diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,113560 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.992557195415909,
+  "global_step": 189000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.5091,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.6233,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 0.5031,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.5441,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 0.6067,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.6e-05,
+      "loss": 0.6633,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 0.4737,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 0.5007,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000144,
+      "loss": 0.5474,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016,
+      "loss": 0.5314,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 0.5054,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000192,
+      "loss": 0.4627,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020800000000000001,
+      "loss": 0.477,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022400000000000002,
+      "loss": 0.7119,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024,
+      "loss": 0.5165,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025600000000000004,
+      "loss": 0.4349,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 0.5688,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000288,
+      "loss": 0.6779,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000304,
+      "loss": 0.4452,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032,
+      "loss": 0.4647,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000336,
+      "loss": 0.4946,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035200000000000005,
+      "loss": 0.5094,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036800000000000005,
+      "loss": 0.5185,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000384,
+      "loss": 0.5356,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004,
+      "loss": 0.5815,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00041600000000000003,
+      "loss": 0.5643,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00043200000000000004,
+      "loss": 0.4796,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044800000000000005,
+      "loss": 0.445,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000464,
+      "loss": 0.4843,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00048,
+      "loss": 0.5037,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000496,
+      "loss": 0.5654,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005120000000000001,
+      "loss": 0.5071,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000528,
+      "loss": 0.5664,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005440000000000001,
+      "loss": 0.5739,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00056,
+      "loss": 0.5219,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000576,
+      "loss": 0.6144,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000592,
+      "loss": 0.5183,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000608,
+      "loss": 0.568,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006240000000000001,
+      "loss": 0.5101,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00064,
+      "loss": 0.5742,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000656,
+      "loss": 0.5648,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000672,
+      "loss": 0.443,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000688,
+      "loss": 0.5076,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007040000000000001,
+      "loss": 0.5114,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00072,
+      "loss": 0.5598,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007360000000000001,
+      "loss": 0.4879,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000752,
+      "loss": 0.5685,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000768,
+      "loss": 0.656,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000784,
+      "loss": 0.6048,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008,
+      "loss": 0.4491,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007999575992707075,
+      "loss": 0.6598,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000799915198541415,
+      "loss": 0.5416,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007998727978121224,
+      "loss": 0.5244,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007998303970828299,
+      "loss": 0.5654,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007997879963535373,
+      "loss": 0.5205,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007997455956242447,
+      "loss": 0.5595,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007997031948949523,
+      "loss": 0.5699,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007996607941656597,
+      "loss": 0.5969,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007996183934363671,
+      "loss": 0.5695,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007995759927070746,
+      "loss": 0.5458,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007995335919777821,
+      "loss": 0.5963,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007994911912484895,
+      "loss": 0.4507,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000799448790519197,
+      "loss": 0.5426,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007994063897899044,
+      "loss": 0.5186,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007993639890606119,
+      "loss": 0.585,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007993215883313193,
+      "loss": 0.6048,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007992791876020268,
+      "loss": 0.5359,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007992367868727342,
+      "loss": 0.5294,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007991943861434417,
+      "loss": 0.5608,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007991519854141492,
+      "loss": 0.582,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007991095846848566,
+      "loss": 0.494,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000799067183955564,
+      "loss": 0.5367,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007990247832262716,
+      "loss": 0.5862,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000798982382496979,
+      "loss": 0.5445,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007989399817676864,
+      "loss": 0.5066,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007988975810383938,
+      "loss": 0.5879,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007988551803091014,
+      "loss": 0.5377,
+      "step": 770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007988127795798088,
+      "loss": 0.7732,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007987703788505162,
+      "loss": 0.5179,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007987279781212237,
+      "loss": 0.5247,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007986855773919312,
+      "loss": 0.5853,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007986431766626387,
+      "loss": 0.5689,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007986007759333461,
+      "loss": 0.5968,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007985583752040535,
+      "loss": 0.6044,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000798515974474761,
+      "loss": 0.4833,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007984735737454685,
+      "loss": 0.4947,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007984311730161759,
+      "loss": 0.5956,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007983887722868833,
+      "loss": 0.576,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007983463715575909,
+      "loss": 0.6091,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007983039708282983,
+      "loss": 0.6454,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007982615700990057,
+      "loss": 0.6375,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007982191693697132,
+      "loss": 0.6126,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007981767686404207,
+      "loss": 0.6092,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007981343679111281,
+      "loss": 0.5463,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007980919671818356,
+      "loss": 0.5512,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000798049566452543,
+      "loss": 0.6774,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007980071657232505,
+      "loss": 0.4869,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007979647649939579,
+      "loss": 0.6152,
+      "step": 980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007979223642646654,
+      "loss": 0.5422,
+      "step": 990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007978799635353728,
+      "loss": 0.5472,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007978375628060803,
+      "loss": 0.5053,
+      "step": 1010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007977951620767878,
+      "loss": 0.5338,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007977527613474952,
+      "loss": 0.5285,
+      "step": 1030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007977103606182026,
+      "loss": 0.5798,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007976679598889102,
+      "loss": 0.5947,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007976255591596176,
+      "loss": 0.5286,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000797583158430325,
+      "loss": 0.57,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007975407577010325,
+      "loss": 0.4947,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007974983569717399,
+      "loss": 0.5894,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007974559562424474,
+      "loss": 0.5931,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007974135555131549,
+      "loss": 0.6035,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007973711547838623,
+      "loss": 0.5528,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007973287540545698,
+      "loss": 0.5854,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007972863533252773,
+      "loss": 0.59,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007972439525959847,
+      "loss": 0.5969,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007972015518666921,
+      "loss": 0.5385,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007971591511373996,
+      "loss": 0.5943,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007971167504081071,
+      "loss": 0.6122,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007970743496788145,
+      "loss": 0.6109,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000797031948949522,
+      "loss": 0.6666,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007969895482202294,
+      "loss": 0.5948,
+      "step": 1210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007969471474909369,
+      "loss": 0.6014,
+      "step": 1220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007969047467616444,
+      "loss": 0.6087,
+      "step": 1230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007968623460323518,
+      "loss": 0.48,
+      "step": 1240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007968199453030592,
+      "loss": 0.5745,
+      "step": 1250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007967775445737668,
+      "loss": 0.6438,
+      "step": 1260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007967351438444741,
+      "loss": 0.5564,
+      "step": 1270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007966927431151816,
+      "loss": 0.5707,
+      "step": 1280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000796650342385889,
+      "loss": 0.5371,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007966079416565965,
+      "loss": 0.5599,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000796565540927304,
+      "loss": 0.5122,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007965231401980114,
+      "loss": 0.6212,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007964807394687189,
+      "loss": 0.5456,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007964383387394264,
+      "loss": 0.5815,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007963959380101338,
+      "loss": 0.5772,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007963535372808412,
+      "loss": 0.5782,
+      "step": 1360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007963111365515487,
+      "loss": 0.6232,
+      "step": 1370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007962687358222562,
+      "loss": 0.5326,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007962263350929636,
+      "loss": 0.4819,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007961839343636711,
+      "loss": 0.6134,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007961415336343785,
+      "loss": 0.6849,
+      "step": 1410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000796099132905086,
+      "loss": 0.5251,
+      "step": 1420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007960567321757935,
+      "loss": 0.6464,
+      "step": 1430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007960143314465009,
+      "loss": 0.6525,
+      "step": 1440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007959719307172083,
+      "loss": 0.7665,
+      "step": 1450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007959295299879159,
+      "loss": 0.5674,
+      "step": 1460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007958871292586233,
+      "loss": 0.5564,
+      "step": 1470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007958447285293307,
+      "loss": 0.5398,
+      "step": 1480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007958023278000382,
+      "loss": 0.5429,
+      "step": 1490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007957599270707457,
+      "loss": 0.5074,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007957175263414531,
+      "loss": 0.5954,
+      "step": 1510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007956751256121606,
+      "loss": 0.59,
+      "step": 1520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000795632724882868,
+      "loss": 0.6186,
+      "step": 1530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007955903241535755,
+      "loss": 0.6087,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000795547923424283,
+      "loss": 0.5917,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007955055226949904,
+      "loss": 0.5571,
+      "step": 1560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007954631219656978,
+      "loss": 0.4796,
+      "step": 1570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007954207212364054,
+      "loss": 0.5097,
+      "step": 1580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007953783205071127,
+      "loss": 0.5406,
+      "step": 1590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007953359197778202,
+      "loss": 0.5884,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007952935190485277,
+      "loss": 0.6562,
+      "step": 1610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007952511183192351,
+      "loss": 0.5142,
+      "step": 1620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007952087175899426,
+      "loss": 0.626,
+      "step": 1630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007951663168606501,
+      "loss": 0.5783,
+      "step": 1640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007951239161313574,
+      "loss": 0.6906,
+      "step": 1650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000795081515402065,
+      "loss": 0.5203,
+      "step": 1660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007950391146727725,
+      "loss": 0.5111,
+      "step": 1670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007949967139434798,
+      "loss": 0.5662,
+      "step": 1680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007949543132141873,
+      "loss": 0.6169,
+      "step": 1690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007949119124848948,
+      "loss": 0.7241,
+      "step": 1700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007948695117556022,
+      "loss": 0.5481,
+      "step": 1710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007948271110263097,
+      "loss": 0.577,
+      "step": 1720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007947847102970172,
+      "loss": 0.6324,
+      "step": 1730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007947423095677246,
+      "loss": 0.6097,
+      "step": 1740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007946999088384321,
+      "loss": 0.5152,
+      "step": 1750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007946575081091396,
+      "loss": 0.6197,
+      "step": 1760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007946151073798469,
+      "loss": 0.4844,
+      "step": 1770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007945727066505545,
+      "loss": 0.7067,
+      "step": 1780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000794530305921262,
+      "loss": 0.7044,
+      "step": 1790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007944879051919693,
+      "loss": 0.4668,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007944455044626768,
+      "loss": 0.6165,
+      "step": 1810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007944031037333842,
+      "loss": 0.5812,
+      "step": 1820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007943607030040917,
+      "loss": 0.6261,
+      "step": 1830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007943183022747992,
+      "loss": 0.6095,
+      "step": 1840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007942759015455066,
+      "loss": 0.5262,
+      "step": 1850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007942335008162141,
+      "loss": 0.583,
+      "step": 1860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007941911000869216,
+      "loss": 0.5823,
+      "step": 1870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000794148699357629,
+      "loss": 0.6352,
+      "step": 1880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007941062986283364,
+      "loss": 0.528,
+      "step": 1890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007940638978990439,
+      "loss": 0.6448,
+      "step": 1900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007940214971697513,
+      "loss": 0.6355,
+      "step": 1910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007939790964404588,
+      "loss": 0.6835,
+      "step": 1920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007939366957111663,
+      "loss": 0.6171,
+      "step": 1930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007938942949818737,
+      "loss": 0.5404,
+      "step": 1940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007938518942525812,
+      "loss": 0.7288,
+      "step": 1950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007938094935232887,
+      "loss": 0.4581,
+      "step": 1960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000793767092793996,
+      "loss": 0.6394,
+      "step": 1970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007937246920647035,
+      "loss": 0.5071,
+      "step": 1980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007936822913354111,
+      "loss": 0.5476,
+      "step": 1990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007936398906061184,
+      "loss": 0.5368,
+      "step": 2000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007935974898768259,
+      "loss": 0.6201,
+      "step": 2010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007935550891475334,
+      "loss": 0.6034,
+      "step": 2020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007935126884182408,
+      "loss": 0.5521,
+      "step": 2030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007934702876889483,
+      "loss": 0.5972,
+      "step": 2040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007934278869596558,
+      "loss": 0.6811,
+      "step": 2050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007933854862303631,
+      "loss": 0.4694,
+      "step": 2060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007933430855010707,
+      "loss": 0.5801,
+      "step": 2070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007933006847717782,
+      "loss": 0.6109,
+      "step": 2080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007932582840424855,
+      "loss": 0.5923,
+      "step": 2090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000793215883313193,
+      "loss": 0.5756,
+      "step": 2100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007931734825839006,
+      "loss": 0.6424,
+      "step": 2110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007931310818546079,
+      "loss": 0.5795,
+      "step": 2120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007930886811253154,
+      "loss": 0.4616,
+      "step": 2130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007930462803960229,
+      "loss": 0.6285,
+      "step": 2140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007930038796667303,
+      "loss": 0.7408,
+      "step": 2150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007929614789374378,
+      "loss": 0.6614,
+      "step": 2160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007929190782081453,
+      "loss": 0.5848,
+      "step": 2170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007928766774788526,
+      "loss": 0.5521,
+      "step": 2180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007928342767495602,
+      "loss": 0.5437,
+      "step": 2190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007927918760202675,
+      "loss": 0.5712,
+      "step": 2200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000792749475290975,
+      "loss": 0.606,
+      "step": 2210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007927070745616825,
+      "loss": 0.6201,
+      "step": 2220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00079266467383239,
+      "loss": 0.5566,
+      "step": 2230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007926222731030974,
+      "loss": 0.5775,
+      "step": 2240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007925798723738049,
+      "loss": 0.6357,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007925374716445122,
+      "loss": 0.4891,
+      "step": 2260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007924950709152198,
+      "loss": 0.5384,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007924526701859273,
+      "loss": 0.5898,
+      "step": 2280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007924102694566346,
+      "loss": 0.5556,
+      "step": 2290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007923678687273421,
+      "loss": 0.7283,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007923254679980497,
+      "loss": 0.508,
+      "step": 2310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000792283067268757,
+      "loss": 0.4856,
+      "step": 2320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007922406665394645,
+      "loss": 0.5468,
+      "step": 2330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000792198265810172,
+      "loss": 0.5032,
+      "step": 2340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007921558650808794,
+      "loss": 0.5627,
+      "step": 2350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007921134643515869,
+      "loss": 0.6195,
+      "step": 2360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007920710636222944,
+      "loss": 0.5602,
+      "step": 2370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007920286628930017,
+      "loss": 0.6204,
+      "step": 2380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007919862621637093,
+      "loss": 0.5698,
+      "step": 2390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007919438614344168,
+      "loss": 0.5808,
+      "step": 2400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007919014607051241,
+      "loss": 0.6035,
+      "step": 2410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007918590599758316,
+      "loss": 0.6023,
+      "step": 2420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007918166592465391,
+      "loss": 0.5229,
+      "step": 2430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007917742585172465,
+      "loss": 0.596,
+      "step": 2440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000791731857787954,
+      "loss": 0.6354,
+      "step": 2450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007916894570586615,
+      "loss": 0.5765,
+      "step": 2460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007916470563293689,
+      "loss": 0.6014,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007916046556000764,
+      "loss": 0.6596,
+      "step": 2480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007915622548707839,
+      "loss": 0.6122,
+      "step": 2490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007915198541414912,
+      "loss": 0.5848,
+      "step": 2500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007914774534121987,
+      "loss": 0.5931,
+      "step": 2510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007914350526829062,
+      "loss": 0.5904,
+      "step": 2520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007913926519536136,
+      "loss": 0.5554,
+      "step": 2530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007913502512243211,
+      "loss": 0.5614,
+      "step": 2540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007913078504950286,
+      "loss": 0.4867,
+      "step": 2550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000791265449765736,
+      "loss": 0.5689,
+      "step": 2560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007912230490364435,
+      "loss": 0.544,
+      "step": 2570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007911806483071509,
+      "loss": 0.663,
+      "step": 2580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007911382475778583,
+      "loss": 0.559,
+      "step": 2590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007910958468485659,
+      "loss": 0.5476,
+      "step": 2600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007910534461192733,
+      "loss": 0.6038,
+      "step": 2610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007910110453899807,
+      "loss": 0.6968,
+      "step": 2620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007909686446606882,
+      "loss": 0.4827,
+      "step": 2630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007909262439313957,
+      "loss": 0.668,
+      "step": 2640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007908838432021031,
+      "loss": 0.5254,
+      "step": 2650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007908414424728106,
+      "loss": 0.6104,
+      "step": 2660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000790799041743518,
+      "loss": 0.6409,
+      "step": 2670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007907566410142255,
+      "loss": 0.6134,
+      "step": 2680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000790714240284933,
+      "loss": 0.7575,
+      "step": 2690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007906718395556403,
+      "loss": 0.5912,
+      "step": 2700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007906294388263478,
+      "loss": 0.5487,
+      "step": 2710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007905870380970554,
+      "loss": 0.6304,
+      "step": 2720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007905446373677627,
+      "loss": 0.5316,
+      "step": 2730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007905022366384702,
+      "loss": 0.6676,
+      "step": 2740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007904598359091777,
+      "loss": 0.5876,
+      "step": 2750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007904174351798851,
+      "loss": 0.602,
+      "step": 2760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007903750344505926,
+      "loss": 0.497,
+      "step": 2770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007903326337213001,
+      "loss": 0.6571,
+      "step": 2780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007902902329920074,
+      "loss": 0.5879,
+      "step": 2790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000790247832262715,
+      "loss": 0.6539,
+      "step": 2800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007902054315334224,
+      "loss": 0.6027,
+      "step": 2810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007901630308041298,
+      "loss": 0.5757,
+      "step": 2820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007901206300748373,
+      "loss": 0.6731,
+      "step": 2830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007900782293455448,
+      "loss": 0.4728,
+      "step": 2840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007900358286162522,
+      "loss": 0.5114,
+      "step": 2850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007899934278869597,
+      "loss": 0.6053,
+      "step": 2860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007899510271576671,
+      "loss": 0.6404,
+      "step": 2870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007899086264283746,
+      "loss": 0.5954,
+      "step": 2880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007898662256990821,
+      "loss": 0.5363,
+      "step": 2890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007898238249697895,
