
+      "loss": 1.5756,
+      "step": 38990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6895222652081803e-05,
+      "loss": 1.6013,
+      "step": 39000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6883696144495454e-05,
+      "loss": 1.6076,
+      "step": 39010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.687217156519648e-05,
+      "loss": 1.5992,
+      "step": 39020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.686064891692294e-05,
+      "loss": 1.5968,
+      "step": 39030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.684912820241239e-05,
+      "loss": 1.6165,
+      "step": 39040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6837609424401953e-05,
+      "loss": 1.6549,
+      "step": 39050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6826092585628273e-05,
+      "loss": 1.6706,
+      "step": 39060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6814577688827556e-05,
+      "loss": 1.6727,
+      "step": 39070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.680306473673552e-05,
+      "loss": 1.6914,
+      "step": 39080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6791553732087447e-05,
+      "loss": 1.6731,
+      "step": 39090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6780044677618144e-05,
+      "loss": 1.7149,
+      "step": 39100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.676853757606194e-05,
+      "loss": 1.664,
+      "step": 39110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.675703243015273e-05,
+      "loss": 1.6827,
+      "step": 39120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6745529242623924e-05,
+      "loss": 1.6748,
+      "step": 39130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6734028016208463e-05,
+      "loss": 1.6684,
+      "step": 39140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.672252875363885e-05,
+      "loss": 1.6592,
+      "step": 39150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6711031457647088e-05,
+      "loss": 1.6551,
+      "step": 39160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6699536130964724e-05,
+      "loss": 1.6744,
+      "step": 39170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6688042776322855e-05,
+      "loss": 1.6602,
+      "step": 39180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.667655139645208e-05,
+      "loss": 1.6684,
+      "step": 39190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.666506199408256e-05,
+      "loss": 1.6067,
+      "step": 39200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6653574571943952e-05,
+      "loss": 1.5655,
+      "step": 39210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.664208913276547e-05,
+      "loss": 1.571,
+      "step": 39220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6630605679275846e-05,
+      "loss": 1.5416,
+      "step": 39230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6619124214203347e-05,
+      "loss": 1.573,
+      "step": 39240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6607644740275744e-05,
+      "loss": 1.5577,
+      "step": 39250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6596167260220375e-05,
+      "loss": 1.5479,
+      "step": 39260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6584691776764072e-05,
+      "loss": 1.6363,
+      "step": 39270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.65732182926332e-05,
+      "loss": 1.5497,
+      "step": 39280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.656174681055365e-05,
+      "loss": 1.5819,
+      "step": 39290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6550277333250856e-05,
+      "loss": 1.5544,
+      "step": 39300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6538809863449733e-05,
+      "loss": 1.6114,
+      "step": 39310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6527344403874773e-05,
+      "loss": 1.5593,
+      "step": 39320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.651588095724994e-05,
+      "loss": 1.5932,
+      "step": 39330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6504419526298747e-05,
+      "loss": 1.5799,
+      "step": 39340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6492960113744233e-05,
+      "loss": 1.6312,
+      "step": 39350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6481502722308945e-05,
+      "loss": 1.5758,
+      "step": 39360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6470047354714933e-05,
+      "loss": 1.5695,
+      "step": 39370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.645859401368382e-05,
+      "loss": 1.5225,
+      "step": 39380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.644714270193669e-05,
+      "loss": 1.5564,
+      "step": 39390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6435693422194164e-05,
+      "loss": 1.6017,
+      "step": 39400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6424246177176406e-05,
+      "loss": 1.6071,
+      "step": 39410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.641280096960305e-05,
+      "loss": 1.6136,
+      "step": 39420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.640135780219329e-05,
+      "loss": 1.614,
+      "step": 39430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6389916677665813e-05,
+      "loss": 1.5882,
+      "step": 39440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6378477598738808e-05,
+      "loss": 1.6441,
+      "step": 39450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.636704056813002e-05,
+      "loss": 1.5937,
+      "step": 39460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6355605588556663e-05,
+      "loss": 1.6104,
+      "step": 39470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6344172662735476e-05,
+      "loss": 1.6768,
+      "step": 39480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.633274179338274e-05,
+      "loss": 1.5862,
+      "step": 39490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.63213129832142e-05,
+      "loss": 1.5866,
+      "step": 39500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6309886234945147e-05,
+      "loss": 1.534,
+      "step": 39510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6298461551290364e-05,
+      "loss": 1.5384,
+      "step": 39520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.628703893496416e-05,
+      "loss": 1.5564,
+      "step": 39530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.627561838868033e-05,
+      "loss": 1.5615,
+      "step": 39540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.62641999151522e-05,
+      "loss": 1.5446,
+      "step": 39550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6252783517092592e-05,
+      "loss": 1.5877,
+      "step": 39560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6241369197213822e-05,
+      "loss": 1.5591,
+      "step": 39570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6229956958227748e-05,
+      "loss": 1.5765,
+      "step": 39580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6218546802845708e-05,
+      "loss": 1.5556,
+      "step": 39590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6207138733778527e-05,
+      "loss": 1.5382,
+      "step": 39600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.619573275373658e-05,
+      "loss": 1.599,
+      "step": 39610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.618432886542972e-05,
+      "loss": 1.567,
+      "step": 39620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.617292707156729e-05,
+      "loss": 1.5705,
+      "step": 39630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6161527374858166e-05,
+      "loss": 1.5521,
+      "step": 39640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6150129778010698e-05,
+      "loss": 1.5673,
+      "step": 39650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6138734283732766e-05,
+      "loss": 1.5601,
+      "step": 39660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6127340894731717e-05,
+      "loss": 1.5915,
+      "step": 39670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6115949613714422e-05,
+      "loss": 1.6511,
+      "step": 39680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6104560443387245e-05,
+      "loss": 1.5667,
+      "step": 39690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.609317338645605e-05,
+      "loss": 1.5551,
+      "step": 39700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.608178844562618e-05,
+      "loss": 1.5782,
+      "step": 39710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.607040562360252e-05,
+      "loss": 1.6231,
+      "step": 39720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.60590249230894e-05,
+      "loss": 1.6085,
+      "step": 39730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.604764634679067e-05,
+      "loss": 1.5623,
+      "step": 39740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.603626989740969e-05,
+      "loss": 1.5714,
+      "step": 39750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.602489557764929e-05,
+      "loss": 1.5566,
+      "step": 39760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6013523390211793e-05,
+      "loss": 1.5595,
+      "step": 39770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6002153337799045e-05,
+      "loss": 1.6104,
+      "step": 39780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5990785423112356e-05,
+      "loss": 1.5597,
+      "step": 39790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5979419648852527e-05,
+      "loss": 1.5618,
+      "step": 39800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5968056017719885e-05,
+      "loss": 1.526,
+      "step": 39810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.59566945324142e-05,
+      "loss": 1.557,
+      "step": 39820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.594533519563477e-05,
+      "loss": 1.571,
+      "step": 39830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.593397801008036e-05,
+      "loss": 1.5794,
+      "step": 39840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5922622978449248e-05,
+      "loss": 1.5606,
+      "step": 39850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5911270103439158e-05,
+      "loss": 1.5906,
+      "step": 39860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.589991938774736e-05,
+      "loss": 1.5902,
+      "step": 39870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5888570834070548e-05,
+      "loss": 1.6501,
+      "step": 39880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.587722444510496e-05,
+      "loss": 1.6391,
+      "step": 39890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5865880223546288e-05,
+      "loss": 1.5878,
+      "step": 39900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.585453817208969e-05,
+      "loss": 1.5549,
+      "step": 39910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5843198293429867e-05,
+      "loss": 1.5613,
+      "step": 39920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5831860590260954e-05,
+      "loss": 1.559,
+      "step": 39930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5820525065276574e-05,
+      "loss": 1.5626,
+      "step": 39940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5809191721169865e-05,
+      "loss": 1.5549,
+      "step": 39950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5797860560633412e-05,
+      "loss": 1.5813,
+      "step": 39960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5786531586359286e-05,
+      "loss": 1.5855,
+      "step": 39970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5775204801039074e-05,
+      "loss": 1.5706,
+      "step": 39980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5763880207363788e-05,
+      "loss": 1.554,
+      "step": 39990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.575255780802396e-05,
+      "loss": 1.5782,
+      "step": 40000
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 1.5893032550811768,
+      "eval_runtime": 82.1174,
+      "eval_samples_per_second": 36.533,
+      "eval_steps_per_second": 4.567,
+      "step": 40000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5741237605709587e-05,
+      "loss": 1.5483,
+      "step": 40010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.572991960311015e-05,
+      "loss": 1.6209,
+      "step": 40020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.571860380291459e-05,
+      "loss": 1.5402,
+      "step": 40030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.570729020781135e-05,
+      "loss": 1.5681,
+      "step": 40040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5695978820488327e-05,
+      "loss": 1.5641,
+      "step": 40050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5684669643632897e-05,
+      "loss": 1.5574,
+      "step": 40060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5673362679931935e-05,
+      "loss": 1.584,
+      "step": 40070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5662057932071768e-05,
+      "loss": 1.5969,
+      "step": 40080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5650755402738172e-05,
+      "loss": 1.5696,
+      "step": 40090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5639455094616462e-05,
+      "loss": 1.5702,
+      "step": 40100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5628157010391363e-05,
+      "loss": 1.5483,
+      "step": 40110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5616861152747108e-05,
+      "loss": 1.5312,
+      "step": 40120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5605567524367393e-05,
+      "loss": 1.5911,
+      "step": 40130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5594276127935357e-05,
+      "loss": 1.5652,
+      "step": 40140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5582986966133655e-05,
+      "loss": 1.5772,
+      "step": 40150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5571700041644384e-05,
+      "loss": 1.5528,
+      "step": 40160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5560415357149103e-05,
+      "loss": 1.5573,
+      "step": 40170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.554913291532886e-05,
+      "loss": 1.5711,
+      "step": 40180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.553785271886416e-05,
+      "loss": 1.5924,
+      "step": 40190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.552657477043496e-05,
+      "loss": 1.6108,
+      "step": 40200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.551529907272072e-05,
+      "loss": 1.6392,
+      "step": 40210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.550402562840033e-05,
+      "loss": 1.6086,
+      "step": 40220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5492754440152146e-05,
+      "loss": 1.6047,
+      "step": 40230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.548148551065402e-05,
+      "loss": 1.555,
+      "step": 40240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5470218842583238e-05,
+      "loss": 1.5682,
+      "step": 40250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5458954438616547e-05,
+      "loss": 1.5691,
+      "step": 40260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5447692301430182e-05,
+      "loss": 1.5594,
+      "step": 40270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.543643243369982e-05,
+      "loss": 1.5894,
+      "step": 40280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5425174838100587e-05,
+      "loss": 1.5827,
+      "step": 40290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5413919517307107e-05,
+      "loss": 1.5524,
+      "step": 40300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5402666473993425e-05,
+      "loss": 1.5529,
+      "step": 40310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5391415710833066e-05,
+      "loss": 1.5768,
+      "step": 40320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.538016723049901e-05,
+      "loss": 1.5348,
+      "step": 40330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.536892103566369e-05,
+      "loss": 1.5794,
+      "step": 40340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5357677128999005e-05,
+      "loss": 1.5712,
+      "step": 40350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.534643551317631e-05,
+      "loss": 1.5625,
+      "step": 40360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5335196190866384e-05,
+      "loss": 1.574,
+      "step": 40370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5323959164739515e-05,
+      "loss": 1.5474,
+      "step": 40380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5312724437465408e-05,
+      "loss": 1.5503,
+      "step": 40390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.530149201171322e-05,
+      "loss": 1.5434,
+      "step": 40400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5290261890151595e-05,
+      "loss": 1.5673,
+      "step": 40410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5279034075448593e-05,
+      "loss": 1.5651,
+      "step": 40420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5267808570271733e-05,
+      "loss": 1.5404,
+      "step": 40430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5256585377288014e-05,
+      "loss": 1.5651,
+      "step": 40440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5245364499163844e-05,
+      "loss": 1.5476,
+      "step": 40450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5234145938565115e-05,
+      "loss": 1.5626,
+      "step": 40460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5222929698157146e-05,
+      "loss": 1.5785,
+      "step": 40470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5211715780604718e-05,
+      "loss": 1.6041,
+      "step": 40480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.520050418857205e-05,
+      "loss": 1.6076,
+      "step": 40490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5189294924722825e-05,
+      "loss": 1.5756,
+      "step": 40500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5178087991720158e-05,
+      "loss": 1.5833,
+      "step": 40510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.51668833922266e-05,
+      "loss": 1.5662,
+      "step": 40520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.515568112890418e-05,
+      "loss": 1.5742,
+      "step": 40530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5144481204414351e-05,
+      "loss": 1.5812,
+      "step": 40540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5133283621417998e-05,
+      "loss": 1.5908,
+      "step": 40550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5122088382575483e-05,
+      "loss": 1.5723,
+      "step": 40560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5110895490546578e-05,
+      "loss": 1.5963,
+      "step": 40570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5099704947990528e-05,
+      "loss": 1.6236,
+      "step": 40580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5088516757565995e-05,
+      "loss": 1.6409,
+      "step": 40590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5077330921931083e-05,
+      "loss": 1.6617,
+      "step": 40600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5066147443743361e-05,
+      "loss": 1.6618,
+      "step": 40610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5054966325659814e-05,
+      "loss": 1.6597,
+      "step": 40620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.504378757033687e-05,
+      "loss": 1.6585,
+      "step": 40630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5032611180430406e-05,
+      "loss": 1.645,
+      "step": 40640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5021437158595733e-05,
+      "loss": 1.6876,
+      "step": 40650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5010265507487581e-05,
+      "loss": 1.672,
+      "step": 40660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4999096229760156e-05,
+      "loss": 1.6322,
+      "step": 40670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4987929328067064e-05,
+      "loss": 1.6327,
+      "step": 40680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.497676480506135e-05,
+      "loss": 1.6568,
+      "step": 40690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4965602663395528e-05,
+      "loss": 1.64,
+      "step": 40700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4954442905721502e-05,
+      "loss": 1.6379,
+      "step": 40710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4943285534690626e-05,
+      "loss": 1.6612,
+      "step": 40720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.493213055295371e-05,
+      "loss": 1.5794,
+      "step": 40730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4920977963160965e-05,
+      "loss": 1.6017,
+      "step": 40740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4909827767962033e-05,
+      "loss": 1.5533,
+      "step": 40750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4898679970006023e-05,
+      "loss": 1.5505,
+      "step": 40760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4887534571941431e-05,
+      "loss": 1.585,
+      "step": 40770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4876391576416215e-05,
+      "loss": 1.5731,
+      "step": 40780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4865250986077744e-05,
+      "loss": 1.5595,
+      "step": 40790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4854112803572824e-05,
+      "loss": 1.5926,
+      "step": 40800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4842977031547683e-05,
+      "loss": 1.5848,
+      "step": 40810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4831843672647988e-05,
+      "loss": 1.5626,
+      "step": 40820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4820712729518806e-05,
+      "loss": 1.5733,
+      "step": 40830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4809584204804671e-05,
+      "loss": 1.5231,
+      "step": 40840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.479845810114951e-05,
+      "loss": 1.5251,
+      "step": 40850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4787334421196668e-05,
+      "loss": 1.5359,
+      "step": 40860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4776213167588965e-05,
+      "loss": 1.5273,
+      "step": 40870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4765094342968582e-05,
+      "loss": 1.5995,
+      "step": 40880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4753977949977154e-05,
+      "loss": 1.5298,
+      "step": 40890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4742863991255756e-05,
+      "loss": 1.5485,
+      "step": 40900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4731752469444845e-05,
+      "loss": 1.5373,
+      "step": 40910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4720643387184314e-05,
+      "loss": 1.6521,
+      "step": 40920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.47095367471135e-05,
+      "loss": 1.5349,
+      "step": 40930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4698432551871124e-05,
+      "loss": 1.5442,
+      "step": 40940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.468733080409535e-05,
+      "loss": 1.5967,
+      "step": 40950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4676231506423752e-05,
+      "loss": 1.6176,
+      "step": 40960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4665134661493327e-05,
+      "loss": 1.5479,
+      "step": 40970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.465404027194047e-05,
+      "loss": 1.555,
+      "step": 40980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4642948340401028e-05,
+      "loss": 1.6044,
+      "step": 40990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4631858869510223e-05,
+      "loss": 1.6422,
+      "step": 41000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4620771861902737e-05,
+      "loss": 1.6331,
+      "step": 41010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4609687320212631e-05,
+      "loss": 1.5515,
+      "step": 41020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.459860524707338e-05,
+      "loss": 1.549,
+      "step": 41030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.45875256451179e-05,
+      "loss": 1.5286,
+      "step": 41040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4576448516978514e-05,
+      "loss": 1.5512,
+      "step": 41050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4565373865286913e-05,
+      "loss": 1.5312,
+      "step": 41060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4554301692674278e-05,
+      "loss": 1.5537,
+      "step": 41070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4543232001771123e-05,
+      "loss": 1.5345,
+      "step": 41080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.453216479520742e-05,
+      "loss": 1.5743,
+      "step": 41090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4521100075612543e-05,
+      "loss": 1.5569,
+      "step": 41100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4510037845615266e-05,
+      "loss": 1.5499,
+      "step": 41110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4498978107843758e-05,
+      "loss": 1.5511,
+      "step": 41120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4487920864925648e-05,
+      "loss": 1.5941,
+      "step": 41130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4476866119487906e-05,
+      "loss": 1.5664,
+      "step": 41140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4465813874156953e-05,
+      "loss": 1.6154,
+      "step": 41150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4454764131558602e-05,
+      "loss": 1.6256,
+      "step": 41160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4443716894318074e-05,
+      "loss": 1.575,
+      "step": 41170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.443267216505999e-05,
+      "loss": 1.5828,
+      "step": 41180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4421629946408383e-05,
+      "loss": 1.5404,
+      "step": 41190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.441059024098669e-05,
+      "loss": 1.5541,
+      "step": 41200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4399553051417719e-05,
+      "loss": 1.558,
+      "step": 41210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4388518380323748e-05,
+      "loss": 1.561,
+      "step": 41220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4377486230326372e-05,
+      "loss": 1.5945,
+      "step": 41230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4366456604046674e-05,
+      "loss": 1.5328,
+      "step": 41240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4355429504105066e-05,
+      "loss": 1.5309,
+      "step": 41250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4344404933121392e-05,
+      "loss": 1.5405,
+      "step": 41260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4333382893714894e-05,
+      "loss": 1.5362,
+      "step": 41270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4322363388504224e-05,
+      "loss": 1.5636,
+      "step": 41280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.431134642010738e-05,
+      "loss": 1.5355,
+      "step": 41290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4300331991141841e-05,
+      "loss": 1.5594,
+      "step": 41300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4289320104224404e-05,
+      "loss": 1.5534,
+      "step": 41310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4278310761971306e-05,
+      "loss": 1.562,
+      "step": 41320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4267303966998166e-05,
+      "loss": 1.5794,
+      "step": 41330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4256299721920003e-05,
+      "loss": 1.5461,
+      "step": 41340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4245298029351223e-05,
+      "loss": 1.5464,
+      "step": 41350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4234298891905634e-05,
+      "loss": 1.5218,
+      "step": 41360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4223302312196441e-05,
+      "loss": 1.5672,
+      "step": 41370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4212308292836201e-05,
+      "loss": 1.5578,
+      "step": 41380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4201316836436939e-05,
+      "loss": 1.5666,
+      "step": 41390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4190327945609994e-05,
+      "loss": 1.575,
+      "step": 41400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4179341622966136e-05,
+      "loss": 1.5484,
+      "step": 41410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4168357871115522e-05,
+      "loss": 1.5433,
+      "step": 41420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4157376692667702e-05,
+      "loss": 1.5534,
+      "step": 41430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4146398090231572e-05,
+      "loss": 1.563,
+      "step": 41440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4135422066415493e-05,
+      "loss": 1.5708,
+      "step": 41450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.412444862382713e-05,
+      "loss": 1.6022,
+      "step": 41460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4113477765073612e-05,
+      "loss": 1.6043,
+      "step": 41470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4102509492761396e-05,
+      "loss": 1.5763,
+      "step": 41480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4091543809496347e-05,
+      "loss": 1.5666,
+      "step": 41490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4080580717883719e-05,
+      "loss": 1.5547,
+      "step": 41500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4069620220528138e-05,
+      "loss": 1.5456,
+      "step": 41510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4058662320033624e-05,
+      "loss": 1.511,
+      "step": 41520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.404770701900358e-05,
+      "loss": 1.5496,
+      "step": 41530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4036754320040796e-05,
+      "loss": 1.4655,
+      "step": 41540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4025804225747402e-05,
+      "loss": 1.5148,
+      "step": 41550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4014856738724985e-05,
+      "loss": 1.5204,
+      "step": 41560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4003911861574443e-05,
+      "loss": 1.5484,
+      "step": 41570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3992969596896088e-05,
+      "loss": 1.498,
+      "step": 41580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3982029947289607e-05,
+      "loss": 1.522,
+      "step": 41590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3971092915354073e-05,
+      "loss": 1.5372,
+      "step": 41600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3960158503687895e-05,
+      "loss": 1.5364,
+      "step": 41610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3949226714888936e-05,
+      "loss": 1.5487,
+      "step": 41620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3938297551554359e-05,
+      "loss": 1.4983,
+      "step": 41630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3927371016280743e-05,
+      "loss": 1.532,
+      "step": 41640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3916447111664043e-05,
+      "loss": 1.5035,
+      "step": 41650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3905525840299582e-05,
+      "loss": 1.5163,
+      "step": 41660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3894607204782051e-05,
+      "loss": 1.5296,
+      "step": 41670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3883691207705519e-05,
+      "loss": 1.4876,
+      "step": 41680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3872777851663438e-05,
+      "loss": 1.4934,
+      "step": 41690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3861867139248618e-05,
+      "loss": 1.4925,
+      "step": 41700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3850959073053262e-05,
+      "loss": 1.5343,
+      "step": 41710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.38400536556689e-05,
+      "loss": 1.5292,
+      "step": 41720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3829150889686495e-05,
+      "loss": 1.5309,
+      "step": 41730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3818250777696329e-05,
+      "loss": 1.5062,
+      "step": 41740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3807353322288068e-05,
+      "loss": 1.5219,
+      "step": 41750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3796458526050765e-05,
+      "loss": 1.5074,
+      "step": 41760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.378556639157283e-05,
+      "loss": 1.5528,
+      "step": 41770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.377467692144201e-05,
+      "loss": 1.4984,
+      "step": 41780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3763790118245479e-05,
+      "loss": 1.5192,
+      "step": 41790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3752905984569724e-05,
+      "loss": 1.5097,
+      "step": 41800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.374202452300063e-05,
+      "loss": 1.5143,
+      "step": 41810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3731145736123427e-05,
+      "loss": 1.5071,
+      "step": 41820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3720269626522726e-05,
+      "loss": 1.5275,
+      "step": 41830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.370939619678249e-05,
+      "loss": 1.519,
+      "step": 41840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3698525449486053e-05,
+      "loss": 1.5167,
+      "step": 41850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3687657387216116e-05,
+      "loss": 1.5147,
+      "step": 41860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3676792012554706e-05,
+      "loss": 1.5113,
+      "step": 41870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3665929328083279e-05,
+      "loss": 1.5001,
+      "step": 41880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3655069336382587e-05,
+      "loss": 1.5319,
+      "step": 41890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3644212040032773e-05,
+      "loss": 1.5315,
+      "step": 41900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3633357441613334e-05,
+      "loss": 1.531,
+      "step": 41910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3622505543703131e-05,
+      "loss": 1.5634,
+      "step": 41920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3611656348880377e-05,
+      "loss": 1.5178,
+      "step": 41930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3600809859722658e-05,
+      "loss": 1.5812,
+      "step": 41940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.358996607880687e-05,
+      "loss": 1.5584,
+      "step": 41950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3579125008709342e-05,
+      "loss": 1.5125,
+      "step": 41960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3568286652005685e-05,
+      "loss": 1.5227,
+      "step": 41970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3557451011270908e-05,
+      "loss": 1.5133,
+      "step": 41980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3546618089079363e-05,
+      "loss": 1.5033,
+      "step": 41990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3535787888004759e-05,
+      "loss": 1.518,
+      "step": 42000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3524960410620157e-05,
+      "loss": 1.5011,
+      "step": 42010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3514135659497968e-05,
+      "loss": 1.5244,
+      "step": 42020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3503313637209963e-05,
+      "loss": 1.5127,
+      "step": 42030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3492494346327239e-05,
+      "loss": 1.5169,
+      "step": 42040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.34816777894203e-05,
+      "loss": 1.5429,
+      "step": 42050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3470863969058933e-05,
+      "loss": 1.5287,
+      "step": 42060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3460052887812319e-05,
+      "loss": 1.5074,
+      "step": 42070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3449244548248977e-05,
+      "loss": 1.505,
+      "step": 42080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3438438952936778e-05,
+      "loss": 1.501,
+      "step": 42090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3427636104442915e-05,
+      "loss": 1.4436,
+      "step": 42100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.341683600533398e-05,
+      "loss": 1.4839,
+      "step": 42110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3406038658175856e-05,
+      "loss": 1.5189,
+      "step": 42120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3395244065533813e-05,
+      "loss": 1.5291,
+      "step": 42130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3384452229972446e-05,
+      "loss": 1.4994,
+      "step": 42140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3373663154055702e-05,
+      "loss": 1.5158,
+      "step": 42150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3362876840346872e-05,
+      "loss": 1.5063,
+      "step": 42160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3352093291408585e-05,