+      "loss": 0.5591,
+      "step": 2900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007897814242404969,
+      "loss": 0.4779,
+      "step": 2910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007897390235112045,
+      "loss": 0.5933,
+      "step": 2920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007896966227819119,
+      "loss": 0.6104,
+      "step": 2930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007896542220526193,
+      "loss": 0.6662,
+      "step": 2940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007896118213233268,
+      "loss": 0.5963,
+      "step": 2950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007895694205940343,
+      "loss": 0.5619,
+      "step": 2960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007895270198647417,
+      "loss": 0.6362,
+      "step": 2970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007894846191354492,
+      "loss": 0.6881,
+      "step": 2980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007894422184061566,
+      "loss": 0.5541,
+      "step": 2990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007893998176768641,
+      "loss": 0.5718,
+      "step": 3000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007893574169475716,
+      "loss": 0.653,
+      "step": 3010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000789315016218279,
+      "loss": 0.6233,
+      "step": 3020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007892726154889864,
+      "loss": 0.6744,
+      "step": 3030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007892302147596939,
+      "loss": 0.6081,
+      "step": 3040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007891878140304014,
+      "loss": 0.5098,
+      "step": 3050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007891454133011088,
+      "loss": 0.5829,
+      "step": 3060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007891030125718163,
+      "loss": 0.5751,
+      "step": 3070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007890606118425238,
+      "loss": 0.6398,
+      "step": 3080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007890182111132312,
+      "loss": 0.51,
+      "step": 3090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007889758103839387,
+      "loss": 0.5058,
+      "step": 3100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000788933409654646,
+      "loss": 0.551,
+      "step": 3110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007888910089253535,
+      "loss": 0.6021,
+      "step": 3120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000788848608196061,
+      "loss": 0.5232,
+      "step": 3130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007888062074667685,
+      "loss": 0.5788,
+      "step": 3140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007887638067374759,
+      "loss": 0.6711,
+      "step": 3150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007887214060081834,
+      "loss": 0.6358,
+      "step": 3160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007886790052788909,
+      "loss": 0.6255,
+      "step": 3170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007886366045495983,
+      "loss": 0.6984,
+      "step": 3180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007885942038203057,
+      "loss": 0.6655,
+      "step": 3190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007885518030910131,
+      "loss": 0.6883,
+      "step": 3200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007885094023617207,
+      "loss": 0.6157,
+      "step": 3210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007884670016324281,
+      "loss": 0.5707,
+      "step": 3220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007884246009031355,
+      "loss": 0.6344,
+      "step": 3230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000788382200173843,
+      "loss": 0.5579,
+      "step": 3240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007883397994445505,
+      "loss": 0.6221,
+      "step": 3250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007882973987152579,
+      "loss": 0.6806,
+      "step": 3260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007882549979859654,
+      "loss": 0.6406,
+      "step": 3270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007882125972566729,
+      "loss": 0.6485,
+      "step": 3280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007881701965273803,
+      "loss": 0.6076,
+      "step": 3290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007881277957980878,
+      "loss": 0.7057,
+      "step": 3300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007880853950687952,
+      "loss": 0.6198,
+      "step": 3310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007880429943395026,
+      "loss": 0.694,
+      "step": 3320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007880005936102102,
+      "loss": 0.6509,
+      "step": 3330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007879581928809176,
+      "loss": 0.6122,
+      "step": 3340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000787915792151625,
+      "loss": 0.6295,
+      "step": 3350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007878733914223325,
+      "loss": 0.7357,
+      "step": 3360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00078783099069304,
+      "loss": 0.5658,
+      "step": 3370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007877885899637474,
+      "loss": 0.692,
+      "step": 3380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007877461892344549,
+      "loss": 0.6278,
+      "step": 3390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007877037885051623,
+      "loss": 0.6803,
+      "step": 3400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007876613877758698,
+      "loss": 0.5734,
+      "step": 3410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007876189870465772,
+      "loss": 0.6874,
+      "step": 3420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007875765863172847,
+      "loss": 0.5183,
+      "step": 3430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007875341855879921,
+      "loss": 0.5669,
+      "step": 3440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007874917848586996,
+      "loss": 0.6534,
+      "step": 3450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007874493841294071,
+      "loss": 0.6402,
+      "step": 3460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007874069834001145,
+      "loss": 0.5163,
+      "step": 3470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007873645826708219,
+      "loss": 0.6744,
+      "step": 3480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007873221819415295,
+      "loss": 0.6753,
+      "step": 3490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007872797812122369,
+      "loss": 0.5066,
+      "step": 3500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007872373804829443,
+      "loss": 0.5706,
+      "step": 3510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007871949797536518,
+      "loss": 0.6045,
+      "step": 3520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007871525790243593,
+      "loss": 0.5469,
+      "step": 3530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007871101782950667,
+      "loss": 0.5884,
+      "step": 3540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007870677775657742,
+      "loss": 0.4794,
+      "step": 3550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007870253768364816,
+      "loss": 0.5506,
+      "step": 3560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007869829761071891,
+      "loss": 0.6138,
+      "step": 3570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007869405753778966,
+      "loss": 0.7031,
+      "step": 3580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000786898174648604,
+      "loss": 0.6951,
+      "step": 3590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007868557739193114,
+      "loss": 0.6586,
+      "step": 3600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000786813373190019,
+      "loss": 0.5815,
+      "step": 3610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007867709724607264,
+      "loss": 0.7233,
+      "step": 3620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007867285717314338,
+      "loss": 0.6076,
+      "step": 3630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007866861710021412,
+      "loss": 0.5623,
+      "step": 3640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007866437702728487,
+      "loss": 0.6109,
+      "step": 3650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007866013695435562,
+      "loss": 0.6527,
+      "step": 3660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007865589688142636,
+      "loss": 0.5235,
+      "step": 3670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007865165680849711,
+      "loss": 0.5523,
+      "step": 3680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007864741673556786,
+      "loss": 0.6074,
+      "step": 3690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000786431766626386,
+      "loss": 0.6749,
+      "step": 3700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007863893658970935,
+      "loss": 0.584,
+      "step": 3710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007863469651678009,
+      "loss": 0.5507,
+      "step": 3720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007863045644385083,
+      "loss": 0.5906,
+      "step": 3730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007862621637092158,
+      "loss": 0.5458,
+      "step": 3740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007862197629799233,
+      "loss": 0.6944,
+      "step": 3750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007861773622506307,
+      "loss": 0.5771,
+      "step": 3760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007861349615213382,
+      "loss": 0.5659,
+      "step": 3770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007860925607920457,
+      "loss": 0.5658,
+      "step": 3780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007860501600627531,
+      "loss": 0.7003,
+      "step": 3790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007860077593334605,
+      "loss": 0.5329,
+      "step": 3800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007859653586041681,
+      "loss": 0.5499,
+      "step": 3810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007859229578748755,
+      "loss": 0.6325,
+      "step": 3820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007858805571455829,
+      "loss": 0.6094,
+      "step": 3830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007858381564162904,
+      "loss": 0.7074,
+      "step": 3840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007857957556869978,
+      "loss": 0.6479,
+      "step": 3850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007857533549577053,
+      "loss": 0.6194,
+      "step": 3860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007857109542284128,
+      "loss": 0.5824,
+      "step": 3870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007856685534991202,
+      "loss": 0.7079,
+      "step": 3880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007856261527698277,
+      "loss": 0.5648,
+      "step": 3890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007855837520405352,
+      "loss": 0.7314,
+      "step": 3900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007855413513112426,
+      "loss": 0.6453,
+      "step": 3910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00078549895058195,
+      "loss": 0.555,
+      "step": 3920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007854565498526575,
+      "loss": 0.4925,
+      "step": 3930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000785414149123365,
+      "loss": 0.5157,
+      "step": 3940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007853717483940724,
+      "loss": 0.5987,
+      "step": 3950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007853293476647799,
+      "loss": 0.7067,
+      "step": 3960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007852869469354873,
+      "loss": 0.436,
+      "step": 3970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007852445462061948,
+      "loss": 0.643,
+      "step": 3980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007852021454769023,
+      "loss": 0.6058,
+      "step": 3990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007851597447476097,
+      "loss": 0.6332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007851173440183171,
+      "loss": 0.5234,
+      "step": 4010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007850749432890247,
+      "loss": 0.4634,
+      "step": 4020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007850325425597321,
+      "loss": 0.5965,
+      "step": 4030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007849901418304395,
+      "loss": 0.6243,
+      "step": 4040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000784947741101147,
+      "loss": 0.6651,
+      "step": 4050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007849053403718544,
+      "loss": 0.6369,
+      "step": 4060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007848629396425619,
+      "loss": 0.5679,
+      "step": 4070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007848205389132694,
+      "loss": 0.6032,
+      "step": 4080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007847781381839767,
+      "loss": 0.5698,
+      "step": 4090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007847357374546843,
+      "loss": 0.6337,
+      "step": 4100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007846933367253918,
+      "loss": 0.6551,
+      "step": 4110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007846509359960991,
+      "loss": 0.573,
+      "step": 4120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007846085352668066,
+      "loss": 0.6287,
+      "step": 4130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007845661345375142,
+      "loss": 0.6004,
+      "step": 4140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007845237338082215,
+      "loss": 0.5903,
+      "step": 4150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000784481333078929,
+      "loss": 0.5827,
+      "step": 4160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007844389323496364,
+      "loss": 0.5596,
+      "step": 4170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007843965316203439,
+      "loss": 0.5913,
+      "step": 4180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007843541308910514,
+      "loss": 0.5476,
+      "step": 4190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007843117301617588,
+      "loss": 0.5515,
+      "step": 4200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007842693294324662,
+      "loss": 0.5908,
+      "step": 4210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007842269287031738,
+      "loss": 0.5775,
+      "step": 4220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007841845279738812,
+      "loss": 0.6579,
+      "step": 4230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007841421272445886,
+      "loss": 0.6182,
+      "step": 4240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007840997265152961,
+      "loss": 0.6763,
+      "step": 4250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007840573257860036,
+      "loss": 0.5892,
+      "step": 4260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000784014925056711,
+      "loss": 0.7075,
+      "step": 4270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007839725243274185,
+      "loss": 0.5989,
+      "step": 4280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007839301235981259,
+      "loss": 0.7097,
+      "step": 4290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007838877228688334,
+      "loss": 0.5612,
+      "step": 4300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007838453221395409,
+      "loss": 0.6322,
+      "step": 4310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007838029214102483,
+      "loss": 0.5991,
+      "step": 4320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007837605206809557,
+      "loss": 0.5368,
+      "step": 4330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007837181199516633,
+      "loss": 0.6674,
+      "step": 4340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007836757192223706,
+      "loss": 0.5854,
+      "step": 4350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007836333184930781,
+      "loss": 0.644,
+      "step": 4360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007835909177637856,
+      "loss": 0.6525,
+      "step": 4370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000783548517034493,
+      "loss": 0.5781,
+      "step": 4380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007835061163052005,
+      "loss": 0.6052,
+      "step": 4390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000783463715575908,
+      "loss": 0.5652,
+      "step": 4400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007834213148466153,
+      "loss": 0.6408,
+      "step": 4410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007833789141173229,
+      "loss": 0.6128,
+      "step": 4420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007833365133880304,
+      "loss": 0.5614,
+      "step": 4430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007832941126587377,
+      "loss": 0.5681,
+      "step": 4440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007832517119294452,
+      "loss": 0.6266,
+      "step": 4450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007832093112001527,
+      "loss": 0.6191,
+      "step": 4460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007831669104708601,
+      "loss": 0.5644,
+      "step": 4470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007831245097415676,
+      "loss": 0.6388,
+      "step": 4480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007830821090122751,
+      "loss": 0.574,
+      "step": 4490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007830397082829825,
+      "loss": 0.616,
+      "step": 4500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00078299730755369,
+      "loss": 0.5824,
+      "step": 4510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007829549068243975,
+      "loss": 0.567,
+      "step": 4520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007829125060951048,
+      "loss": 0.5877,
+      "step": 4530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007828701053658123,
+      "loss": 0.6796,
+      "step": 4540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007828277046365199,
+      "loss": 0.6504,
+      "step": 4550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007827853039072272,
+      "loss": 0.5854,
+      "step": 4560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007827429031779347,
+      "loss": 0.6635,
+      "step": 4570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007827005024486422,
+      "loss": 0.6903,
+      "step": 4580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007826581017193496,
+      "loss": 0.7791,
+      "step": 4590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007826157009900571,
+      "loss": 0.6974,
+      "step": 4600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007825733002607646,
+      "loss": 0.5247,
+      "step": 4610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007825308995314719,
+      "loss": 0.7485,
+      "step": 4620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007824884988021795,
+      "loss": 0.6153,
+      "step": 4630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000782446098072887,
+      "loss": 0.5295,
+      "step": 4640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007824036973435943,
+      "loss": 0.5912,
+      "step": 4650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007823612966143018,
+      "loss": 0.6401,
+      "step": 4660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007823188958850092,
+      "loss": 0.6331,
+      "step": 4670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007822764951557167,
+      "loss": 0.6078,
+      "step": 4680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007822340944264242,
+      "loss": 0.6459,
+      "step": 4690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007821916936971315,
+      "loss": 0.6144,
+      "step": 4700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007821492929678391,
+      "loss": 0.6134,
+      "step": 4710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007821068922385466,
+      "loss": 0.5599,
+      "step": 4720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007820644915092539,
+      "loss": 0.5863,
+      "step": 4730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007820220907799614,
+      "loss": 0.5725,
+      "step": 4740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000781979690050669,
+      "loss": 0.5583,
+      "step": 4750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007819372893213763,
+      "loss": 0.7017,
+      "step": 4760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007818948885920838,
+      "loss": 0.4631,
+      "step": 4770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007818524878627913,
+      "loss": 0.6202,
+      "step": 4780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007818100871334987,
+      "loss": 0.6626,
+      "step": 4790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007817676864042062,
+      "loss": 0.595,
+      "step": 4800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007817252856749137,
+      "loss": 0.5791,
+      "step": 4810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000781682884945621,
+      "loss": 0.5932,
+      "step": 4820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007816404842163286,
+      "loss": 0.5971,
+      "step": 4830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007815980834870361,
+      "loss": 0.6643,
+      "step": 4840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007815556827577434,
+      "loss": 0.6145,
+      "step": 4850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007815132820284509,
+      "loss": 0.6516,
+      "step": 4860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007814708812991585,
+      "loss": 0.6677,
+      "step": 4870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007814284805698658,
+      "loss": 0.6253,
+      "step": 4880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007813860798405733,
+      "loss": 0.6701,
+      "step": 4890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007813436791112808,
+      "loss": 0.638,
+      "step": 4900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007813012783819882,
+      "loss": 0.5344,
+      "step": 4910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007812588776526957,
+      "loss": 0.6352,
+      "step": 4920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007812164769234032,
+      "loss": 0.6374,
+      "step": 4930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007811740761941105,
+      "loss": 0.563,
+      "step": 4940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007811316754648181,
+      "loss": 0.6436,
+      "step": 4950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007810892747355255,
+      "loss": 0.5727,
+      "step": 4960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007810468740062329,
+      "loss": 0.6346,
+      "step": 4970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007810044732769404,
+      "loss": 0.5473,
+      "step": 4980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007809620725476479,
+      "loss": 0.5893,
+      "step": 4990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007809196718183553,
+      "loss": 0.5588,
+      "step": 5000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007808772710890628,
+      "loss": 0.5835,
+      "step": 5010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007808348703597701,
+      "loss": 0.6042,
+      "step": 5020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007807924696304777,
+      "loss": 0.5172,
+      "step": 5030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007807500689011852,
+      "loss": 0.6127,
+      "step": 5040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007807076681718925,
+      "loss": 0.6139,
+      "step": 5050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007806652674426,
+      "loss": 0.6042,
+      "step": 5060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007806228667133075,
+      "loss": 0.5899,
+      "step": 5070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000780580465984015,
+      "loss": 0.5925,
+      "step": 5080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007805380652547224,
+      "loss": 0.6487,
+      "step": 5090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007804956645254299,
+      "loss": 0.6454,
+      "step": 5100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007804532637961373,
+      "loss": 0.4416,
+      "step": 5110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007804108630668448,
+      "loss": 0.5783,
+      "step": 5120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007803684623375523,
+      "loss": 0.6036,
+      "step": 5130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007803260616082596,
+      "loss": 0.6761,
+      "step": 5140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007802836608789671,
+      "loss": 0.552,
+      "step": 5150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007802412601496747,
+      "loss": 0.4905,
+      "step": 5160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000780198859420382,
+      "loss": 0.5744,
+      "step": 5170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007801564586910895,
+      "loss": 0.638,
+      "step": 5180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000780114057961797,
+      "loss": 0.6112,
+      "step": 5190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007800716572325044,
+      "loss": 0.5127,
+      "step": 5200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007800292565032119,
+      "loss": 0.6041,
+      "step": 5210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007799868557739194,
+      "loss": 0.7574,
+      "step": 5220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007799444550446267,
+      "loss": 0.6615,
+      "step": 5230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007799020543153343,
+      "loss": 0.6202,
+      "step": 5240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007798596535860418,
+      "loss": 0.5936,
+      "step": 5250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007798172528567491,
+      "loss": 0.535,
+      "step": 5260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007797748521274566,
+      "loss": 0.695,
+      "step": 5270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007797324513981641,
+      "loss": 0.6237,
+      "step": 5280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007796900506688715,
+      "loss": 0.673,
+      "step": 5290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000779647649939579,
+      "loss": 0.5976,
+      "step": 5300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007796052492102865,
+      "loss": 0.6039,
+      "step": 5310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007795628484809939,
+      "loss": 0.5627,
+      "step": 5320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007795204477517014,
+      "loss": 0.5844,
+      "step": 5330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007794780470224088,
+      "loss": 0.7201,
+      "step": 5340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007794356462931162,
+      "loss": 0.5847,
+      "step": 5350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007793932455638238,
+      "loss": 0.6687,
+      "step": 5360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007793508448345312,
+      "loss": 0.635,
+      "step": 5370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007793084441052386,
+      "loss": 0.4912,
+      "step": 5380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007792660433759461,
+      "loss": 0.6875,
+      "step": 5390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007792236426466536,
+      "loss": 0.5787,
+      "step": 5400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000779181241917361,
+      "loss": 0.578,
+      "step": 5410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007791388411880685,
+      "loss": 0.5157,
+      "step": 5420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007790964404587759,
+      "loss": 0.5754,
+      "step": 5430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007790540397294834,
+      "loss": 0.6248,
+      "step": 5440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007790116390001909,
+      "loss": 0.6231,
+      "step": 5450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007789692382708983,
+      "loss": 0.6599,
+      "step": 5460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007789268375416057,
+      "loss": 0.601,
+      "step": 5470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007788844368123133,
+      "loss": 0.542,
+      "step": 5480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007788420360830207,
+      "loss": 0.5781,
+      "step": 5490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007787996353537281,
+      "loss": 0.5232,
+      "step": 5500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007787572346244356,
+      "loss": 0.6865,
+      "step": 5510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000778714833895143,