+      "loss": 1.5338,
+      "step": 42170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3341312509802822e-05,
+      "loss": 1.5137,
+      "step": 42180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.33305344980909e-05,
+      "loss": 1.493,
+      "step": 42190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3319759258833492e-05,
+      "loss": 1.5025,
+      "step": 42200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.330898679459057e-05,
+      "loss": 1.4278,
+      "step": 42210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3298217107921515e-05,
+      "loss": 1.5214,
+      "step": 42220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3287450201384983e-05,
+      "loss": 1.4947,
+      "step": 42230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3276686077539008e-05,
+      "loss": 1.5312,
+      "step": 42240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3265924738940949e-05,
+      "loss": 1.5203,
+      "step": 42250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3255166188147514e-05,
+      "loss": 1.5252,
+      "step": 42260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3244410427714715e-05,
+      "loss": 1.533,
+      "step": 42270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3233657460197963e-05,
+      "loss": 1.5381,
+      "step": 42280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3222907288151943e-05,
+      "loss": 1.5358,
+      "step": 42290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.321215991413071e-05,
+      "loss": 1.5302,
+      "step": 42300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3201415340687646e-05,
+      "loss": 1.5134,
+      "step": 42310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3190673570375466e-05,
+      "loss": 1.3953,
+      "step": 42320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3179934605746224e-05,
+      "loss": 1.5472,
+      "step": 42330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3169198449351303e-05,
+      "loss": 1.5218,
+      "step": 42340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.315846510374143e-05,
+      "loss": 1.5374,
+      "step": 42350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3147734571466625e-05,
+      "loss": 1.5279,
+      "step": 42360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3137006855076306e-05,
+      "loss": 1.5498,
+      "step": 42370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.312628195711915e-05,
+      "loss": 1.5365,
+      "step": 42380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3115559880143236e-05,
+      "loss": 1.5384,
+      "step": 42390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3104840626695903e-05,
+      "loss": 1.4761,
+      "step": 42400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.309412419932387e-05,
+      "loss": 1.5717,
+      "step": 42410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.308341060057316e-05,
+      "loss": 1.5447,
+      "step": 42420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3072699832989144e-05,
+      "loss": 1.5621,
+      "step": 42430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.306199189911648e-05,
+      "loss": 1.5802,
+      "step": 42440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3051286801499212e-05,
+      "loss": 1.5573,
+      "step": 42450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3040584542680656e-05,
+      "loss": 1.5567,
+      "step": 42460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3029885125203482e-05,
+      "loss": 1.5613,
+      "step": 42470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3019188551609676e-05,
+      "loss": 1.5234,
+      "step": 42480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3008494824440557e-05,
+      "loss": 1.5484,
+      "step": 42490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.299780394623676e-05,
+      "loss": 1.5234,
+      "step": 42500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.298711591953824e-05,
+      "loss": 1.5403,
+      "step": 42510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2976430746884292e-05,
+      "loss": 1.5185,
+      "step": 42520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.296574843081349e-05,
+      "loss": 1.5198,
+      "step": 42530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2955068973863805e-05,
+      "loss": 1.5377,
+      "step": 42540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2944392378572445e-05,
+      "loss": 1.5315,
+      "step": 42550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2933718647475992e-05,
+      "loss": 1.5469,
+      "step": 42560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2923047783110332e-05,
+      "loss": 1.5341,
+      "step": 42570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2912379788010676e-05,
+      "loss": 1.5386,
+      "step": 42580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.290171466471152e-05,
+      "loss": 1.53,
+      "step": 42590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2891052415746747e-05,
+      "loss": 1.5182,
+      "step": 42600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2880393043649468e-05,
+      "loss": 1.5076,
+      "step": 42610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2869736550952203e-05,
+      "loss": 1.5436,
+      "step": 42620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2859082940186711e-05,
+      "loss": 1.5497,
+      "step": 42630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.284843221388411e-05,
+      "loss": 1.515,
+      "step": 42640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2837784374574819e-05,
+      "loss": 1.5285,
+      "step": 42650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.282713942478857e-05,
+      "loss": 1.5388,
+      "step": 42660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2816497367054411e-05,
+      "loss": 1.5157,
+      "step": 42670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.280585820390071e-05,
+      "loss": 1.52,
+      "step": 42680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.279522193785514e-05,
+      "loss": 1.6367,
+      "step": 42690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2784588571444667e-05,
+      "loss": 1.5226,
+      "step": 42700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2773958107195616e-05,
+      "loss": 1.5139,
+      "step": 42710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2763330547633572e-05,
+      "loss": 1.5306,
+      "step": 42720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2752705895283457e-05,
+      "loss": 1.5232,
+      "step": 42730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2742084152669497e-05,
+      "loss": 1.5334,
+      "step": 42740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.273146532231524e-05,
+      "loss": 1.4799,
+      "step": 42750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2720849406743496e-05,
+      "loss": 1.6232,
+      "step": 42760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2710236408476456e-05,
+      "loss": 1.5305,
+      "step": 42770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2699626330035541e-05,
+      "loss": 1.5135,
+      "step": 42780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2689019173941535e-05,
+      "loss": 1.5218,
+      "step": 42790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2678414942714498e-05,
+      "loss": 1.5411,
+      "step": 42800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2667813638873807e-05,
+      "loss": 1.5246,
+      "step": 42810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2657215264938144e-05,
+      "loss": 1.5164,
+      "step": 42820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2646619823425484e-05,
+      "loss": 1.6018,
+      "step": 42830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2636027316853119e-05,
+      "loss": 1.5423,
+      "step": 42840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2625437747737634e-05,
+      "loss": 1.5627,
+      "step": 42850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2614851118594934e-05,
+      "loss": 1.5739,
+      "step": 42860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2604267431940176e-05,
+      "loss": 1.5762,
+      "step": 42870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2593686690287899e-05,
+      "loss": 1.5671,
+      "step": 42880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2583108896151858e-05,
+      "loss": 1.6074,
+      "step": 42890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2572534052045165e-05,
+      "loss": 1.5652,
+      "step": 42900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2561962160480204e-05,
+      "loss": 1.5241,
+      "step": 42910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2551393223968683e-05,
+      "loss": 1.6026,
+      "step": 42920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2540827245021552e-05,
+      "loss": 1.5431,
+      "step": 42930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2530264226149146e-05,
+      "loss": 1.5524,
+      "step": 42940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2519704169861015e-05,
+      "loss": 1.5595,
+      "step": 42950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2509147078666048e-05,
+      "loss": 1.591,
+      "step": 42960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2498592955072415e-05,
+      "loss": 1.5608,
+      "step": 42970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2488041801587592e-05,
+      "loss": 1.5184,
+      "step": 42980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2477493620718337e-05,
+      "loss": 1.5482,
+      "step": 42990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2466948414970711e-05,
+      "loss": 1.509,
+      "step": 43000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2456406186850072e-05,
+      "loss": 1.5322,
+      "step": 43010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.244586693886104e-05,
+      "loss": 1.536,
+      "step": 43020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.243533067350758e-05,
+      "loss": 1.5403,
+      "step": 43030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2424797393292895e-05,
+      "loss": 1.5284,
+      "step": 43040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2414267100719513e-05,
+      "loss": 1.5206,
+      "step": 43050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2403739798289238e-05,
+      "loss": 1.3462,
+      "step": 43060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2393215488503169e-05,
+      "loss": 1.3785,
+      "step": 43070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2382694173861693e-05,
+      "loss": 1.534,
+      "step": 43080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.237217585686449e-05,
+      "loss": 1.5222,
+      "step": 43090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.23616605400105e-05,
+      "loss": 1.5202,
+      "step": 43100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2351148225798007e-05,
+      "loss": 1.5179,
+      "step": 43110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2340638916724517e-05,
+      "loss": 1.4851,
+      "step": 43120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2330132615286869e-05,
+      "loss": 1.5359,
+      "step": 43130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.231962932398116e-05,
+      "loss": 1.5241,
+      "step": 43140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2309129045302791e-05,
+      "loss": 1.497,
+      "step": 43150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2298631781746437e-05,
+      "loss": 1.5224,
+      "step": 43160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2288137535806057e-05,
+      "loss": 1.5101,
+      "step": 43170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2277646309974905e-05,
+      "loss": 1.4912,
+      "step": 43180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2267158106745477e-05,
+      "loss": 1.5201,
+      "step": 43190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2256672928609622e-05,
+      "loss": 1.5572,
+      "step": 43200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2246190778058398e-05,
+      "loss": 1.5244,
+      "step": 43210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2235711657582183e-05,
+      "loss": 1.5068,
+      "step": 43220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2225235569670628e-05,
+      "loss": 1.5245,
+      "step": 43230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2214762516812667e-05,
+      "loss": 1.5043,
+      "step": 43240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2204292501496506e-05,
+      "loss": 1.5127,
+      "step": 43250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2193825526209628e-05,
+      "loss": 1.4951,
+      "step": 43260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.218336159343881e-05,
+      "loss": 1.5176,
+      "step": 43270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2172900705670069e-05,
+      "loss": 1.4977,
+      "step": 43280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2162442865388751e-05,
+      "loss": 1.5078,
+      "step": 43290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.215198807507942e-05,
+      "loss": 1.5037,
+      "step": 43300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2141536337225987e-05,
+      "loss": 1.5089,
+      "step": 43310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2131087654311557e-05,
+      "loss": 1.5426,
+      "step": 43320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2120642028818568e-05,
+      "loss": 1.5138,
+      "step": 43330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2110199463228706e-05,
+      "loss": 1.5503,
+      "step": 43340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.209975996002295e-05,
+      "loss": 1.5818,
+      "step": 43350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2089323521681503e-05,
+      "loss": 1.5444,
+      "step": 43360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2078890150683919e-05,
+      "loss": 1.5192,
+      "step": 43370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2068459849508947e-05,
+      "loss": 1.5143,
+      "step": 43380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2058032620634652e-05,
+      "loss": 1.5189,
+      "step": 43390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.204760846653835e-05,
+      "loss": 1.5124,
+      "step": 43400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2037187389696633e-05,
+      "loss": 1.5127,
+      "step": 43410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2026769392585363e-05,
+      "loss": 1.5022,
+      "step": 43420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2016354477679667e-05,
+      "loss": 1.521,
+      "step": 43430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2005942647453953e-05,
+      "loss": 1.4332,
+      "step": 43440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.199553390438185e-05,
+      "loss": 1.5107,
+      "step": 43450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1985128250936331e-05,
+      "loss": 1.527,
+      "step": 43460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1974725689589559e-05,
+      "loss": 1.5226,
+      "step": 43470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1964326222813007e-05,
+      "loss": 1.5223,
+      "step": 43480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.19539298530774e-05,
+      "loss": 1.5141,
+      "step": 43490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1943536582852738e-05,
+      "loss": 1.5161,
+      "step": 43500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1933146414608245e-05,
+      "loss": 1.5118,
+      "step": 43510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1922759350812476e-05,
+      "loss": 1.4786,
+      "step": 43520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.191237539393317e-05,
+      "loss": 1.5104,
+      "step": 43530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1901994546437411e-05,
+      "loss": 1.5261,
+      "step": 43540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1891616810791467e-05,
+      "loss": 1.5086,
+      "step": 43550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1881242189460915e-05,
+      "loss": 1.5117,
+      "step": 43560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1870870684910573e-05,
+      "loss": 1.5155,
+      "step": 43570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1860502299604528e-05,
+      "loss": 1.5052,
+      "step": 43580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.185013703600612e-05,
+      "loss": 1.5181,
+      "step": 43590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1839774896577946e-05,
+      "loss": 1.5074,
+      "step": 43600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1829415883781878e-05,
+      "loss": 1.4969,
+      "step": 43610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1819060000078993e-05,
+      "loss": 1.5028,
+      "step": 43620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1808707247929712e-05,
+      "loss": 1.5016,
+      "step": 43630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1798357629793624e-05,
+      "loss": 1.5091,
+      "step": 43640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1788011148129626e-05,
+      "loss": 1.5267,
+      "step": 43650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1777667805395856e-05,
+      "loss": 1.4964,
+      "step": 43660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1767327604049714e-05,
+      "loss": 1.5434,
+      "step": 43670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1756990546547813e-05,
+      "loss": 1.5185,
+      "step": 43680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1746656635346093e-05,
+      "loss": 1.5269,
+      "step": 43690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1736325872899678e-05,
+      "loss": 1.5073,
+      "step": 43700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1725998261662977e-05,
+      "loss": 1.5338,
+      "step": 43710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1715673804089647e-05,
+      "loss": 1.5157,
+      "step": 43720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1705352502632588e-05,
+      "loss": 1.5234,
+      "step": 43730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1695034359743961e-05,
+      "loss": 1.5081,
+      "step": 43740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1684719377875168e-05,
+      "loss": 1.5071,
+      "step": 43750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1674407559476863e-05,
+      "loss": 1.5125,
+      "step": 43760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1664098906998947e-05,
+      "loss": 1.5049,
+      "step": 43770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.165379342289058e-05,
+      "loss": 1.5121,
+      "step": 43780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1643491109600129e-05,
+      "loss": 1.5084,
+      "step": 43790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.163319196957528e-05,
+      "loss": 1.5171,
+      "step": 43800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1622896005262893e-05,
+      "loss": 1.4973,
+      "step": 43810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1612603219109108e-05,
+      "loss": 1.5074,
+      "step": 43820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1602313613559313e-05,
+      "loss": 1.5245,
+      "step": 43830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1592027191058136e-05,
+      "loss": 1.5254,
+      "step": 43840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1581743954049418e-05,
+      "loss": 1.5041,
+      "step": 43850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1571463904976313e-05,
+      "loss": 1.4813,
+      "step": 43860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1561187046281141e-05,
+      "loss": 1.502,
+      "step": 43870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1550913380405509e-05,
+      "loss": 1.511,
+      "step": 43880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1540642909790258e-05,
+      "loss": 1.5054,
+      "step": 43890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1530375636875464e-05,
+      "loss": 1.5367,
+      "step": 43900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.152011156410045e-05,
+      "loss": 1.4836,
+      "step": 43910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1509850693903768e-05,
+      "loss": 1.5129,
+      "step": 43920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1499593028723226e-05,
+      "loss": 1.538,
+      "step": 43930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1489338570995836e-05,
+      "loss": 1.5445,
+      "step": 43940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1479087323157909e-05,
+      "loss": 1.5766,
+      "step": 43950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1468839287644926e-05,
+      "loss": 1.5463,
+      "step": 43960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1458594466891644e-05,
+      "loss": 1.5008,
+      "step": 43970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1448352863332054e-05,
+      "loss": 1.4844,
+      "step": 43980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.143811447939937e-05,
+      "loss": 1.4891,
+      "step": 43990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1427879317526047e-05,
+      "loss": 1.4914,
+      "step": 44000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1417647380143786e-05,
+      "loss": 1.4975,
+      "step": 44010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1407418669683487e-05,
+      "loss": 1.5058,
+      "step": 44020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1397193188575339e-05,
+      "loss": 1.4873,
+      "step": 44030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1386970939248703e-05,
+      "loss": 1.4879,
+      "step": 44040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1376751924132214e-05,
+      "loss": 1.4851,
+      "step": 44050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1366536145653724e-05,
+      "loss": 1.4301,
+      "step": 44060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1356323606240321e-05,
+      "loss": 1.5252,
+      "step": 44070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1346114308318315e-05,
+      "loss": 1.506,
+      "step": 44080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1335908254313251e-05,
+      "loss": 1.4936,
+      "step": 44090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1325705446649917e-05,
+      "loss": 1.4777,
+      "step": 44100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1315505887752287e-05,
+      "loss": 1.5018,
+      "step": 44110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1305309580043627e-05,
+      "loss": 1.4953,
+      "step": 44120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.129511652594637e-05,
+      "loss": 1.4971,
+      "step": 44130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.128492672788221e-05,
+      "loss": 1.4569,
+      "step": 44140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1274740188272061e-05,
+      "loss": 1.4674,
+      "step": 44150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1264556909536068e-05,
+      "loss": 1.4537,
+      "step": 44160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1254376894093569e-05,
+      "loss": 1.4497,
+      "step": 44170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1244200144363184e-05,
+      "loss": 1.4994,
+      "step": 44180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1234026662762704e-05,
+      "loss": 1.4864,
+      "step": 44190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.122385645170917e-05,
+      "loss": 1.4853,
+      "step": 44200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1213689513618842e-05,
+      "loss": 1.484,
+      "step": 44210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1203525850907198e-05,
+      "loss": 1.5006,
+      "step": 44220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1193365465988948e-05,
+      "loss": 1.4791,
+      "step": 44230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1183208361278007e-05,
+      "loss": 1.4832,
+      "step": 44240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1173054539187527e-05,
+      "loss": 1.463,
+      "step": 44250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1162904002129873e-05,
+      "loss": 1.475,
+      "step": 44260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1152756752516633e-05,
+      "loss": 1.4567,
+      "step": 44270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.114261279275859e-05,
+      "loss": 1.5045,
+      "step": 44280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1132472125265794e-05,
+      "loss": 1.5168,
+      "step": 44290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1122334752447463e-05,
+      "loss": 1.5157,
+      "step": 44300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1112200676712064e-05,
+      "loss": 1.5085,
+      "step": 44310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1102069900467269e-05,
+      "loss": 1.4969,
+      "step": 44320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1091942426119975e-05,
+      "loss": 1.5195,
+      "step": 44330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1081818256076263e-05,
+      "loss": 1.5465,
+      "step": 44340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1071697392741484e-05,
+      "loss": 1.5344,
+      "step": 44350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1061579838520153e-05,
+      "loss": 1.4999,
+      "step": 44360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1051465595816025e-05,
+      "loss": 1.4783,
+      "step": 44370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1041354667032056e-05,
+      "loss": 1.4925,
+      "step": 44380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1031247054570427e-05,
+      "loss": 1.4775,
+      "step": 44390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.102114276083252e-05,
+      "loss": 1.4643,
+      "step": 44400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1011041788218937e-05,
+      "loss": 1.4822,
+      "step": 44410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1000944139129494e-05,
+      "loss": 1.4745,
+      "step": 44420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0990849815963181e-05,
+      "loss": 1.4768,
+      "step": 44430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0980758821118267e-05,
+      "loss": 1.4961,
+      "step": 44440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0970671156992152e-05,
+      "loss": 1.506,
+      "step": 44450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.096058682598152e-05,
+      "loss": 1.4897,
+      "step": 44460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0950505830482196e-05,
+      "loss": 1.4805,
+      "step": 44470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0940428172889258e-05,
+      "loss": 1.4837,
+      "step": 44480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0930353855596973e-05,
+      "loss": 1.4574,
+      "step": 44490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0920282880998823e-05,
+      "loss": 1.5058,
+      "step": 44500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0910215251487462e-05,
+      "loss": 1.4799,
+      "step": 44510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.090015096945482e-05,
+      "loss": 1.478,
+      "step": 44520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0890090037291956e-05,
+      "loss": 1.4905,
+      "step": 44530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0880032457389177e-05,
+      "loss": 1.4843,
+      "step": 44540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0869978232135985e-05,
+      "loss": 1.5049,
+      "step": 44550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.085992736392108e-05,
+      "loss": 1.508,
+      "step": 44560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0849879855132365e-05,
+      "loss": 1.4822,
+      "step": 44570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.083983570815695e-05,
+      "loss": 1.4636,
+      "step": 44580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0829794925381151e-05,
+      "loss": 1.4644,
+      "step": 44590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0819757509190448e-05,
+      "loss": 1.5176,
+      "step": 44600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.080972346196959e-05,
+      "loss": 1.5004,
+      "step": 44610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0799692786102458e-05,
+      "loss": 1.4904,
+      "step": 44620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0789665483972166e-05,
+      "loss": 1.4863,
+      "step": 44630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.077964155796102e-05,
+      "loss": 1.4809,
+      "step": 44640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0769621010450537e-05,
+      "loss": 1.478,
+      "step": 44650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0759603843821387e-05,
+      "loss": 1.498,
+      "step": 44660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0749590060453504e-05,
+      "loss": 1.4965,
+      "step": 44670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0739579662725951e-05,
+      "loss": 1.4947,
+      "step": 44680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.072957265301705e-05,
+      "loss": 1.4781,
+      "step": 44690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0719569033704261e-05,
+      "loss": 1.4702,
+      "step": 44700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0709568807164273e-05,
+      "loss": 1.4836,
+      "step": 44710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0699571975772954e-05,
+      "loss": 1.4885,
+      "step": 44720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0689578541905376e-05,
+      "loss": 1.5019,
+      "step": 44730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0679588507935798e-05,
+      "loss": 1.4694,
+      "step": 44740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0669601876237672e-05,
+      "loss": 1.5008,
+      "step": 44750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0659618649183647e-05,
+      "loss": 1.4867,
+      "step": 44760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0649638829145537e-05,
+      "loss": 1.4744,
+      "step": 44770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0639662418494399e-05,
+      "loss": 1.4877,
+      "step": 44780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.062968941960042e-05,
+      "loss": 1.4988,
+      "step": 44790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0619719834833016e-05,
+      "loss": 1.4964,
+      "step": 44800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0609753666560781e-05,
+      "loss": 1.4822,
+      "step": 44810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0599790917151506e-05,
+      "loss": 1.4774,
+      "step": 44820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.058983158897213e-05,
+      "loss": 1.4701,
+      "step": 44830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0579875684388849e-05,
+      "loss": 1.4151,
+      "step": 44840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.056992320576698e-05,
+      "loss": 1.4966,
+      "step": 44850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0559974155471059e-05,
+      "loss": 1.5007,
+      "step": 44860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.05500285358648e-05,
+      "loss": 1.494,
+      "step": 44870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0540086349311105e-05,
+      "loss": 1.512,
+      "step": 44880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0530147598172058e-05,
+      "loss": 1.5136,
+      "step": 44890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0520212284808927e-05,
+      "loss": 1.5578,
+      "step": 44900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0510280411582163e-05,
+      "loss": 1.5213,
+      "step": 44910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.05003519808514e-05,
+      "loss": 1.5169,
+      "step": 44920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0490426994975464e-05,
+      "loss": 1.4906,
+      "step": 44930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0480505456312323e-05,
+      "loss": 1.4947,
+      "step": 44940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0470587367219195e-05,
+      "loss": 1.5001,
+      "step": 44950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0460672730052406e-05,
+      "loss": 1.4882,
+      "step": 44960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0450761547167512e-05,
+      "loss": 1.5152,
+      "step": 44970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0440853820919228e-05,
+      "loss": 1.4759,
+      "step": 44980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0430949553661457e-05,
+      "loss": 1.4706,
+      "step": 44990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.042104874774725e-05,
+      "loss": 1.4779,
+      "step": 45000
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 1.5883851051330566,
+      "eval_runtime": 81.9958,
+      "eval_samples_per_second": 36.587,
+      "eval_steps_per_second": 4.573,
+      "step": 45000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0411151405528896e-05,
+      "loss": 1.4955,
+      "step": 45010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.04012575293578e-05,
+      "loss": 1.4941,
+      "step": 45020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0391367121584573e-05,
+      "loss": 1.5202,
+      "step": 45030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0381480184558998e-05,
+      "loss": 1.4977,
+      "step": 45040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0371596720630036e-05,
+      "loss": 1.5057,
+      "step": 45050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0361716732145818e-05,
+      "loss": 1.5014,
+      "step": 45060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0351840221453651e-05,
+      "loss": 1.4775,
+      "step": 45070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0341967190900027e-05,
+      "loss": 1.4694,
+      "step": 45080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0332097642830569e-05,
+      "loss": 1.4855,
+      "step": 45090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0322231579590142e-05,
+      "loss": 1.4839,
+      "step": 45100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.031236900352271e-05,
+      "loss": 1.5023,
+      "step": 45110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0302509916971473e-05,
+      "loss": 1.4768,
+      "step": 45120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0292654322278753e-05,
+      "loss": 1.5087,
+      "step": 45130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0282802221786066e-05,
+      "loss": 1.5031,
+      "step": 45140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0272953617834095e-05,
+      "loss": 1.4925,
+      "step": 45150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0263108512762698e-05,
+      "loss": 1.4866,
+      "step": 45160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0253266908910868e-05,
+      "loss": 1.4695,
+      "step": 45170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0243428808616831e-05,
+      "loss": 1.4682,
+      "step": 45180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0233594214217912e-05,
+      "loss": 1.4931,
+      "step": 45190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0223763128050645e-05,
+      "loss": 1.5851,
+      "step": 45200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0213935552450715e-05,
+      "loss": 1.4837,
+      "step": 45210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.020411148975298e-05,
+      "loss": 1.4793,
+      "step": 45220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.019429094229146e-05,
+      "loss": 1.4906,
+      "step": 45230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0184473912399337e-05,
+      "loss": 1.5011,
+      "step": 45240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.017466040240897e-05,
+      "loss": 1.4972,
+      "step": 45250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0164850414651847e-05,
+      "loss": 1.5116,
+      "step": 45260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0155043951458679e-05,
+      "loss": 1.5027,
+      "step": 45270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0145241015159279e-05,
+      "loss": 1.4907,
+      "step": 45280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0135441608082654e-05,
+      "loss": 1.4682,
+      "step": 45290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0125645732556968e-05,
+      "loss": 1.4876,
+      "step": 45300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0115853390909555e-05,
+      "loss": 1.4773,
+      "step": 45310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.010606458546687e-05,
+      "loss": 1.5428,
+      "step": 45320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0096279318554591e-05,
+      "loss": 1.4599,
+      "step": 45330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0086497592497488e-05,
+      "loss": 1.4766,
+      "step": 45340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0076719409619558e-05,
+      "loss": 1.4779,
+      "step": 45350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.006694477224389e-05,
+      "loss": 1.4641,
+      "step": 45360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0057173682692777e-05,
+      "loss": 1.4916,
+      "step": 45370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0047406143287649e-05,
+      "loss": 1.5116,
+      "step": 45380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.00376421563491e-05,
+      "loss": 1.4928,
+      "step": 45390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0027881724196875e-05,
+      "loss": 1.4831,
+      "step": 45400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0018124849149873e-05,
+      "loss": 1.4732,
+      "step": 45410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.000837153352617e-05,
+      "loss": 1.4853,
+      "step": 45420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.998621779642944e-06,
+      "loss": 1.5264,
+      "step": 45430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.988875589816599e-06,
+      "loss": 1.5323,
+      "step": 45440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.979132966362625e-06,
+      "loss": 1.4899,
+      "step": 45450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.969393911595703e-06,
+      "loss": 1.5472,
+      "step": 45460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.959658427829658e-06,
+      "loss": 1.5268,
+      "step": 45470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.949926517377467e-06,
+      "loss": 1.5686,
+      "step": 45480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.940198182551253e-06,
+      "loss": 1.5553,
+      "step": 45490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.930473425662293e-06,
+      "loss": 1.5395,
+      "step": 45500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.920752249021022e-06,
+      "loss": 1.5152,
+      "step": 45510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.911034654936997e-06,
+      "loss": 1.576,
+      "step": 45520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.901320645718966e-06,
+      "loss": 1.5625,
+      "step": 45530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.891610223674789e-06,
+      "loss": 1.5382,
+      "step": 45540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.881903391111486e-06,
+      "loss": 1.5183,
+      "step": 45550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.872200150335228e-06,
+      "loss": 1.4914,
+      "step": 45560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.862500503651334e-06,
+      "loss": 1.5182,
+      "step": 45570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.852804453364259e-06,
+      "loss": 1.4673,
+      "step": 45580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.843112001777624e-06,
+      "loss": 1.4805,
+      "step": 45590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.833423151194152e-06,
+      "loss": 1.5006,
+      "step": 45600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.823737903915772e-06,
+      "loss": 1.5299,
+      "step": 45610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.8140562622435e-06,
+      "loss": 1.5052,
+      "step": 45620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.804378228477529e-06,
+      "loss": 1.4895,
+      "step": 45630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.794703804917183e-06,
+      "loss": 1.4627,
+      "step": 45640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.785032993860933e-06,
+      "loss": 1.4412,
+      "step": 45650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.775365797606387e-06,
+      "loss": 1.3176,
+      "step": 45660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.765702218450298e-06,
+      "loss": 1.4935,
+      "step": 45670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.756042258688566e-06,
+      "loss": 1.491,
+      "step": 45680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.746385920616196e-06,
+      "loss": 1.5005,
+      "step": 45690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.736733206527396e-06,
+      "loss": 1.5159,
+      "step": 45700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.727084118715452e-06,
+      "loss": 1.4647,
+      "step": 45710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.717438659472819e-06,
+      "loss": 1.459,
+      "step": 45720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.707796831091087e-06,
+      "loss": 1.468,
+      "step": 45730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.698158635860988e-06,
+      "loss": 1.4579,
+      "step": 45740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.688524076072355e-06,
+      "loss": 1.4844,
+      "step": 45750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.678893154014224e-06,
+      "loss": 1.4929,
+      "step": 45760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.669265871974698e-06,
+      "loss": 1.4861,
+      "step": 45770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.659642232241058e-06,
+      "loss": 1.4981,
+      "step": 45780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.650022237099707e-06,
+      "loss": 1.4822,
+      "step": 45790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.640405888836182e-06,
+      "loss": 1.4686,
+      "step": 45800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.630793189735152e-06,
+      "loss": 1.4509,
+      "step": 45810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.621184142080422e-06,
+      "loss": 1.4935,
+      "step": 45820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.611578748154928e-06,
+      "loss": 1.4957,
+      "step": 45830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.60197701024074e-06,
+      "loss": 1.5428,
+      "step": 45840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.592378930619064e-06,
+      "loss": 1.508,
+      "step": 45850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.582784511570206e-06,
+      "loss": 1.4563,
+      "step": 45860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.573193755373663e-06,
+      "loss": 1.4817,
+      "step": 45870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.563606664307997e-06,
+      "loss": 1.4697,
+      "step": 45880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.554023240650939e-06,
+      "loss": 1.4858,
+      "step": 45890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.544443486679333e-06,
+      "loss": 1.4932,
+      "step": 45900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.534867404669171e-06,
+      "loss": 1.4996,
+      "step": 45910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.525294996895526e-06,
+      "loss": 1.4784,
+      "step": 45920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.515726265632665e-06,
+      "loss": 1.4554,
+      "step": 45930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.506161213153924e-06,
+      "loss": 1.4681,
+      "step": 45940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.49659984173179e-06,
+      "loss": 1.3982,
+      "step": 45950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.48704215363788e-06,
+      "loss": 1.4924,
+      "step": 45960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.477488151142917e-06,
+      "loss": 1.5032,
+      "step": 45970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.467937836516768e-06,
+      "loss": 1.512,
+      "step": 45980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.458391212028414e-06,
+      "loss": 1.5217,
+      "step": 45990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.448848279945968e-06,
+      "loss": 1.4993,
+      "step": 46000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.439309042536633e-06,
+      "loss": 1.4774,
+      "step": 46010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.429773502066796e-06,
+      "loss": 1.4808,
+      "step": 46020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.420241660801895e-06,
+      "loss": 1.4738,
+      "step": 46030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.410713521006554e-06,
+      "loss": 1.4941,
+      "step": 46040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.401189084944464e-06,
+      "loss": 1.4982,
+      "step": 46050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.39166835487847e-06,
+      "loss": 1.4676,
+      "step": 46060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.382151333070522e-06,
+      "loss": 1.4917,
+      "step": 46070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.372638021781701e-06,
+      "loss": 1.492,
+      "step": 46080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.363128423272178e-06,
+      "loss": 1.4877,
+      "step": 46090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.353622539801288e-06,
+      "loss": 1.4719,
+      "step": 46100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.344120373627438e-06,
+      "loss": 1.5018,
+      "step": 46110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.33462192700818e-06,
+      "loss": 1.5436,
+      "step": 46120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.32512720220017e-06,
+      "loss": 1.5147,
+      "step": 46130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.315636201459185e-06,
+      "loss": 1.5054,
+      "step": 46140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.306148927040115e-06,
+      "loss": 1.4983,
+      "step": 46150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.296665381196967e-06,
+      "loss": 1.5413,
+      "step": 46160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.287185566182869e-06,
+      "loss": 1.5103,
+      "step": 46170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.277709484250024e-06,
+      "loss": 1.5451,
+      "step": 46180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.268237137649818e-06,
+      "loss": 1.5703,
+      "step": 46190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.25876852863268e-06,
+      "loss": 1.5122,
+      "step": 46200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.249303659448192e-06,
+      "loss": 1.5228,
+      "step": 46210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.239842532345036e-06,
+      "loss": 1.502,
+      "step": 46220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.23038514957101e-06,
+      "loss": 1.5609,
+      "step": 46230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.220931513372996e-06,
+      "loss": 1.4846,
+      "step": 46240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.21148162599704e-06,
+      "loss": 1.4585,
+      "step": 46250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.202035489688233e-06,
+      "loss": 1.4693,
+      "step": 46260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.192593106690837e-06,
+      "loss": 1.4675,
+      "step": 46270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.183154479248165e-06,
+      "loss": 1.4985,
+      "step": 46280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.17371960960268e-06,
+      "loss": 1.511,
+      "step": 46290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.164288499995929e-06,
+      "loss": 1.5003,
+      "step": 46300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.15486115266858e-06,
+      "loss": 1.4743,
+      "step": 46310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.145437569860399e-06,
+      "loss": 1.473,
+      "step": 46320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.136017753810255e-06,
+      "loss": 1.5164,
+      "step": 46330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.12660170675614e-06,
+      "loss": 1.2634,
+      "step": 46340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.117189430935111e-06,
+      "loss": 1.2658,
+      "step": 46350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.107780928583387e-06,
+      "loss": 1.4907,
+      "step": 46360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.098376201936231e-06,
+      "loss": 1.4938,
+      "step": 46370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.088975253228049e-06,
+      "loss": 1.4837,
+      "step": 46380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.079578084692336e-06,
+      "loss": 1.5032,
+      "step": 46390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.070184698561698e-06,
+      "loss": 1.5036,
+      "step": 46400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.060795097067807e-06,
+      "loss": 1.4836,
+      "step": 46410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.051409282441497e-06,
+      "loss": 1.4657,
+      "step": 46420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.042027256912646e-06,
+      "loss": 1.4568,
+      "step": 46430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.03264902271026e-06,
+      "loss": 1.5213,
+      "step": 46440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.023274582062438e-06,
+      "loss": 1.4912,
+      "step": 46450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.013903937196377e-06,
+      "loss": 1.4793,
+      "step": 46460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.004537090338377e-06,
+      "loss": 1.4794,
+      "step": 46470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.995174043713828e-06,
+      "loss": 1.4688,
+      "step": 46480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.985814799547224e-06,
+      "loss": 1.4745,
+      "step": 46490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.97645936006215e-06,
+      "loss": 1.4903,
+      "step": 46500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.9671077274813e-06,
+      "loss": 1.463,
+      "step": 46510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.957759904026427e-06,
+      "loss": 1.4662,
+      "step": 46520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.948415891918438e-06,
+      "loss": 1.4628,
+      "step": 46530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.93907569337728e-06,
+      "loss": 1.511,
+      "step": 46540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.92973931062202e-06,
+      "loss": 1.4924,
+      "step": 46550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.920406745870816e-06,
+      "loss": 1.4794,
+      "step": 46560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.911078001340924e-06,
+      "loss": 1.4765,
+      "step": 46570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.901753079248665e-06,
+      "loss": 1.4915,
+      "step": 46580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.892431981809499e-06,
+      "loss": 1.4789,
+      "step": 46590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.883114711237931e-06,
+      "loss": 1.4648,
+      "step": 46600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.87380126974758e-06,
+      "loss": 1.4777,
+      "step": 46610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.864491659551157e-06,
+      "loss": 1.4521,
+      "step": 46620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.855185882860454e-06,
+      "loss": 1.4541,
+      "step": 46630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.845883941886358e-06,
+      "loss": 1.4562,
+      "step": 46640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.836585838838837e-06,
+      "loss": 1.4695,
+      "step": 46650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.827291575926966e-06,
+      "loss": 1.478,
+      "step": 46660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.818001155358869e-06,
+      "loss": 1.4659,
+      "step": 46670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.808714579341814e-06,
+      "loss": 1.4782,
+      "step": 46680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.799431850082099e-06,
+      "loss": 1.4849,
+      "step": 46690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.790152969785143e-06,
+      "loss": 1.4688,
+      "step": 46700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.780877940655441e-06,
+      "loss": 1.4743,
+      "step": 46710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.771606764896568e-06,
+      "loss": 1.4887,
+      "step": 46720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.762339444711192e-06,
+      "loss": 1.4866,
+      "step": 46730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.75307598230107e-06,
+      "loss": 1.4962,
+      "step": 46740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.743816379867007e-06,
+      "loss": 1.5155,
+      "step": 46750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.73456063960895e-06,
+      "loss": 1.5367,
+      "step": 46760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.72530876372587e-06,
+      "loss": 1.507,
+      "step": 46770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.716060754415861e-06,
+      "loss": 1.496,
+      "step": 46780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.706816613876076e-06,
+      "loss": 1.4854,
+      "step": 46790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.697576344302758e-06,
+      "loss": 1.5139,
+      "step": 46800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.688339947891228e-06,
+      "loss": 1.4783,
+      "step": 46810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.67910742683589e-06,
+      "loss": 1.488,
+      "step": 46820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.66987878333023e-06,
+      "loss": 1.4473,
+      "step": 46830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.66065401956678e-06,
+      "loss": 1.4943,
+      "step": 46840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.651433137737217e-06,
+      "loss": 1.4954,
+      "step": 46850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.642216140032228e-06,
+      "loss": 1.4675,
+      "step": 46860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.633003028641611e-06,
+      "loss": 1.4741,
+      "step": 46870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.623793805754241e-06,
+      "loss": 1.4591,
+      "step": 46880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.614588473558066e-06,
+      "loss": 1.4226,
+      "step": 46890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.605387034240087e-06,
+      "loss": 1.485,
+      "step": 46900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.596189489986428e-06,
+      "loss": 1.4972,
+      "step": 46910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.586995842982236e-06,
+      "loss": 1.487,
+      "step": 46920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.577806095411767e-06,
+      "loss": 1.4927,
+      "step": 46930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.568620249458336e-06,
+      "loss": 1.508,
+      "step": 46940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.559438307304338e-06,
+      "loss": 1.4723,
+      "step": 46950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.550260271131236e-06,
+      "loss": 1.4673,
+      "step": 46960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.541086143119561e-06,
+      "loss": 1.4407,
+      "step": 46970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.531915925448925e-06,
+      "loss": 1.4962,
+      "step": 46980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.522749620298004e-06,
+      "loss": 1.4612,
+      "step": 46990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.513587229844556e-06,
+      "loss": 1.4714,
+      "step": 47000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.504428756265375e-06,
+      "loss": 1.4823,
+      "step": 47010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.495274201736384e-06,
+      "loss": 1.4978,
+      "step": 47020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.486123568432511e-06,
+      "loss": 1.4868,
+      "step": 47030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.476976858527794e-06,
+      "loss": 1.4599,
+      "step": 47040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.467834074195321e-06,
+      "loss": 1.4624,
+      "step": 47050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.458695217607267e-06,
+      "loss": 1.4739,
+      "step": 47060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.449560290934833e-06,
+      "loss": 1.4979,
+      "step": 47070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.440429296348343e-06,
+      "loss": 1.4847,
+      "step": 47080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.431302236017132e-06,
+      "loss": 1.4802,
+      "step": 47090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.422179112109635e-06,
+      "loss": 1.4948,
+      "step": 47100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.413059926793343e-06,
+      "loss": 1.5017,
+      "step": 47110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.403944682234808e-06,
+      "loss": 1.4524,
+      "step": 47120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.394833380599649e-06,
+      "loss": 1.441,
+      "step": 47130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.385726024052542e-06,
+      "loss": 1.4815,
+      "step": 47140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.376622614757245e-06,
+      "loss": 1.4987,
+      "step": 47150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.36752315487654e-06,
+      "loss": 1.4666,
+      "step": 47160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.35842764657232e-06,
+      "loss": 1.4762,
+      "step": 47170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.349336092005488e-06,
+      "loss": 1.4914,
+      "step": 47180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.340248493336061e-06,
+      "loss": 1.4832,
+      "step": 47190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.331164852723066e-06,
+      "loss": 1.4713,
+      "step": 47200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.322085172324623e-06,
+      "loss": 1.455,
+      "step": 47210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.3130094542979e-06,
+      "loss": 1.4844,
+      "step": 47220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.303937700799124e-06,
+      "loss": 1.4817,
+      "step": 47230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.294869913983564e-06,
+      "loss": 1.4883,
+      "step": 47240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.285806096005593e-06,
+      "loss": 1.4827,
+      "step": 47250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.276746249018585e-06,
+      "loss": 1.5028,
+      "step": 47260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.267690375175007e-06,
+      "loss": 1.5385,
+      "step": 47270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.258638476626368e-06,
+      "loss": 1.4914,
+      "step": 47280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.249590555523237e-06,
+      "loss": 1.4922,
+      "step": 47290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.24054661401524e-06,
+      "loss": 1.5184,
+      "step": 47300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.231506654251048e-06,
+      "loss": 1.477,
+      "step": 47310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.222470678378407e-06,
+      "loss": 1.4718,
+      "step": 47320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.213438688544073e-06,
+      "loss": 1.4732,
+      "step": 47330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.204410686893918e-06,
+      "loss": 1.4546,
+      "step": 47340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.195386675572809e-06,
+      "loss": 1.4826,
+      "step": 47350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.186366656724692e-06,
+      "loss": 1.472,
+      "step": 47360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.177350632492567e-06,
+      "loss": 1.4703,
+      "step": 47370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.168338605018483e-06,
+      "loss": 1.4656,
+      "step": 47380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.15933057644351e-06,
+      "loss": 1.4873,
+      "step": 47390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.15032654890783e-06,
+      "loss": 1.4679,
+      "step": 47400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.1413265245506e-06,
+      "loss": 1.5016,
+      "step": 47410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.1323305055101e-06,
+      "loss": 1.4804,
+      "step": 47420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.123338493923593e-06,
+      "loss": 1.4869,
+      "step": 47430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.114350491927433e-06,
+      "loss": 1.4836,
+      "step": 47440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.105366501657002e-06,
+      "loss": 1.4748,
+      "step": 47450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.096386525246738e-06,
+      "loss": 1.4752,
+      "step": 47460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.087410564830116e-06,
+      "loss": 1.4715,
+      "step": 47470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.078438622539666e-06,
+      "loss": 1.5017,
+      "step": 47480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.06947070050697e-06,
+      "loss": 1.4867,
+      "step": 47490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.060506800862613e-06,
+      "loss": 1.4781,
+      "step": 47500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.051546925736298e-06,
+      "loss": 1.4666,
+      "step": 47510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.042591077256695e-06,
+      "loss": 1.4919,
+      "step": 47520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.033639257551568e-06,
+      "loss": 1.475,
+      "step": 47530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.024691468747706e-06,
+      "loss": 1.4831,
+      "step": 47540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.015747712970948e-06,
+      "loss": 1.4788,
+      "step": 47550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.006807992346144e-06,
+      "loss": 1.4679,
+      "step": 47560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.997872308997248e-06,
+      "loss": 1.4902,
+      "step": 47570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.988940665047187e-06,
+      "loss": 1.4785,
+      "step": 47580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.980013062617972e-06,
+      "loss": 1.4556,
+      "step": 47590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.971089503830637e-06,
+      "loss": 1.4901,
+      "step": 47600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.96216999080526e-06,
+      "loss": 1.4582,
+      "step": 47610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.953254525660952e-06,
+      "loss": 1.4817,
+      "step": 47620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.944343110515875e-06,
+      "loss": 1.4782,
+      "step": 47630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.935435747487214e-06,
+      "loss": 1.4891,
+      "step": 47640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.9265324386912e-06,
+      "loss": 1.4592,
+      "step": 47650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.917633186243106e-06,
+      "loss": 1.4818,
+      "step": 47660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.908737992257209e-06,
+      "loss": 1.5071,
+      "step": 47670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.89984685884688e-06,
+      "loss": 1.4964,
+      "step": 47680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.89095978812447e-06,
+      "loss": 1.4745,
+      "step": 47690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.882076782201389e-06,
+      "loss": 1.4741,
+      "step": 47700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.873197843188085e-06,
+      "loss": 1.4745,
+      "step": 47710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.86432297319403e-06,
+      "loss": 1.454,
+      "step": 47720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.855452174327732e-06,
+      "loss": 1.4666,
+      "step": 47730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.846585448696738e-06,
+      "loss": 1.4895,
+      "step": 47740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.837722798407626e-06,
+      "loss": 1.456,
+      "step": 47750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.828864225565977e-06,
+      "loss": 1.4764,
+      "step": 47760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.820009732276465e-06,
+      "loss": 1.4636,
+      "step": 47770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.811159320642728e-06,
+      "loss": 1.4285,
+      "step": 47780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.802312992767477e-06,
+      "loss": 1.5169,
+      "step": 47790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.793470750752435e-06,
+      "loss": 1.4799,
+      "step": 47800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.784632596698374e-06,
+      "loss": 1.5157,
+      "step": 47810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.775798532705047e-06,
+      "loss": 1.4954,
+      "step": 47820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.76696856087131e-06,
+      "loss": 1.5439,
+      "step": 47830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.758142683294974e-06,
+      "loss": 1.5205,
+      "step": 47840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.749320902072918e-06,
+      "loss": 1.514,
+      "step": 47850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.74050321930104e-06,
+      "loss": 1.4671,
+      "step": 47860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.73168963707426e-06,
+      "loss": 1.4976,
+      "step": 47870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.722880157486529e-06,
+      "loss": 1.4714,
+      "step": 47880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.714074782630818e-06,
+      "loss": 1.4916,
+      "step": 47890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.705273514599126e-06,
+      "loss": 1.489,
+      "step": 47900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.696476355482473e-06,
+      "loss": 1.4979,
+      "step": 47910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.687683307370919e-06,
+      "loss": 1.4943,
+      "step": 47920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.678894372353504e-06,
+      "loss": 1.4786,
+      "step": 47930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.670109552518356e-06,
+      "loss": 1.46,
+      "step": 47940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.661328849952565e-06,
+      "loss": 1.4619,
+      "step": 47950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.652552266742272e-06,
+      "loss": 1.5009,
+      "step": 47960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.643779804972642e-06,
+      "loss": 1.4769,
+      "step": 47970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.635011466727853e-06,
+      "loss": 1.4845,
+      "step": 47980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.626247254091087e-06,
+      "loss": 1.4777,
+      "step": 47990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.6174871691445935e-06,
+      "loss": 1.4806,
+      "step": 48000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.608731213969586e-06,
+      "loss": 1.4884,
+      "step": 48010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.599979390646325e-06,
+      "loss": 1.4958,
+      "step": 48020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.591231701254092e-06,
+      "loss": 1.4857,
+      "step": 48030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.58248814787118e-06,
+      "loss": 1.4582,
+      "step": 48040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.573748732574895e-06,
+      "loss": 1.4682,
+      "step": 48050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.56501345744157e-06,
+      "loss": 1.4969,
+      "step": 48060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.556282324546551e-06,
+      "loss": 1.4768,
+      "step": 48070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.547555335964179e-06,
+      "loss": 1.482,
+      "step": 48080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.5388324937678554e-06,
+      "loss": 1.4827,
+      "step": 48090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.5301138000299445e-06,
+      "loss": 1.4932,
+      "step": 48100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.521399256821879e-06,
+      "loss": 1.4832,
+      "step": 48110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.512688866214051e-06,
+      "loss": 1.4903,
+      "step": 48120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.503982630275904e-06,
+      "loss": 1.496,
+      "step": 48130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.4952805510758835e-06,
+      "loss": 1.4717,
+      "step": 48140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.486582630681452e-06,
+      "loss": 1.4708,
+      "step": 48150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.477888871159056e-06,
+      "loss": 1.614,
+      "step": 48160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.469199274574204e-06,
+      "loss": 1.4598,
+      "step": 48170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.460513842991368e-06,
+      "loss": 1.4795,
+      "step": 48180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.451832578474058e-06,
+      "loss": 1.5045,
+      "step": 48190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.443155483084782e-06,
+      "loss": 1.4814,
+      "step": 48200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.434482558885064e-06,
+      "loss": 1.4969,
+      "step": 48210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.4258138079354335e-06,
+      "loss": 1.477,
+      "step": 48220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.4171492322954285e-06,
+      "loss": 1.4691,
+      "step": 48230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.408488834023605e-06,
+      "loss": 1.4571,
+      "step": 48240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.399832615177496e-06,
+      "loss": 1.5241,
+      "step": 48250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.391180577813692e-06,
+      "loss": 1.4803,
+      "step": 48260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.382532723987736e-06,
+      "loss": 1.4696,
+      "step": 48270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.373889055754213e-06,
+      "loss": 1.4719,
+      "step": 48280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3652495751667e-06,
+      "loss": 1.4654,
+      "step": 48290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.356614284277791e-06,
+      "loss": 1.4712,
+      "step": 48300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.347983185139048e-06,
+      "loss": 1.4928,
+      "step": 48310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.339356279801104e-06,
+      "loss": 1.4756,
+      "step": 48320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3307335703135195e-06,
+      "loss": 1.4933,
+      "step": 48330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.322115058724926e-06,
+      "loss": 1.5256,
+      "step": 48340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.313500747082902e-06,
+      "loss": 1.5002,
+      "step": 48350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3048906374340645e-06,
+      "loss": 1.4994,
+      "step": 48360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.296284731824015e-06,
+      "loss": 1.5227,
+      "step": 48370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.287683032297369e-06,
+      "loss": 1.5546,
+      "step": 48380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.279085540897726e-06,
+      "loss": 1.537,
+      "step": 48390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.270492259667699e-06,
+      "loss": 1.5325,
+      "step": 48400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.261903190648906e-06,
+      "loss": 1.5425,
+      "step": 48410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.253318335881931e-06,
+      "loss": 1.5262,
+      "step": 48420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.244737697406409e-06,
+      "loss": 1.5214,
+      "step": 48430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.2361612772609246e-06,
+      "loss": 1.5081,
+      "step": 48440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.227589077483085e-06,
+      "loss": 1.5809,
+      "step": 48450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.219021100109494e-06,
+      "loss": 1.4774,
+      "step": 48460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.210457347175756e-06,
+      "loss": 1.4893,
+      "step": 48470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.2018978207164355e-06,
+      "loss": 1.5304,
+      "step": 48480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.193342522765159e-06,
+      "loss": 1.5455,
+      "step": 48490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.184791455354486e-06,
+      "loss": 1.5436,
+      "step": 48500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.176244620516001e-06,
+      "loss": 1.5235,
+      "step": 48510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.167702020280282e-06,
+      "loss": 1.4938,
+      "step": 48520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.159163656676893e-06,
+      "loss": 1.4499,
+      "step": 48530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.150629531734398e-06,
+      "loss": 1.4967,
+      "step": 48540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.142099647480349e-06,