+      "loss": 0.664,
+      "step": 5520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007786724331658505,
+      "loss": 0.6189,
+      "step": 5530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000778630032436558,
+      "loss": 0.6569,
+      "step": 5540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007785876317072653,
+      "loss": 0.5482,
+      "step": 5550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007785452309779729,
+      "loss": 0.6417,
+      "step": 5560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007785028302486803,
+      "loss": 0.6044,
+      "step": 5570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007784604295193877,
+      "loss": 0.5786,
+      "step": 5580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007784180287900952,
+      "loss": 0.6359,
+      "step": 5590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007783756280608027,
+      "loss": 0.5617,
+      "step": 5600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007783332273315101,
+      "loss": 0.7343,
+      "step": 5610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007782908266022176,
+      "loss": 0.5628,
+      "step": 5620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000778248425872925,
+      "loss": 0.5783,
+      "step": 5630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007782060251436325,
+      "loss": 0.6466,
+      "step": 5640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00077816362441434,
+      "loss": 0.4999,
+      "step": 5650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007781212236850474,
+      "loss": 0.6204,
+      "step": 5660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007780788229557548,
+      "loss": 0.5278,
+      "step": 5670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007780364222264623,
+      "loss": 0.5897,
+      "step": 5680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007779940214971698,
+      "loss": 0.5271,
+      "step": 5690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007779516207678772,
+      "loss": 0.6375,
+      "step": 5700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007779092200385847,
+      "loss": 0.658,
+      "step": 5710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007778668193092922,
+      "loss": 0.5876,
+      "step": 5720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007778244185799996,
+      "loss": 0.6133,
+      "step": 5730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007777820178507071,
+      "loss": 0.5796,
+      "step": 5740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007777396171214145,
+      "loss": 0.5593,
+      "step": 5750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007776972163921219,
+      "loss": 0.5959,
+      "step": 5760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007776548156628295,
+      "loss": 0.5671,
+      "step": 5770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007776124149335369,
+      "loss": 0.5223,
+      "step": 5780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007775700142042443,
+      "loss": 0.7194,
+      "step": 5790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007775276134749518,
+      "loss": 0.6266,
+      "step": 5800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007774852127456593,
+      "loss": 0.5637,
+      "step": 5810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007774428120163667,
+      "loss": 0.5577,
+      "step": 5820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007774004112870742,
+      "loss": 0.5734,
+      "step": 5830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007773580105577817,
+      "loss": 0.57,
+      "step": 5840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007773156098284891,
+      "loss": 0.6176,
+      "step": 5850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007772732090991966,
+      "loss": 0.6607,
+      "step": 5860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000777230808369904,
+      "loss": 0.5944,
+      "step": 5870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007771884076406114,
+      "loss": 0.7022,
+      "step": 5880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007771460069113189,
+      "loss": 0.6541,
+      "step": 5890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007771036061820264,
+      "loss": 0.5973,
+      "step": 5900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007770612054527338,
+      "loss": 0.5962,
+      "step": 5910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007770188047234413,
+      "loss": 0.6234,
+      "step": 5920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007769764039941488,
+      "loss": 0.6357,
+      "step": 5930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007769340032648562,
+      "loss": 0.5536,
+      "step": 5940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007768916025355636,
+      "loss": 0.6613,
+      "step": 5950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000776849201806271,
+      "loss": 0.6475,
+      "step": 5960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007768068010769786,
+      "loss": 0.596,
+      "step": 5970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000776764400347686,
+      "loss": 0.5162,
+      "step": 5980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007767219996183935,
+      "loss": 0.6393,
+      "step": 5990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007766795988891009,
+      "loss": 0.5547,
+      "step": 6000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007766371981598084,
+      "loss": 0.6589,
+      "step": 6010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007765947974305159,
+      "loss": 0.5306,
+      "step": 6020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007765523967012233,
+      "loss": 0.6374,
+      "step": 6030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007765099959719307,
+      "loss": 0.5988,
+      "step": 6040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007764675952426383,
+      "loss": 0.5984,
+      "step": 6050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007764251945133457,
+      "loss": 0.6709,
+      "step": 6060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007763827937840531,
+      "loss": 0.5298,
+      "step": 6070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007763403930547605,
+      "loss": 0.6266,
+      "step": 6080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007762979923254681,
+      "loss": 0.5356,
+      "step": 6090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007762555915961755,
+      "loss": 0.6789,
+      "step": 6100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007762131908668829,
+      "loss": 0.5834,
+      "step": 6110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007761707901375904,
+      "loss": 0.7174,
+      "step": 6120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007761283894082979,
+      "loss": 0.6057,
+      "step": 6130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007760859886790053,
+      "loss": 0.5919,
+      "step": 6140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007760435879497128,
+      "loss": 0.547,
+      "step": 6150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007760011872204202,
+      "loss": 0.6068,
+      "step": 6160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007759587864911277,
+      "loss": 0.593,
+      "step": 6170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007759163857618352,
+      "loss": 0.6016,
+      "step": 6180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007758739850325426,
+      "loss": 0.5751,
+      "step": 6190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00077583158430325,
+      "loss": 0.6359,
+      "step": 6200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007757891835739575,
+      "loss": 0.5639,
+      "step": 6210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000775746782844665,
+      "loss": 0.5254,
+      "step": 6220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007757043821153724,
+      "loss": 0.6913,
+      "step": 6230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007756619813860798,
+      "loss": 0.5439,
+      "step": 6240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007756195806567874,
+      "loss": 0.6438,
+      "step": 6250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007755771799274948,
+      "loss": 0.6363,
+      "step": 6260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007755347791982022,
+      "loss": 0.633,
+      "step": 6270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007754923784689097,
+      "loss": 0.5649,
+      "step": 6280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007754499777396172,
+      "loss": 0.6257,
+      "step": 6290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007754075770103246,
+      "loss": 0.5736,
+      "step": 6300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007753651762810321,
+      "loss": 0.5462,
+      "step": 6310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007753227755517395,
+      "loss": 0.6175,
+      "step": 6320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000775280374822447,
+      "loss": 0.6101,
+      "step": 6330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007752379740931545,
+      "loss": 0.5474,
+      "step": 6340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007751955733638619,
+      "loss": 0.7151,
+      "step": 6350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007751531726345693,
+      "loss": 0.6211,
+      "step": 6360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007751107719052769,
+      "loss": 0.574,
+      "step": 6370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007750683711759843,
+      "loss": 0.6345,
+      "step": 6380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007750259704466917,
+      "loss": 0.6718,
+      "step": 6390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007749835697173992,
+      "loss": 0.5703,
+      "step": 6400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007749411689881066,
+      "loss": 0.7077,
+      "step": 6410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007748987682588141,
+      "loss": 0.5817,
+      "step": 6420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007748563675295216,
+      "loss": 0.6521,
+      "step": 6430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000774813966800229,
+      "loss": 0.6118,
+      "step": 6440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007747715660709365,
+      "loss": 0.5046,
+      "step": 6450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000774729165341644,
+      "loss": 0.6397,
+      "step": 6460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007746867646123514,
+      "loss": 0.6057,
+      "step": 6470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007746443638830588,
+      "loss": 0.5903,
+      "step": 6480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007746019631537662,
+      "loss": 0.5182,
+      "step": 6490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007745595624244737,
+      "loss": 0.6889,
+      "step": 6500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007745171616951812,
+      "loss": 0.5525,
+      "step": 6510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007744747609658886,
+      "loss": 0.5793,
+      "step": 6520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007744323602365961,
+      "loss": 0.4952,
+      "step": 6530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007743899595073036,
+      "loss": 0.5174,
+      "step": 6540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000774347558778011,
+      "loss": 0.5308,
+      "step": 6550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007743051580487184,
+      "loss": 0.5757,
+      "step": 6560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007742627573194259,
+      "loss": 0.6377,
+      "step": 6570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007742203565901334,
+      "loss": 0.5611,
+      "step": 6580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007741779558608408,
+      "loss": 0.5419,
+      "step": 6590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007741355551315483,
+      "loss": 0.593,
+      "step": 6600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007740931544022557,
+      "loss": 0.6184,
+      "step": 6610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007740507536729632,
+      "loss": 0.6429,
+      "step": 6620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007740083529436707,
+      "loss": 0.5284,
+      "step": 6630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007739659522143781,
+      "loss": 0.6045,
+      "step": 6640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007739235514850855,
+      "loss": 0.601,
+      "step": 6650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007738811507557931,
+      "loss": 0.6243,
+      "step": 6660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007738387500265005,
+      "loss": 0.6177,
+      "step": 6670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007737963492972079,
+      "loss": 0.5173,
+      "step": 6680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007737539485679154,
+      "loss": 0.6091,
+      "step": 6690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007737115478386229,
+      "loss": 0.6482,
+      "step": 6700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007736691471093303,
+      "loss": 0.7593,
+      "step": 6710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007736267463800378,
+      "loss": 0.6338,
+      "step": 6720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007735843456507452,
+      "loss": 0.6173,
+      "step": 6730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007735419449214527,
+      "loss": 0.5905,
+      "step": 6740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007734995441921602,
+      "loss": 0.5481,
+      "step": 6750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007734571434628676,
+      "loss": 0.589,
+      "step": 6760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000773414742733575,
+      "loss": 0.5981,
+      "step": 6770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007733723420042826,
+      "loss": 0.6859,
+      "step": 6780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00077332994127499,
+      "loss": 0.5649,
+      "step": 6790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007732875405456974,
+      "loss": 0.6478,
+      "step": 6800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007732451398164049,
+      "loss": 0.5209,
+      "step": 6810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007732027390871123,
+      "loss": 0.6234,
+      "step": 6820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007731603383578198,
+      "loss": 0.6824,
+      "step": 6830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007731179376285273,
+      "loss": 0.5669,
+      "step": 6840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007730755368992346,
+      "loss": 0.6288,
+      "step": 6850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007730331361699422,
+      "loss": 0.5869,
+      "step": 6860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007729907354406497,
+      "loss": 0.5,
+      "step": 6870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000772948334711357,
+      "loss": 0.6136,
+      "step": 6880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007729059339820645,
+      "loss": 0.6823,
+      "step": 6890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007728635332527721,
+      "loss": 0.5519,
+      "step": 6900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007728211325234794,
+      "loss": 0.5885,
+      "step": 6910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007727787317941869,
+      "loss": 0.6006,
+      "step": 6920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007727363310648944,
+      "loss": 0.7368,
+      "step": 6930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007726939303356018,
+      "loss": 0.5831,
+      "step": 6940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007726515296063093,
+      "loss": 0.562,
+      "step": 6950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007726091288770168,
+      "loss": 0.6877,
+      "step": 6960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007725667281477241,
+      "loss": 0.6029,
+      "step": 6970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007725243274184317,
+      "loss": 0.56,
+      "step": 6980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007724819266891392,
+      "loss": 0.6132,
+      "step": 6990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007724395259598465,
+      "loss": 0.5266,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000772397125230554,
+      "loss": 0.6888,
+      "step": 7010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007723547245012614,
+      "loss": 0.645,
+      "step": 7020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007723123237719689,
+      "loss": 0.6033,
+      "step": 7030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007722699230426764,
+      "loss": 0.5054,
+      "step": 7040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007722275223133838,
+      "loss": 0.5736,
+      "step": 7050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007721851215840913,
+      "loss": 0.6939,
+      "step": 7060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007721427208547988,
+      "loss": 0.6571,
+      "step": 7070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007721003201255062,
+      "loss": 0.6274,
+      "step": 7080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007720579193962136,
+      "loss": 0.5362,
+      "step": 7090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007720155186669211,
+      "loss": 0.519,
+      "step": 7100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007719731179376285,
+      "loss": 0.6264,
+      "step": 7110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000771930717208336,
+      "loss": 0.6095,
+      "step": 7120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007718883164790435,
+      "loss": 0.5955,
+      "step": 7130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007718459157497509,
+      "loss": 0.6268,
+      "step": 7140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007718035150204584,
+      "loss": 0.5433,
+      "step": 7150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007717611142911659,
+      "loss": 0.691,
+      "step": 7160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007717187135618732,
+      "loss": 0.5381,
+      "step": 7170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007716763128325807,
+      "loss": 0.626,
+      "step": 7180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007716339121032883,
+      "loss": 0.6366,
+      "step": 7190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007715915113739956,
+      "loss": 0.636,
+      "step": 7200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007715491106447031,
+      "loss": 0.564,
+      "step": 7210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007715067099154106,
+      "loss": 0.7273,
+      "step": 7220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000771464309186118,
+      "loss": 0.7586,
+      "step": 7230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007714219084568255,
+      "loss": 0.6252,
+      "step": 7240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000771379507727533,
+      "loss": 0.6642,
+      "step": 7250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007713371069982403,
+      "loss": 0.5657,
+      "step": 7260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007712947062689479,
+      "loss": 0.5254,
+      "step": 7270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007712523055396554,
+      "loss": 0.5148,
+      "step": 7280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007712099048103627,
+      "loss": 0.689,
+      "step": 7290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007711675040810702,
+      "loss": 0.5358,
+      "step": 7300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007711251033517778,
+      "loss": 0.4977,
+      "step": 7310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007710827026224851,
+      "loss": 0.5207,
+      "step": 7320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007710403018931926,
+      "loss": 0.6051,
+      "step": 7330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007709979011639001,
+      "loss": 0.6496,
+      "step": 7340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007709555004346075,
+      "loss": 0.809,
+      "step": 7350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000770913099705315,
+      "loss": 0.6206,
+      "step": 7360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007708706989760225,
+      "loss": 0.5263,
+      "step": 7370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007708282982467298,
+      "loss": 0.6258,
+      "step": 7380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007707858975174374,
+      "loss": 0.605,
+      "step": 7390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007707434967881449,
+      "loss": 0.6053,
+      "step": 7400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007707010960588522,
+      "loss": 0.7063,
+      "step": 7410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007706586953295597,
+      "loss": 0.6702,
+      "step": 7420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007706162946002672,
+      "loss": 0.6451,
+      "step": 7430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007705738938709746,
+      "loss": 0.6194,
+      "step": 7440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007705314931416821,
+      "loss": 0.4984,
+      "step": 7450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007704890924123894,
+      "loss": 0.7647,
+      "step": 7460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000770446691683097,
+      "loss": 0.6545,
+      "step": 7470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007704042909538045,
+      "loss": 0.6128,
+      "step": 7480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007703618902245118,
+      "loss": 0.5173,
+      "step": 7490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007703194894952193,
+      "loss": 0.7232,
+      "step": 7500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007702770887659269,
+      "loss": 0.573,
+      "step": 7510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007702346880366342,
+      "loss": 0.6052,
+      "step": 7520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007701922873073417,
+      "loss": 0.6709,
+      "step": 7530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007701498865780492,
+      "loss": 0.6687,
+      "step": 7540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007701074858487566,
+      "loss": 0.5686,
+      "step": 7550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007700650851194641,
+      "loss": 0.5664,
+      "step": 7560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007700226843901716,
+      "loss": 0.6329,
+      "step": 7570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007699802836608789,
+      "loss": 0.589,
+      "step": 7580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007699378829315865,
+      "loss": 0.6654,
+      "step": 7590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000769895482202294,
+      "loss": 0.56,
+      "step": 7600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007698530814730013,
+      "loss": 0.6502,
+      "step": 7610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007698106807437088,
+      "loss": 0.6265,
+      "step": 7620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007697682800144163,
+      "loss": 0.565,
+      "step": 7630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007697258792851237,
+      "loss": 0.5124,
+      "step": 7640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007696834785558312,
+      "loss": 0.529,
+      "step": 7650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007696410778265387,
+      "loss": 0.6316,
+      "step": 7660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007695986770972461,
+      "loss": 0.5952,
+      "step": 7670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007695562763679536,
+      "loss": 0.6284,
+      "step": 7680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007695138756386611,
+      "loss": 0.7009,
+      "step": 7690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007694714749093684,
+      "loss": 0.5168,
+      "step": 7700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007694290741800759,
+      "loss": 0.6853,
+      "step": 7710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007693866734507834,
+      "loss": 0.5777,
+      "step": 7720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007693442727214908,
+      "loss": 0.624,
+      "step": 7730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007693018719921983,
+      "loss": 0.5905,
+      "step": 7740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007692594712629058,
+      "loss": 0.5593,
+      "step": 7750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007692170705336132,
+      "loss": 0.6197,
+      "step": 7760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007691746698043207,
+      "loss": 0.5957,
+      "step": 7770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007691322690750281,
+      "loss": 0.5709,
+      "step": 7780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007690898683457356,
+      "loss": 0.6195,
+      "step": 7790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007690474676164431,
+      "loss": 0.621,
+      "step": 7800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007690050668871505,
+      "loss": 0.5229,
+      "step": 7810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007689626661578579,
+      "loss": 0.5913,
+      "step": 7820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007689202654285654,
+      "loss": 0.5522,
+      "step": 7830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007688778646992729,
+      "loss": 0.6063,
+      "step": 7840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007688354639699803,
+      "loss": 0.6173,
+      "step": 7850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007687930632406878,
+      "loss": 0.5803,
+      "step": 7860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007687506625113953,
+      "loss": 0.6865,
+      "step": 7870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007687082617821027,
+      "loss": 0.5696,
+      "step": 7880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007686658610528102,
+      "loss": 0.5863,
+      "step": 7890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007686234603235175,
+      "loss": 0.7458,
+      "step": 7900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000768581059594225,
+      "loss": 0.5145,
+      "step": 7910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007685386588649326,
+      "loss": 0.7213,
+      "step": 7920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00076849625813564,
+      "loss": 0.638,
+      "step": 7930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007684538574063474,
+      "loss": 0.659,
+      "step": 7940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007684114566770549,
+      "loss": 0.6848,
+      "step": 7950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007683690559477623,
+      "loss": 0.6549,
+      "step": 7960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007683266552184698,
+      "loss": 0.5442,
+      "step": 7970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007682842544891773,
+      "loss": 0.5427,
+      "step": 7980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007682418537598846,
+      "loss": 0.6002,
+      "step": 7990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007681994530305922,
+      "loss": 0.6659,
+      "step": 8000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007681570523012997,
+      "loss": 0.6429,
+      "step": 8010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000768114651572007,
+      "loss": 0.6183,
+      "step": 8020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007680722508427145,
+      "loss": 0.5832,
+      "step": 8030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000768029850113422,
+      "loss": 0.6756,
+      "step": 8040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007679874493841294,
+      "loss": 0.6405,
+      "step": 8050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007679450486548369,
+      "loss": 0.6598,
+      "step": 8060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007679026479255443,
+      "loss": 0.6456,
+      "step": 8070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007678602471962518,
+      "loss": 0.5152,
+      "step": 8080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007678178464669593,
+      "loss": 0.6053,
+      "step": 8090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007677754457376667,
+      "loss": 0.5916,
+      "step": 8100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007677330450083741,
+      "loss": 0.542,
+      "step": 8110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007676906442790817,
+      "loss": 0.5207,
+      "step": 8120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007676482435497891,
+      "loss": 0.6265,
+      "step": 8130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007676058428204965,