+      "loss": 1.4689,
+      "step": 48550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.133574005941293e-06,
+      "loss": 1.4718,
+      "step": 48560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.125052609142768e-06,
+      "loss": 1.4661,
+      "step": 48570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.116535459109313e-06,
+      "loss": 1.492,
+      "step": 48580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.1080225578644276e-06,
+      "loss": 1.499,
+      "step": 48590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.099513907430652e-06,
+      "loss": 1.4844,
+      "step": 48600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.091009509829466e-06,
+      "loss": 1.4861,
+      "step": 48610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.0825093670813695e-06,
+      "loss": 1.4741,
+      "step": 48620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.0740134812058405e-06,
+      "loss": 1.2331,
+      "step": 48630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.065521854221358e-06,
+      "loss": 1.3972,
+      "step": 48640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.057034488145356e-06,
+      "loss": 1.4636,
+      "step": 48650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.0485513849943095e-06,
+      "loss": 1.4798,
+      "step": 48660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.04007254678363e-06,
+      "loss": 1.4766,
+      "step": 48670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.03159797552774e-06,
+      "loss": 1.484,
+      "step": 48680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.023127673240049e-06,
+      "loss": 1.5094,
+      "step": 48690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.014661641932946e-06,
+      "loss": 1.4542,
+      "step": 48700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.0061998836178085e-06,
+      "loss": 1.4556,
+      "step": 48710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.997742400304997e-06,
+      "loss": 1.4898,
+      "step": 48720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.989289194003865e-06,
+      "loss": 1.4871,
+      "step": 48730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.980840266722716e-06,
+      "loss": 1.4616,
+      "step": 48740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.972395620468896e-06,
+      "loss": 1.4628,
+      "step": 48750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.963955257248677e-06,
+      "loss": 1.4765,
+      "step": 48760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.955519179067349e-06,
+      "loss": 1.476,
+      "step": 48770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.947087387929163e-06,
+      "loss": 1.4587,
+      "step": 48780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.93865988583737e-06,
+      "loss": 1.4651,
+      "step": 48790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.930236674794186e-06,
+      "loss": 1.4909,
+      "step": 48800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.921817756800822e-06,
+      "loss": 1.4651,
+      "step": 48810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.9134031338574404e-06,
+      "loss": 1.4697,
+      "step": 48820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.904992807963237e-06,
+      "loss": 1.449,
+      "step": 48830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.8965867811163234e-06,
+      "loss": 1.4761,
+      "step": 48840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.888185055313834e-06,
+      "loss": 1.4746,
+      "step": 48850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.879787632551862e-06,
+      "loss": 1.4634,
+      "step": 48860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.871394514825488e-06,
+      "loss": 1.4616,
+      "step": 48870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.8630057041287665e-06,
+      "loss": 1.4629,
+      "step": 48880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.854621202454722e-06,
+      "loss": 1.476,
+      "step": 48890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.84624101179537e-06,
+      "loss": 1.463,
+      "step": 48900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.837865134141674e-06,
+      "loss": 1.4489,
+      "step": 48910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.8294935714836195e-06,
+      "loss": 1.4666,
+      "step": 48920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.821126325810112e-06,
+      "loss": 1.4869,
+      "step": 48930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.812763399109073e-06,
+      "loss": 1.4637,
+      "step": 48940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.804404793367378e-06,
+      "loss": 1.476,
+      "step": 48950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.796050510570887e-06,
+      "loss": 1.4861,
+      "step": 48960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.787700552704409e-06,
+      "loss": 1.4865,
+      "step": 48970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.779354921751774e-06,
+      "loss": 1.4967,
+      "step": 48980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.771013619695726e-06,
+      "loss": 1.511,
+      "step": 48990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.762676648518018e-06,
+      "loss": 1.5213,
+      "step": 49000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.754344010199362e-06,
+      "loss": 1.4955,
+      "step": 49010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.746015706719447e-06,
+      "loss": 1.4646,
+      "step": 49020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.737691740056923e-06,
+      "loss": 1.4994,
+      "step": 49030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.729372112189417e-06,
+      "loss": 1.4643,
+      "step": 49040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.721056825093522e-06,
+      "loss": 1.4806,
+      "step": 49050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.7127458807448e-06,
+      "loss": 1.4764,
+      "step": 49060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.704439281117789e-06,
+      "loss": 1.4937,
+      "step": 49070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.696137028185965e-06,
+      "loss": 1.4644,
+      "step": 49080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.687839123921821e-06,
+      "loss": 1.4555,
+      "step": 49090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.679545570296772e-06,
+      "loss": 1.4319,
+      "step": 49100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.6712563692812204e-06,
+      "loss": 1.4405,
+      "step": 49110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.66297152284453e-06,
+      "loss": 1.4739,
+      "step": 49120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.65469103295504e-06,
+      "loss": 1.4516,
+      "step": 49130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.6464149015800255e-06,
+      "loss": 1.4757,
+      "step": 49140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.638143130685773e-06,
+      "loss": 1.4993,
+      "step": 49150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.629875722237483e-06,
+      "loss": 1.4639,
+      "step": 49160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.621612678199354e-06,
+      "loss": 1.4662,
+      "step": 49170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.61335400053453e-06,
+      "loss": 1.4645,
+      "step": 49180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.60509969120513e-06,
+      "loss": 1.4652,
+      "step": 49190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.596849752172224e-06,
+      "loss": 1.4663,
+      "step": 49200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.5886041853958525e-06,
+      "loss": 1.4706,
+      "step": 49210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.580362992835013e-06,
+      "loss": 1.4915,
+      "step": 49220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.57212617644766e-06,
+      "loss": 1.4755,
+      "step": 49230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.563893738190726e-06,
+      "loss": 1.4755,
+      "step": 49240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.555665680020063e-06,
+      "loss": 1.4735,
+      "step": 49250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.547442003890542e-06,
+      "loss": 1.4604,
+      "step": 49260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.539222711755935e-06,
+      "loss": 1.4827,
+      "step": 49270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.531007805569006e-06,
+      "loss": 1.479,
+      "step": 49280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.522797287281468e-06,
+      "loss": 1.4815,
+      "step": 49290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.514591158844002e-06,
+      "loss": 1.4686,
+      "step": 49300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.506389422206214e-06,
+      "loss": 1.4664,
+      "step": 49310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.498192079316712e-06,
+      "loss": 1.4808,
+      "step": 49320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.48999913212302e-06,
+      "loss": 1.4857,
+      "step": 49330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.481810582571643e-06,
+      "loss": 1.4781,
+      "step": 49340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.473626432608029e-06,
+      "loss": 1.4531,
+      "step": 49350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.465446684176587e-06,
+      "loss": 1.4776,
+      "step": 49360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.457271339220678e-06,
+      "loss": 1.4729,
+      "step": 49370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.449100399682611e-06,
+      "loss": 1.4553,
+      "step": 49380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.4409338675036676e-06,
+      "loss": 1.5116,
+      "step": 49390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.432771744624047e-06,
+      "loss": 1.4892,
+      "step": 49400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.424614032982945e-06,
+      "loss": 1.4913,
+      "step": 49410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.41646073451847e-06,
+      "loss": 1.4847,
+      "step": 49420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.408311851167706e-06,
+      "loss": 1.5246,
+      "step": 49430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.400167384866676e-06,
+      "loss": 1.507,
+      "step": 49440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.392027337550363e-06,
+      "loss": 1.4707,
+      "step": 49450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3838917111526956e-06,
+      "loss": 1.5054,
+      "step": 49460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.375760507606557e-06,
+      "loss": 1.4718,
+      "step": 49470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.367633728843753e-06,
+      "loss": 1.4743,
+      "step": 49480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.359511376795088e-06,
+      "loss": 1.4673,
+      "step": 49490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.351393453390267e-06,
+      "loss": 1.4863,
+      "step": 49500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.343279960557966e-06,
+      "loss": 1.4704,
+      "step": 49510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3351709002258095e-06,
+      "loss": 1.452,
+      "step": 49520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.327066274320359e-06,
+      "loss": 1.4491,
+      "step": 49530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3189660847671274e-06,
+      "loss": 1.4798,
+      "step": 49540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3108703334905785e-06,
+      "loss": 1.4789,
+      "step": 49550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.302779022414121e-06,
+      "loss": 1.4441,
+      "step": 49560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.29469215346008e-06,
+      "loss": 1.4639,
+      "step": 49570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.2866097285497835e-06,
+      "loss": 1.4839,
+      "step": 49580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.2785317496034474e-06,
+      "loss": 1.4668,
+      "step": 49590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.270458218540259e-06,
+      "loss": 1.4655,
+      "step": 49600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.262389137278343e-06,
+      "loss": 1.5048,
+      "step": 49610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.254324507734777e-06,
+      "loss": 1.4745,
+      "step": 49620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.246264331825549e-06,
+      "loss": 1.4797,
+      "step": 49630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.23820861146564e-06,
+      "loss": 1.4662,
+      "step": 49640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.2301573485689215e-06,
+      "loss": 1.4983,
+      "step": 49650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.222110545048237e-06,
+      "loss": 1.469,
+      "step": 49660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.214068202815362e-06,
+      "loss": 1.4489,
+      "step": 49670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.206030323781012e-06,
+      "loss": 1.4613,
+      "step": 49680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.197996909854842e-06,
+      "loss": 1.4845,
+      "step": 49690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.189967962945442e-06,
+      "loss": 1.5037,
+      "step": 49700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.1819434849603505e-06,
+      "loss": 1.4888,
+      "step": 49710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.173923477806037e-06,
+      "loss": 1.4575,
+      "step": 49720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.165907943387919e-06,
+      "loss": 1.4659,
+      "step": 49730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.1578968836103175e-06,
+      "loss": 1.4899,
+      "step": 49740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.149890300376549e-06,
+      "loss": 1.472,
+      "step": 49750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.141888195588807e-06,
+      "loss": 1.4453,
+      "step": 49760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.13389057114826e-06,
+      "loss": 1.5032,
+      "step": 49770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.125897428954994e-06,
+      "loss": 1.4644,
+      "step": 49780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.11790877090804e-06,
+      "loss": 1.4639,
+      "step": 49790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.109924598905353e-06,
+      "loss": 1.4739,
+      "step": 49800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.101944914843843e-06,
+      "loss": 1.4716,
+      "step": 49810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.093969720619319e-06,
+      "loss": 1.4968,
+      "step": 49820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.085999018126554e-06,
+      "loss": 1.5054,
+      "step": 49830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.078032809259238e-06,
+      "loss": 1.4639,
+      "step": 49840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.070071095910004e-06,
+      "loss": 1.4769,
+      "step": 49850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.06211387997041e-06,
+      "loss": 1.4926,
+      "step": 49860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.054161163330946e-06,
+      "loss": 1.4912,
+      "step": 49870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.046212947881038e-06,
+      "loss": 1.5049,
+      "step": 49880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.038269235509025e-06,
+      "loss": 1.4705,
+      "step": 49890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.030330028102213e-06,
+      "loss": 1.5136,
+      "step": 49900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.022395327546787e-06,
+      "loss": 1.5132,
+      "step": 49910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.014465135727917e-06,
+      "loss": 1.4923,
+      "step": 49920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0065394545296524e-06,
+      "loss": 1.4701,
+      "step": 49930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.998618285835001e-06,
+      "loss": 1.5028,
+      "step": 49940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.990701631525886e-06,
+      "loss": 1.4903,
+      "step": 49950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9827894934831624e-06,
+      "loss": 1.4864,
+      "step": 49960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.974881873586616e-06,
+      "loss": 1.4934,
+      "step": 49970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.966978773714949e-06,
+      "loss": 1.4822,
+      "step": 49980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.959080195745803e-06,
+      "loss": 1.4776,
+      "step": 49990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.95118614155572e-06,
+      "loss": 1.4725,
+      "step": 50000
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 1.5840998888015747,
+      "eval_runtime": 82.3816,
+      "eval_samples_per_second": 36.416,
+      "eval_steps_per_second": 4.552,
+      "step": 50000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.943296613020211e-06,
+      "loss": 1.4755,
+      "step": 50010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.935411612013664e-06,
+      "loss": 1.4956,
+      "step": 50020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9275311404094205e-06,
+      "loss": 1.4811,
+      "step": 50030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.919655200079741e-06,
+      "loss": 1.496,
+      "step": 50040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.911783792895809e-06,
+      "loss": 1.4275,
+      "step": 50050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.903916920727712e-06,
+      "loss": 1.4923,
+      "step": 50060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.896054585444502e-06,
+      "loss": 1.4898,
+      "step": 50070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.888196788914107e-06,
+      "loss": 1.4462,
+      "step": 50080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.880343533003407e-06,
+      "loss": 1.4748,
+      "step": 50090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.872494819578195e-06,
+      "loss": 1.4669,
+      "step": 50100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.864650650503181e-06,
+      "loss": 1.4858,
+      "step": 50110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.856811027641998e-06,
+      "loss": 1.4693,
+      "step": 50120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.848975952857199e-06,
+      "loss": 1.491,
+      "step": 50130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.841145428010258e-06,
+      "loss": 1.4601,
+      "step": 50140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.833319454961566e-06,
+      "loss": 1.4842,
+      "step": 50150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8254980355704396e-06,
+      "loss": 1.4795,
+      "step": 50160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.817681171695086e-06,
+      "loss": 1.4354,
+      "step": 50170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.809868865192677e-06,
+      "loss": 1.4989,
+      "step": 50180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.802061117919258e-06,
+      "loss": 1.5115,
+      "step": 50190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.794257931729816e-06,
+      "loss": 1.4828,
+      "step": 50200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.786459308478246e-06,
+      "loss": 1.4782,
+      "step": 50210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.778665250017365e-06,
+      "loss": 1.467,
+      "step": 50220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.770875758198885e-06,
+      "loss": 1.4982,
+      "step": 50230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.763090834873472e-06,
+      "loss": 1.482,
+      "step": 50240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.755310481890664e-06,
+      "loss": 1.4769,
+      "step": 50250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7475347010989385e-06,
+      "loss": 1.5394,
+      "step": 50260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.739763494345682e-06,
+      "loss": 1.4819,
+      "step": 50270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7319968634771935e-06,
+      "loss": 1.4658,
+      "step": 50280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.724234810338683e-06,
+      "loss": 1.4804,
+      "step": 50290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7164773367742755e-06,
+      "loss": 1.473,
+      "step": 50300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.708724444627015e-06,
+      "loss": 1.48,
+      "step": 50310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7009761357388224e-06,
+      "loss": 1.4694,
+      "step": 50320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6932324119505886e-06,
+      "loss": 1.4761,
+      "step": 50330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6854932751020635e-06,
+      "loss": 1.5036,
+      "step": 50340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.677758727031929e-06,
+      "loss": 1.5378,
+      "step": 50350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.670028769577776e-06,
+      "loss": 1.5155,
+      "step": 50360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.662303404576103e-06,
+      "loss": 1.5137,
+      "step": 50370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6545826338623185e-06,
+      "loss": 1.5696,
+      "step": 50380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.646866459270744e-06,
+      "loss": 1.5355,
+      "step": 50390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.639154882634582e-06,
+      "loss": 1.5573,
+      "step": 50400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.631447905785991e-06,
+      "loss": 1.5321,
+      "step": 50410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.623745530555993e-06,
+      "loss": 1.523,
+      "step": 50420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.616047758774537e-06,
+      "loss": 1.4869,
+      "step": 50430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.608354592270474e-06,
+      "loss": 1.5374,
+      "step": 50440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.600666032871563e-06,
+      "loss": 1.5196,
+      "step": 50450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.592982082404469e-06,
+      "loss": 1.5224,
+      "step": 50460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5853027426947564e-06,
+      "loss": 1.5262,
+      "step": 50470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5776280155669076e-06,
+      "loss": 1.5345,
+      "step": 50480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.569957902844275e-06,
+      "loss": 1.4899,
+      "step": 50490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.562292406349173e-06,
+      "loss": 1.5037,
+      "step": 50500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5546315279027565e-06,
+      "loss": 1.5225,
+      "step": 50510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.546975269325125e-06,
+      "loss": 1.5435,
+      "step": 50520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.539323632435265e-06,
+      "loss": 1.5001,
+      "step": 50530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.531676619051077e-06,
+      "loss": 1.4946,
+      "step": 50540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5240342309893275e-06,
+      "loss": 1.4655,
+      "step": 50550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.51639647006574e-06,
+      "loss": 1.4741,
+      "step": 50560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.508763338094891e-06,
+      "loss": 1.4562,
+      "step": 50570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5011348368902785e-06,
+      "loss": 1.4809,
+      "step": 50580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.493510968264298e-06,
+      "loss": 1.482,
+      "step": 50590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.485891734028242e-06,
+      "loss": 1.499,
+      "step": 50600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4782771359923045e-06,
+      "loss": 1.4791,
+      "step": 50610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.470667175965577e-06,
+      "loss": 1.4516,
+      "step": 50620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.463061855756049e-06,
+      "loss": 1.4586,
+      "step": 50630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.455461177170604e-06,
+      "loss": 1.2635,
+      "step": 50640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.447865142015035e-06,
+      "loss": 1.4854,
+      "step": 50650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.440273752094005e-06,
+      "loss": 1.4919,
+      "step": 50660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.432687009211116e-06,
+      "loss": 1.4165,
+      "step": 50670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.425104915168819e-06,
+      "loss": 1.4546,
+      "step": 50680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.417527471768494e-06,
+      "loss": 1.4885,
+      "step": 50690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4099546808104e-06,
+      "loss": 1.4557,
+      "step": 50700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.402386544093707e-06,
+      "loss": 1.4634,
+      "step": 50710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.394823063416446e-06,
+      "loss": 1.4505,
+      "step": 50720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.387264240575588e-06,
+      "loss": 1.488,
+      "step": 50730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.379710077366956e-06,
+      "loss": 1.4901,
+      "step": 50740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.372160575585286e-06,
+      "loss": 1.4724,
+      "step": 50750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.364615737024209e-06,
+      "loss": 1.4867,
+      "step": 50760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.357075563476238e-06,
+      "loss": 1.4728,
+      "step": 50770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.349540056732782e-06,
+      "loss": 1.4688,
+      "step": 50780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.342009218584146e-06,
+      "loss": 1.4778,
+      "step": 50790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.334483050819522e-06,
+      "loss": 1.4729,
+      "step": 50800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.326961555226976e-06,
+      "loss": 1.4942,
+      "step": 50810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.319444733593504e-06,
+      "loss": 1.4733,
+      "step": 50820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.311932587704943e-06,
+      "loss": 1.481,
+      "step": 50830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.304425119346068e-06,
+      "loss": 1.4531,
+      "step": 50840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2969223303005e-06,
+      "loss": 1.4811,
+      "step": 50850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.28942422235077e-06,
+      "loss": 1.4929,
+      "step": 50860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.281930797278295e-06,
+      "loss": 1.4633,
+      "step": 50870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.274442056863385e-06,
+      "loss": 1.478,
+      "step": 50880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.266958002885208e-06,
+      "loss": 1.493,
+      "step": 50890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.259478637121867e-06,
+      "loss": 1.4426,
+      "step": 50900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2520039613503076e-06,
+      "loss": 1.5749,
+      "step": 50910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.24453397734638e-06,
+      "loss": 1.4814,
+      "step": 50920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.237068686884819e-06,
+      "loss": 1.486,
+      "step": 50930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.229608091739247e-06,
+      "loss": 1.4897,
+      "step": 50940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.222152193682159e-06,
+      "loss": 1.4756,
+      "step": 50950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.21470099448495e-06,
+      "loss": 1.4609,
+      "step": 50960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2072544959178905e-06,
+      "loss": 1.5108,
+      "step": 50970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.19981269975012e-06,
+      "loss": 1.4782,
+      "step": 50980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.192375607749697e-06,
+      "loss": 1.4608,
+      "step": 50990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.184943221683524e-06,
+      "loss": 1.4532,
+      "step": 51000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.177515543317407e-06,
+      "loss": 1.4633,
+      "step": 51010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.170092574416027e-06,
+      "loss": 1.4678,
+      "step": 51020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.162674316742955e-06,
+      "loss": 1.4905,
+      "step": 51030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.155260772060619e-06,
+      "loss": 1.4889,
+      "step": 51040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.147851942130363e-06,
+      "loss": 1.5124,
+      "step": 51050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.140447828712372e-06,
+      "loss": 1.5236,
+      "step": 51060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.133048433565748e-06,
+      "loss": 1.5395,
+      "step": 51070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.125653758448443e-06,
+      "loss": 1.5118,
+      "step": 51080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.118263805117302e-06,
+      "loss": 1.5657,
+      "step": 51090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.11087857532804e-06,
+      "loss": 1.513,
+      "step": 51100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.103498070835258e-06,
+      "loss": 1.51,
+      "step": 51110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0961222933924326e-06,
+      "loss": 1.5447,
+      "step": 51120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.088751244751913e-06,
+      "loss": 1.4895,
+      "step": 51130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0813849266649335e-06,
+      "loss": 1.5375,
+      "step": 51140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.074023340881581e-06,
+      "loss": 1.4948,
+      "step": 51150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.066666489150859e-06,
+      "loss": 1.4658,
+      "step": 51160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.059314373220606e-06,
+      "loss": 1.4589,
+      "step": 51170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.051966994837557e-06,
+      "loss": 1.4557,
+      "step": 51180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0446243557473155e-06,
+      "loss": 1.4644,
+      "step": 51190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.037286457694368e-06,
+      "loss": 1.4702,
+      "step": 51200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.029953302422047e-06,
+      "loss": 1.4857,
+      "step": 51210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.022624891672606e-06,
+      "loss": 1.4653,
+      "step": 51220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.015301227187119e-06,
+      "loss": 1.4971,
+      "step": 51230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.007982310705567e-06,
+      "loss": 1.4748,
+      "step": 51240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.000668143966791e-06,
+      "loss": 1.4976,
+      "step": 51250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9933587287085055e-06,
+      "loss": 1.4955,
+      "step": 51260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.986054066667295e-06,
+      "loss": 1.5041,
+      "step": 51270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.978754159578617e-06,
+      "loss": 1.4695,
+      "step": 51280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.971459009176796e-06,
+      "loss": 1.496,
+      "step": 51290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.964168617195028e-06,
+      "loss": 1.5291,
+      "step": 51300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.956882985365388e-06,
+      "loss": 1.5725,
+      "step": 51310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9496021154187885e-06,
+      "loss": 1.5869,
+      "step": 51320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.942326009085057e-06,
+      "loss": 1.5913,
+      "step": 51330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.935054668092853e-06,
+      "loss": 1.5562,
+      "step": 51340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.927788094169714e-06,
+      "loss": 1.5068,
+      "step": 51350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9205262890420515e-06,
+      "loss": 1.476,
+      "step": 51360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.913269254435146e-06,
+      "loss": 1.4974,
+      "step": 51370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.906016992073115e-06,
+      "loss": 1.5013,
+      "step": 51380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.898769503678996e-06,
+      "loss": 1.4621,
+      "step": 51390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.891526790974637e-06,
+      "loss": 1.4486,
+      "step": 51400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.884288855680788e-06,
+      "loss": 1.4742,
+      "step": 51410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8770556995170465e-06,
+      "loss": 1.4658,
+      "step": 51420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8698273242018835e-06,
+      "loss": 1.4826,
+      "step": 51430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.862603731452631e-06,
+      "loss": 1.4749,
+      "step": 51440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.855384922985482e-06,
+      "loss": 1.4647,
+      "step": 51450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8481709005155e-06,
+      "loss": 1.5001,
+      "step": 51460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.840961665756591e-06,
+      "loss": 1.4815,
+      "step": 51470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8337572204215645e-06,
+      "loss": 1.4699,
+      "step": 51480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.826557566222048e-06,
+      "loss": 1.4558,
+      "step": 51490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.819362704868552e-06,
+      "loss": 1.4667,
+      "step": 51500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.812172638070447e-06,
+      "loss": 1.4347,
+      "step": 51510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.804987367535965e-06,
+      "loss": 1.4585,
+      "step": 51520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.797806894972198e-06,
+      "loss": 1.4926,
+      "step": 51530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.790631222085101e-06,
+      "loss": 1.4799,
+      "step": 51540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.783460350579463e-06,
+      "loss": 1.4829,
+      "step": 51550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.776294282158983e-06,
+      "loss": 1.4903,
+      "step": 51560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.76913301852617e-06,
+      "loss": 1.4846,
+      "step": 51570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.761976561382417e-06,
+      "loss": 1.4837,
+      "step": 51580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.754824912427966e-06,
+      "loss": 1.4699,
+      "step": 51590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.747678073361924e-06,
+      "loss": 1.4608,
+      "step": 51600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.740536045882249e-06,
+      "loss": 1.4909,
+      "step": 51610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7333988316857595e-06,
+      "loss": 1.4851,
+      "step": 51620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.726266432468135e-06,
+      "loss": 1.4976,
+      "step": 51630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.719138849923882e-06,
+      "loss": 1.4899,
+      "step": 51640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.712016085746418e-06,
+      "loss": 1.5098,
+      "step": 51650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.704898141627956e-06,
+      "loss": 1.4991,
+      "step": 51660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.697785019259604e-06,
+      "loss": 1.4902,
+      "step": 51670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.690676720331308e-06,
+      "loss": 1.5154,
+      "step": 51680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.683573246531881e-06,
+      "loss": 1.5273,
+      "step": 51690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.676474599548958e-06,
+      "loss": 1.5253,
+      "step": 51700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6693807810690774e-06,
+      "loss": 1.5098,
+      "step": 51710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.662291792777579e-06,
+      "loss": 1.4836,
+      "step": 51720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6552076363586914e-06,
+      "loss": 1.4605,
+      "step": 51730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.648128313495478e-06,
+      "loss": 1.4865,
+      "step": 51740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.64105382586986e-06,
+      "loss": 1.4925,
+      "step": 51750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6339841751626065e-06,
+      "loss": 1.5029,
+      "step": 51760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.626919363053339e-06,
+      "loss": 1.4843,
+      "step": 51770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.619859391220532e-06,
+      "loss": 1.4805,
+      "step": 51780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.612804261341505e-06,
+      "loss": 1.4954,
+      "step": 51790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.605753975092436e-06,
+      "loss": 1.4977,
+      "step": 51800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.59870853414833e-06,
+      "loss": 1.462,
+      "step": 51810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.591667940183078e-06,
+      "loss": 1.4538,
+      "step": 51820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.584632194869382e-06,
+      "loss": 1.4682,
+      "step": 51830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.577601299878812e-06,
+      "loss": 1.5604,
+      "step": 51840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5705752568817816e-06,
+      "loss": 1.4856,
+      "step": 51850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5635540675475595e-06,
+      "loss": 1.4553,
+      "step": 51860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.556537733544236e-06,
+      "loss": 1.4897,
+      "step": 51870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5495262565387915e-06,
+      "loss": 1.4756,
+      "step": 51880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.542519638197002e-06,
+      "loss": 1.4618,
+      "step": 51890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.535517880183526e-06,
+      "loss": 1.4847,
+      "step": 51900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.528520984161852e-06,
+      "loss": 1.4869,
+      "step": 51910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.521528951794316e-06,
+      "loss": 1.4897,
+      "step": 51920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.514541784742099e-06,
+      "loss": 1.4824,
+      "step": 51930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.507559484665227e-06,
+      "loss": 1.4629,
+      "step": 51940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.500582053222574e-06,
+      "loss": 1.4682,
+      "step": 51950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4936094920718315e-06,
+      "loss": 1.5411,
+      "step": 51960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4866418028695845e-06,
+      "loss": 1.4954,
+      "step": 51970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.479678987271199e-06,
+      "loss": 1.4566,
+      "step": 51980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.472721046930942e-06,
+      "loss": 1.4761,
+      "step": 51990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.465767983501878e-06,
+      "loss": 1.4734,
+      "step": 52000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4588197986359324e-06,
+      "loss": 1.4791,
+      "step": 52010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.45187649398387e-06,
+      "loss": 1.4852,
+      "step": 52020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.444938071195293e-06,
+      "loss": 1.4687,
+      "step": 52030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.438004531918649e-06,
+      "loss": 1.4596,
+      "step": 52040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.431075877801225e-06,
+      "loss": 1.4706,
+      "step": 52050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.424152110489135e-06,
+      "loss": 1.4752,
+      "step": 52060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.417233231627344e-06,
+      "loss": 1.4563,
+      "step": 52070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.410319242859654e-06,
+      "loss": 1.4526,
+      "step": 52080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.403410145828704e-06,
+      "loss": 1.4754,
+      "step": 52090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.396505942175971e-06,
+      "loss": 1.4715,
+      "step": 52100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.389606633541771e-06,
+      "loss": 1.4799,
+      "step": 52110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.382712221565258e-06,
+      "loss": 1.4921,
+      "step": 52120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.375822707884402e-06,
+      "loss": 1.5106,
+      "step": 52130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.368938094136055e-06,
+      "loss": 1.4828,
+      "step": 52140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.362058381955855e-06,
+      "loss": 1.4662,
+      "step": 52150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.355183572978308e-06,
+      "loss": 1.479,
+      "step": 52160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.34831366883674e-06,
+      "loss": 1.471,
+      "step": 52170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.341448671163317e-06,
+      "loss": 1.5148,
+      "step": 52180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.334588581589041e-06,
+      "loss": 1.4908,
+      "step": 52190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3277334017437445e-06,
+      "loss": 1.4553,
+      "step": 52200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.320883133256093e-06,
+      "loss": 1.4699,
+      "step": 52210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.314037777753591e-06,
+      "loss": 1.4779,
+      "step": 52220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.30719733686257e-06,
+      "loss": 1.4804,
+      "step": 52230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.300361812208187e-06,
+      "loss": 1.4856,
+      "step": 52240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.293531205414456e-06,
+      "loss": 1.4632,
+      "step": 52250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.286705518104189e-06,
+      "loss": 1.4798,
+      "step": 52260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.279884751899055e-06,
+      "loss": 1.4386,
+      "step": 52270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.273068908419542e-06,
+      "loss": 1.463,
+      "step": 52280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.26625798928498e-06,
+      "loss": 1.4722,
+      "step": 52290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2594519961135e-06,
+      "loss": 1.5278,
+      "step": 52300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.252650930522109e-06,
+      "loss": 1.4915,
+      "step": 52310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2458547941266e-06,
+      "loss": 1.5032,
+      "step": 52320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.239063588541617e-06,
+      "loss": 1.5165,
+      "step": 52330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.232277315380631e-06,
+      "loss": 1.4951,
+      "step": 52340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.225495976255933e-06,
+      "loss": 1.5162,
+      "step": 52350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.218719572778648e-06,
+      "loss": 1.5337,
+      "step": 52360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.21194810655873e-06,
+      "loss": 1.5214,
+      "step": 52370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.205181579204964e-06,
+      "loss": 1.4992,
+      "step": 52380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.198419992324931e-06,
+      "loss": 1.4882,
+      "step": 52390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.191663347525094e-06,
+      "loss": 1.4752,
+      "step": 52400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.184911646410686e-06,
+      "loss": 1.4666,
+      "step": 52410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.178164890585795e-06,
+      "loss": 1.4433,
+      "step": 52420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.171423081653333e-06,
+      "loss": 1.4737,
+      "step": 52430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.164686221215028e-06,
+      "loss": 1.4779,
+      "step": 52440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.15795431087144e-06,
+      "loss": 1.4896,
+      "step": 52450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.151227352221953e-06,
+      "loss": 1.4879,
+      "step": 52460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1445053468647525e-06,
+      "loss": 1.4986,
+      "step": 52470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.137788296396894e-06,
+      "loss": 1.4891,
+      "step": 52480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1310762024142025e-06,
+      "loss": 1.4652,
+      "step": 52490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.124369066511363e-06,
+      "loss": 1.4622,
+      "step": 52500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.117666890281868e-06,
+      "loss": 1.471,
+      "step": 52510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.110969675318032e-06,
+      "loss": 1.4636,
+      "step": 52520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.104277423210995e-06,
+      "loss": 1.4621,
+      "step": 52530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0975901355507136e-06,
+      "loss": 1.4678,
+      "step": 52540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.090907813925971e-06,
+      "loss": 1.4633,
+      "step": 52550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.084230459924351e-06,
+      "loss": 1.467,
+      "step": 52560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.077558075132298e-06,
+      "loss": 1.4819,
+      "step": 52570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.070890661135029e-06,
+      "loss": 1.4894,
+      "step": 52580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.064228219516608e-06,
+      "loss": 1.4698,
+      "step": 52590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.057570751859912e-06,
+      "loss": 1.473,
+      "step": 52600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.050918259746642e-06,
+      "loss": 1.4624,
+      "step": 52610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.044270744757289e-06,
+      "loss": 1.4665,
+      "step": 52620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.037628208471209e-06,
+      "loss": 1.4703,
+      "step": 52630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.030990652466529e-06,
+      "loss": 1.4539,
+      "step": 52640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.02435807832022e-06,
+      "loss": 1.4844,
+      "step": 52650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.017730487608063e-06,
+      "loss": 1.4782,
+      "step": 52660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0111078819046495e-06,
+      "loss": 1.4803,
+      "step": 52670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.004490262783397e-06,
+      "loss": 1.4925,
+      "step": 52680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.99787763181653e-06,
+      "loss": 1.4868,
+      "step": 52690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.991269990575089e-06,
+      "loss": 1.4717,
+      "step": 52700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.984667340628934e-06,
+      "loss": 1.4561,
+      "step": 52710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.978069683546737e-06,
+      "loss": 1.4471,
+      "step": 52720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971477020895967e-06,
+      "loss": 1.4712,
+      "step": 52730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.964889354242943e-06,
+      "loss": 1.4862,
+      "step": 52740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.958306685152763e-06,
+      "loss": 1.4806,
+      "step": 52750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.951729015189351e-06,
+      "loss": 1.4773,
+      "step": 52760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.945156345915446e-06,
+      "loss": 1.4777,
+      "step": 52770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.938588678892602e-06,
+      "loss": 1.4824,
+      "step": 52780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.932026015681156e-06,
+      "loss": 1.4854,
+      "step": 52790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.925468357840306e-06,
+      "loss": 1.4886,
+      "step": 52800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.918915706928017e-06,
+      "loss": 1.474,
+      "step": 52810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.912368064501082e-06,
+      "loss": 1.4633,
+      "step": 52820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.905825432115104e-06,
+      "loss": 1.4565,
+      "step": 52830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.899287811324498e-06,
+      "loss": 1.4744,
+      "step": 52840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.89275520368248e-06,
+      "loss": 1.4453,
+      "step": 52850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.88622761074108e-06,
+      "loss": 1.4691,
+      "step": 52860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.879705034051148e-06,
+      "loss": 1.4915,
+      "step": 52870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.873187475162307e-06,
+      "loss": 1.4806,
+      "step": 52880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8666749356230385e-06,
+      "loss": 1.4768,
+      "step": 52890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.860167416980578e-06,
+      "loss": 1.4903,
+      "step": 52900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.853664920781019e-06,
+      "loss": 1.49,
+      "step": 52910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8471674485692195e-06,
+      "loss": 1.4669,
+      "step": 52920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8406750018888685e-06,
+      "loss": 1.4733,
+      "step": 52930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8341875822824555e-06,
+      "loss": 1.4584,
+      "step": 52940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.827705191291278e-06,
+      "loss": 1.4969,
+      "step": 52950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.821227830455418e-06,
+      "loss": 1.4829,
+      "step": 52960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.814755501313802e-06,
+      "loss": 1.5063,
+      "step": 52970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.808288205404123e-06,
+      "loss": 1.5382,
+      "step": 52980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.801825944262899e-06,
+      "loss": 1.5021,
+      "step": 52990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7953687194254463e-06,
+      "loss": 1.51,
+      "step": 53000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7889165324258844e-06,
+      "loss": 1.5057,
+      "step": 53010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.782469384797141e-06,
+      "loss": 1.4953,
+      "step": 53020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7760272780709343e-06,
+      "loss": 1.5023,
+      "step": 53030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7695902137778073e-06,
+      "loss": 1.4957,
+      "step": 53040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.763158193447064e-06,
+      "loss": 1.4929,
+      "step": 53050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7567312186068656e-06,
+      "loss": 1.4784,
+      "step": 53060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7503092907841263e-06,
+      "loss": 1.4954,
+      "step": 53070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7438924115045876e-06,
+      "loss": 1.4652,
+      "step": 53080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7374805822927823e-06,
+      "loss": 1.4784,
+      "step": 53090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.73107380467205e-06,
+      "loss": 1.4951,
+      "step": 53100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7246720801645125e-06,
+      "loss": 1.481,
+      "step": 53110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7182754102911223e-06,
+      "loss": 1.4982,
+      "step": 53120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.711883796571594e-06,
+      "loss": 1.5036,
+      "step": 53130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7054972405244803e-06,
+      "loss": 1.4755,
+      "step": 53140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.699115743667092e-06,
+      "loss": 1.4792,
+      "step": 53150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6927393075155693e-06,
+      "loss": 1.4755,
+      "step": 53160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6863679335848335e-06,
+      "loss": 1.443,
+      "step": 53170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6800016233886123e-06,
+      "loss": 1.4836,
+      "step": 53180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.673640378439422e-06,
+      "loss": 1.5109,
+      "step": 53190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6672842002485803e-06,
+      "loss": 1.492,
+      "step": 53200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.66093309032621e-06,
+      "loss": 1.5158,
+      "step": 53210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.654587050181199e-06,
+      "loss": 1.5022,
+      "step": 53220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.648246081321277e-06,
+      "loss": 1.4729,
+      "step": 53230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.641910185252925e-06,
+      "loss": 1.4934,
+      "step": 53240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6355793634814478e-06,
+      "loss": 1.4687,
+      "step": 53250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6292536175109276e-06,
+      "loss": 1.5378,
+      "step": 53260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622932948844257e-06,
+      "loss": 1.537,
+      "step": 53270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.616617358983096e-06,
+      "loss": 1.5674,
+      "step": 53280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6103068494279356e-06,
+      "loss": 1.5454,
+      "step": 53290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.604001421678024e-06,
+      "loss": 1.5621,
+      "step": 53300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5977010772314245e-06,
+      "loss": 1.5634,
+      "step": 53310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5914058175849807e-06,
+      "loss": 1.5594,
+      "step": 53320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5851156442343353e-06,
+      "loss": 1.5741,
+      "step": 53330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.578830558673918e-06,
+      "loss": 1.5657,
+      "step": 53340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5725505623969564e-06,
+      "loss": 1.5853,
+      "step": 53350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5662756568954598e-06,
+      "loss": 1.5727,
+      "step": 53360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.560005843660233e-06,
+      "loss": 1.5969,
+      "step": 53370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.553741124180879e-06,
+      "loss": 1.5711,
+      "step": 53380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.547481499945762e-06,
+      "loss": 1.5814,
+      "step": 53390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5412269724420827e-06,
+      "loss": 1.5811,
+      "step": 53400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.534977543155782e-06,
+      "loss": 1.5698,
+      "step": 53410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5287332135716196e-06,
+      "loss": 1.5744,
+      "step": 53420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5224939851731363e-06,
+      "loss": 1.5695,
+      "step": 53430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.516259859442664e-06,
+      "loss": 1.5621,
+      "step": 53440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5100308378613027e-06,
+      "loss": 1.5563,
+      "step": 53450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.503806921908981e-06,
+      "loss": 1.5139,
+      "step": 53460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.497588113064365e-06,
+      "loss": 1.5141,
+      "step": 53470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.491374412804943e-06,
+      "loss": 1.4896,
+      "step": 53480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4851658226069753e-06,
+      "loss": 1.4853,
+      "step": 53490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.478962343945513e-06,
+      "loss": 1.5165,
+      "step": 53500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.472763978294391e-06,
+      "loss": 1.5623,
+      "step": 53510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.466570727126228e-06,
+      "loss": 1.5295,
+      "step": 53520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.460382591912434e-06,
+      "loss": 1.4999,
+      "step": 53530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.454199574123179e-06,
+      "loss": 1.5121,
+      "step": 53540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.448021675227464e-06,
+      "loss": 1.5069,
+      "step": 53550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.44184889669302e-06,
+      "loss": 1.5101,
+      "step": 53560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.435681239986416e-06,
+      "loss": 1.5458,
+      "step": 53570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.429518706572951e-06,
+      "loss": 1.5296,
+      "step": 53580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.423361297916744e-06,
+      "loss": 1.5202,
+      "step": 53590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4172090154806796e-06,
+      "loss": 1.4876,
+      "step": 53600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.411061860726436e-06,
+      "loss": 1.4769,
+      "step": 53610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4049198351144535e-06,
+      "loss": 1.4701,
+      "step": 53620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.398782940103984e-06,
+      "loss": 1.4828,
+      "step": 53630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.392651177153028e-06,
+      "loss": 1.4836,
+      "step": 53640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3865245477183867e-06,
+      "loss": 1.4957,
+      "step": 53650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3804030532556373e-06,
+      "loss": 1.509,
+      "step": 53660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3742866952191394e-06,
+      "loss": 1.4779,
+      "step": 53670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3681754750620243e-06,
+      "loss": 1.4791,
+      "step": 53680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.362069394236209e-06,
+      "loss": 1.4602,
+      "step": 53690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3559684541923958e-06,
+      "loss": 1.4724,
+      "step": 53700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.34987265638004e-06,
+      "loss": 1.4471,
+      "step": 53710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3437820022474143e-06,
+      "loss": 1.4668,
+      "step": 53720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.337696493241535e-06,
+      "loss": 1.4773,
+      "step": 53730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.331616130808213e-06,
+      "loss": 1.4918,
+      "step": 53740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.325540916392031e-06,
+      "loss": 1.4921,
+      "step": 53750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.319470851436357e-06,
+      "loss": 1.4536,
+      "step": 53760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3134059373833126e-06,
+      "loss": 1.4726,
+      "step": 53770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3073461756738367e-06,
+      "loss": 1.4617,
+      "step": 53780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3012915677475913e-06,
+      "loss": 1.4737,
+      "step": 53790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.295242115043068e-06,
+      "loss": 1.4729,
+      "step": 53800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2891978189974915e-06,
+      "loss": 1.4921,
+      "step": 53810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.283158681046883e-06,
+      "loss": 1.4789,
+      "step": 53820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2771247026260306e-06,
+      "loss": 1.5095,
+      "step": 53830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2710958851685012e-06,
+      "loss": 1.4844,
+      "step": 53840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2650722301066306e-06,
+      "loss": 1.4664,
+      "step": 53850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.259053738871534e-06,
+      "loss": 1.4621,
+      "step": 53860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2530404128930964e-06,
+      "loss": 1.487,
+      "step": 53870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2470322535999657e-06,
+      "loss": 1.4638,
+      "step": 53880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.241029262419587e-06,
+      "loss": 1.4855,
+      "step": 53890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2350314407781523e-06,
+      "loss": 1.4661,
+      "step": 53900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2290387901006413e-06,
+      "loss": 1.4909,
+      "step": 53910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.223051311810796e-06,
+      "loss": 1.4953,
+      "step": 53920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2170690073311386e-06,
+      "loss": 1.4794,
+      "step": 53930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.211091878082942e-06,
+      "loss": 1.4757,
+      "step": 53940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2051199254862886e-06,
+      "loss": 1.4564,
+      "step": 53950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1991531509599855e-06,
+      "loss": 1.4654,
+      "step": 53960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1931915559216372e-06,
+      "loss": 1.4879,
+      "step": 53970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1872351417876145e-06,
+      "loss": 1.4954,
+      "step": 53980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1812839099730514e-06,
+      "loss": 1.4728,
+      "step": 53990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1753378618918557e-06,
+      "loss": 1.5115,
+      "step": 54000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1693969989566945e-06,
+      "loss": 1.4837,
+      "step": 54010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1634613225790173e-06,
+      "loss": 1.4774,
+      "step": 54020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1575308341690314e-06,
+      "loss": 1.4913,
+      "step": 54030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1516055351357183e-06,
+      "loss": 1.505,
+      "step": 54040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1456854268868046e-06,
+      "loss": 1.5045,
+      "step": 54050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.139770510828824e-06,
+      "loss": 1.5235,
+      "step": 54060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.133860788367041e-06,
+      "loss": 1.538,
+      "step": 54070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1279562609054987e-06,
+      "loss": 1.5155,
+      "step": 54080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.122056929847009e-06,
+      "loss": 1.5252,
+      "step": 54090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1161627965931527e-06,
+      "loss": 1.5092,
+      "step": 54100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.110273862544255e-06,
+      "loss": 1.5066,
+      "step": 54110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1043901290994382e-06,
+      "loss": 1.4899,
+      "step": 54120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.098511597656556e-06,
+      "loss": 1.4912,
+      "step": 54130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.092638269612247e-06,
+      "loss": 1.4896,
+      "step": 54140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0867701463619085e-06,
+      "loss": 1.4893,
+      "step": 54150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.080907229299704e-06,
+      "loss": 1.4853,
+      "step": 54160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0750495198185484e-06,
+      "loss": 1.5088,
+      "step": 54170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0691970193101357e-06,
+      "loss": 1.4546,
+      "step": 54180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.063349729164919e-06,
+      "loss": 1.475,
+      "step": 54190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0575076507720874e-06,
+      "loss": 1.4698,
+      "step": 54200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.051670785519639e-06,
+      "loss": 1.4663,
+      "step": 54210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.045839134794293e-06,
+      "loss": 1.4958,
+      "step": 54220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0400126999815453e-06,
+      "loss": 1.5144,
+      "step": 54230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0341914824656537e-06,
+      "loss": 1.5114,
+      "step": 54240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0283754836296362e-06,
+      "loss": 1.4832,
+      "step": 54250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0225647048552686e-06,
+      "loss": 1.4687,
+      "step": 54260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.016759147523085e-06,
+      "loss": 1.5204,
+      "step": 54270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0109588130123824e-06,
+      "loss": 1.5385,
+      "step": 54280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0051637027012165e-06,
+      "loss": 1.5527,
+      "step": 54290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9993738179664055e-06,
+      "loss": 1.5715,
+      "step": 54300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.99358916018351e-06,
+      "loss": 1.5795,
+      "step": 54310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9878097307268677e-06,
+      "loss": 1.5761,
+      "step": 54320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.982035530969565e-06,
+      "loss": 1.5532,
+      "step": 54330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.97626656228345e-06,
+      "loss": 1.5826,
+      "step": 54340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9705028260391203e-06,
+      "loss": 1.5647,
+      "step": 54350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9647443236059476e-06,
+      "loss": 1.5685,
+      "step": 54360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9589910563520283e-06,
+      "loss": 1.5711,
+      "step": 54370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9532430256442563e-06,
+      "loss": 1.5612,
+      "step": 54380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9475002328482465e-06,
+      "loss": 1.5673,
+      "step": 54390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.941762679328386e-06,
+      "loss": 1.5545,
+      "step": 54400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9360303664478146e-06,
+      "loss": 1.5784,
+      "step": 54410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9303032955684264e-06,
+      "loss": 1.5757,
+      "step": 54420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9245814680508716e-06,
+      "loss": 1.5635,
+      "step": 54430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.918864885254549e-06,
+      "loss": 1.5525,
+      "step": 54440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9131535485376253e-06,
+      "loss": 1.5299,
+      "step": 54450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.907447459256993e-06,
+      "loss": 1.5315,
+      "step": 54460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9017466187683364e-06,
+      "loss": 1.4961,
+      "step": 54470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8960510284260554e-06,
+      "loss": 1.4871,
+      "step": 54480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.890360689583335e-06,
+      "loss": 1.5083,
+      "step": 54490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.884675603592085e-06,
+      "loss": 1.4622,
+      "step": 54500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8789957718029837e-06,
+      "loss": 1.4939,
+      "step": 54510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8733211955654567e-06,
+      "loss": 1.4899,
+      "step": 54520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8676518762276867e-06,
+      "loss": 1.4765,
+      "step": 54530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8619878151365826e-06,
+      "loss": 1.4784,
+      "step": 54540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8563290136378486e-06,
+      "loss": 1.4604,
+      "step": 54550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.850675473075898e-06,
+      "loss": 1.4745,
+      "step": 54560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8450271947939117e-06,
+      "loss": 1.4924,
+      "step": 54570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8393841801338194e-06,
+      "loss": 1.4574,
+      "step": 54580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8337464304363043e-06,
+      "loss": 1.479,
+      "step": 54590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8281139470407863e-06,
+      "loss": 1.4636,
+      "step": 54600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.822486731285448e-06,
+      "loss": 1.463,
+      "step": 54610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.81686478450722e-06,
+      "loss": 1.4934,
+      "step": 54620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8112481080417557e-06,
+      "loss": 1.4953,
+      "step": 54630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8056367032234966e-06,
+      "loss": 1.5016,
+      "step": 54640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8000305713856007e-06,
+      "loss": 1.504,
+      "step": 54650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.794429713859986e-06,
+      "loss": 1.4807,
+      "step": 54660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7888341319773135e-06,
+      "loss": 1.4723,
+      "step": 54670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7832438270670005e-06,
+      "loss": 1.5035,
+      "step": 54680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.777658800457186e-06,
+      "loss": 1.475,
+      "step": 54690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7720790534747936e-06,
+      "loss": 1.4958,
+      "step": 54700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.766504587445448e-06,
+      "loss": 1.4911,
+      "step": 54710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7609354036935623e-06,
+      "loss": 1.4868,
+      "step": 54720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.755371503542259e-06,
+      "loss": 1.5188,
+      "step": 54730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7498128883134273e-06,
+      "loss": 1.5067,
+      "step": 54740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7442595593276917e-06,
+      "loss": 1.5195,
+      "step": 54750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7387115179044244e-06,
+      "loss": 1.5164,
+      "step": 54760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7331687653617377e-06,
+      "loss": 1.4779,
+      "step": 54770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7276313030164922e-06,
+      "loss": 1.5089,
+      "step": 54780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7220991321842944e-06,
+      "loss": 1.4609,
+      "step": 54790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7165722541794697e-06,
+      "loss": 1.4831,
+      "step": 54800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7110506703151295e-06,
+      "loss": 1.4776,
+      "step": 54810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.705534381903085e-06,
+      "loss": 1.4824,
+      "step": 54820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.700023390253911e-06,
+      "loss": 1.4935,
+      "step": 54830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.694517696676921e-06,
+      "loss": 1.4624,
+      "step": 54840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6890173024801706e-06,
+      "loss": 1.4705,
+      "step": 54850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.683522208970443e-06,
+      "loss": 1.4675,
+      "step": 54860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.678032417453294e-06,
+      "loss": 1.5274,
+      "step": 54870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6725479292329797e-06,
+      "loss": 1.4628,
+      "step": 54880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.667068745612525e-06,
+      "loss": 1.4946,
+      "step": 54890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6615948678936826e-06,
+      "loss": 1.4843,
+      "step": 54900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6561262973769456e-06,
+      "loss": 1.4765,
+      "step": 54910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6506630353615502e-06,
+      "loss": 1.4877,
+      "step": 54920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.645205083145469e-06,