+      "loss": 0.5757,
+      "step": 8140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000767563442091204,
+      "loss": 0.6004,
+      "step": 8150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007675210413619115,
+      "loss": 0.5795,
+      "step": 8160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007674786406326189,
+      "loss": 0.5835,
+      "step": 8170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007674362399033264,
+      "loss": 0.5072,
+      "step": 8180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007673938391740338,
+      "loss": 0.6108,
+      "step": 8190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007673514384447413,
+      "loss": 0.6604,
+      "step": 8200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007673090377154488,
+      "loss": 0.6534,
+      "step": 8210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007672666369861562,
+      "loss": 0.6648,
+      "step": 8220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007672242362568636,
+      "loss": 0.605,
+      "step": 8230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007671818355275711,
+      "loss": 0.6421,
+      "step": 8240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007671394347982786,
+      "loss": 0.6396,
+      "step": 8250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000767097034068986,
+      "loss": 0.7274,
+      "step": 8260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007670546333396935,
+      "loss": 0.7185,
+      "step": 8270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000767012232610401,
+      "loss": 0.5677,
+      "step": 8280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007669698318811084,
+      "loss": 0.5207,
+      "step": 8290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007669274311518159,
+      "loss": 0.5518,
+      "step": 8300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007668850304225233,
+      "loss": 0.6022,
+      "step": 8310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007668426296932308,
+      "loss": 0.5682,
+      "step": 8320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007668002289639383,
+      "loss": 0.5575,
+      "step": 8330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007667578282346457,
+      "loss": 0.541,
+      "step": 8340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007667154275053531,
+      "loss": 0.726,
+      "step": 8350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007666730267760606,
+      "loss": 0.5154,
+      "step": 8360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000766630626046768,
+      "loss": 0.512,
+      "step": 8370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007665882253174755,
+      "loss": 0.7272,
+      "step": 8380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007665458245881829,
+      "loss": 0.654,
+      "step": 8390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007665034238588905,
+      "loss": 0.6862,
+      "step": 8400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007664610231295979,
+      "loss": 0.6486,
+      "step": 8410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007664186224003053,
+      "loss": 0.6841,
+      "step": 8420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007663762216710127,
+      "loss": 0.6623,
+      "step": 8430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007663338209417202,
+      "loss": 0.7007,
+      "step": 8440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007662914202124277,
+      "loss": 0.6529,
+      "step": 8450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007662490194831351,
+      "loss": 0.5428,
+      "step": 8460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007662066187538426,
+      "loss": 0.5843,
+      "step": 8470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007661642180245501,
+      "loss": 0.5791,
+      "step": 8480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007661218172952575,
+      "loss": 0.655,
+      "step": 8490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000766079416565965,
+      "loss": 0.6185,
+      "step": 8500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007660370158366724,
+      "loss": 0.5906,
+      "step": 8510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007659946151073798,
+      "loss": 0.5526,
+      "step": 8520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007659522143780874,
+      "loss": 0.564,
+      "step": 8530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007659098136487948,
+      "loss": 0.6993,
+      "step": 8540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007658674129195022,
+      "loss": 0.5585,
+      "step": 8550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007658250121902097,
+      "loss": 0.5103,
+      "step": 8560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007657826114609172,
+      "loss": 0.5449,
+      "step": 8570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007657402107316246,
+      "loss": 0.4888,
+      "step": 8580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007656978100023321,
+      "loss": 0.6759,
+      "step": 8590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007656554092730395,
+      "loss": 0.572,
+      "step": 8600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000765613008543747,
+      "loss": 0.6625,
+      "step": 8610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007655706078144545,
+      "loss": 0.6337,
+      "step": 8620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007655282070851619,
+      "loss": 0.653,
+      "step": 8630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007654858063558693,
+      "loss": 0.6106,
+      "step": 8640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007654434056265768,
+      "loss": 0.6629,
+      "step": 8650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007654010048972843,
+      "loss": 0.6565,
+      "step": 8660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007653586041679917,
+      "loss": 0.5824,
+      "step": 8670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007653162034386991,
+      "loss": 0.5425,
+      "step": 8680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007652738027094067,
+      "loss": 0.6052,
+      "step": 8690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007652314019801141,
+      "loss": 0.5812,
+      "step": 8700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007651890012508215,
+      "loss": 0.6381,
+      "step": 8710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000765146600521529,
+      "loss": 0.6073,
+      "step": 8720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007651041997922365,
+      "loss": 0.5368,
+      "step": 8730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007650617990629439,
+      "loss": 0.7608,
+      "step": 8740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007650193983336514,
+      "loss": 0.4941,
+      "step": 8750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007649769976043588,
+      "loss": 0.632,
+      "step": 8760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007649345968750663,
+      "loss": 0.7651,
+      "step": 8770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007648921961457738,
+      "loss": 0.5062,
+      "step": 8780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007648497954164812,
+      "loss": 0.5611,
+      "step": 8790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007648073946871886,
+      "loss": 0.627,
+      "step": 8800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007647649939578962,
+      "loss": 0.6366,
+      "step": 8810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007647225932286036,
+      "loss": 0.6125,
+      "step": 8820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000764680192499311,
+      "loss": 0.6978,
+      "step": 8830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007646377917700184,
+      "loss": 0.5558,
+      "step": 8840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000764595391040726,
+      "loss": 0.64,
+      "step": 8850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007645529903114334,
+      "loss": 0.6389,
+      "step": 8860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007645105895821409,
+      "loss": 0.6405,
+      "step": 8870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007644681888528483,
+      "loss": 0.6921,
+      "step": 8880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007644257881235558,
+      "loss": 0.6883,
+      "step": 8890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007643833873942633,
+      "loss": 0.6687,
+      "step": 8900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007643409866649707,
+      "loss": 0.6088,
+      "step": 8910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007642985859356781,
+      "loss": 0.6154,
+      "step": 8920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007642561852063857,
+      "loss": 0.6552,
+      "step": 8930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007642137844770931,
+      "loss": 0.5808,
+      "step": 8940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007641713837478005,
+      "loss": 0.5133,
+      "step": 8950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007641289830185079,
+      "loss": 0.5386,
+      "step": 8960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007640865822892154,
+      "loss": 0.6349,
+      "step": 8970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007640441815599229,
+      "loss": 0.5621,
+      "step": 8980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007640017808306303,
+      "loss": 0.6052,
+      "step": 8990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007639593801013377,
+      "loss": 0.6937,
+      "step": 9000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007639169793720453,
+      "loss": 0.6184,
+      "step": 9010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007638745786427527,
+      "loss": 0.7246,
+      "step": 9020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007638321779134601,
+      "loss": 0.6943,
+      "step": 9030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007637897771841676,
+      "loss": 0.4638,
+      "step": 9040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000763747376454875,
+      "loss": 0.5698,
+      "step": 9050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007637049757255825,
+      "loss": 0.6173,
+      "step": 9060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00076366257499629,
+      "loss": 0.6143,
+      "step": 9070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007636201742669974,
+      "loss": 0.591,
+      "step": 9080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007635777735377049,
+      "loss": 0.5914,
+      "step": 9090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007635353728084124,
+      "loss": 0.7357,
+      "step": 9100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007634929720791198,
+      "loss": 0.6463,
+      "step": 9110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007634505713498272,
+      "loss": 0.6637,
+      "step": 9120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007634081706205347,
+      "loss": 0.5815,
+      "step": 9130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007633657698912422,
+      "loss": 0.6009,
+      "step": 9140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007633233691619496,
+      "loss": 0.5554,
+      "step": 9150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007632809684326571,
+      "loss": 0.6193,
+      "step": 9160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007632385677033645,
+      "loss": 0.6058,
+      "step": 9170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000763196166974072,
+      "loss": 0.5501,
+      "step": 9180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007631537662447795,
+      "loss": 0.5933,
+      "step": 9190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007631113655154869,
+      "loss": 0.6403,
+      "step": 9200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007630689647861943,
+      "loss": 0.631,
+      "step": 9210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007630265640569019,
+      "loss": 0.6228,
+      "step": 9220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007629841633276093,
+      "loss": 0.5897,
+      "step": 9230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007629417625983167,
+      "loss": 0.5963,
+      "step": 9240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007628993618690242,
+      "loss": 0.6296,
+      "step": 9250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007628569611397316,
+      "loss": 0.6764,
+      "step": 9260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007628145604104391,
+      "loss": 0.6023,
+      "step": 9270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007627721596811466,
+      "loss": 0.7838,
+      "step": 9280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000762729758951854,
+      "loss": 0.6439,
+      "step": 9290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007626873582225615,
+      "loss": 0.5452,
+      "step": 9300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000762644957493269,
+      "loss": 0.7239,
+      "step": 9310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007626025567639763,
+      "loss": 0.6873,
+      "step": 9320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007625601560346838,
+      "loss": 0.608,
+      "step": 9330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007625177553053914,
+      "loss": 0.7264,
+      "step": 9340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007624753545760987,
+      "loss": 0.5253,
+      "step": 9350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007624329538468062,
+      "loss": 0.5611,
+      "step": 9360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007623905531175136,
+      "loss": 0.6189,
+      "step": 9370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007623481523882211,
+      "loss": 0.6604,
+      "step": 9380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007623057516589286,
+      "loss": 0.6627,
+      "step": 9390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000762263350929636,
+      "loss": 0.6491,
+      "step": 9400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007622209502003434,
+      "loss": 0.5831,
+      "step": 9410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000762178549471051,
+      "loss": 0.7038,
+      "step": 9420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007621361487417584,
+      "loss": 0.673,
+      "step": 9430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007620937480124658,
+      "loss": 0.5855,
+      "step": 9440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007620513472831733,
+      "loss": 0.632,
+      "step": 9450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007620089465538808,
+      "loss": 0.6441,
+      "step": 9460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007619665458245882,
+      "loss": 0.5871,
+      "step": 9470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007619241450952957,
+      "loss": 0.7076,
+      "step": 9480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007618817443660031,
+      "loss": 0.6391,
+      "step": 9490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007618393436367106,
+      "loss": 0.5737,
+      "step": 9500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007617969429074181,
+      "loss": 0.6174,
+      "step": 9510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007617545421781255,
+      "loss": 0.6152,
+      "step": 9520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007617121414488329,
+      "loss": 0.5467,
+      "step": 9530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007616697407195405,
+      "loss": 0.61,
+      "step": 9540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007616273399902479,
+      "loss": 0.5968,
+      "step": 9550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007615849392609553,
+      "loss": 0.6215,
+      "step": 9560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007615425385316628,
+      "loss": 0.5158,
+      "step": 9570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007615001378023702,
+      "loss": 0.6269,
+      "step": 9580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007614577370730777,
+      "loss": 0.6019,
+      "step": 9590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007614153363437852,
+      "loss": 0.5946,
+      "step": 9600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007613729356144925,
+      "loss": 0.6059,
+      "step": 9610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007613305348852001,
+      "loss": 0.6349,
+      "step": 9620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007612881341559076,
+      "loss": 0.6576,
+      "step": 9630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007612457334266149,
+      "loss": 0.6598,
+      "step": 9640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007612033326973224,
+      "loss": 0.6276,
+      "step": 9650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007611609319680299,
+      "loss": 0.5596,
+      "step": 9660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007611185312387373,
+      "loss": 0.5251,
+      "step": 9670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007610761305094448,
+      "loss": 0.6251,
+      "step": 9680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007610337297801523,
+      "loss": 0.5493,
+      "step": 9690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007609913290508597,
+      "loss": 0.6198,
+      "step": 9700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007609489283215672,
+      "loss": 0.5803,
+      "step": 9710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007609065275922747,
+      "loss": 0.6944,
+      "step": 9720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000760864126862982,
+      "loss": 0.6522,
+      "step": 9730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007608217261336895,
+      "loss": 0.7118,
+      "step": 9740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007607793254043971,
+      "loss": 0.561,
+      "step": 9750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007607369246751044,
+      "loss": 0.6794,
+      "step": 9760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007606945239458119,
+      "loss": 0.6302,
+      "step": 9770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007606521232165194,
+      "loss": 0.6144,
+      "step": 9780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007606097224872268,
+      "loss": 0.6316,
+      "step": 9790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007605673217579343,
+      "loss": 0.6788,
+      "step": 9800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007605249210286418,
+      "loss": 0.7701,
+      "step": 9810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007604825202993492,
+      "loss": 0.5893,
+      "step": 9820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007604401195700567,
+      "loss": 0.6793,
+      "step": 9830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007603977188407642,
+      "loss": 0.6783,
+      "step": 9840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007603553181114715,
+      "loss": 0.6744,
+      "step": 9850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000760312917382179,
+      "loss": 0.6276,
+      "step": 9860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007602705166528864,
+      "loss": 0.5408,
+      "step": 9870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007602281159235939,
+      "loss": 0.487,
+      "step": 9880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007601857151943014,
+      "loss": 0.675,
+      "step": 9890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007601433144650088,
+      "loss": 0.5671,
+      "step": 9900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007601009137357163,
+      "loss": 0.7367,
+      "step": 9910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007600585130064238,
+      "loss": 0.5942,
+      "step": 9920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007600161122771311,
+      "loss": 0.6035,
+      "step": 9930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007599737115478386,
+      "loss": 0.6635,
+      "step": 9940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007599313108185462,
+      "loss": 0.5992,
+      "step": 9950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007598889100892535,
+      "loss": 0.6101,
+      "step": 9960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000759846509359961,
+      "loss": 0.6378,
+      "step": 9970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007598041086306685,
+      "loss": 0.6128,
+      "step": 9980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007597617079013759,
+      "loss": 0.5432,
+      "step": 9990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007597193071720834,
+      "loss": 0.5998,
+      "step": 10000
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.6361307501792908,
+      "eval_runtime": 337.8435,
+      "eval_samples_per_second": 15.555,
+      "eval_steps_per_second": 3.889,
+      "step": 10000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007596769064427909,
+      "loss": 0.6544,
+      "step": 10010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007596345057134982,
+      "loss": 0.6739,
+      "step": 10020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007595921049842058,
+      "loss": 0.6349,
+      "step": 10030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007595497042549133,
+      "loss": 0.6995,
+      "step": 10040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007595073035256206,
+      "loss": 0.5983,
+      "step": 10050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007594649027963281,
+      "loss": 0.6614,
+      "step": 10060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007594225020670357,
+      "loss": 0.6029,
+      "step": 10070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000759380101337743,
+      "loss": 0.7124,
+      "step": 10080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007593377006084505,
+      "loss": 0.6148,
+      "step": 10090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000759295299879158,
+      "loss": 0.5817,
+      "step": 10100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007592528991498654,
+      "loss": 0.5396,
+      "step": 10110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007592104984205729,
+      "loss": 0.63,
+      "step": 10120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007591680976912804,
+      "loss": 0.6856,
+      "step": 10130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007591256969619877,
+      "loss": 0.5574,
+      "step": 10140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007590832962326953,
+      "loss": 0.621,
+      "step": 10150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007590408955034028,
+      "loss": 0.6546,
+      "step": 10160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007589984947741101,
+      "loss": 0.628,
+      "step": 10170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007589560940448176,
+      "loss": 0.6099,
+      "step": 10180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007589136933155251,
+      "loss": 0.5651,
+      "step": 10190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007588712925862325,
+      "loss": 0.7003,
+      "step": 10200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00075882889185694,
+      "loss": 0.6097,
+      "step": 10210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007587864911276473,
+      "loss": 0.6286,
+      "step": 10220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007587440903983549,
+      "loss": 0.5699,
+      "step": 10230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007587016896690624,
+      "loss": 0.5349,
+      "step": 10240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007586592889397697,
+      "loss": 0.7123,
+      "step": 10250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007586168882104772,
+      "loss": 0.5462,
+      "step": 10260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007585744874811847,
+      "loss": 0.6152,
+      "step": 10270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007585320867518921,
+      "loss": 0.6061,
+      "step": 10280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007584896860225996,
+      "loss": 0.5084,
+      "step": 10290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007584472852933071,
+      "loss": 0.5812,
+      "step": 10300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007584048845640145,
+      "loss": 0.6344,
+      "step": 10310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000758362483834722,
+      "loss": 0.5583,
+      "step": 10320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007583200831054295,
+      "loss": 0.5861,
+      "step": 10330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007582776823761368,
+      "loss": 0.6502,
+      "step": 10340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007582352816468444,
+      "loss": 0.5275,
+      "step": 10350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007581928809175519,
+      "loss": 0.5731,
+      "step": 10360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007581504801882592,
+      "loss": 0.6453,
+      "step": 10370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007581080794589667,
+      "loss": 0.5865,
+      "step": 10380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007580656787296742,
+      "loss": 0.714,
+      "step": 10390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007580232780003816,
+      "loss": 0.7056,
+      "step": 10400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007579808772710891,
+      "loss": 0.5485,
+      "step": 10410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007579384765417966,
+      "loss": 0.5998,
+      "step": 10420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000757896075812504,
+      "loss": 0.6239,
+      "step": 10430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007578536750832115,
+      "loss": 0.5806,
+      "step": 10440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000757811274353919,
+      "loss": 0.6128,
+      "step": 10450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007577688736246263,
+      "loss": 0.5991,
+      "step": 10460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007577264728953338,
+      "loss": 0.6084,
+      "step": 10470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007576840721660414,
+      "loss": 0.564,
+      "step": 10480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007576416714367487,
+      "loss": 0.5061,
+      "step": 10490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007575992707074562,
+      "loss": 0.6631,
+      "step": 10500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007575568699781637,
+      "loss": 0.6126,
+      "step": 10510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007575144692488711,
+      "loss": 0.5286,
+      "step": 10520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007574720685195786,
+      "loss": 0.7224,
+      "step": 10530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000757429667790286,
+      "loss": 0.6591,
+      "step": 10540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007573872670609934,
+      "loss": 0.53,
+      "step": 10550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000757344866331701,
+      "loss": 0.591,
+      "step": 10560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007573024656024084,
+      "loss": 0.6099,
+      "step": 10570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007572600648731158,
+      "loss": 0.7179,
+      "step": 10580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007572176641438233,
+      "loss": 0.5985,
+      "step": 10590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007571752634145308,
+      "loss": 0.6045,
+      "step": 10600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007571328626852382,
+      "loss": 0.5584,
+      "step": 10610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007570904619559457,
+      "loss": 0.7706,
+      "step": 10620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000757048061226653,
+      "loss": 0.5897,
+      "step": 10630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007570056604973606,
+      "loss": 0.6053,
+      "step": 10640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007569632597680681,
+      "loss": 0.6208,
+      "step": 10650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007569208590387755,
+      "loss": 0.4623,
+      "step": 10660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007568784583094829,
+      "loss": 0.6884,
+      "step": 10670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007568360575801905,
+      "loss": 0.6803,
+      "step": 10680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007567936568508979,
+      "loss": 0.6405,
+      "step": 10690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007567512561216053,
+      "loss": 0.5893,
+      "step": 10700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007567088553923128,
+      "loss": 0.6655,
+      "step": 10710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007566664546630203,
+      "loss": 0.5431,
+      "step": 10720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007566240539337277,
+      "loss": 0.5302,