+      "loss": 1.4913,
+      "step": 54930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6397524420254116e-06,
+      "loss": 1.4506,
+      "step": 54940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6343051132968284e-06,
+      "loss": 1.4755,
+      "step": 54950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6288630982539114e-06,
+      "loss": 1.4976,
+      "step": 54960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6234263981895685e-06,
+      "loss": 1.4658,
+      "step": 54970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6179950143954827e-06,
+      "loss": 1.4677,
+      "step": 54980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6125689481620393e-06,
+      "loss": 1.4633,
+      "step": 54990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6071482007783753e-06,
+      "loss": 1.4836,
+      "step": 55000
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 1.5818219184875488,
+      "eval_runtime": 82.083,
+      "eval_samples_per_second": 36.548,
+      "eval_steps_per_second": 4.569,
+      "step": 55000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.601732773532367e-06,
+      "loss": 1.4945,
+      "step": 55010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.596322667710624e-06,
+      "loss": 1.4528,
+      "step": 55020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5909178845984734e-06,
+      "loss": 1.4597,
+      "step": 55030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5855184254800185e-06,
+      "loss": 1.4065,
+      "step": 55040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5801242916380557e-06,
+      "loss": 1.4753,
+      "step": 55050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5747354843541403e-06,
+      "loss": 1.4535,
+      "step": 55060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.569352004908554e-06,
+      "loss": 1.4683,
+      "step": 55070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5639738545803147e-06,
+      "loss": 1.4666,
+      "step": 55080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.558601034647176e-06,
+      "loss": 1.4752,
+      "step": 55090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5532335463856222e-06,
+      "loss": 1.4936,
+      "step": 55100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5478713910708745e-06,
+      "loss": 1.4668,
+      "step": 55110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.54251456997687e-06,
+      "loss": 1.4776,
+      "step": 55120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.537163084376315e-06,
+      "loss": 1.4783,
+      "step": 55130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5318169355406096e-06,
+      "loss": 1.5109,
+      "step": 55140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5264761247399045e-06,
+      "loss": 1.4534,
+      "step": 55150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.521140653243087e-06,
+      "loss": 1.4642,
+      "step": 55160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.51581052231776e-06,
+      "loss": 1.4675,
+      "step": 55170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.510485733230275e-06,
+      "loss": 1.483,
+      "step": 55180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5051662872457056e-06,
+      "loss": 1.4908,
+      "step": 55190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.499852185627841e-06,
+      "loss": 1.4799,
+      "step": 55200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4945434296392374e-06,
+      "loss": 1.4626,
+      "step": 55210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4892400205411475e-06,
+      "loss": 1.4683,
+      "step": 55220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4839419595935637e-06,
+      "loss": 1.4457,
+      "step": 55230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4786492480552174e-06,
+      "loss": 1.4914,
+      "step": 55240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4733618871835536e-06,
+      "loss": 1.5014,
+      "step": 55250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.468079878234761e-06,
+      "loss": 1.5026,
+      "step": 55260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.462803222463744e-06,
+      "loss": 1.4685,
+      "step": 55270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.457531921124151e-06,
+      "loss": 1.4845,
+      "step": 55280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4522659754683293e-06,
+      "loss": 1.4814,
+      "step": 55290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4470053867473946e-06,
+      "loss": 1.4769,
+      "step": 55300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4417501562111506e-06,
+      "loss": 1.4792,
+      "step": 55310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.436500285108151e-06,
+      "loss": 1.4822,
+      "step": 55320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.431255774685673e-06,
+      "loss": 1.486,
+      "step": 55330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.42601662618972e-06,
+      "loss": 1.463,
+      "step": 55340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4207828408650046e-06,
+      "loss": 1.4841,
+      "step": 55350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.415554419955002e-06,
+      "loss": 1.4941,
+      "step": 55360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4103313647018704e-06,
+      "loss": 1.4911,
+      "step": 55370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4051136763465264e-06,
+      "loss": 1.4931,
+      "step": 55380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3999013561285922e-06,
+      "loss": 1.5412,
+      "step": 55390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3946944052864213e-06,
+      "loss": 1.5073,
+      "step": 55400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3894928250570962e-06,
+      "loss": 1.4784,
+      "step": 55410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3842966166764123e-06,
+      "loss": 1.4648,
+      "step": 55420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3791057813789015e-06,
+      "loss": 1.4795,
+      "step": 55430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.373920320397807e-06,
+      "loss": 1.4737,
+      "step": 55440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3687402349651084e-06,
+      "loss": 1.5094,
+      "step": 55450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3635655263114843e-06,
+      "loss": 1.4934,
+      "step": 55460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3583961956663735e-06,
+      "loss": 1.4758,
+      "step": 55470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3532322442579023e-06,
+      "loss": 1.4553,
+      "step": 55480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.348073673312934e-06,
+      "loss": 1.5223,
+      "step": 55490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.342920484057054e-06,
+      "loss": 1.4714,
+      "step": 55500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.33777267771457e-06,
+      "loss": 1.4807,
+      "step": 55510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3326302555084943e-06,
+      "loss": 1.4728,
+      "step": 55520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3274932186605954e-06,
+      "loss": 1.4819,
+      "step": 55530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3223615683913207e-06,
+      "loss": 1.4787,
+      "step": 55540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.317235305919868e-06,
+      "loss": 1.458,
+      "step": 55550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3121144324641437e-06,
+      "loss": 1.4782,
+      "step": 55560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3069989492407738e-06,
+      "loss": 1.5676,
+      "step": 55570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3018888574651037e-06,
+      "loss": 1.4751,
+      "step": 55580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2967841583512033e-06,
+      "loss": 1.4754,
+      "step": 55590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.291684853111856e-06,
+      "loss": 1.481,
+      "step": 55600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2865909429585552e-06,
+      "loss": 1.4759,
+      "step": 55610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2815024291015367e-06,
+      "loss": 1.4698,
+      "step": 55620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.276419312749725e-06,
+      "loss": 1.4414,
+      "step": 55630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.271341595110793e-06,
+      "loss": 1.4347,
+      "step": 55640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.266269277391103e-06,
+      "loss": 1.4929,
+      "step": 55650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2612023607957474e-06,
+      "loss": 1.4615,
+      "step": 55660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.256140846528537e-06,
+      "loss": 1.476,
+      "step": 55670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.251084735792003e-06,
+      "loss": 1.4852,
+      "step": 55680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2460340297873665e-06,
+      "loss": 1.5124,
+      "step": 55690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2409887297146047e-06,
+      "loss": 1.4725,
+      "step": 55700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.235948836772378e-06,
+      "loss": 1.4448,
+      "step": 55710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2309143521580784e-06,
+      "loss": 1.52,
+      "step": 55720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2258852770678075e-06,
+      "loss": 1.4537,
+      "step": 55730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2208616126963816e-06,
+      "loss": 1.4882,
+      "step": 55740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.215843360237338e-06,
+      "loss": 1.473,
+      "step": 55750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2108305208829144e-06,
+      "loss": 1.4748,
+      "step": 55760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.205823095824083e-06,
+      "loss": 1.492,
+      "step": 55770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2008210862505013e-06,
+      "loss": 1.4789,
+      "step": 55780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1958244933505734e-06,
+      "loss": 1.457,
+      "step": 55790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.190833318311389e-06,
+      "loss": 1.4145,
+      "step": 55800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.185847562318763e-06,
+      "loss": 1.4016,
+      "step": 55810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1808672265572232e-06,
+      "loss": 1.5047,
+      "step": 55820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1758923122100123e-06,
+      "loss": 1.4851,
+      "step": 55830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1709228204590622e-06,
+      "loss": 1.5013,
+      "step": 55840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.165958752485059e-06,
+      "loss": 1.5312,
+      "step": 55850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.161000109467354e-06,
+      "loss": 1.5337,
+      "step": 55860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1560468925840487e-06,
+      "loss": 1.499,
+      "step": 55870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.151099103011925e-06,
+      "loss": 1.5017,
+      "step": 55880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1461567419264944e-06,
+      "loss": 1.4837,
+      "step": 55890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1412198105019738e-06,
+      "loss": 1.4676,
+      "step": 55900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1362883099112855e-06,
+      "loss": 1.4883,
+      "step": 55910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.131362241326068e-06,
+      "loss": 1.4909,
+      "step": 55920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.126441605916665e-06,
+      "loss": 1.4762,
+      "step": 55930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1215264048521338e-06,
+      "loss": 1.4679,
+      "step": 55940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1166166393002284e-06,
+      "loss": 1.4802,
+      "step": 55950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1117123104274327e-06,
+      "loss": 1.4682,
+      "step": 55960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1068134193989203e-06,
+      "loss": 1.4743,
+      "step": 55970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1019199673785766e-06,
+      "loss": 1.4565,
+      "step": 55980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0970319555290025e-06,
+      "loss": 1.4782,
+      "step": 55990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.092149385011502e-06,
+      "loss": 1.4957,
+      "step": 56000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.087272256986078e-06,
+      "loss": 1.4825,
+      "step": 56010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0824005726114592e-06,
+      "loss": 1.4591,
+      "step": 56020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0775343330450612e-06,
+      "loss": 1.4867,
+      "step": 56030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.072673539443018e-06,
+      "loss": 1.4664,
+      "step": 56040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0678181929601663e-06,
+      "loss": 1.4842,
+      "step": 56050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0629682947500506e-06,
+      "loss": 1.4746,
+      "step": 56060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0581238459649156e-06,
+      "loss": 1.4875,
+      "step": 56070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0532848477557182e-06,
+      "loss": 1.5043,
+      "step": 56080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0484513012721168e-06,
+      "loss": 1.473,
+      "step": 56090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0436232076624706e-06,
+      "loss": 1.4668,
+      "step": 56100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0388005680738594e-06,
+      "loss": 1.4794,
+      "step": 56110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0339833836520357e-06,
+      "loss": 1.4773,
+      "step": 56120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0291716555414952e-06,
+      "loss": 1.4871,
+      "step": 56130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0243653848854042e-06,
+      "loss": 1.4795,
+      "step": 56140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0195645728256525e-06,
+      "loss": 1.4831,
+      "step": 56150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.014769220502824e-06,
+      "loss": 1.4704,
+      "step": 56160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0099793290562135e-06,
+      "loss": 1.4807,
+      "step": 56170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0051948996237996e-06,
+      "loss": 1.4514,
+      "step": 56180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0004159333422924e-06,
+      "loss": 1.4764,
+      "step": 56190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.995642431347078e-06,
+      "loss": 1.4826,
+      "step": 56200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9908743947722577e-06,
+      "loss": 1.4854,
+      "step": 56210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.986111824750628e-06,
+      "loss": 1.4933,
+      "step": 56220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.981354722413692e-06,
+      "loss": 1.4825,
+      "step": 56230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9766030888916513e-06,
+      "loss": 1.451,
+      "step": 56240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.971856925313409e-06,
+      "loss": 1.4993,
+      "step": 56250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9671162328065734e-06,
+      "loss": 1.4853,
+      "step": 56260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.96238101249743e-06,
+      "loss": 1.5184,
+      "step": 56270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.957651265511004e-06,
+      "loss": 1.5228,
+      "step": 56280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9529269929709798e-06,
+      "loss": 1.5102,
+      "step": 56290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9482081959997704e-06,
+      "loss": 1.4951,
+      "step": 56300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9434948757184714e-06,
+      "loss": 1.5029,
+      "step": 56310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9387870332468833e-06,
+      "loss": 1.501,
+      "step": 56320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9340846697035066e-06,
+      "loss": 1.4929,
+      "step": 56330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9293877862055415e-06,
+      "loss": 1.4982,
+      "step": 56340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9246963838688683e-06,
+      "loss": 1.4866,
+      "step": 56350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9200104638081e-06,
+      "loss": 1.4976,
+      "step": 56360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.915330027136511e-06,
+      "loss": 1.5074,
+      "step": 56370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9106550749660945e-06,
+      "loss": 1.4742,
+      "step": 56380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9059856084075316e-06,
+      "loss": 1.4736,
+      "step": 56390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.90132162857021e-06,
+      "loss": 1.4597,
+      "step": 56400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8966631365622017e-06,
+      "loss": 1.48,
+      "step": 56410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8920101334902795e-06,
+      "loss": 1.4928,
+      "step": 56420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8873626204599233e-06,
+      "loss": 1.5158,
+      "step": 56430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.88272059857528e-06,
+      "loss": 1.4891,
+      "step": 56440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8780840689392288e-06,
+      "loss": 1.4735,
+      "step": 56450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8734530326533134e-06,
+      "loss": 1.5343,
+      "step": 56460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8688274908177872e-06,
+      "loss": 1.5748,
+      "step": 56470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8642074445315987e-06,
+      "loss": 1.5613,
+      "step": 56480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8595928948923842e-06,
+      "loss": 1.5513,
+      "step": 56490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.85498384299648e-06,
+      "loss": 1.5737,
+      "step": 56500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8503802899389105e-06,
+      "loss": 1.569,
+      "step": 56510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8457822368134037e-06,
+      "loss": 1.5671,
+      "step": 56520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8411896847123633e-06,
+      "loss": 1.5859,
+      "step": 56530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8366026347269084e-06,
+      "loss": 1.5768,
+      "step": 56540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8320210879468307e-06,
+      "loss": 1.5597,
+      "step": 56550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8274450454606268e-06,
+      "loss": 1.5799,
+      "step": 56560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8228745083554794e-06,
+      "loss": 1.5591,
+      "step": 56570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.818309477717267e-06,
+      "loss": 1.5755,
+      "step": 56580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8137499546305614e-06,
+      "loss": 1.5762,
+      "step": 56590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8091959401786234e-06,
+      "loss": 1.5526,
+      "step": 56600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8046474354433934e-06,
+      "loss": 1.5328,
+      "step": 56610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8001044415055318e-06,
+      "loss": 1.5285,
+      "step": 56620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7955669594443587e-06,
+      "loss": 1.4867,
+      "step": 56630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7910349903379002e-06,
+      "loss": 1.4941,
+      "step": 56640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7865085352628736e-06,
+      "loss": 1.4921,
+      "step": 56650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7819875952946818e-06,
+      "loss": 1.4873,
+      "step": 56660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7774721715074188e-06,
+      "loss": 1.4881,
+      "step": 56670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7729622649738652e-06,
+      "loss": 1.4851,
+      "step": 56680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.768457876765503e-06,
+      "loss": 1.4649,
+      "step": 56690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7639590079524759e-06,
+      "loss": 1.4728,
+      "step": 56700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.759465659603654e-06,
+      "loss": 1.4739,
+      "step": 56710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7549778327865613e-06,
+      "loss": 1.5082,
+      "step": 56720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7504955285674279e-06,
+      "loss": 1.5347,
+      "step": 56730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7460187480111716e-06,
+      "loss": 1.5097,
+      "step": 56740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7415474921813973e-06,
+      "loss": 1.5348,
+      "step": 56750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.73708176214038e-06,
+      "loss": 1.4791,
+      "step": 56760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7326215589491184e-06,
+      "loss": 1.516,
+      "step": 56770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7281668836672537e-06,
+      "loss": 1.5425,
+      "step": 56780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7237177373531587e-06,
+      "loss": 1.5109,
+      "step": 56790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.719274121063852e-06,
+      "loss": 1.4886,
+      "step": 56800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7148360358550665e-06,
+      "loss": 1.4978,
+      "step": 56810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7104034827812088e-06,
+      "loss": 1.4812,
+      "step": 56820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.705976462895373e-06,
+      "loss": 1.4446,
+      "step": 56830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7015549772493394e-06,
+      "loss": 1.4751,
+      "step": 56840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.697139026893571e-06,
+      "loss": 1.4899,
+      "step": 56850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6927286128772258e-06,
+      "loss": 1.4905,
+      "step": 56860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6883237362481264e-06,
+      "loss": 1.4655,
+      "step": 56870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6839243980528052e-06,
+      "loss": 1.4738,
+      "step": 56880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6795305993364568e-06,
+      "loss": 1.484,
+      "step": 56890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6751423411429711e-06,
+      "loss": 1.4821,
+      "step": 56900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.670759624514917e-06,
+      "loss": 1.4774,
+      "step": 56910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.666382450493556e-06,
+      "loss": 1.4952,
+      "step": 56920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6620108201188116e-06,
+      "loss": 1.4808,
+      "step": 56930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6576447344293228e-06,
+      "loss": 1.4524,
+      "step": 56940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6532841944623762e-06,
+      "loss": 1.4955,
+      "step": 56950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6489292012539682e-06,
+      "loss": 1.496,
+      "step": 56960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6445797558387598e-06,
+      "loss": 1.481,
+      "step": 56970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6402358592501055e-06,
+      "loss": 1.5019,
+      "step": 56980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6358975125200348e-06,
+      "loss": 1.4621,
+      "step": 56990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6315647166792597e-06,
+      "loss": 1.4764,
+      "step": 57000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6272374727571787e-06,
+      "loss": 1.4566,
+      "step": 57010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6229157817818613e-06,
+      "loss": 1.4808,
+      "step": 57020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6185996447800695e-06,
+      "loss": 1.4942,
+      "step": 57030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.61428906277723e-06,
+      "loss": 1.4714,
+      "step": 57040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.609984036797471e-06,
+      "loss": 1.4754,
+      "step": 57050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6056845678635801e-06,
+      "loss": 1.4689,
+      "step": 57060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6013906569970372e-06,
+      "loss": 1.4641,
+      "step": 57070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5971023052179957e-06,
+      "loss": 1.4712,
+      "step": 57080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.592819513545296e-06,
+      "loss": 1.4912,
+      "step": 57090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5885422829964442e-06,
+      "loss": 1.4833,
+      "step": 57100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5842706145876434e-06,
+      "loss": 1.4845,
+      "step": 57110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5800045093337546e-06,
+      "loss": 1.4533,
+      "step": 57120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5757439682483332e-06,
+      "loss": 1.4735,
+      "step": 57130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5714889923436061e-06,
+      "loss": 1.5135,
+      "step": 57140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5672395826304782e-06,
+      "loss": 1.5223,
+      "step": 57150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5629957401185307e-06,
+      "loss": 1.5294,
+      "step": 57160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.558757465816027e-06,
+      "loss": 1.5174,
+      "step": 57170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5545247607299085e-06,
+      "loss": 1.5103,
+      "step": 57180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5502976258657765e-06,
+      "loss": 1.5058,
+      "step": 57190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5460760622279386e-06,
+      "loss": 1.4975,
+      "step": 57200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5418600708193509e-06,
+      "loss": 1.5155,
+      "step": 57210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5376496526416567e-06,
+      "loss": 1.4885,
+      "step": 57220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5334448086951808e-06,
+      "loss": 1.4862,
+      "step": 57230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5292455399789158e-06,
+      "loss": 1.5048,
+      "step": 57240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5250518474905328e-06,
+      "loss": 1.4688,
+      "step": 57250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5208637322263796e-06,
+      "loss": 1.4772,
+      "step": 57260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5166811951814686e-06,
+      "loss": 1.4754,
+      "step": 57270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5125042373495073e-06,
+      "loss": 1.483,
+      "step": 57280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5083328597228553e-06,
+      "loss": 1.5114,
+      "step": 57290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.504167063292561e-06,
+      "loss": 1.4899,
+      "step": 57300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5000068490483438e-06,
+      "loss": 1.4794,
+      "step": 57310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.495852217978591e-06,
+      "loss": 1.5322,
+      "step": 57320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4917031710703715e-06,
+      "loss": 1.5527,
+      "step": 57330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4875597093094245e-06,
+      "loss": 1.5719,
+      "step": 57340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4834218336801625e-06,
+      "loss": 1.5578,
+      "step": 57350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.479289545165663e-06,
+      "loss": 1.5642,
+      "step": 57360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4751628447476933e-06,
+      "loss": 1.5596,
+      "step": 57370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4710417334066777e-06,
+      "loss": 1.5927,
+      "step": 57380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4669262121217164e-06,
+      "loss": 1.5612,
+      "step": 57390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4628162818705827e-06,
+      "loss": 1.5725,
+      "step": 57400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4587119436297314e-06,
+      "loss": 1.56,
+      "step": 57410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4546131983742634e-06,
+      "loss": 1.5703,
+      "step": 57420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.45052004707798e-06,
+      "loss": 1.5746,
+      "step": 57430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4464324907133342e-06,
+      "loss": 1.5527,
+      "step": 57440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4423505302514573e-06,
+      "loss": 1.5185,
+      "step": 57450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4382741666621485e-06,
+      "loss": 1.5198,
+      "step": 57460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4342034009138776e-06,
+      "loss": 1.481,
+      "step": 57470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.430138233973785e-06,
+      "loss": 1.4912,
+      "step": 57480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4260786668076841e-06,
+      "loss": 1.4835,
+      "step": 57490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4220247003800534e-06,
+      "loss": 1.4882,
+      "step": 57500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.417976335654042e-06,
+      "loss": 1.4571,
+      "step": 57510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4139335735914694e-06,
+      "loss": 1.4826,
+      "step": 57520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.409896415152817e-06,
+      "loss": 1.4773,
+      "step": 57530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4058648612972515e-06,
+      "loss": 1.4894,
+      "step": 57540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4018389129825866e-06,
+      "loss": 1.4846,
+      "step": 57550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3978185711653213e-06,
+      "loss": 1.4791,
+      "step": 57560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3938038368006139e-06,
+      "loss": 1.4728,
+      "step": 57570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3897947108422988e-06,
+      "loss": 1.4598,
+      "step": 57580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.385791194242858e-06,
+      "loss": 1.4758,
+      "step": 57590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.38179328795347e-06,
+      "loss": 1.4943,
+      "step": 57600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3778009929239583e-06,
+      "loss": 1.484,
+      "step": 57610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3738143101028168e-06,
+      "loss": 1.4688,
+      "step": 57620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3698332404372154e-06,
+      "loss": 1.5061,
+      "step": 57630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3658577848729842e-06,
+      "loss": 1.4936,
+      "step": 57640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3618879443546173e-06,
+      "loss": 1.4905,
+      "step": 57650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.357923719825277e-06,
+      "loss": 1.4966,
+      "step": 57660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3539651122267932e-06,
+      "loss": 1.5101,
+      "step": 57670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.350012122499658e-06,
+      "loss": 1.5311,
+      "step": 57680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3460647515830344e-06,
+      "loss": 1.5072,
+      "step": 57690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3421230004147383e-06,
+      "loss": 1.4929,
+      "step": 57700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3381868699312683e-06,
+      "loss": 1.4991,
+      "step": 57710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3342563610677733e-06,
+      "loss": 1.4514,
+      "step": 57720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3303314747580676e-06,
+      "loss": 1.4825,
+      "step": 57730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3264122119346412e-06,
+      "loss": 1.525,
+      "step": 57740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3224985735286382e-06,
+      "loss": 1.4915,
+      "step": 57750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3185905604698596e-06,
+      "loss": 1.4697,
+      "step": 57760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3146881736867928e-06,
+      "loss": 1.4962,
+      "step": 57770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.310791414106563e-06,
+      "loss": 1.5408,
+      "step": 57780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.306900282654977e-06,
+      "loss": 1.4683,
+      "step": 57790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3030147802564925e-06,
+      "loss": 1.46,
+      "step": 57800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2991349078342403e-06,
+      "loss": 1.4924,
+      "step": 57810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2952606663100026e-06,
+      "loss": 1.4726,
+      "step": 57820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.291392056604232e-06,
+      "loss": 1.4741,
+      "step": 57830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2875290796360462e-06,
+      "loss": 1.526,
+      "step": 57840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2836717363232026e-06,
+      "loss": 1.4906,
+      "step": 57850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2798200275821542e-06,
+      "loss": 1.4854,
+      "step": 57860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2759739543279886e-06,
+      "loss": 1.4757,
+      "step": 57870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2721335174744636e-06,
+      "loss": 1.4842,
+      "step": 57880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2682987179339994e-06,
+      "loss": 1.4606,
+      "step": 57890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2644695566176728e-06,
+      "loss": 1.4539,
+      "step": 57900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2606460344352256e-06,
+      "loss": 1.5026,
+      "step": 57910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2568281522950642e-06,
+      "loss": 1.4798,
+      "step": 57920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.25301591110423e-06,
+      "loss": 1.466,
+      "step": 57930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.249209311768465e-06,
+      "loss": 1.4546,
+      "step": 57940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2454083551921347e-06,
+      "loss": 1.4925,
+      "step": 57950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2416130422782778e-06,
+      "loss": 1.4667,
+      "step": 57960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2378233739285982e-06,
+      "loss": 1.4917,
+      "step": 57970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2340393510434505e-06,
+      "loss": 1.5175,
+      "step": 57980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2302609745218492e-06,
+      "loss": 1.4727,
+      "step": 57990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2264882452614673e-06,
+      "loss": 1.4647,
+      "step": 58000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2227211641586438e-06,
+      "loss": 1.4716,
+      "step": 58010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2189597321083573e-06,
+      "loss": 1.4826,
+      "step": 58020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2152039500042705e-06,
+      "loss": 1.4546,
+      "step": 58030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2114538187386809e-06,
+      "loss": 1.4504,
+      "step": 58040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2077093392025508e-06,
+      "loss": 1.4671,
+      "step": 58050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2039705122855043e-06,
+      "loss": 1.5355,
+      "step": 58060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2002373388758203e-06,
+      "loss": 1.4845,
+      "step": 58070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.196509819860428e-06,
+      "loss": 1.4896,
+      "step": 58080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1927879561249272e-06,
+      "loss": 1.4623,
+      "step": 58090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.189071748553558e-06,
+      "loss": 1.5255,
+      "step": 58100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.185361198029228e-06,