+      "step": 10730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007565816532044352,
+      "loss": 0.6615,
+      "step": 10740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007565392524751425,
+      "loss": 0.622,
+      "step": 10750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007564968517458501,
+      "loss": 0.6325,
+      "step": 10760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007564544510165576,
+      "loss": 0.5363,
+      "step": 10770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000756412050287265,
+      "loss": 0.5953,
+      "step": 10780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007563696495579724,
+      "loss": 0.6159,
+      "step": 10790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007563272488286799,
+      "loss": 0.6203,
+      "step": 10800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007562848480993873,
+      "loss": 0.65,
+      "step": 10810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007562424473700948,
+      "loss": 0.522,
+      "step": 10820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007562000466408022,
+      "loss": 0.5774,
+      "step": 10830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007561576459115097,
+      "loss": 0.7589,
+      "step": 10840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007561152451822172,
+      "loss": 0.6015,
+      "step": 10850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007560728444529246,
+      "loss": 0.6641,
+      "step": 10860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000756030443723632,
+      "loss": 0.6256,
+      "step": 10870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007559880429943396,
+      "loss": 0.5818,
+      "step": 10880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000755945642265047,
+      "loss": 0.6654,
+      "step": 10890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007559032415357544,
+      "loss": 0.5756,
+      "step": 10900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007558608408064619,
+      "loss": 0.686,
+      "step": 10910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007558184400771694,
+      "loss": 0.5812,
+      "step": 10920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007557760393478768,
+      "loss": 0.5893,
+      "step": 10930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007557336386185843,
+      "loss": 0.6391,
+      "step": 10940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007556912378892917,
+      "loss": 0.6423,
+      "step": 10950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007556488371599992,
+      "loss": 0.5752,
+      "step": 10960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007556064364307067,
+      "loss": 0.5765,
+      "step": 10970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007555640357014141,
+      "loss": 0.5854,
+      "step": 10980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007555216349721215,
+      "loss": 0.6918,
+      "step": 10990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000755479234242829,
+      "loss": 0.5194,
+      "step": 11000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007554368335135365,
+      "loss": 0.7056,
+      "step": 11010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007553944327842439,
+      "loss": 0.6055,
+      "step": 11020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007553520320549514,
+      "loss": 0.5804,
+      "step": 11030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007553096313256589,
+      "loss": 0.6088,
+      "step": 11040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007552672305963663,
+      "loss": 0.6886,
+      "step": 11050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007552248298670738,
+      "loss": 0.6164,
+      "step": 11060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007551824291377812,
+      "loss": 0.6539,
+      "step": 11070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007551400284084886,
+      "loss": 0.6394,
+      "step": 11080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007550976276791962,
+      "loss": 0.5296,
+      "step": 11090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007550552269499036,
+      "loss": 0.5987,
+      "step": 11100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000755012826220611,
+      "loss": 0.6983,
+      "step": 11110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007549704254913185,
+      "loss": 0.6313,
+      "step": 11120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000754928024762026,
+      "loss": 0.6948,
+      "step": 11130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007548856240327334,
+      "loss": 0.5854,
+      "step": 11140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007548432233034408,
+      "loss": 0.623,
+      "step": 11150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007548008225741483,
+      "loss": 0.5919,
+      "step": 11160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007547584218448558,
+      "loss": 0.6124,
+      "step": 11170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007547160211155632,
+      "loss": 0.6113,
+      "step": 11180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007546736203862707,
+      "loss": 0.6173,
+      "step": 11190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007546312196569781,
+      "loss": 0.5483,
+      "step": 11200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007545888189276856,
+      "loss": 0.507,
+      "step": 11210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000754546418198393,
+      "loss": 0.6492,
+      "step": 11220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007545040174691005,
+      "loss": 0.6521,
+      "step": 11230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007544616167398079,
+      "loss": 0.6518,
+      "step": 11240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007544192160105155,
+      "loss": 0.4971,
+      "step": 11250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007543768152812229,
+      "loss": 0.7763,
+      "step": 11260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007543344145519303,
+      "loss": 0.5626,
+      "step": 11270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007542920138226377,
+      "loss": 0.5704,
+      "step": 11280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007542496130933453,
+      "loss": 0.6156,
+      "step": 11290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007542072123640527,
+      "loss": 0.6445,
+      "step": 11300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007541648116347601,
+      "loss": 0.6429,
+      "step": 11310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007541224109054676,
+      "loss": 0.5081,
+      "step": 11320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007540800101761751,
+      "loss": 0.5698,
+      "step": 11330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007540376094468825,
+      "loss": 0.614,
+      "step": 11340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00075399520871759,
+      "loss": 0.6224,
+      "step": 11350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007539528079882974,
+      "loss": 0.6508,
+      "step": 11360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007539104072590049,
+      "loss": 0.5994,
+      "step": 11370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007538680065297124,
+      "loss": 0.5786,
+      "step": 11380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007538256058004198,
+      "loss": 0.5275,
+      "step": 11390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007537832050711272,
+      "loss": 0.6075,
+      "step": 11400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007537408043418347,
+      "loss": 0.5497,
+      "step": 11410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007536984036125422,
+      "loss": 0.4839,
+      "step": 11420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007536560028832496,
+      "loss": 0.642,
+      "step": 11430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000753613602153957,
+      "loss": 0.5893,
+      "step": 11440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007535712014246646,
+      "loss": 0.6042,
+      "step": 11450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000753528800695372,
+      "loss": 0.6363,
+      "step": 11460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007534863999660794,
+      "loss": 0.6551,
+      "step": 11470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007534439992367869,
+      "loss": 0.5657,
+      "step": 11480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007534015985074944,
+      "loss": 0.6136,
+      "step": 11490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007533591977782018,
+      "loss": 0.6537,
+      "step": 11500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007533167970489093,
+      "loss": 0.6369,
+      "step": 11510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007532743963196167,
+      "loss": 0.5707,
+      "step": 11520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007532319955903242,
+      "loss": 0.6284,
+      "step": 11530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007531895948610317,
+      "loss": 0.6201,
+      "step": 11540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007531471941317391,
+      "loss": 0.588,
+      "step": 11550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007531047934024465,
+      "loss": 0.5642,
+      "step": 11560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007530623926731541,
+      "loss": 0.6118,
+      "step": 11570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007530199919438615,
+      "loss": 0.5893,
+      "step": 11580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007529775912145689,
+      "loss": 0.6033,
+      "step": 11590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007529351904852764,
+      "loss": 0.7123,
+      "step": 11600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007528927897559838,
+      "loss": 0.6357,
+      "step": 11610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007528503890266913,
+      "loss": 0.5915,
+      "step": 11620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007528079882973988,
+      "loss": 0.5912,
+      "step": 11630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007527655875681062,
+      "loss": 0.6112,
+      "step": 11640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007527231868388137,
+      "loss": 0.5619,
+      "step": 11650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007526807861095212,
+      "loss": 0.6299,
+      "step": 11660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007526383853802286,
+      "loss": 0.5745,
+      "step": 11670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000752595984650936,
+      "loss": 0.5975,
+      "step": 11680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007525535839216434,
+      "loss": 0.6203,
+      "step": 11690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000752511183192351,
+      "loss": 0.6461,
+      "step": 11700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007524687824630584,
+      "loss": 0.6729,
+      "step": 11710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007524263817337658,
+      "loss": 0.7198,
+      "step": 11720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007523839810044733,
+      "loss": 0.5475,
+      "step": 11730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007523415802751808,
+      "loss": 0.6529,
+      "step": 11740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007522991795458882,
+      "loss": 0.7057,
+      "step": 11750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007522567788165956,
+      "loss": 0.5945,
+      "step": 11760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007522143780873031,
+      "loss": 0.6071,
+      "step": 11770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007521719773580107,
+      "loss": 0.6047,
+      "step": 11780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000752129576628718,
+      "loss": 0.6428,
+      "step": 11790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007520871758994255,
+      "loss": 0.5286,
+      "step": 11800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007520447751701329,
+      "loss": 0.5209,
+      "step": 11810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007520023744408404,
+      "loss": 0.7729,
+      "step": 11820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007519599737115479,
+      "loss": 0.6752,
+      "step": 11830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007519175729822553,
+      "loss": 0.6177,
+      "step": 11840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007518751722529628,
+      "loss": 0.5883,
+      "step": 11850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007518327715236703,
+      "loss": 0.6387,
+      "step": 11860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007517903707943777,
+      "loss": 0.6131,
+      "step": 11870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007517479700650851,
+      "loss": 0.5791,
+      "step": 11880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007517055693357926,
+      "loss": 0.5556,
+      "step": 11890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007516631686065001,
+      "loss": 0.5669,
+      "step": 11900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007516207678772075,
+      "loss": 0.6502,
+      "step": 11910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000751578367147915,
+      "loss": 0.5086,
+      "step": 11920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007515359664186224,
+      "loss": 0.5305,
+      "step": 11930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007514935656893299,
+      "loss": 0.5678,
+      "step": 11940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007514511649600374,
+      "loss": 0.5955,
+      "step": 11950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007514087642307448,
+      "loss": 0.7589,
+      "step": 11960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007513663635014522,
+      "loss": 0.698,
+      "step": 11970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007513239627721598,
+      "loss": 0.5837,
+      "step": 11980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007512815620428672,
+      "loss": 0.537,
+      "step": 11990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007512391613135746,
+      "loss": 0.7069,
+      "step": 12000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007511967605842821,
+      "loss": 0.6812,
+      "step": 12010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007511543598549895,
+      "loss": 0.6666,
+      "step": 12020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000751111959125697,
+      "loss": 0.6355,
+      "step": 12030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007510695583964045,
+      "loss": 0.4884,
+      "step": 12040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007510271576671118,
+      "loss": 0.6222,
+      "step": 12050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007509847569378194,
+      "loss": 0.5423,
+      "step": 12060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007509423562085269,
+      "loss": 0.6097,
+      "step": 12070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007508999554792342,
+      "loss": 0.6327,
+      "step": 12080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007508575547499417,
+      "loss": 0.5862,
+      "step": 12090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007508151540206493,
+      "loss": 0.5047,
+      "step": 12100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007507727532913566,
+      "loss": 0.591,
+      "step": 12110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007507303525620641,
+      "loss": 0.6165,
+      "step": 12120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007506879518327716,
+      "loss": 0.6532,
+      "step": 12130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000750645551103479,
+      "loss": 0.6097,
+      "step": 12140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007506031503741865,
+      "loss": 0.6374,
+      "step": 12150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000750560749644894,
+      "loss": 0.508,
+      "step": 12160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007505183489156013,
+      "loss": 0.5172,
+      "step": 12170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007504759481863089,
+      "loss": 0.7089,
+      "step": 12180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007504335474570164,
+      "loss": 0.6947,
+      "step": 12190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007503911467277237,
+      "loss": 0.5599,
+      "step": 12200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007503487459984312,
+      "loss": 0.6658,
+      "step": 12210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007503063452691386,
+      "loss": 0.5542,
+      "step": 12220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007502639445398461,
+      "loss": 0.5889,
+      "step": 12230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007502215438105536,
+      "loss": 0.6069,
+      "step": 12240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000750179143081261,
+      "loss": 0.6269,
+      "step": 12250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007501367423519685,
+      "loss": 0.5793,
+      "step": 12260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000750094341622676,
+      "loss": 0.6086,
+      "step": 12270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007500519408933834,
+      "loss": 0.6366,
+      "step": 12280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007500095401640908,
+      "loss": 0.5846,
+      "step": 12290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007499671394347984,
+      "loss": 0.6141,
+      "step": 12300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007499247387055058,
+      "loss": 0.6024,
+      "step": 12310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007498823379762132,
+      "loss": 0.6157,
+      "step": 12320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007498399372469207,
+      "loss": 0.6804,
+      "step": 12330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007497975365176281,
+      "loss": 0.5805,
+      "step": 12340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007497551357883356,
+      "loss": 0.633,
+      "step": 12350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007497127350590431,
+      "loss": 0.6117,
+      "step": 12360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007496703343297504,
+      "loss": 0.6397,
+      "step": 12370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000749627933600458,
+      "loss": 0.683,
+      "step": 12380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007495855328711655,
+      "loss": 0.6531,
+      "step": 12390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007495431321418728,
+      "loss": 0.6204,
+      "step": 12400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007495007314125803,
+      "loss": 0.6636,
+      "step": 12410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007494583306832878,
+      "loss": 0.6741,
+      "step": 12420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007494159299539952,
+      "loss": 0.5562,
+      "step": 12430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007493735292247027,
+      "loss": 0.6214,
+      "step": 12440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007493311284954102,
+      "loss": 0.6067,
+      "step": 12450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007492887277661176,
+      "loss": 0.5956,
+      "step": 12460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007492463270368251,
+      "loss": 0.5562,
+      "step": 12470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007492039263075326,
+      "loss": 0.6638,
+      "step": 12480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007491615255782399,
+      "loss": 0.664,
+      "step": 12490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007491191248489474,
+      "loss": 0.684,
+      "step": 12500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000749076724119655,
+      "loss": 0.5465,
+      "step": 12510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007490343233903623,
+      "loss": 0.7079,
+      "step": 12520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007489919226610698,
+      "loss": 0.6375,
+      "step": 12530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007489495219317773,
+      "loss": 0.5776,
+      "step": 12540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007489071212024847,
+      "loss": 0.6072,
+      "step": 12550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007488647204731922,
+      "loss": 0.7531,
+      "step": 12560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007488223197438997,
+      "loss": 0.639,
+      "step": 12570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000748779919014607,
+      "loss": 0.5889,
+      "step": 12580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007487375182853146,
+      "loss": 0.5962,
+      "step": 12590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007486951175560221,
+      "loss": 0.5892,
+      "step": 12600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007486527168267294,
+      "loss": 0.6007,
+      "step": 12610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007486103160974369,
+      "loss": 0.63,
+      "step": 12620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007485679153681445,
+      "loss": 0.5749,
+      "step": 12630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007485255146388518,
+      "loss": 0.6368,
+      "step": 12640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007484831139095593,
+      "loss": 0.5417,
+      "step": 12650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007484407131802666,
+      "loss": 0.641,
+      "step": 12660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007483983124509742,
+      "loss": 0.5266,
+      "step": 12670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007483559117216817,
+      "loss": 0.6357,
+      "step": 12680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000748313510992389,
+      "loss": 0.5623,
+      "step": 12690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007482711102630965,
+      "loss": 0.5317,
+      "step": 12700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007482287095338041,
+      "loss": 0.5722,
+      "step": 12710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007481863088045114,
+      "loss": 0.5847,
+      "step": 12720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007481439080752189,
+      "loss": 0.6959,
+      "step": 12730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007481015073459264,
+      "loss": 0.758,
+      "step": 12740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007480591066166338,
+      "loss": 0.6306,
+      "step": 12750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007480167058873413,
+      "loss": 0.6342,
+      "step": 12760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007479743051580488,
+      "loss": 0.6282,
+      "step": 12770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007479319044287561,
+      "loss": 0.7212,
+      "step": 12780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007478895036994637,
+      "loss": 0.6278,
+      "step": 12790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007478471029701712,
+      "loss": 0.6269,
+      "step": 12800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007478047022408785,
+      "loss": 0.567,
+      "step": 12810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000747762301511586,
+      "loss": 0.6161,
+      "step": 12820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007477199007822936,
+      "loss": 0.6015,
+      "step": 12830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007476775000530009,
+      "loss": 0.6014,
+      "step": 12840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007476350993237084,
+      "loss": 0.6351,
+      "step": 12850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007475926985944159,
+      "loss": 0.6446,
+      "step": 12860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007475502978651233,
+      "loss": 0.5391,
+      "step": 12870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007475078971358308,
+      "loss": 0.6163,
+      "step": 12880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007474654964065383,
+      "loss": 0.5519,
+      "step": 12890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007474230956772456,
+      "loss": 0.5735,
+      "step": 12900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007473806949479532,
+      "loss": 0.5562,
+      "step": 12910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007473382942186607,
+      "loss": 0.6746,
+      "step": 12920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000747295893489368,
+      "loss": 0.5621,
+      "step": 12930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007472534927600755,
+      "loss": 0.6022,
+      "step": 12940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000747211092030783,
+      "loss": 0.5434,
+      "step": 12950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007471686913014904,
+      "loss": 0.6853,
+      "step": 12960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007471262905721979,
+      "loss": 0.6753,
+      "step": 12970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007470838898429053,
+      "loss": 0.58,
+      "step": 12980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007470414891136128,
+      "loss": 0.5815,
+      "step": 12990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007469990883843203,
+      "loss": 0.6307,
+      "step": 13000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007469566876550277,
+      "loss": 0.6604,
+      "step": 13010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007469142869257351,
+      "loss": 0.6728,
+      "step": 13020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007468718861964426,
+      "loss": 0.6022,
+      "step": 13030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007468294854671501,
+      "loss": 0.5398,
+      "step": 13040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007467870847378575,
+      "loss": 0.6341,
+      "step": 13050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000746744684008565,
+      "loss": 0.6798,
+      "step": 13060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007467022832792725,
+      "loss": 0.5191,
+      "step": 13070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007466598825499799,
+      "loss": 0.576,
+      "step": 13080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007466174818206874,
+      "loss": 0.5796,
+      "step": 13090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007465750810913947,
+      "loss": 0.6727,
+      "step": 13100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007465326803621022,
+      "loss": 0.5564,
+      "step": 13110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007464902796328098,
+      "loss": 0.5987,
+      "step": 13120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007464478789035171,
+      "loss": 0.6462,
+      "step": 13130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007464054781742246,
+      "loss": 0.6555,
+      "step": 13140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007463630774449321,
+      "loss": 0.7378,
+      "step": 13150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007463206767156395,
+      "loss": 0.5343,
+      "step": 13160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000746278275986347,
+      "loss": 0.6108,
+      "step": 13170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007462358752570545,
+      "loss": 0.5885,
+      "step": 13180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007461934745277618,
+      "loss": 0.7035,
+      "step": 13190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007461510737984694,
+      "loss": 0.644,
+      "step": 13200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007461086730691769,
+      "loss": 0.4994,
+      "step": 13210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007460662723398842,
+      "loss": 0.6407,
+      "step": 13220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007460238716105917,
+      "loss": 0.6107,
+      "step": 13230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007459814708812993,
+      "loss": 0.6146,
+      "step": 13240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007459390701520066,
+      "loss": 0.6068,
+      "step": 13250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007458966694227141,
+      "loss": 0.6297,
+      "step": 13260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007458542686934215,
+      "loss": 0.6471,
+      "step": 13270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000745811867964129,
+      "loss": 0.6486,
+      "step": 13280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007457694672348365,
+      "loss": 0.6516,
+      "step": 13290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007457270665055439,
+      "loss": 0.6181,
+      "step": 13300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007456846657762513,
+      "loss": 0.6484,
+      "step": 13310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007456422650469589,
+      "loss": 0.5871,