+      "loss": 1.4769,
+      "step": 58110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1816563054334956e-06,
+      "loss": 1.4769,
+      "step": 58120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1779570716465765e-06,
+      "loss": 1.4789,
+      "step": 58130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.17426349754734e-06,
+      "loss": 1.4852,
+      "step": 58140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1705755840133198e-06,
+      "loss": 1.4699,
+      "step": 58150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1668933319206904e-06,
+      "loss": 1.5018,
+      "step": 58160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1632167421442907e-06,
+      "loss": 1.4898,
+      "step": 58170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1595458155576165e-06,
+      "loss": 1.493,
+      "step": 58180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.155880553032801e-06,
+      "loss": 1.4799,
+      "step": 58190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1522209554406639e-06,
+      "loss": 1.468,
+      "step": 58200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1485670236506429e-06,
+      "loss": 1.4563,
+      "step": 58210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1449187585308524e-06,
+      "loss": 1.5323,
+      "step": 58220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1412761609480515e-06,
+      "loss": 1.4679,
+      "step": 58230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.137639231767665e-06,
+      "loss": 1.4691,
+      "step": 58240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1340079718537483e-06,
+      "loss": 1.4826,
+      "step": 58250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.130382382069034e-06,
+      "loss": 1.4616,
+      "step": 58260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1267624632748913e-06,
+      "loss": 1.4722,
+      "step": 58270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1231482163313518e-06,
+      "loss": 1.4939,
+      "step": 58280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1195396420970893e-06,
+      "loss": 1.4896,
+      "step": 58290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.115936741429438e-06,
+      "loss": 1.4689,
+      "step": 58300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.112339515184385e-06,
+      "loss": 1.4842,
+      "step": 58310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1087479642165666e-06,
+      "loss": 1.4849,
+      "step": 58320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1051620893792692e-06,
+      "loss": 1.4478,
+      "step": 58330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.101581891524428e-06,
+      "loss": 1.4759,
+      "step": 58340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0980073715026406e-06,
+      "loss": 1.458,
+      "step": 58350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0944385301631388e-06,
+      "loss": 1.49,
+      "step": 58360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0908753683538302e-06,
+      "loss": 1.5237,
+      "step": 58370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.087317886921241e-06,
+      "loss": 1.532,
+      "step": 58380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0837660867105748e-06,
+      "loss": 1.5088,
+      "step": 58390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0802199685656711e-06,
+      "loss": 1.5121,
+      "step": 58400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0766795333290308e-06,
+      "loss": 1.4911,
+      "step": 58410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.073144781841784e-06,
+      "loss": 1.4799,
+      "step": 58420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0696157149437363e-06,
+      "loss": 1.4472,
+      "step": 58430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.066092333473323e-06,
+      "loss": 1.4861,
+      "step": 58440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0625746382676411e-06,
+      "loss": 1.4737,
+      "step": 58450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0590626301624274e-06,
+      "loss": 1.4676,
+      "step": 58460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0555563099920728e-06,
+      "loss": 1.4854,
+      "step": 58470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0520556785896169e-06,
+      "loss": 1.4786,
+      "step": 58480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.048560736786744e-06,
+      "loss": 1.4862,
+      "step": 58490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0450714854137955e-06,
+      "loss": 1.5,
+      "step": 58500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0415879252997447e-06,
+      "loss": 1.4651,
+      "step": 58510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0381100572722347e-06,
+      "loss": 1.4922,
+      "step": 58520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0346378821575354e-06,
+      "loss": 1.4866,
+      "step": 58530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0311714007805728e-06,
+      "loss": 1.4729,
+      "step": 58540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0277106139649245e-06,
+      "loss": 1.4843,
+      "step": 58550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0242555225328105e-06,
+      "loss": 1.4845,
+      "step": 58560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0208061273050934e-06,
+      "loss": 1.4822,
+      "step": 58570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0173624291012956e-06,
+      "loss": 1.4663,
+      "step": 58580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0139244287395655e-06,
+      "loss": 1.4864,
+      "step": 58590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0104921270367274e-06,
+      "loss": 1.472,
+      "step": 58600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0070655248082178e-06,
+      "loss": 1.495,
+      "step": 58610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.003644622868144e-06,
+      "loss": 1.4775,
+      "step": 58620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0002294220292474e-06,
+      "loss": 1.4781,
+      "step": 58630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.968199231029207e-07,
+      "loss": 1.4643,
+      "step": 58640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.934161268991966e-07,
+      "loss": 1.4911,
+      "step": 58650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.900180342267584e-07,
+      "loss": 1.4793,
+      "step": 58660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.86625645892933e-07,
+      "loss": 1.4958,
+      "step": 58670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.83238962703681e-07,
+      "loss": 1.4782,
+      "step": 58680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.798579854636308e-07,
+      "loss": 1.472,
+      "step": 58690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.76482714976032e-07,
+      "loss": 1.4873,
+      "step": 58700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.731131520427926e-07,
+      "loss": 1.4984,
+      "step": 58710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.697492974644561e-07,
+      "loss": 1.5191,
+      "step": 58720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.66391152040219e-07,
+      "loss": 1.5424,
+      "step": 58730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.6303871656791e-07,
+      "loss": 1.5048,
+      "step": 58740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.596919918440111e-07,
+      "loss": 1.4939,
+      "step": 58750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.563509786636477e-07,
+      "loss": 1.4922,
+      "step": 58760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.530156778205735e-07,
+      "loss": 1.5091,
+      "step": 58770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.496860901072047e-07,
+      "loss": 1.485,
+      "step": 58780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.463622163145891e-07,
+      "loss": 1.4946,
+      "step": 58790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.430440572324173e-07,
+      "loss": 1.4904,
+      "step": 58800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.397316136490225e-07,
+      "loss": 1.4602,
+      "step": 58810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.364248863513863e-07,
+      "loss": 1.4937,
+      "step": 58820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.331238761251276e-07,
+      "loss": 1.4808,
+      "step": 58830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.298285837545051e-07,
+      "loss": 1.4918,
+      "step": 58840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.265390100224147e-07,
+      "loss": 1.4981,
+      "step": 58850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.232551557104119e-07,
+      "loss": 1.4816,
+      "step": 58860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.199770215986725e-07,
+      "loss": 1.5277,
+      "step": 58870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.167046084660236e-07,
+      "loss": 1.5451,
+      "step": 58880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.134379170899349e-07,
+      "loss": 1.5308,
+      "step": 58890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.101769482465078e-07,
+      "loss": 1.5452,
+      "step": 58900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.069217027104921e-07,
+      "loss": 1.5654,
+      "step": 58910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.036721812552773e-07,
+      "loss": 1.5596,
+      "step": 58920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.004283846528932e-07,
+      "loss": 1.5862,
+      "step": 58930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.971903136739984e-07,
+      "loss": 1.5782,
+      "step": 58940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.939579690879107e-07,
+      "loss": 1.5804,
+      "step": 58950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.907313516625687e-07,
+      "loss": 1.5497,
+      "step": 58960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.875104621645619e-07,
+      "loss": 1.567,
+      "step": 58970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.842953013591171e-07,
+      "loss": 1.5717,
+      "step": 58980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.810858700100955e-07,
+      "loss": 1.5554,
+      "step": 58990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.778821688799982e-07,
+      "loss": 1.527,
+      "step": 59000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.746841987299747e-07,
+      "loss": 1.5014,
+      "step": 59010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.714919603197979e-07,
+      "loss": 1.4956,
+      "step": 59020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.683054544078889e-07,
+      "loss": 1.4669,
+      "step": 59030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.651246817513031e-07,
+      "loss": 1.475,
+      "step": 59040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.619496431057361e-07,
+      "loss": 1.4877,
+      "step": 59050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.587803392255178e-07,
+      "loss": 1.4631,
+      "step": 59060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.556167708636181e-07,
+      "loss": 1.4749,
+      "step": 59070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.524589387716442e-07,
+      "loss": 1.4815,
+      "step": 59080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.493068436998431e-07,
+      "loss": 1.4599,
+      "step": 59090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.461604863970907e-07,
+      "loss": 1.4643,
+      "step": 59100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.430198676109058e-07,
+      "loss": 1.4999,
+      "step": 59110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.398849880874471e-07,
+      "loss": 1.4747,
+      "step": 59120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.367558485714994e-07,
+      "loss": 1.4855,
+      "step": 59130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.336324498064929e-07,
+      "loss": 1.4629,
+      "step": 59140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.305147925344898e-07,
+      "loss": 1.4995,
+      "step": 59150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.274028774961945e-07,
+      "loss": 1.4684,
+      "step": 59160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.242967054309297e-07,
+      "loss": 1.4817,
+      "step": 59170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.211962770766801e-07,
+      "loss": 1.4839,
+      "step": 59180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.181015931700425e-07,
+      "loss": 1.5003,
+      "step": 59190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.150126544462594e-07,
+      "loss": 1.5047,
+      "step": 59200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.119294616392075e-07,
+      "loss": 1.5243,
+      "step": 59210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.088520154813984e-07,
+      "loss": 1.4819,
+      "step": 59220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.057803167039774e-07,
+      "loss": 1.4853,
+      "step": 59230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.027143660367248e-07,
+      "loss": 1.4807,
+      "step": 59240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.99654164208058e-07,
+      "loss": 1.4947,
+      "step": 59250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.965997119450174e-07,
+      "loss": 1.5098,
+      "step": 59260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.935510099732979e-07,
+      "loss": 1.4555,
+      "step": 59270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.905080590172031e-07,
+      "loss": 1.4103,
+      "step": 59280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.874708597996966e-07,
+      "loss": 1.3874,
+      "step": 59290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.84439413042351e-07,
+      "loss": 1.3984,
+      "step": 59300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.814137194653876e-07,
+      "loss": 1.4038,
+      "step": 59310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.783937797876589e-07,
+      "loss": 1.4096,
+      "step": 59320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.753795947266468e-07,
+      "loss": 1.3863,
+      "step": 59330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.723711649984588e-07,
+      "loss": 1.3856,
+      "step": 59340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.693684913178595e-07,
+      "loss": 1.3745,
+      "step": 59350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.66371574398217e-07,
+      "loss": 1.3847,
+      "step": 59360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.633804149515505e-07,
+      "loss": 1.403,
+      "step": 59370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.603950136885057e-07,
+      "loss": 1.4057,
+      "step": 59380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.574153713183568e-07,
+      "loss": 1.4055,
+      "step": 59390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.544414885490125e-07,
+      "loss": 1.4041,
+      "step": 59400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.514733660870188e-07,
+      "loss": 1.3832,
+      "step": 59410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.485110046375476e-07,
+      "loss": 1.3858,
+      "step": 59420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.455544049043972e-07,
+      "loss": 1.3809,
+      "step": 59430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.426035675900083e-07,
+      "loss": 1.3727,
+      "step": 59440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.396584933954426e-07,
+      "loss": 1.3982,
+      "step": 59450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.367191830203957e-07,
+      "loss": 1.3747,
+      "step": 59460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.337856371631979e-07,
+      "loss": 1.367,
+      "step": 59470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.308578565208113e-07,
+      "loss": 1.3827,
+      "step": 59480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.2793584178881e-07,
+      "loss": 1.3906,
+      "step": 59490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.250195936614246e-07,
+      "loss": 1.3266,
+      "step": 59500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.221091128314956e-07,
+      "loss": 1.3763,
+      "step": 59510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.192043999905085e-07,
+      "loss": 1.388,
+      "step": 59520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.163054558285615e-07,
+      "loss": 1.3726,
+      "step": 59530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.134122810343979e-07,
+      "loss": 1.409,
+      "step": 59540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.10524876295382e-07,
+      "loss": 1.4022,
+      "step": 59550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.076432422975065e-07,
+      "loss": 1.4211,
+      "step": 59560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.04767379725399e-07,
+      "loss": 1.3742,
+      "step": 59570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.018972892623127e-07,
+      "loss": 1.4245,
+      "step": 59580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.990329715901328e-07,
+      "loss": 1.4452,
+      "step": 59590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.961744273893567e-07,
+      "loss": 1.3943,
+      "step": 59600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.933216573391382e-07,
+      "loss": 1.4327,
+      "step": 59610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.904746621172325e-07,
+      "loss": 1.3746,
+      "step": 59620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.876334424000403e-07,
+      "loss": 1.4087,
+      "step": 59630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.847979988625825e-07,
+      "loss": 1.3783,
+      "step": 59640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.81968332178512e-07,
+      "loss": 1.4068,
+      "step": 59650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.791444430200994e-07,
+      "loss": 1.3688,
+      "step": 59660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.76326332058258e-07,
+      "loss": 1.3779,
+      "step": 59670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.735139999625162e-07,
+      "loss": 1.3718,
+      "step": 59680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.70707447401031e-07,
+      "loss": 1.4244,
+      "step": 59690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.679066750405943e-07,
+      "loss": 1.3695,
+      "step": 59700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.651116835466154e-07,
+      "loss": 1.4015,
+      "step": 59710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.623224735831352e-07,
+      "loss": 1.4227,
+      "step": 59720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.595390458128181e-07,
+      "loss": 1.3377,
+      "step": 59730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.567614008969574e-07,
+      "loss": 1.3937,
+      "step": 59740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.539895394954726e-07,
+      "loss": 1.394,
+      "step": 59750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.512234622669089e-07,
+      "loss": 1.3969,
+      "step": 59760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.484631698684296e-07,
+      "loss": 1.3727,
+      "step": 59770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.457086629558407e-07,
+      "loss": 1.3957,
+      "step": 59780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.429599421835519e-07,
+      "loss": 1.4139,
+      "step": 59790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.402170082046188e-07,
+      "loss": 1.41,
+      "step": 59800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.37479861670709e-07,
+      "loss": 1.395,
+      "step": 59810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.347485032321216e-07,
+      "loss": 1.398,
+      "step": 59820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.320229335377736e-07,
+      "loss": 1.4142,
+      "step": 59830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.293031532352167e-07,
+      "loss": 1.3971,
+      "step": 59840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.265891629706172e-07,
+      "loss": 1.3751,
+      "step": 59850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.238809633887705e-07,
+      "loss": 1.3398,
+      "step": 59860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.211785551330979e-07,
+      "loss": 1.4401,
+      "step": 59870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.184819388456442e-07,
+      "loss": 1.4284,
+      "step": 59880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.157911151670748e-07,
+      "loss": 1.4288,
+      "step": 59890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.131060847366782e-07,
+      "loss": 1.4708,
+      "step": 59900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.104268481923775e-07,
+      "loss": 1.4601,
+      "step": 59910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.077534061706997e-07,
+      "loss": 1.4352,
+      "step": 59920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.050857593068171e-07,
+      "loss": 1.4373,
+      "step": 59930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.024239082345062e-07,
+      "loss": 1.4235,
+      "step": 59940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.997678535861778e-07,
+      "loss": 1.411,
+      "step": 59950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.97117595992866e-07,
+      "loss": 1.4528,
+      "step": 59960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.94473136084217e-07,
+      "loss": 1.4501,
+      "step": 59970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.918344744885146e-07,
+      "loss": 1.4422,
+      "step": 59980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.892016118326543e-07,
+      "loss": 1.4527,
+      "step": 59990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.865745487421498e-07,
+      "loss": 1.4451,
+      "step": 60000
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 1.5818288326263428,
+      "eval_runtime": 81.9976,
+      "eval_samples_per_second": 36.586,
+      "eval_steps_per_second": 4.573,
+      "step": 60000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.839532858411573e-07,
+      "loss": 1.4384,
+      "step": 60010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.813378237524314e-07,
+      "loss": 1.4485,
+      "step": 60020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.787281630973612e-07,
+      "loss": 1.4151,
+      "step": 60030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.761243044959535e-07,
+      "loss": 1.4175,
+      "step": 60040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.735262485668441e-07,
+      "loss": 1.4268,
+      "step": 60050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.709339959272781e-07,
+      "loss": 1.4373,
+      "step": 60060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.683475471931294e-07,
+      "loss": 1.4243,
+      "step": 60070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.657669029788954e-07,
+      "loss": 1.4327,
+      "step": 60080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.631920638976828e-07,
+      "loss": 1.4457,
+      "step": 60090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.606230305612359e-07,
+      "loss": 1.4328,
+      "step": 60100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.580598035799051e-07,
+      "loss": 1.4268,
+      "step": 60110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.555023835626671e-07,
+      "loss": 1.4811,
+      "step": 60120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.529507711171194e-07,
+      "loss": 1.4301,
+      "step": 60130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.504049668494826e-07,
+      "loss": 1.437,
+      "step": 60140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.478649713645867e-07,
+      "loss": 1.429,
+      "step": 60150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.453307852658962e-07,
+      "loss": 1.4005,
+      "step": 60160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.42802409155485e-07,
+      "loss": 1.3891,
+      "step": 60170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.40279843634045e-07,
+      "loss": 1.3829,
+      "step": 60180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.377630893008995e-07,
+      "loss": 1.3939,
+      "step": 60190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.352521467539811e-07,
+      "loss": 1.4158,
+      "step": 60200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.327470165898435e-07,
+      "loss": 1.4018,
+      "step": 60210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.302476994036631e-07,
+      "loss": 1.41,
+      "step": 60220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.27754195789229e-07,
+      "loss": 1.3976,
+      "step": 60230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.252665063389561e-07,
+      "loss": 1.4135,
+      "step": 60240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.227846316438745e-07,
+      "loss": 1.4189,
+      "step": 60250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.203085722936263e-07,
+      "loss": 1.4156,
+      "step": 60260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.178383288764882e-07,
+      "loss": 1.4066,
+      "step": 60270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.15373901979338e-07,
+      "loss": 1.4267,
+      "step": 60280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.129152921876823e-07,
+      "loss": 1.3942,
+      "step": 60290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.104625000856428e-07,
+      "loss": 1.3961,
+      "step": 60300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.080155262559588e-07,
+      "loss": 1.3971,
+      "step": 60310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.05574371279982e-07,
+      "loss": 1.3823,
+      "step": 60320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.031390357376959e-07,
+      "loss": 1.3918,
+      "step": 60330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.007095202076845e-07,
+      "loss": 1.3773,
+      "step": 60340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.982858252671613e-07,
+      "loss": 1.3921,
+      "step": 60350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.95867951491949e-07,
+      "loss": 1.3728,
+      "step": 60360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.934558994564964e-07,
+      "loss": 1.3783,
+      "step": 60370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.910496697338562e-07,
+      "loss": 1.3877,
+      "step": 60380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.886492628957101e-07,
+      "loss": 1.4156,
+      "step": 60390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.862546795123518e-07,
+      "loss": 1.4009,
+      "step": 60400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.838659201526874e-07,
+      "loss": 1.4033,
+      "step": 60410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.814829853842462e-07,
+      "loss": 1.402,
+      "step": 60420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.79105875773167e-07,
+      "loss": 1.4631,
+      "step": 60430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.767345918842175e-07,
+      "loss": 1.3907,
+      "step": 60440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7436913428075824e-07,
+      "loss": 1.3474,
+      "step": 60450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.720095035247896e-07,
+      "loss": 1.392,
+      "step": 60460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6965570017691305e-07,
+      "loss": 1.4169,
+      "step": 60470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6730772479635367e-07,
+      "loss": 1.3749,
+      "step": 60480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.649655779409373e-07,
+      "loss": 1.3936,
+      "step": 60490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6262926016713005e-07,
+      "loss": 1.3213,
+      "step": 60500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6029877202999064e-07,
+      "loss": 1.4039,
+      "step": 60510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.579741140832039e-07,
+      "loss": 1.4118,
+      "step": 60520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5565528687906423e-07,
+      "loss": 1.3979,
+      "step": 60530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.533422909684837e-07,
+      "loss": 1.3974,
+      "step": 60540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5103512690098937e-07,
+      "loss": 1.4182,
+      "step": 60550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4873379522472056e-07,
+      "loss": 1.4022,
+      "step": 60560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.464382964864372e-07,
+      "loss": 1.3883,
+      "step": 60570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4414863123150017e-07,
+      "loss": 1.4029,
+      "step": 60580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.418648000038994e-07,
+      "loss": 1.4081,
+      "step": 60590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.395868033462286e-07,
+      "loss": 1.3972,
+      "step": 60600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.373146417996965e-07,
+      "loss": 1.4307,
+      "step": 60610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.350483159041352e-07,
+      "loss": 1.4386,
+      "step": 60620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3278782619797786e-07,
+      "loss": 1.4653,
+      "step": 60630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.305331732182727e-07,
+      "loss": 1.431,
+      "step": 60640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.282843575006912e-07,
+      "loss": 1.4682,
+      "step": 60650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2604137957950595e-07,
+      "loss": 1.4537,
+      "step": 60660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.23804239987613e-07,
+      "loss": 1.4513,
+      "step": 60670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.215729392565149e-07,
+      "loss": 1.4493,
+      "step": 60680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1934747791632645e-07,
+      "loss": 1.4327,
+      "step": 60690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.171278564957776e-07,
+      "loss": 1.4541,
+      "step": 60700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.149140755222103e-07,
+      "loss": 1.4529,
+      "step": 60710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1270613552158166e-07,
+      "loss": 1.4189,
+      "step": 60720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1050403701845543e-07,
+      "loss": 1.4135,
+      "step": 60730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0830778053601307e-07,
+      "loss": 1.4151,
+      "step": 60740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0611736659603995e-07,
+      "loss": 1.4222,
+      "step": 60750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.039327957189476e-07,
+      "loss": 1.4165,
+      "step": 60760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.017540684237431e-07,
+      "loss": 1.4077,
+      "step": 60770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.995811852280568e-07,
+      "loss": 1.4474,
+      "step": 60780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.974141466481257e-07,
+      "loss": 1.4193,
+      "step": 60790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.952529531987992e-07,
+      "loss": 1.4244,
+      "step": 60800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.930976053935359e-07,
+      "loss": 1.423,
+      "step": 60810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.909481037444096e-07,
+      "loss": 1.433,
+      "step": 60820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.888044487621034e-07,
+      "loss": 1.4563,
+      "step": 60830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.866666409559072e-07,
+      "loss": 1.4197,
+      "step": 60840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8453468083373135e-07,
+      "loss": 1.4426,
+      "step": 60850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.824085689020873e-07,
+      "loss": 1.438,
+      "step": 60860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.802883056661044e-07,
+      "loss": 1.4409,
+      "step": 60870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.78173891629513e-07,
+      "loss": 1.3888,
+      "step": 60880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.76065327294664e-07,
+      "loss": 1.4263,
+      "step": 60890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7396261316251513e-07,
+      "loss": 1.4283,
+      "step": 60900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.718657497326333e-07,
+      "loss": 1.452,
+      "step": 60910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.697747375031896e-07,
+      "loss": 1.4101,
+      "step": 60920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6768957697097814e-07,
+      "loss": 1.4018,
+      "step": 60930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6561026863139167e-07,
+      "loss": 1.4052,
+      "step": 60940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.63536812978435e-07,
+      "loss": 1.3939,
+      "step": 60950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6146921050472794e-07,
+      "loss": 1.3988,
+      "step": 60960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.594074617014914e-07,
+      "loss": 1.3868,
+      "step": 60970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.573515670585642e-07,
+      "loss": 1.4025,
+      "step": 60980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5530152706438614e-07,
+      "loss": 1.4671,
+      "step": 60990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.532573422060148e-07,
+      "loss": 1.3759,
+      "step": 61000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5121901296910067e-07,
+      "loss": 1.3779,
+      "step": 61010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.491865398379257e-07,
+      "loss": 1.4063,
+      "step": 61020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4715992329536487e-07,
+      "loss": 1.3951,
+      "step": 61030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4513916382290237e-07,
+      "loss": 1.3976,
+      "step": 61040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.431242619006375e-07,
+      "loss": 1.3962,
+      "step": 61050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.411152180072791e-07,
+      "loss": 1.4307,
+      "step": 61060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.391120326201286e-07,
+      "loss": 1.4205,
+      "step": 61070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.371147062151164e-07,
+      "loss": 1.4065,
+      "step": 61080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.351232392667658e-07,
+      "loss": 1.4099,
+      "step": 61090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.331376322482149e-07,
+      "loss": 1.4129,
+      "step": 61100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3115788563120865e-07,
+      "loss": 1.3809,
+      "step": 61110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.291839998860985e-07,
+      "loss": 1.3808,
+      "step": 61120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2721597548184544e-07,
+      "loss": 1.3901,
+      "step": 61130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.252538128860144e-07,
+      "loss": 1.3959,
+      "step": 61140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.232975125647825e-07,
+      "loss": 1.3829,
+      "step": 61150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2134707498293057e-07,
+      "loss": 1.3828,
+      "step": 61160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1940250060384924e-07,
+      "loss": 1.3865,
+      "step": 61170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.174637898895272e-07,
+      "loss": 1.4056,
+      "step": 61180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1553094330057377e-07,
+      "loss": 1.3985,
+      "step": 61190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1360396129619665e-07,
+      "loss": 1.4118,
+      "step": 61200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1168284433421293e-07,
+      "loss": 1.3853,
+      "step": 61210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.097675928710408e-07,
+      "loss": 1.4015,
+      "step": 61220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0785820736171924e-07,
+      "loss": 1.3923,
+      "step": 61230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0595468825987127e-07,
+      "loss": 1.3785,
+      "step": 61240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0405703601774913e-07,
+      "loss": 1.4135,
+      "step": 61250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.021652510861922e-07,
+      "loss": 1.4261,
+      "step": 61260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0027933391466313e-07,
+      "loss": 1.4145,
+      "step": 61270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9839928495121463e-07,
+      "loss": 1.4218,
+      "step": 61280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9652510464251716e-07,
+      "loss": 1.4515,