+      "step": 13320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007455998643176663,
+      "loss": 0.691,
+      "step": 13330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007455574635883737,
+      "loss": 0.5649,
+      "step": 13340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007455150628590812,
+      "loss": 0.5493,
+      "step": 13350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007454726621297887,
+      "loss": 0.6166,
+      "step": 13360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007454302614004961,
+      "loss": 0.5412,
+      "step": 13370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007453878606712036,
+      "loss": 0.71,
+      "step": 13380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000745345459941911,
+      "loss": 0.5896,
+      "step": 13390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007453030592126185,
+      "loss": 0.6035,
+      "step": 13400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000745260658483326,
+      "loss": 0.618,
+      "step": 13410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007452182577540334,
+      "loss": 0.5391,
+      "step": 13420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007451758570247408,
+      "loss": 0.5666,
+      "step": 13430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007451334562954484,
+      "loss": 0.6124,
+      "step": 13440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007450910555661558,
+      "loss": 0.6676,
+      "step": 13450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007450486548368632,
+      "loss": 0.6854,
+      "step": 13460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007450062541075707,
+      "loss": 0.668,
+      "step": 13470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007449638533782782,
+      "loss": 0.6037,
+      "step": 13480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007449214526489856,
+      "loss": 0.7304,
+      "step": 13490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007448790519196931,
+      "loss": 0.6132,
+      "step": 13500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007448366511904005,
+      "loss": 0.643,
+      "step": 13510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000744794250461108,
+      "loss": 0.5895,
+      "step": 13520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007447518497318155,
+      "loss": 0.6195,
+      "step": 13530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007447094490025229,
+      "loss": 0.5637,
+      "step": 13540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007446670482732303,
+      "loss": 0.6548,
+      "step": 13550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007446246475439378,
+      "loss": 0.6058,
+      "step": 13560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007445822468146453,
+      "loss": 0.6111,
+      "step": 13570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007445398460853527,
+      "loss": 0.6542,
+      "step": 13580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007444974453560601,
+      "loss": 0.6391,
+      "step": 13590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007444550446267677,
+      "loss": 0.676,
+      "step": 13600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007444126438974751,
+      "loss": 0.8043,
+      "step": 13610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007443702431681825,
+      "loss": 0.5862,
+      "step": 13620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00074432784243889,
+      "loss": 0.5719,
+      "step": 13630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007442854417095974,
+      "loss": 0.5825,
+      "step": 13640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007442430409803049,
+      "loss": 0.5895,
+      "step": 13650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007442006402510123,
+      "loss": 0.6612,
+      "step": 13660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007441582395217198,
+      "loss": 0.6712,
+      "step": 13670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007441158387924273,
+      "loss": 0.648,
+      "step": 13680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007440734380631347,
+      "loss": 0.5609,
+      "step": 13690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007440310373338422,
+      "loss": 0.6083,
+      "step": 13700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007439886366045496,
+      "loss": 0.5155,
+      "step": 13710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000743946235875257,
+      "loss": 0.6379,
+      "step": 13720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007439038351459646,
+      "loss": 0.6733,
+      "step": 13730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000743861434416672,
+      "loss": 0.6182,
+      "step": 13740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007438190336873794,
+      "loss": 0.7276,
+      "step": 13750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007437766329580869,
+      "loss": 0.6636,
+      "step": 13760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007437342322287944,
+      "loss": 0.5745,
+      "step": 13770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007436918314995018,
+      "loss": 0.7242,
+      "step": 13780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007436494307702093,
+      "loss": 0.6628,
+      "step": 13790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007436070300409168,
+      "loss": 0.5946,
+      "step": 13800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007435646293116242,
+      "loss": 0.5858,
+      "step": 13810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007435222285823317,
+      "loss": 0.5847,
+      "step": 13820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007434798278530391,
+      "loss": 0.606,
+      "step": 13830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007434374271237465,
+      "loss": 0.6959,
+      "step": 13840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007433950263944541,
+      "loss": 0.7436,
+      "step": 13850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007433526256651615,
+      "loss": 0.5982,
+      "step": 13860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007433102249358689,
+      "loss": 0.6031,
+      "step": 13870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007432678242065764,
+      "loss": 0.5581,
+      "step": 13880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007432254234772839,
+      "loss": 0.6424,
+      "step": 13890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007431830227479913,
+      "loss": 0.5949,
+      "step": 13900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007431406220186987,
+      "loss": 0.731,
+      "step": 13910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007430982212894062,
+      "loss": 0.6413,
+      "step": 13920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007430558205601137,
+      "loss": 0.5959,
+      "step": 13930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007430134198308211,
+      "loss": 0.6236,
+      "step": 13940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007429710191015286,
+      "loss": 0.5956,
+      "step": 13950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000742928618372236,
+      "loss": 0.6389,
+      "step": 13960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007428862176429435,
+      "loss": 0.7018,
+      "step": 13970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000742843816913651,
+      "loss": 0.5867,
+      "step": 13980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007428014161843584,
+      "loss": 0.562,
+      "step": 13990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007427590154550658,
+      "loss": 0.706,
+      "step": 14000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007427166147257734,
+      "loss": 0.6874,
+      "step": 14010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007426742139964808,
+      "loss": 0.561,
+      "step": 14020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007426318132671882,
+      "loss": 0.7459,
+      "step": 14030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007425894125378957,
+      "loss": 0.6497,
+      "step": 14040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007425470118086032,
+      "loss": 0.7178,
+      "step": 14050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007425046110793106,
+      "loss": 0.5097,
+      "step": 14060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000742462210350018,
+      "loss": 0.5795,
+      "step": 14070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007424198096207255,
+      "loss": 0.6202,
+      "step": 14080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000742377408891433,
+      "loss": 0.4939,
+      "step": 14090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007423350081621405,
+      "loss": 0.6458,
+      "step": 14100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007422926074328479,
+      "loss": 0.7222,
+      "step": 14110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007422502067035553,
+      "loss": 0.613,
+      "step": 14120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007422078059742629,
+      "loss": 0.627,
+      "step": 14130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007421654052449703,
+      "loss": 0.6498,
+      "step": 14140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007421230045156777,
+      "loss": 0.6762,
+      "step": 14150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007420806037863851,
+      "loss": 0.5524,
+      "step": 14160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007420382030570926,
+      "loss": 0.6521,
+      "step": 14170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007419958023278001,
+      "loss": 0.6267,
+      "step": 14180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007419534015985075,
+      "loss": 0.6594,
+      "step": 14190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007419110008692149,
+      "loss": 0.5393,
+      "step": 14200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007418686001399225,
+      "loss": 0.6549,
+      "step": 14210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007418261994106299,
+      "loss": 0.6943,
+      "step": 14220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007417837986813373,
+      "loss": 0.6557,
+      "step": 14230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007417413979520448,
+      "loss": 0.5756,
+      "step": 14240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007416989972227522,
+      "loss": 0.6499,
+      "step": 14250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007416565964934597,
+      "loss": 0.5379,
+      "step": 14260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007416141957641672,
+      "loss": 0.6383,
+      "step": 14270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007415717950348746,
+      "loss": 0.5249,
+      "step": 14280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007415293943055821,
+      "loss": 0.6645,
+      "step": 14290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007414869935762896,
+      "loss": 0.6242,
+      "step": 14300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000741444592846997,
+      "loss": 0.6758,
+      "step": 14310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007414021921177044,
+      "loss": 0.6297,
+      "step": 14320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000741359791388412,
+      "loss": 0.7331,
+      "step": 14330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007413173906591194,
+      "loss": 0.5878,
+      "step": 14340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007412749899298268,
+      "loss": 0.6685,
+      "step": 14350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007412325892005343,
+      "loss": 0.6424,
+      "step": 14360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007411901884712417,
+      "loss": 0.5827,
+      "step": 14370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007411477877419492,
+      "loss": 0.6836,
+      "step": 14380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007411053870126567,
+      "loss": 0.6079,
+      "step": 14390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007410629862833641,
+      "loss": 0.6642,
+      "step": 14400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007410205855540716,
+      "loss": 0.5933,
+      "step": 14410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007409781848247791,
+      "loss": 0.5551,
+      "step": 14420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007409357840954865,
+      "loss": 0.6871,
+      "step": 14430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007408933833661939,
+      "loss": 0.6032,
+      "step": 14440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007408509826369014,
+      "loss": 0.688,
+      "step": 14450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007408085819076089,
+      "loss": 0.7228,
+      "step": 14460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007407661811783163,
+      "loss": 0.6899,
+      "step": 14470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007407237804490238,
+      "loss": 0.5755,
+      "step": 14480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007406813797197312,
+      "loss": 0.6081,
+      "step": 14490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007406389789904387,
+      "loss": 0.6162,
+      "step": 14500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007405965782611462,
+      "loss": 0.7299,
+      "step": 14510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007405541775318535,
+      "loss": 0.5963,
+      "step": 14520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000740511776802561,
+      "loss": 0.6647,
+      "step": 14530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007404693760732686,
+      "loss": 0.6602,
+      "step": 14540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007404269753439759,
+      "loss": 0.6536,
+      "step": 14550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007403845746146834,
+      "loss": 0.565,
+      "step": 14560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007403421738853908,
+      "loss": 0.5678,
+      "step": 14570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007402997731560983,
+      "loss": 0.6437,
+      "step": 14580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007402573724268058,
+      "loss": 0.5605,
+      "step": 14590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007402149716975132,
+      "loss": 0.5891,
+      "step": 14600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007401725709682206,
+      "loss": 0.7367,
+      "step": 14610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007401301702389282,
+      "loss": 0.6042,
+      "step": 14620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007400877695096356,
+      "loss": 0.6288,
+      "step": 14630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000740045368780343,
+      "loss": 0.5818,
+      "step": 14640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007400029680510505,
+      "loss": 0.6382,
+      "step": 14650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000739960567321758,
+      "loss": 0.5987,
+      "step": 14660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007399181665924654,
+      "loss": 0.6191,
+      "step": 14670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007398757658631729,
+      "loss": 0.6835,
+      "step": 14680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007398333651338803,
+      "loss": 0.6823,
+      "step": 14690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007397909644045878,
+      "loss": 0.6701,
+      "step": 14700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007397485636752953,
+      "loss": 0.489,
+      "step": 14710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007397061629460027,
+      "loss": 0.6434,
+      "step": 14720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007396637622167101,
+      "loss": 0.6888,
+      "step": 14730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007396213614874177,
+      "loss": 0.5761,
+      "step": 14740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007395789607581251,
+      "loss": 0.605,
+      "step": 14750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007395365600288325,
+      "loss": 0.6319,
+      "step": 14760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00073949415929954,
+      "loss": 0.692,
+      "step": 14770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007394517585702474,
+      "loss": 0.6627,
+      "step": 14780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007394093578409549,
+      "loss": 0.5894,
+      "step": 14790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007393669571116624,
+      "loss": 0.6542,
+      "step": 14800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007393245563823697,
+      "loss": 0.6613,
+      "step": 14810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007392821556530773,
+      "loss": 0.5248,
+      "step": 14820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007392397549237848,
+      "loss": 0.5872,
+      "step": 14830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007391973541944921,
+      "loss": 0.6033,
+      "step": 14840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007391549534651996,
+      "loss": 0.5938,
+      "step": 14850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007391125527359072,
+      "loss": 0.6402,
+      "step": 14860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007390701520066145,
+      "loss": 0.6271,
+      "step": 14870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000739027751277322,
+      "loss": 0.643,
+      "step": 14880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007389853505480295,
+      "loss": 0.6084,
+      "step": 14890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007389429498187369,
+      "loss": 0.5807,
+      "step": 14900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007389005490894444,
+      "loss": 0.6567,
+      "step": 14910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007388581483601519,
+      "loss": 0.6591,
+      "step": 14920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007388157476308592,
+      "loss": 0.5358,
+      "step": 14930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007387733469015668,
+      "loss": 0.597,
+      "step": 14940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007387309461722743,
+      "loss": 0.6052,
+      "step": 14950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007386885454429816,
+      "loss": 0.5493,
+      "step": 14960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007386461447136891,
+      "loss": 0.645,
+      "step": 14970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007386037439843966,
+      "loss": 0.6871,
+      "step": 14980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000738561343255104,
+      "loss": 0.7201,
+      "step": 14990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007385189425258115,
+      "loss": 0.604,
+      "step": 15000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000738476541796519,
+      "loss": 0.6668,
+      "step": 15010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007384341410672264,
+      "loss": 0.5855,
+      "step": 15020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007383917403379339,
+      "loss": 0.6276,
+      "step": 15030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007383493396086414,
+      "loss": 0.5422,
+      "step": 15040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007383069388793487,
+      "loss": 0.6336,
+      "step": 15050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007382645381500562,
+      "loss": 0.6637,
+      "step": 15060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007382221374207638,
+      "loss": 0.6478,
+      "step": 15070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007381797366914711,
+      "loss": 0.5825,
+      "step": 15080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007381373359621786,
+      "loss": 0.645,
+      "step": 15090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000738094935232886,
+      "loss": 0.6042,
+      "step": 15100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007380525345035935,
+      "loss": 0.748,
+      "step": 15110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000738010133774301,
+      "loss": 0.5841,
+      "step": 15120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007379677330450083,
+      "loss": 0.5532,
+      "step": 15130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007379253323157158,
+      "loss": 0.561,
+      "step": 15140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007378829315864234,
+      "loss": 0.666,
+      "step": 15150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007378405308571307,
+      "loss": 0.6019,
+      "step": 15160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007377981301278382,
+      "loss": 0.5926,
+      "step": 15170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007377557293985457,
+      "loss": 0.5976,
+      "step": 15180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007377133286692531,
+      "loss": 0.5917,
+      "step": 15190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007376709279399606,
+      "loss": 0.5789,
+      "step": 15200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007376285272106681,
+      "loss": 0.5984,
+      "step": 15210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007375861264813754,
+      "loss": 0.5242,
+      "step": 15220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000737543725752083,
+      "loss": 0.5126,
+      "step": 15230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007375013250227905,
+      "loss": 0.6455,
+      "step": 15240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007374589242934978,
+      "loss": 0.7547,
+      "step": 15250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007374165235642053,
+      "loss": 0.6853,
+      "step": 15260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007373741228349129,
+      "loss": 0.6655,
+      "step": 15270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007373317221056202,
+      "loss": 0.5715,
+      "step": 15280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007372893213763277,
+      "loss": 0.6872,
+      "step": 15290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007372469206470352,
+      "loss": 0.5997,
+      "step": 15300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007372045199177426,
+      "loss": 0.6356,
+      "step": 15310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007371621191884501,
+      "loss": 0.6224,
+      "step": 15320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007371197184591576,
+      "loss": 0.5724,
+      "step": 15330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007370773177298649,
+      "loss": 0.7233,
+      "step": 15340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007370349170005725,
+      "loss": 0.5913,
+      "step": 15350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00073699251627128,
+      "loss": 0.5893,
+      "step": 15360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007369501155419873,
+      "loss": 0.6799,
+      "step": 15370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007369077148126948,
+      "loss": 0.6619,
+      "step": 15380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007368653140834024,
+      "loss": 0.632,
+      "step": 15390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007368229133541097,
+      "loss": 0.5894,
+      "step": 15400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007367805126248172,
+      "loss": 0.6487,
+      "step": 15410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007367381118955245,
+      "loss": 0.6319,
+      "step": 15420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007366957111662321,
+      "loss": 0.6552,
+      "step": 15430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007366533104369396,
+      "loss": 0.6324,
+      "step": 15440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000736610909707647,
+      "loss": 0.6085,
+      "step": 15450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007365685089783544,
+      "loss": 0.548,
+      "step": 15460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000736526108249062,
+      "loss": 0.6562,
+      "step": 15470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007364837075197694,
+      "loss": 0.6543,
+      "step": 15480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007364413067904768,
+      "loss": 0.6441,
+      "step": 15490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007363989060611843,
+      "loss": 0.6044,
+      "step": 15500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007363565053318918,
+      "loss": 0.6424,
+      "step": 15510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007363141046025992,
+      "loss": 0.6432,
+      "step": 15520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007362717038733067,
+      "loss": 0.5247,
+      "step": 15530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000736229303144014,
+      "loss": 0.5856,
+      "step": 15540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007361869024147216,
+      "loss": 0.5221,
+      "step": 15550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007361445016854291,
+      "loss": 0.6036,
+      "step": 15560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007361021009561364,
+      "loss": 0.6727,
+      "step": 15570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007360597002268439,
+      "loss": 0.5766,
+      "step": 15580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007360172994975514,
+      "loss": 0.5932,
+      "step": 15590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007359748987682588,
+      "loss": 0.7062,
+      "step": 15600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007359324980389663,
+      "loss": 0.5923,
+      "step": 15610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007358900973096738,
+      "loss": 0.4876,
+      "step": 15620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007358476965803812,
+      "loss": 0.614,
+      "step": 15630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007358052958510887,
+      "loss": 0.5965,
+      "step": 15640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007357628951217962,
+      "loss": 0.6867,
+      "step": 15650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007357204943925035,
+      "loss": 0.632,
+      "step": 15660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000735678093663211,
+      "loss": 0.5183,
+      "step": 15670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007356356929339186,
+      "loss": 0.578,
+      "step": 15680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007355932922046259,
+      "loss": 0.7461,
+      "step": 15690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007355508914753334,
+      "loss": 0.5942,
+      "step": 15700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007355084907460409,
+      "loss": 0.5023,
+      "step": 15710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007354660900167483,
+      "loss": 0.5746,
+      "step": 15720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007354236892874558,
+      "loss": 0.5986,
+      "step": 15730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007353812885581632,
+      "loss": 0.5728,
+      "step": 15740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007353388878288706,
+      "loss": 0.6462,
+      "step": 15750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007352964870995782,
+      "loss": 0.604,
+      "step": 15760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007352540863702856,
+      "loss": 0.6255,
+      "step": 15770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000735211685640993,
+      "loss": 0.7649,
+      "step": 15780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007351692849117005,
+      "loss": 0.5214,
+      "step": 15790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000735126884182408,
+      "loss": 0.6155,
+      "step": 15800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007350844834531154,
+      "loss": 0.698,
+      "step": 15810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007350420827238229,
+      "loss": 0.6477,
+      "step": 15820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007349996819945304,
+      "loss": 0.6177,
+      "step": 15830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007349572812652378,
+      "loss": 0.6556,
+      "step": 15840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007349148805359453,
+      "loss": 0.6655,
+      "step": 15850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007348724798066527,
+      "loss": 0.5396,
+      "step": 15860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007348300790773601,
+      "loss": 0.631,
+      "step": 15870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007347876783480677,
+      "loss": 0.6247,
+      "step": 15880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007347452776187751,
+      "loss": 0.6959,
+      "step": 15890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007347028768894825,
+      "loss": 0.6202,
+      "step": 15900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00073466047616019,
+      "loss": 0.573,
+      "step": 15910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007346180754308975,
+      "loss": 0.6539,
+      "step": 15920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007345756747016049,
+      "loss": 0.5947,
+      "step": 15930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007345332739723124,
+      "loss": 0.7541,
+      "step": 15940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007344908732430197,
+      "loss": 0.6038,
+      "step": 15950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007344484725137273,
+      "loss": 0.7052,
+      "step": 15960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007344060717844348,
+      "loss": 0.5246,
+      "step": 15970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007343636710551421,
+      "loss": 0.5291,
+      "step": 15980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007343212703258496,
+      "loss": 0.6291,
+      "step": 15990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007342788695965572,
+      "loss": 0.5963,
+      "step": 16000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007342364688672645,
+      "loss": 0.5817,
+      "step": 16010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000734194068137972,
+      "loss": 0.5988,
+      "step": 16020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007341516674086794,
+      "loss": 0.5936,
+      "step": 16030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000734109266679387,
+      "loss": 0.5444,
+      "step": 16040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007340668659500944,
+      "loss": 0.5372,
+      "step": 16050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007340244652208018,
+      "loss": 0.6743,
+      "step": 16060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007339820644915092,
+      "loss": 0.6394,
+      "step": 16070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007339396637622168,
+      "loss": 0.5999,
+      "step": 16080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007338972630329242,
+      "loss": 0.5811,
+      "step": 16090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007338548623036316,
+      "loss": 0.6296,
+      "step": 16100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007338124615743391,
+      "loss": 0.6402,
+      "step": 16110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007337700608450466,
+      "loss": 0.605,
+      "step": 16120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000733727660115754,
+      "loss": 0.6436,
+      "step": 16130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007336852593864615,
+      "loss": 0.6466,
+      "step": 16140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007336428586571689,
+      "loss": 0.7211,
+      "step": 16150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007336004579278764,
+      "loss": 0.6473,
+      "step": 16160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007335580571985839,
+      "loss": 0.5904,
+      "step": 16170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007335156564692913,
+      "loss": 0.6351,
+      "step": 16180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007334732557399987,
+      "loss": 0.7375,
+      "step": 16190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007334308550107062,
+      "loss": 0.6142,
+      "step": 16200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007333884542814137,
+      "loss": 0.6657,
+      "step": 16210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007333460535521211,
+      "loss": 0.6089,
+      "step": 16220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007333036528228286,
+      "loss": 0.6561,
+      "step": 16230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007332612520935361,
+      "loss": 0.6024,
+      "step": 16240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007332188513642435,
+      "loss": 0.6674,
+      "step": 16250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000733176450634951,
+      "loss": 0.6292,
+      "step": 16260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007331340499056584,
+      "loss": 0.6667,
+      "step": 16270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007330916491763658,
+      "loss": 0.6541,
+      "step": 16280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007330492484470734,
+      "loss": 0.521,
+      "step": 16290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007330068477177808,
+      "loss": 0.6272,
+      "step": 16300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007329644469884882,
+      "loss": 0.6003,
+      "step": 16310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007329220462591957,
+      "loss": 0.6223,
+      "step": 16320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007328796455299032,
+      "loss": 0.5614,
+      "step": 16330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007328372448006106,
+      "loss": 0.6057,
+      "step": 16340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000732794844071318,
+      "loss": 0.675,
+      "step": 16350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007327524433420256,
+      "loss": 0.6642,
+      "step": 16360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000732710042612733,
+      "loss": 0.6538,
+      "step": 16370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007326676418834404,
+      "loss": 0.703,
+      "step": 16380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007326252411541479,
+      "loss": 0.5156,
+      "step": 16390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007325828404248553,
+      "loss": 0.5663,
+      "step": 16400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007325404396955628,
+      "loss": 0.6556,
+      "step": 16410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007324980389662703,
+      "loss": 0.6383,
+      "step": 16420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007324556382369777,
+      "loss": 0.5925,
+      "step": 16430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007324132375076852,
+      "loss": 0.5572,
+      "step": 16440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007323708367783927,
+      "loss": 0.4902,
+      "step": 16450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007323284360491001,
+      "loss": 0.6336,
+      "step": 16460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007322860353198075,
+      "loss": 0.6811,
+      "step": 16470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007322436345905149,
+      "loss": 0.6521,
+      "step": 16480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007322012338612225,
+      "loss": 0.5733,
+      "step": 16490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007321588331319299,
+      "loss": 0.694,
+      "step": 16500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007321164324026373,
+      "loss": 0.5726,
+      "step": 16510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007320740316733448,
+      "loss": 0.5779,
+      "step": 16520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007320316309440523,
+      "loss": 0.5764,
+      "step": 16530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007319892302147597,
+      "loss": 0.6296,
+      "step": 16540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007319468294854672,
+      "loss": 0.5741,
+      "step": 16550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007319044287561746,
+      "loss": 0.7486,
+      "step": 16560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007318620280268821,
+      "loss": 0.6184,
+      "step": 16570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007318196272975896,
+      "loss": 0.6578,
+      "step": 16580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000731777226568297,
+      "loss": 0.5588,
+      "step": 16590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007317348258390044,
+      "loss": 0.571,
+      "step": 16600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000731692425109712,
+      "loss": 0.5996,
+      "step": 16610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007316500243804194,
+      "loss": 0.6088,
+      "step": 16620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007316076236511268,
+      "loss": 0.6024,
+      "step": 16630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007315652229218342,
+      "loss": 0.607,
+      "step": 16640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007315228221925418,
+      "loss": 0.6863,
+      "step": 16650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007314804214632492,
+      "loss": 0.6516,
+      "step": 16660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007314380207339566,
+      "loss": 0.5095,
+      "step": 16670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007313956200046641,
+      "loss": 0.6025,
+      "step": 16680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007313532192753716,
+      "loss": 0.5959,
+      "step": 16690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000731310818546079,
+      "loss": 0.5485,
+      "step": 16700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007312684178167865,
+      "loss": 0.6573,
+      "step": 16710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007312260170874939,
+      "loss": 0.5794,
+      "step": 16720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007311836163582014,
+      "loss": 0.5564,
+      "step": 16730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007311412156289089,
+      "loss": 0.6392,
+      "step": 16740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007310988148996163,
+      "loss": 0.5766,
+      "step": 16750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007310564141703237,
+      "loss": 0.7061,
+      "step": 16760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007310140134410313,
+      "loss": 0.7205,
+      "step": 16770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007309716127117387,
+      "loss": 0.5958,
+      "step": 16780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007309292119824461,
+      "loss": 0.6157,
+      "step": 16790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007308868112531536,
+      "loss": 0.5907,
+      "step": 16800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007308444105238611,
+      "loss": 0.6223,
+      "step": 16810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007308020097945685,
+      "loss": 0.6177,
+      "step": 16820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000730759609065276,
+      "loss": 0.6976,
+      "step": 16830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007307172083359834,
+      "loss": 0.5278,
+      "step": 16840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007306748076066909,
+      "loss": 0.5984,
+      "step": 16850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007306324068773984,
+      "loss": 0.6882,
+      "step": 16860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007305900061481058,
+      "loss": 0.5681,
+      "step": 16870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007305476054188132,
+      "loss": 0.604,
+      "step": 16880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007305052046895208,
+      "loss": 0.5527,
+      "step": 16890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007304628039602282,
+      "loss": 0.5396,
+      "step": 16900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007304204032309356,
+      "loss": 0.6247,
+      "step": 16910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000730378002501643,
+      "loss": 0.6183,
+      "step": 16920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007303356017723505,
+      "loss": 0.6076,
+      "step": 16930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000730293201043058,
+      "loss": 0.7084,
+      "step": 16940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007302508003137655,
+      "loss": 0.7182,
+      "step": 16950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007302083995844728,
+      "loss": 0.6645,
+      "step": 16960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007301659988551804,
+      "loss": 0.6664,
+      "step": 16970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007301235981258879,
+      "loss": 0.6011,
+      "step": 16980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007300811973965952,
+      "loss": 0.5842,
+      "step": 16990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007300387966673027,
+      "loss": 0.4616,
+      "step": 17000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007299963959380101,
+      "loss": 0.5858,
+      "step": 17010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007299539952087176,
+      "loss": 0.6305,
+      "step": 17020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007299115944794251,
+      "loss": 0.6119,
+      "step": 17030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007298691937501325,
+      "loss": 0.697,
+      "step": 17040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00072982679302084,
+      "loss": 0.6877,
+      "step": 17050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007297843922915475,
+      "loss": 0.6244,
+      "step": 17060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007297419915622549,
+      "loss": 0.6343,
+      "step": 17070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007296995908329623,
+      "loss": 0.6527,
+      "step": 17080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007296571901036698,
+      "loss": 0.6032,
+      "step": 17090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007296147893743773,
+      "loss": 0.6217,
+      "step": 17100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007295723886450847,
+      "loss": 0.5998,
+      "step": 17110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007295299879157922,
+      "loss": 0.571,
+      "step": 17120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007294875871864996,
+      "loss": 0.6792,
+      "step": 17130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007294451864572071,
+      "loss": 0.6104,
+      "step": 17140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007294027857279146,
+      "loss": 0.5999,
+      "step": 17150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000729360384998622,
+      "loss": 0.5071,
+      "step": 17160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007293179842693294,
+      "loss": 0.6451,
+      "step": 17170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000729275583540037,
+      "loss": 0.6632,
+      "step": 17180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007292331828107444,
+      "loss": 0.6124,
+      "step": 17190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007291907820814518,
+      "loss": 0.5391,
+      "step": 17200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007291483813521593,
+      "loss": 0.6107,
+      "step": 17210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007291059806228668,
+      "loss": 0.5494,
+      "step": 17220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007290635798935742,
+      "loss": 0.5335,
+      "step": 17230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007290211791642817,
+      "loss": 0.5614,
+      "step": 17240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000728978778434989,
+      "loss": 0.6793,
+      "step": 17250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007289363777056966,
+      "loss": 0.5121,
+      "step": 17260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007288939769764041,
+      "loss": 0.5717,
+      "step": 17270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007288515762471114,
+      "loss": 0.657,
+      "step": 17280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007288091755178189,
+      "loss": 0.585,
+      "step": 17290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007287667747885265,
+      "loss": 0.6849,
+      "step": 17300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007287243740592338,
+      "loss": 0.6092,
+      "step": 17310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007286819733299413,
+      "loss": 0.6287,
+      "step": 17320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007286395726006488,
+      "loss": 0.6077,
+      "step": 17330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007285971718713562,
+      "loss": 0.5957,
+      "step": 17340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007285547711420637,
+      "loss": 0.5763,
+      "step": 17350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007285123704127712,
+      "loss": 0.7659,
+      "step": 17360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007284699696834785,
+      "loss": 0.5984,
+      "step": 17370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007284275689541861,
+      "loss": 0.5336,
+      "step": 17380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007283851682248936,
+      "loss": 0.5709,
+      "step": 17390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007283427674956009,
+      "loss": 0.632,
+      "step": 17400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007283003667663084,
+      "loss": 0.6221,
+      "step": 17410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000728257966037016,
+      "loss": 0.5207,
+      "step": 17420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007282155653077233,
+      "loss": 0.6199,
+      "step": 17430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007281731645784308,
+      "loss": 0.5875,
+      "step": 17440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007281307638491382,
+      "loss": 0.6093,
+      "step": 17450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007280883631198457,
+      "loss": 0.6651,
+      "step": 17460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007280459623905532,
+      "loss": 0.569,
+      "step": 17470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007280035616612606,
+      "loss": 0.6293,
+      "step": 17480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000727961160931968,
+      "loss": 0.5955,
+      "step": 17490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007279187602026756,
+      "loss": 0.5986,
+      "step": 17500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000727876359473383,
+      "loss": 0.7122,
+      "step": 17510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007278339587440904,
+      "loss": 0.5792,
+      "step": 17520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007277915580147979,
+      "loss": 0.644,
+      "step": 17530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007277491572855053,
+      "loss": 0.6263,
+      "step": 17540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007277067565562128,
+      "loss": 0.5893,
+      "step": 17550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007276643558269203,
+      "loss": 0.6411,
+      "step": 17560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007276219550976276,
+      "loss": 0.6298,
+      "step": 17570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007275795543683352,
+      "loss": 0.6352,
+      "step": 17580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007275371536390427,
+      "loss": 0.5532,
+      "step": 17590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00072749475290975,
+      "loss": 0.6109,
+      "step": 17600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007274523521804575,
+      "loss": 0.5958,
+      "step": 17610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000727409951451165,
+      "loss": 0.6146,
+      "step": 17620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007273675507218724,
+      "loss": 0.6172,
+      "step": 17630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007273251499925799,
+      "loss": 0.6996,
+      "step": 17640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007272827492632874,
+      "loss": 0.7388,
+      "step": 17650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007272403485339948,
+      "loss": 0.661,
+      "step": 17660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007271979478047023,
+      "loss": 0.5107,
+      "step": 17670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007271555470754098,
+      "loss": 0.5982,
+      "step": 17680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007271131463461171,
+      "loss": 0.6044,
+      "step": 17690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007270707456168246,
+      "loss": 0.7058,
+      "step": 17700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007270283448875322,
+      "loss": 0.6115,
+      "step": 17710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007269859441582395,
+      "loss": 0.6186,
+      "step": 17720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000726943543428947,
+      "loss": 0.5408,
+      "step": 17730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007269011426996545,
+      "loss": 0.7947,
+      "step": 17740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007268587419703619,
+      "loss": 0.5893,
+      "step": 17750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007268163412410694,
+      "loss": 0.6298,
+      "step": 17760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007267739405117769,
+      "loss": 0.5958,
+      "step": 17770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007267315397824842,
+      "loss": 0.5858,
+      "step": 17780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007266891390531918,
+      "loss": 0.6175,
+      "step": 17790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007266467383238993,
+      "loss": 0.6409,
+      "step": 17800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007266043375946066,
+      "loss": 0.6391,
+      "step": 17810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007265619368653141,
+      "loss": 0.5389,
+      "step": 17820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007265195361360217,
+      "loss": 0.6465,
+      "step": 17830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000726477135406729,
+      "loss": 0.6272,
+      "step": 17840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007264347346774365,
+      "loss": 0.6083,
+      "step": 17850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000726392333948144,
+      "loss": 0.6784,
+      "step": 17860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007263499332188514,
+      "loss": 0.6383,
+      "step": 17870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007263075324895589,
+      "loss": 0.5568,
+      "step": 17880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007262651317602662,
+      "loss": 0.6551,
+      "step": 17890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007262227310309737,
+      "loss": 0.645,
+      "step": 17900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007261803303016813,
+      "loss": 0.5597,
+      "step": 17910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007261379295723886,
+      "loss": 0.6426,
+      "step": 17920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007260955288430961,
+      "loss": 0.7612,
+      "step": 17930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007260531281138036,
+      "loss": 0.6687,
+      "step": 17940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000726010727384511,
+      "loss": 0.7261,
+      "step": 17950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007259683266552185,
+      "loss": 0.617,
+      "step": 17960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000725925925925926,
+      "loss": 0.5772,
+      "step": 17970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007258835251966333,
+      "loss": 0.5861,
+      "step": 17980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007258411244673409,
+      "loss": 0.5667,
+      "step": 17990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007257987237380484,
+      "loss": 0.5959,
+      "step": 18000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007257563230087557,
+      "loss": 0.4942,
+      "step": 18010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007257139222794632,
+      "loss": 0.5567,
+      "step": 18020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007256715215501708,
+      "loss": 0.4675,
+      "step": 18030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007256291208208781,
+      "loss": 0.6557,
+      "step": 18040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007255867200915856,
+      "loss": 0.6902,
+      "step": 18050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007255443193622931,
+      "loss": 0.5983,
+      "step": 18060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007255019186330005,
+      "loss": 0.5857,
+      "step": 18070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000725459517903708,
+      "loss": 0.5751,
+      "step": 18080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007254171171744155,
+      "loss": 0.5528,
+      "step": 18090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007253747164451228,
+      "loss": 0.6601,
+      "step": 18100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007253323157158304,
+      "loss": 0.63,
+      "step": 18110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007252899149865379,
+      "loss": 0.6909,
+      "step": 18120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007252475142572452,
+      "loss": 0.5727,
+      "step": 18130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007252051135279527,
+      "loss": 0.5623,
+      "step": 18140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007251627127986602,
+      "loss": 0.5867,
+      "step": 18150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007251203120693676,
+      "loss": 0.4831,
+      "step": 18160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007250779113400751,
+      "loss": 0.6079,
+      "step": 18170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007250355106107825,
+      "loss": 0.5672,
+      "step": 18180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00072499310988149,
+      "loss": 0.6574,
+      "step": 18190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007249507091521975,
+      "loss": 0.628,
+      "step": 18200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007249083084229049,
+      "loss": 0.578,
+      "step": 18210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007248659076936123,
+      "loss": 0.6557,
+      "step": 18220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007248235069643198,
+      "loss": 0.5247,
+      "step": 18230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007247811062350273,
+      "loss": 0.6387,
+      "step": 18240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007247387055057347,
+      "loss": 0.5896,
+      "step": 18250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007246963047764422,
+      "loss": 0.5839,
+      "step": 18260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007246539040471497,
+      "loss": 0.5796,
+      "step": 18270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007246115033178571,
+      "loss": 0.6407,
+      "step": 18280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007245691025885646,
+      "loss": 0.5823,
+      "step": 18290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000724526701859272,
+      "loss": 0.6114,
+      "step": 18300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007244843011299795,
+      "loss": 0.5612,
+      "step": 18310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000724441900400687,
+      "loss": 0.5973,
+      "step": 18320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007243994996713943,
+      "loss": 0.5664,
+      "step": 18330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007243570989421018,
+      "loss": 0.615,
+      "step": 18340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007243146982128093,
+      "loss": 0.6891,
+      "step": 18350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007242722974835167,
+      "loss": 0.7602,
+      "step": 18360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007242298967542242,
+      "loss": 0.5697,
+      "step": 18370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007241874960249317,
+      "loss": 0.511,
+      "step": 18380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007241450952956392,
+      "loss": 0.5168,
+      "step": 18390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007241026945663466,
+      "loss": 0.5934,
+      "step": 18400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007240602938370541,
+      "loss": 0.6624,
+      "step": 18410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007240178931077614,
+      "loss": 0.7241,
+      "step": 18420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007239754923784689,
+      "loss": 0.6129,
+      "step": 18430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007239330916491765,
+      "loss": 0.6398,
+      "step": 18440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007238906909198838,
+      "loss": 0.5879,
+      "step": 18450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007238482901905913,
+      "loss": 0.7035,
+      "step": 18460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007238058894612988,
+      "loss": 0.6507,
+      "step": 18470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007237634887320062,
+      "loss": 0.6584,
+      "step": 18480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007237210880027137,
+      "loss": 0.5595,
+      "step": 18490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007236786872734211,
+      "loss": 0.6436,
+      "step": 18500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007236362865441285,
+      "loss": 0.6588,