+      "step": 61290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9465679343383956e-07,
+      "loss": 1.4562,
+      "step": 61300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.927943517690601e-07,
+      "loss": 1.4442,
+      "step": 61310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.90937780090661e-07,
+      "loss": 1.4528,
+      "step": 61320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.890870788397282e-07,
+      "loss": 1.3946,
+      "step": 61330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8724224845596003e-07,
+      "loss": 1.3639,
+      "step": 61340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.854032893776476e-07,
+      "loss": 1.4428,
+      "step": 61350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.835702020417025e-07,
+      "loss": 1.4382,
+      "step": 61360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.817429868836263e-07,
+      "loss": 1.4403,
+      "step": 61370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7992164433753576e-07,
+      "loss": 1.4348,
+      "step": 61380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7810617483614856e-07,
+      "loss": 1.4391,
+      "step": 61390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.762965788107891e-07,
+      "loss": 1.4449,
+      "step": 61400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7449285669138016e-07,
+      "loss": 1.4161,
+      "step": 61410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.726950089064595e-07,
+      "loss": 1.4413,
+      "step": 61420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7090303588316046e-07,
+      "loss": 1.4198,
+      "step": 61430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6911693804722026e-07,
+      "loss": 1.4386,
+      "step": 61440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6733671582298833e-07,
+      "loss": 1.4534,
+      "step": 61450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6556236963341253e-07,
+      "loss": 1.4405,
+      "step": 61460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.637938999000444e-07,
+      "loss": 1.435,
+      "step": 61470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.620313070430397e-07,
+      "loss": 1.4605,
+      "step": 61480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.602745914811605e-07,
+      "loss": 1.44,
+      "step": 61490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.585237536317703e-07,
+      "loss": 1.4059,
+      "step": 61500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5677879391083625e-07,
+      "loss": 1.398,
+      "step": 61510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.550397127329296e-07,
+      "loss": 1.3822,
+      "step": 61520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.533065105112226e-07,
+      "loss": 1.3863,
+      "step": 61530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5157918765749413e-07,
+      "loss": 1.3917,
+      "step": 61540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4985774458212986e-07,
+      "loss": 1.3953,
+      "step": 61550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4814218169410265e-07,
+      "loss": 1.3809,
+      "step": 61560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.464324994010114e-07,
+      "loss": 1.3828,
+      "step": 61570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.447286981090369e-07,
+      "loss": 1.4038,
+      "step": 61580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.430307782229774e-07,
+      "loss": 1.4187,
+      "step": 61590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4133874014622414e-07,
+      "loss": 1.4067,
+      "step": 61600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3965258428077507e-07,
+      "loss": 1.3923,
+      "step": 61610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3797231102723195e-07,
+      "loss": 1.4125,
+      "step": 61620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3629792078479762e-07,
+      "loss": 1.4093,
+      "step": 61630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.346294139512789e-07,
+      "loss": 1.4122,
+      "step": 61640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3296679092307539e-07,
+      "loss": 1.3996,
+      "step": 61650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3131005209520717e-07,
+      "loss": 1.3988,
+      "step": 61660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2965919786127332e-07,
+      "loss": 1.4143,
+      "step": 61670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2801422861349898e-07,
+      "loss": 1.4023,
+      "step": 61680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2637514474269094e-07,
+      "loss": 1.4071,
+      "step": 61690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2474194663826831e-07,
+      "loss": 1.4228,
+      "step": 61700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2311463468825121e-07,
+      "loss": 1.4071,
+      "step": 61710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2149320927925821e-07,
+      "loss": 1.4311,
+      "step": 61720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.198776707965089e-07,
+      "loss": 1.3902,
+      "step": 61730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.182680196238296e-07,
+      "loss": 1.412,
+      "step": 61740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1666425614364217e-07,
+      "loss": 1.4194,
+      "step": 61750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1506638073697237e-07,
+      "loss": 1.3934,
+      "step": 61760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1347439378344435e-07,
+      "loss": 1.4113,
+      "step": 61770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1188829566128887e-07,
+      "loss": 1.4148,
+      "step": 61780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1030808674733505e-07,
+      "loss": 1.4326,
+      "step": 61790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0873376741700755e-07,
+      "loss": 1.3805,
+      "step": 61800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0716533804434058e-07,
+      "loss": 1.4088,
+      "step": 61810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0560279900196377e-07,
+      "loss": 1.4018,
+      "step": 61820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0404615066110799e-07,
+      "loss": 1.4527,
+      "step": 61830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0249539339159962e-07,
+      "loss": 1.377,
+      "step": 61840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0095052756187726e-07,
+      "loss": 1.4209,
+      "step": 61850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9941155353897234e-07,
+      "loss": 1.4648,
+      "step": 61860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9787847168851458e-07,
+      "loss": 1.4578,
+      "step": 61870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9635128237473765e-07,
+      "loss": 1.4338,
+      "step": 61880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.948299859604791e-07,
+      "loss": 1.458,
+      "step": 61890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9331458280716086e-07,
+      "loss": 1.4915,
+      "step": 61900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9180507327482555e-07,
+      "loss": 1.4371,
+      "step": 61910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.903014577220974e-07,
+      "loss": 1.4461,
+      "step": 61920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8880373650621563e-07,
+      "loss": 1.4315,
+      "step": 61930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8731190998300674e-07,
+      "loss": 1.4689,
+      "step": 61940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8582597850690386e-07,
+      "loss": 1.4423,
+      "step": 61950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.843459424309385e-07,
+      "loss": 1.4246,
+      "step": 61960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.828718021067377e-07,
+      "loss": 1.4385,
+      "step": 61970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8140355788453245e-07,
+      "loss": 1.4478,
+      "step": 61980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7994121011314923e-07,
+      "loss": 1.4384,
+      "step": 61990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.784847591400185e-07,
+      "loss": 1.4597,
+      "step": 62000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7703420531116343e-07,
+      "loss": 1.4349,
+      "step": 62010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7558954897121393e-07,
+      "loss": 1.438,
+      "step": 62020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.741507904633899e-07,
+      "loss": 1.4242,
+      "step": 62030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7271793012951786e-07,
+      "loss": 1.4456,
+      "step": 62040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7129096831001722e-07,
+      "loss": 1.4386,
+      "step": 62050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6986990534390845e-07,
+      "loss": 1.3971,
+      "step": 62060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6845474156881036e-07,
+      "loss": 1.3981,
+      "step": 62070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6704547732094288e-07,
+      "loss": 1.3781,
+      "step": 62080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6564211293511601e-07,
+      "loss": 1.3942,
+      "step": 62090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6424464874475189e-07,
+      "loss": 1.4112,
+      "step": 62100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6285308508185716e-07,
+      "loss": 1.3931,
+      "step": 62110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6146742227704237e-07,
+      "loss": 1.3769,
+      "step": 62120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.600876606595164e-07,
+      "loss": 1.3996,
+      "step": 62130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587138005570893e-07,
+      "loss": 1.3683,
+      "step": 62140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.573458422961638e-07,
+      "loss": 1.399,
+      "step": 62150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5598378620173825e-07,
+      "loss": 1.4099,
+      "step": 62160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5462763259741497e-07,
+      "loss": 1.4101,
+      "step": 62170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5327738180539454e-07,
+      "loss": 1.4113,
+      "step": 62180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5193303414646754e-07,
+      "loss": 1.4135,
+      "step": 62190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5059458994003129e-07,
+      "loss": 1.4212,
+      "step": 62200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.492620495040731e-07,
+      "loss": 1.4206,
+      "step": 62210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4793541315517855e-07,
+      "loss": 1.4057,
+      "step": 62220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4661468120853726e-07,
+      "loss": 1.4111,
+      "step": 62230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4529985397792601e-07,
+      "loss": 1.3654,
+      "step": 62240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4399093177573098e-07,
+      "loss": 1.3989,
+      "step": 62250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.426879149129229e-07,
+      "loss": 1.3995,
+      "step": 62260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.413908036990763e-07,
+      "loss": 1.4072,
+      "step": 62270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.400995984423642e-07,
+      "loss": 1.3334,
+      "step": 62280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.388142994495495e-07,
+      "loss": 1.4196,
+      "step": 62290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3753490702599626e-07,
+      "loss": 1.4255,
+      "step": 62300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3626142147566967e-07,
+      "loss": 1.4019,
+      "step": 62310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3499384310112217e-07,
+      "loss": 1.3982,
+      "step": 62320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.337321722035073e-07,
+      "loss": 1.4296,
+      "step": 62330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.324764090825825e-07,
+      "loss": 1.4362,
+      "step": 62340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3122655403668416e-07,
+      "loss": 1.4522,
+      "step": 62350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2998260736276358e-07,
+      "loss": 1.4464,
+      "step": 62360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.287445693563538e-07,
+      "loss": 1.4694,
+      "step": 62370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2751244031159737e-07,
+      "loss": 1.4561,
+      "step": 62380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2628622052121842e-07,
+      "loss": 1.4392,
+      "step": 62390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2506591027655057e-07,
+      "loss": 1.4576,
+      "step": 62400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2385150986751194e-07,
+      "loss": 1.4429,
+      "step": 62410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2264301958262726e-07,
+      "loss": 1.4554,
+      "step": 62420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2144043970901132e-07,
+      "loss": 1.4523,
+      "step": 62430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2024377053237168e-07,
+      "loss": 1.4419,
+      "step": 62440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1905301233701982e-07,
+      "loss": 1.4516,
+      "step": 62450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1786816540585443e-07,
+      "loss": 1.4611,
+      "step": 62460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1668923002037535e-07,
+      "loss": 1.4471,
+      "step": 62470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1551620646067796e-07,
+      "loss": 1.4405,
+      "step": 62480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1434909500545321e-07,
+      "loss": 1.4404,
+      "step": 62490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1318789593197654e-07,
+      "loss": 1.4283,
+      "step": 62500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1203260951613837e-07,
+      "loss": 1.4621,
+      "step": 62510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.108832360324108e-07,
+      "loss": 1.4535,
+      "step": 62520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0973977575386151e-07,
+      "loss": 1.4431,
+      "step": 62530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0860222895215655e-07,
+      "loss": 1.4353,
+      "step": 62540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0747059589755748e-07,
+      "loss": 1.3918,
+      "step": 62550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0634487685892147e-07,
+      "loss": 1.3883,
+      "step": 62560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0522507210369847e-07,
+      "loss": 1.3874,
+      "step": 62570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0411118189793123e-07,
+      "loss": 1.4125,
+      "step": 62580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0300320650626082e-07,
+      "loss": 1.3829,
+      "step": 62590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.019011461919267e-07,
+      "loss": 1.3893,
+      "step": 62600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0080500121675274e-07,
+      "loss": 1.4034,
+      "step": 62610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.971477184116395e-08,
+      "loss": 1.3971,
+      "step": 62620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.863045832418371e-08,
+      "loss": 1.4241,
+      "step": 62630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.755206092341984e-08,
+      "loss": 1.4013,
+      "step": 62640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.647957989508405e-08,
+      "loss": 1.4248,
+      "step": 62650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.541301549397807e-08,
+      "loss": 1.4075,
+      "step": 62660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.435236797349645e-08,
+      "loss": 1.3861,
+      "step": 62670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.329763758562926e-08,
+      "loss": 1.3799,
+      "step": 62680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.224882458096496e-08,
+      "loss": 1.4213,
+      "step": 62690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.12059292086792e-08,
+      "loss": 1.3838,
+      "step": 62700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.016895171654605e-08,
+      "loss": 1.4087,
+      "step": 62710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.913789235093506e-08,
+      "loss": 1.3861,
+      "step": 62720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.811275135680309e-08,
+      "loss": 1.4005,
+      "step": 62730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.709352897771084e-08,
+      "loss": 1.4065,
+      "step": 62740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.608022545580075e-08,
+      "loss": 1.4185,
+      "step": 62750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.507284103181911e-08,
+      "loss": 1.3899,
+      "step": 62760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.407137594510229e-08,
+      "loss": 1.3754,
+      "step": 62770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.307583043357936e-08,
+      "loss": 1.4166,
+      "step": 62780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.208620473377226e-08,
+      "loss": 1.3867,
+      "step": 62790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.110249908080402e-08,
+      "loss": 1.3962,
+      "step": 62800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.012471370837937e-08,
+      "loss": 1.3822,
+      "step": 62810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.915284884880414e-08,
+      "loss": 1.3901,
+      "step": 62820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.818690473297696e-08,
+      "loss": 1.3883,
+      "step": 62830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.722688159038927e-08,
+      "loss": 1.3986,
+      "step": 62840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.62727796491225e-08,
+      "loss": 1.4002,
+      "step": 62850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.532459913585643e-08,
+      "loss": 1.4006,
+      "step": 62860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.438234027586366e-08,
+      "loss": 1.3797,
+      "step": 62870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.344600329300122e-08,
+      "loss": 1.3898,
+      "step": 62880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.251558840973283e-08,
+      "loss": 1.4037,
+      "step": 62890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.159109584710388e-08,
+      "loss": 1.4244,
+      "step": 62900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.067252582475814e-08,
+      "loss": 1.4302,
+      "step": 62910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.975987856093213e-08,
+      "loss": 1.3964,
+      "step": 62920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.88531542724552e-08,
+      "loss": 1.407,
+      "step": 62930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.795235317474391e-08,
+      "loss": 1.4176,
+      "step": 62940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.705747548181874e-08,
+      "loss": 1.4169,
+      "step": 62950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.61685214062846e-08,
+      "loss": 1.3913,
+      "step": 62960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.528549115933647e-08,
+      "loss": 1.3996,
+      "step": 62970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.440838495077317e-08,
+      "loss": 1.4181,
+      "step": 62980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.353720298897525e-08,
+      "loss": 1.4163,
+      "step": 62990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.267194548091882e-08,
+      "loss": 1.4286,
+      "step": 63000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.18126126321783e-08,
+      "loss": 1.4204,
+      "step": 63010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.095920464690986e-08,
+      "loss": 1.4398,
+      "step": 63020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.011172172787072e-08,
+      "loss": 1.4177,
+      "step": 63030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.927016407641095e-08,
+      "loss": 1.4352,
+      "step": 63040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.843453189246229e-08,
+      "loss": 1.4169,
+      "step": 63050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.760482537456036e-08,
+      "loss": 1.3893,
+      "step": 63060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.678104471982804e-08,
+      "loss": 1.4048,
+      "step": 63070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.596319012398099e-08,
+      "loss": 1.415,
+      "step": 63080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.5151261781324905e-08,
+      "loss": 1.4127,
+      "step": 63090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.4345259884761027e-08,
+      "loss": 1.398,
+      "step": 63100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.354518462578062e-08,
+      "loss": 1.4096,
+      "step": 63110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.275103619446497e-08,
+      "loss": 1.4116,
+      "step": 63120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.196281477949372e-08,
+      "loss": 1.4121,
+      "step": 63130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.118052056813094e-08,
+      "loss": 1.4028,
+      "step": 63140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.04041537462363e-08,
+      "loss": 1.4253,
+      "step": 63150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.963371449825949e-08,
+      "loss": 1.4008,
+      "step": 63160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.886920300724573e-08,
+      "loss": 1.4082,
+      "step": 63170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.811061945482476e-08,
+      "loss": 1.3943,
+      "step": 63180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.735796402122739e-08,
+      "loss": 1.4271,
+      "step": 63190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6611236885268917e-08,
+      "loss": 1.4605,
+      "step": 63200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5870438224357435e-08,
+      "loss": 1.4544,
+      "step": 63210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.513556821449383e-08,
+      "loss": 1.4295,
+      "step": 63220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4406627030268986e-08,
+      "loss": 1.4515,
+      "step": 63230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.368361484486938e-08,
+      "loss": 1.4289,
+      "step": 63240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.296653183006316e-08,
+      "loss": 1.433,
+      "step": 63250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2255378156225154e-08,
+      "loss": 1.4579,
+      "step": 63260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.155015399230633e-08,
+      "loss": 1.4332,
+      "step": 63270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.085085950585876e-08,
+      "loss": 1.4635,
+      "step": 63280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0157494863019006e-08,
+      "loss": 1.4396,
+      "step": 63290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.947006022851918e-08,
+      "loss": 1.4541,
+      "step": 63300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.878855576568141e-08,
+      "loss": 1.4546,
+      "step": 63310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.811298163642063e-08,
+      "loss": 1.4545,
+      "step": 63320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7443338001236226e-08,
+      "loss": 1.4309,
+      "step": 63330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.677962501923149e-08,
+      "loss": 1.4417,
+      "step": 63340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.612184284808306e-08,
+      "loss": 1.434,
+      "step": 63350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.546999164407705e-08,
+      "loss": 1.4322,
+      "step": 63360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.482407156207568e-08,
+      "loss": 1.4426,
+      "step": 63370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.418408275553953e-08,
+      "loss": 1.4393,
+      "step": 63380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.355002537651919e-08,
+      "loss": 1.4401,
+      "step": 63390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2921899575652504e-08,
+      "loss": 1.4294,
+      "step": 63400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2299705502175646e-08,
+      "loss": 1.4236,
+      "step": 63410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1683443303906493e-08,
+      "loss": 1.4327,
+      "step": 63420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1073113127261244e-08,
+      "loss": 1.4443,
+      "step": 63430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.046871511724059e-08,
+      "loss": 1.4341,
+      "step": 63440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.987024941743799e-08,
+      "loss": 1.4922,
+      "step": 63450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9277716170039715e-08,
+      "loss": 1.4402,
+      "step": 63460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8691115515822043e-08,
+      "loss": 1.443,
+      "step": 63470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8110447594148493e-08,
+      "loss": 1.4687,
+      "step": 63480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7535712542978153e-08,
+      "loss": 1.4763,
+      "step": 63490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6966910498854582e-08,
+      "loss": 1.4508,
+      "step": 63500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.640404159691412e-08,
+      "loss": 1.4658,
+      "step": 63510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5847105970888686e-08,
+      "loss": 1.4558,
+      "step": 63520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5296103753094658e-08,
+      "loss": 1.4341,
+      "step": 63530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4751035074438432e-08,
+      "loss": 1.4271,
+      "step": 63540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4211900064416425e-08,
+      "loss": 1.3818,
+      "step": 63550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3678698851123395e-08,
+      "loss": 1.4161,
+      "step": 63560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.315143156123578e-08,
+      "loss": 1.3957,
+      "step": 63570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2630098320020055e-08,
+      "loss": 1.396,
+      "step": 63580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2114699251338243e-08,
+      "loss": 1.4011,
+      "step": 63590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1605234477639625e-08,
+      "loss": 1.4132,
+      "step": 63600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.110170411996626e-08,
+      "loss": 1.3979,
+      "step": 63610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0604108297941905e-08,
+      "loss": 1.4142,
+      "step": 63620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0112447129791433e-08,
+      "loss": 1.4002,
+      "step": 63630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9626720732321414e-08,
+      "loss": 1.3871,
+      "step": 63640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9146929220933973e-08,
+      "loss": 1.4057,
+      "step": 63650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.867307270961849e-08,
+      "loss": 1.406,
+      "step": 63660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.820515131095435e-08,
+      "loss": 1.3913,
+      "step": 63670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7743165136110964e-08,
+      "loss": 1.4131,
+      "step": 63680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.728711429484775e-08,
+      "loss": 1.4186,
+      "step": 63690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6836998895514155e-08,
+      "loss": 1.4121,
+      "step": 63700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.639281904505241e-08,
+      "loss": 1.4079,
+      "step": 63710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5954574848986436e-08,
+      "loss": 1.4214,
+      "step": 63720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5522266411441278e-08,
+      "loss": 1.4273,
+      "step": 63730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5095893835118113e-08,
+      "loss": 1.433,
+      "step": 63740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.467545722132202e-08,
+      "loss": 1.3967,
+      "step": 63750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4260956669939762e-08,
+      "loss": 1.4042,
+      "step": 63760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3852392279445347e-08,
+      "loss": 1.4123,
+      "step": 63770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3449764146911125e-08,
+      "loss": 1.4277,
+      "step": 63780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3053072367991137e-08,
+      "loss": 1.4108,
+      "step": 63790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2662317036932214e-08,
+      "loss": 1.416,
+      "step": 63800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2277498246571206e-08,
+      "loss": 1.4204,
+      "step": 63810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1898616088337755e-08,
+      "loss": 1.4084,
+      "step": 63820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1525670652243192e-08,
+      "loss": 1.4157,
+      "step": 63830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1158662026891642e-08,
+      "loss": 1.4437,
+      "step": 63840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0797590299482796e-08,
+      "loss": 1.4054,
+      "step": 63850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0442455555798037e-08,
+      "loss": 1.4109,
+      "step": 63860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0093257880208762e-08,
+      "loss": 1.4116,
+      "step": 63870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.749997355679164e-09,
+      "loss": 1.4179,
+      "step": 63880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.412674063766225e-09,
+      "loss": 1.3854,
+      "step": 63890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.081288084605844e-09,
+      "loss": 1.4173,
+      "step": 63900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.755839496932261e-09,
+      "loss": 1.4225,
+      "step": 63910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.43632837806696e-09,
+      "loss": 1.4167,
+      "step": 63920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.122754803918665e-09,
+      "loss": 1.4473,
+      "step": 63930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.815118848986113e-09,
+      "loss": 1.4522,
+      "step": 63940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.513420586358067e-09,
+      "loss": 1.4047,
+      "step": 63950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.217660087713296e-09,
+      "loss": 1.4055,
+      "step": 63960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.927837423320593e-09,
+      "loss": 1.4354,
+      "step": 63970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.64395266203599e-09,
+      "loss": 1.4274,
+      "step": 63980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.36600587130276e-09,
+      "loss": 1.4059,
+      "step": 63990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.093997117159744e-09,
+      "loss": 1.3817,
+      "step": 64000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.827926464227473e-09,
+      "loss": 1.4201,
+      "step": 64010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.567793975724822e-09,
+      "loss": 1.4139,
+      "step": 64020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.31359971344958e-09,
+      "loss": 1.4076,
+      "step": 64030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.065343737795103e-09,
+      "loss": 1.4405,
+      "step": 64040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.82302610774199e-09,
+      "loss": 1.4346,
+      "step": 64050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.586646880863632e-09,
+      "loss": 1.4489,
+      "step": 64060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.356206113315109e-09,
+      "loss": 1.4515,
+      "step": 64070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.131703859849845e-09,
+      "loss": 1.4392,
+      "step": 64080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.913140173800178e-09,
+      "loss": 1.4506,
+      "step": 64090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7005151070967916e-09,
+      "loss": 1.451,
+      "step": 64100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4938287102548316e-09,
+      "loss": 1.4501,
+      "step": 64110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.293081032379464e-09,
+      "loss": 1.4621,
+      "step": 64120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0982721211630927e-09,
+      "loss": 1.4442,
+      "step": 64130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9094020228881414e-09,
+      "loss": 1.4632,
+      "step": 64140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7264707824325998e-09,
+      "loss": 1.4479,
+      "step": 64150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5494784432505967e-09,
+      "loss": 1.4508,
+      "step": 64160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3784250473946058e-09,
+      "loss": 1.4454,
+      "step": 64170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.21331063550434e-09,
+      "loss": 1.4473,
+      "step": 64180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.05413524680953e-09,
+      "loss": 1.4416,
+      "step": 64190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.900898919124372e-09,
+      "loss": 1.443,
+      "step": 64200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7536016888558548e-09,
+      "loss": 1.4568,
+      "step": 64210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6122435910037591e-09,
+      "loss": 1.487,
+      "step": 64220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.476824659144005e-09,
+      "loss": 1.4426,
+      "step": 64230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3473449254564064e-09,
+      "loss": 1.466,
+      "step": 64240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2238044206996923e-09,
+      "loss": 1.4406,
+      "step": 64250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1062031742253842e-09,
+      "loss": 1.4513,
+      "step": 64260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.9454121397502e-10,
+      "loss": 1.4563,
+      "step": 64270
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.888185664746029e-10,
+      "loss": 1.4638,
+      "step": 64280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.890352568457049e-10,
+      "loss": 1.4501,
+      "step": 64290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.951913087915874e-10,
+      "loss": 1.442,
+      "step": 64300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.072867446083042e-10,
+      "loss": 1.4548,
+      "step": 64310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.25321585181926e-10,
+      "loss": 1.4435,
+      "step": 64320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.492958499857647e-10,
+      "loss": 1.4338,
+      "step": 64330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.792095570803733e-10,
+      "loss": 1.4099,
+      "step": 64340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1506272311909725e-10,
+      "loss": 1.4111,
+      "step": 64350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.56855363342523e-10,
+      "loss": 1.4059,
+      "step": 64360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.045874915757029e-10,
+      "loss": 1.4159,
+      "step": 64370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5825912023925694e-10,
+      "loss": 1.3885,
+      "step": 64380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.17870260343822e-10,
+      "loss": 1.4094,
+      "step": 64390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.342092147617386e-11,
+      "loss": 1.4084,
+      "step": 64400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.49111118297585e-11,
+      "loss": 1.4197,
+      "step": 64410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2340838171385225e-11,
+      "loss": 1.4317,
+      "step": 64420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5710105866206802e-11,
+      "loss": 1.4178,
+      "step": 64430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.018918866617206e-12,
+      "loss": 1.4153,
+      "step": 64440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.672797122516002e-13,
+      "loss": 1.4375,
+      "step": 64450
+    },
+    {
+      "epoch": 1.0,
+      "step": 64453,
+      "total_flos": 8.622620333309952e+18,
+      "train_loss": 1.6966034523956515,
+      "train_runtime": 1526476.5703,
+      "train_samples_per_second": 10.809,
+      "train_steps_per_second": 0.042
+    }
+  ],
+  "max_steps": 64453,
+  "num_train_epochs": 9223372036854775807,
+  "total_flos": 8.622620333309952e+18,
+  "trial_name": null,
+  "trial_params": null
+}