+      "step": 18510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007235938858148361,
+      "loss": 0.6075,
+      "step": 18520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007235514850855435,
+      "loss": 0.6023,
+      "step": 18530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007235090843562509,
+      "loss": 0.6874,
+      "step": 18540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007234666836269584,
+      "loss": 0.608,
+      "step": 18550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007234242828976659,
+      "loss": 0.6512,
+      "step": 18560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007233818821683733,
+      "loss": 0.5649,
+      "step": 18570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007233394814390808,
+      "loss": 0.6006,
+      "step": 18580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007232970807097882,
+      "loss": 0.5809,
+      "step": 18590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007232546799804957,
+      "loss": 0.5728,
+      "step": 18600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007232122792512032,
+      "loss": 0.5182,
+      "step": 18610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007231698785219106,
+      "loss": 0.6434,
+      "step": 18620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000723127477792618,
+      "loss": 0.5726,
+      "step": 18630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007230850770633256,
+      "loss": 0.6834,
+      "step": 18640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000723042676334033,
+      "loss": 0.6384,
+      "step": 18650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007230002756047404,
+      "loss": 0.5545,
+      "step": 18660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007229578748754479,
+      "loss": 0.4982,
+      "step": 18670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007229154741461554,
+      "loss": 0.6851,
+      "step": 18680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007228730734168628,
+      "loss": 0.6098,
+      "step": 18690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007228306726875703,
+      "loss": 0.6371,
+      "step": 18700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007227882719582777,
+      "loss": 0.5767,
+      "step": 18710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007227458712289852,
+      "loss": 0.6275,
+      "step": 18720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007227034704996927,
+      "loss": 0.5842,
+      "step": 18730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007226610697704,
+      "loss": 0.6876,
+      "step": 18740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007226186690411075,
+      "loss": 0.5441,
+      "step": 18750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000722576268311815,
+      "loss": 0.6453,
+      "step": 18760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007225338675825225,
+      "loss": 0.7198,
+      "step": 18770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007224914668532299,
+      "loss": 0.6716,
+      "step": 18780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007224490661239373,
+      "loss": 0.6273,
+      "step": 18790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007224066653946449,
+      "loss": 0.59,
+      "step": 18800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007223642646653523,
+      "loss": 0.6018,
+      "step": 18810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007223218639360597,
+      "loss": 0.6353,
+      "step": 18820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007222794632067671,
+      "loss": 0.6088,
+      "step": 18830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007222370624774747,
+      "loss": 0.5748,
+      "step": 18840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007221946617481821,
+      "loss": 0.5847,
+      "step": 18850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007221522610188895,
+      "loss": 0.664,
+      "step": 18860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000722109860289597,
+      "loss": 0.6428,
+      "step": 18870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007220674595603045,
+      "loss": 0.5602,
+      "step": 18880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000722025058831012,
+      "loss": 0.6027,
+      "step": 18890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007219826581017194,
+      "loss": 0.5566,
+      "step": 18900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007219402573724268,
+      "loss": 0.6563,
+      "step": 18910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007218978566431343,
+      "loss": 0.6298,
+      "step": 18920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007218554559138418,
+      "loss": 0.7829,
+      "step": 18930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007218130551845492,
+      "loss": 0.6354,
+      "step": 18940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007217706544552566,
+      "loss": 0.6258,
+      "step": 18950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007217282537259641,
+      "loss": 0.6694,
+      "step": 18960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007216858529966716,
+      "loss": 0.5583,
+      "step": 18970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000721643452267379,
+      "loss": 0.6371,
+      "step": 18980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007216010515380865,
+      "loss": 0.574,
+      "step": 18990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000721558650808794,
+      "loss": 0.5414,
+      "step": 19000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007215162500795014,
+      "loss": 0.61,
+      "step": 19010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007214738493502089,
+      "loss": 0.5119,
+      "step": 19020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007214314486209163,
+      "loss": 0.6398,
+      "step": 19030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007213890478916237,
+      "loss": 0.7128,
+      "step": 19040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007213466471623313,
+      "loss": 0.6507,
+      "step": 19050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007213042464330387,
+      "loss": 0.6263,
+      "step": 19060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007212618457037461,
+      "loss": 0.5896,
+      "step": 19070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007212194449744536,
+      "loss": 0.6075,
+      "step": 19080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007211770442451611,
+      "loss": 0.6228,
+      "step": 19090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007211346435158685,
+      "loss": 0.6643,
+      "step": 19100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007210922427865759,
+      "loss": 0.6432,
+      "step": 19110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007210498420572834,
+      "loss": 0.5934,
+      "step": 19120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007210074413279909,
+      "loss": 0.5739,
+      "step": 19130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007209650405986983,
+      "loss": 0.7134,
+      "step": 19140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007209226398694058,
+      "loss": 0.5658,
+      "step": 19150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007208802391401132,
+      "loss": 0.6782,
+      "step": 19160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007208378384108207,
+      "loss": 0.6519,
+      "step": 19170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007207954376815282,
+      "loss": 0.7282,
+      "step": 19180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007207530369522356,
+      "loss": 0.5826,
+      "step": 19190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000720710636222943,
+      "loss": 0.6852,
+      "step": 19200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007206682354936506,
+      "loss": 0.5673,
+      "step": 19210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000720625834764358,
+      "loss": 0.6116,
+      "step": 19220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007205834340350654,
+      "loss": 0.6033,
+      "step": 19230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007205410333057729,
+      "loss": 0.5422,
+      "step": 19240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007204986325764804,
+      "loss": 0.7235,
+      "step": 19250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007204562318471878,
+      "loss": 0.5592,
+      "step": 19260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007204138311178953,
+      "loss": 0.6231,
+      "step": 19270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007203714303886027,
+      "loss": 0.6741,
+      "step": 19280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007203290296593102,
+      "loss": 0.615,
+      "step": 19290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007202866289300177,
+      "loss": 0.6004,
+      "step": 19300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007202442282007251,
+      "loss": 0.5678,
+      "step": 19310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007202018274714325,
+      "loss": 0.531,
+      "step": 19320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00072015942674214,
+      "loss": 0.6414,
+      "step": 19330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007201170260128475,
+      "loss": 0.6653,
+      "step": 19340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007200746252835549,
+      "loss": 0.6284,
+      "step": 19350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007200322245542623,
+      "loss": 0.6153,
+      "step": 19360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007199898238249699,
+      "loss": 0.6316,
+      "step": 19370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007199474230956773,
+      "loss": 0.6466,
+      "step": 19380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007199050223663847,
+      "loss": 0.6101,
+      "step": 19390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007198626216370921,
+      "loss": 0.6005,
+      "step": 19400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007198202209077997,
+      "loss": 0.6207,
+      "step": 19410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007197778201785071,
+      "loss": 0.6286,
+      "step": 19420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007197354194492145,
+      "loss": 0.625,
+      "step": 19430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000719693018719922,
+      "loss": 0.6569,
+      "step": 19440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007196506179906295,
+      "loss": 0.6067,
+      "step": 19450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007196082172613369,
+      "loss": 0.6574,
+      "step": 19460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007195658165320444,
+      "loss": 0.6847,
+      "step": 19470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007195234158027518,
+      "loss": 0.5976,
+      "step": 19480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007194810150734593,
+      "loss": 0.5956,
+      "step": 19490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007194386143441668,
+      "loss": 0.4963,
+      "step": 19500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007193962136148742,
+      "loss": 0.6015,
+      "step": 19510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007193538128855816,
+      "loss": 0.7437,
+      "step": 19520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007193114121562892,
+      "loss": 0.6133,
+      "step": 19530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007192690114269966,
+      "loss": 0.6524,
+      "step": 19540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000719226610697704,
+      "loss": 0.6267,
+      "step": 19550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007191842099684115,
+      "loss": 0.6482,
+      "step": 19560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007191418092391189,
+      "loss": 0.6271,
+      "step": 19570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007190994085098264,
+      "loss": 0.56,
+      "step": 19580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007190570077805339,
+      "loss": 0.603,
+      "step": 19590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007190146070512413,
+      "loss": 0.6183,
+      "step": 19600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007189722063219488,
+      "loss": 0.5323,
+      "step": 19610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007189298055926563,
+      "loss": 0.5964,
+      "step": 19620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007188874048633637,
+      "loss": 0.5781,
+      "step": 19630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007188450041340711,
+      "loss": 0.6134,
+      "step": 19640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007188026034047786,
+      "loss": 0.6201,
+      "step": 19650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007187602026754861,
+      "loss": 0.5976,
+      "step": 19660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007187178019461935,
+      "loss": 0.6749,
+      "step": 19670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000718675401216901,
+      "loss": 0.5905,
+      "step": 19680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007186330004876084,
+      "loss": 0.5686,
+      "step": 19690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007185905997583159,
+      "loss": 0.68,
+      "step": 19700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007185481990290234,
+      "loss": 0.6849,
+      "step": 19710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007185057982997307,
+      "loss": 0.548,
+      "step": 19720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007184633975704382,
+      "loss": 0.5168,
+      "step": 19730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007184209968411458,
+      "loss": 0.6682,
+      "step": 19740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007183785961118531,
+      "loss": 0.631,
+      "step": 19750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007183361953825606,
+      "loss": 0.65,
+      "step": 19760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000718293794653268,
+      "loss": 0.5683,
+      "step": 19770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007182513939239755,
+      "loss": 0.5944,
+      "step": 19780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000718208993194683,
+      "loss": 0.5564,
+      "step": 19790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007181665924653904,
+      "loss": 0.567,
+      "step": 19800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007181241917360978,
+      "loss": 0.5472,
+      "step": 19810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007180817910068054,
+      "loss": 0.5707,
+      "step": 19820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007180393902775128,
+      "loss": 0.6035,
+      "step": 19830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007179969895482202,
+      "loss": 0.5128,
+      "step": 19840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007179545888189277,
+      "loss": 0.5397,
+      "step": 19850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007179121880896353,
+      "loss": 0.678,
+      "step": 19860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007178697873603426,
+      "loss": 0.735,
+      "step": 19870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007178273866310501,
+      "loss": 0.6383,
+      "step": 19880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007177849859017575,
+      "loss": 0.574,
+      "step": 19890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000717742585172465,
+      "loss": 0.6756,
+      "step": 19900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007177001844431725,
+      "loss": 0.587,
+      "step": 19910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007176577837138799,
+      "loss": 0.6241,
+      "step": 19920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007176153829845873,
+      "loss": 0.614,
+      "step": 19930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007175729822552949,
+      "loss": 0.6061,
+      "step": 19940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007175305815260023,
+      "loss": 0.5588,
+      "step": 19950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007174881807967097,
+      "loss": 0.6382,
+      "step": 19960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007174457800674172,
+      "loss": 0.6457,
+      "step": 19970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007174033793381247,
+      "loss": 0.5767,
+      "step": 19980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007173609786088321,
+      "loss": 0.5303,
+      "step": 19990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007173185778795396,
+      "loss": 0.5425,
+      "step": 20000
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.6291659474372864,
+      "eval_runtime": 337.8327,
+      "eval_samples_per_second": 15.555,
+      "eval_steps_per_second": 3.889,
+      "step": 20000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007172761771502469,
+      "loss": 0.5866,
+      "step": 20010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007172337764209545,
+      "loss": 0.6541,
+      "step": 20020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000717191375691662,
+      "loss": 0.6159,
+      "step": 20030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007171489749623693,
+      "loss": 0.6534,
+      "step": 20040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007171065742330768,
+      "loss": 0.6131,
+      "step": 20050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007170641735037844,
+      "loss": 0.6717,
+      "step": 20060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007170217727744917,
+      "loss": 0.5898,
+      "step": 20070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007169793720451992,
+      "loss": 0.6734,
+      "step": 20080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007169369713159067,
+      "loss": 0.5304,
+      "step": 20090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007168945705866141,
+      "loss": 0.6221,
+      "step": 20100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007168521698573216,
+      "loss": 0.646,
+      "step": 20110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007168097691280291,
+      "loss": 0.6391,
+      "step": 20120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007167673683987364,
+      "loss": 0.6085,
+      "step": 20130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000716724967669444,
+      "loss": 0.6449,
+      "step": 20140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007166825669401515,
+      "loss": 0.582,
+      "step": 20150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007166401662108588,
+      "loss": 0.6431,
+      "step": 20160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007165977654815663,
+      "loss": 0.6114,
+      "step": 20170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007165553647522738,
+      "loss": 0.5759,
+      "step": 20180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007165129640229812,
+      "loss": 0.4313,
+      "step": 20190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007164705632936887,
+      "loss": 0.6361,
+      "step": 20200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007164281625643962,
+      "loss": 0.6138,
+      "step": 20210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007163857618351036,
+      "loss": 0.607,
+      "step": 20220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007163433611058111,
+      "loss": 0.5974,
+      "step": 20230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007163009603765186,
+      "loss": 0.5623,
+      "step": 20240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007162585596472259,
+      "loss": 0.5691,
+      "step": 20250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007162161589179334,
+      "loss": 0.5758,
+      "step": 20260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000716173758188641,
+      "loss": 0.5985,
+      "step": 20270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007161313574593483,
+      "loss": 0.6141,
+      "step": 20280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007160889567300558,
+      "loss": 0.592,
+      "step": 20290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007160465560007632,
+      "loss": 0.5901,
+      "step": 20300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007160041552714707,
+      "loss": 0.5923,
+      "step": 20310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007159617545421782,
+      "loss": 0.6363,
+      "step": 20320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007159193538128855,
+      "loss": 0.6146,
+      "step": 20330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007158769530835931,
+      "loss": 0.5526,
+      "step": 20340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007158345523543006,
+      "loss": 0.6905,
+      "step": 20350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007157921516250079,
+      "loss": 0.6717,
+      "step": 20360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007157497508957154,
+      "loss": 0.5869,
+      "step": 20370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007157073501664229,
+      "loss": 0.6797,
+      "step": 20380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007156649494371303,
+      "loss": 0.6152,
+      "step": 20390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007156225487078378,
+      "loss": 0.6495,
+      "step": 20400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007155801479785453,
+      "loss": 0.6464,
+      "step": 20410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007155377472492527,
+      "loss": 0.5877,
+      "step": 20420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007154953465199602,
+      "loss": 0.5404,
+      "step": 20430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007154529457906677,
+      "loss": 0.6085,
+      "step": 20440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000715410545061375,
+      "loss": 0.6544,
+      "step": 20450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007153681443320825,
+      "loss": 0.7411,
+      "step": 20460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007153257436027901,
+      "loss": 0.6482,
+      "step": 20470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007152833428734974,
+      "loss": 0.5788,
+      "step": 20480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007152409421442049,
+      "loss": 0.5449,
+      "step": 20490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007151985414149124,
+      "loss": 0.6632,
+      "step": 20500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007151561406856198,
+      "loss": 0.5706,
+      "step": 20510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007151137399563273,
+      "loss": 0.6137,
+      "step": 20520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007150713392270348,
+      "loss": 0.6483,
+      "step": 20530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007150289384977421,
+      "loss": 0.501,
+      "step": 20540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007149865377684497,
+      "loss": 0.6377,
+      "step": 20550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007149441370391572,
+      "loss": 0.6234,
+      "step": 20560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007149017363098645,
+      "loss": 0.6803,
+      "step": 20570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000714859335580572,
+      "loss": 0.594,
+      "step": 20580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007148169348512796,
+      "loss": 0.5463,
+      "step": 20590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007147745341219869,
+      "loss": 0.593,
+      "step": 20600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007147321333926944,
+      "loss": 0.4335,
+      "step": 20610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007146897326634018,
+      "loss": 0.6657,
+      "step": 20620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007146473319341093,
+      "loss": 0.6455,
+      "step": 20630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007146049312048168,
+      "loss": 0.5391,
+      "step": 20640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007145625304755242,
+      "loss": 0.6609,
+      "step": 20650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007145201297462316,
+      "loss": 0.6325,
+      "step": 20660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007144777290169392,
+      "loss": 0.5776,
+      "step": 20670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007144353282876466,
+      "loss": 0.6096,
+      "step": 20680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000714392927558354,
+      "loss": 0.7119,
+      "step": 20690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007143505268290615,
+      "loss": 0.6445,
+      "step": 20700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000714308126099769,
+      "loss": 0.6244,
+      "step": 20710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007142657253704764,
+      "loss": 0.6829,
+      "step": 20720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007142233246411839,
+      "loss": 0.5619,
+      "step": 20730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007141809239118912,
+      "loss": 0.5975,
+      "step": 20740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007141385231825988,
+      "loss": 0.5039,
+      "step": 20750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007140961224533063,
+      "loss": 0.6294,
+      "step": 20760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007140537217240136,
+      "loss": 0.6314,
+      "step": 20770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007140113209947211,
+      "loss": 0.5863,
+      "step": 20780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007139689202654286,
+      "loss": 0.6469,
+      "step": 20790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000713926519536136,
+      "loss": 0.6072,
+      "step": 20800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007138841188068435,
+      "loss": 0.6536,
+      "step": 20810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000713841718077551,
+      "loss": 0.6049,
+      "step": 20820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007137993173482584,
+      "loss": 0.5867,
+      "step": 20830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007137569166189659,
+      "loss": 0.5829,
+      "step": 20840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007137145158896734,
+      "loss": 0.6715,
+      "step": 20850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007136721151603807,
+      "loss": 0.6044,
+      "step": 20860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007136297144310883,
+      "loss": 0.6245,
+      "step": 20870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007135873137017958,
+      "loss": 0.5775,
+      "step": 20880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007135449129725031,
+      "loss": 0.6605,
+      "step": 20890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007135025122432106,
+      "loss": 0.6844,
+      "step": 20900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007134601115139181,
+      "loss": 0.6052,
+      "step": 20910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007134177107846255,
+      "loss": 0.6249,
+      "step": 20920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000713375310055333,
+      "loss": 0.6222,
+      "step": 20930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007133329093260404,
+      "loss": 0.6625,
+      "step": 20940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007132905085967479,
+      "loss": 0.691,
+      "step": 20950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007132481078674554,
+      "loss": 0.5479,
+      "step": 20960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007132057071381628,
+      "loss": 0.5582,
+      "step": 20970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007131633064088702,
+      "loss": 0.6459,
+      "step": 20980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007131209056795777,
+      "loss": 0.6081,
+      "step": 20990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007130785049502852,
+      "loss": 0.6229,
+      "step": 21000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007130361042209926,
+      "loss": 0.6742,
+      "step": 21010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007129937034917001,
+      "loss": 0.661,
+      "step": 21020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007129513027624076,
+      "loss": 0.5608,
+      "step": 21030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000712908902033115,
+      "loss": 0.6013,
+      "step": 21040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007128665013038225,
+      "loss": 0.5647,
+      "step": 21050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007128241005745299,
+      "loss": 0.6708,
+      "step": 21060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007127816998452373,
+      "loss": 0.6044,
+      "step": 21070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007127392991159449,
+      "loss": 0.5185,
+      "step": 21080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007126968983866523,
+      "loss": 0.5722,
+      "step": 21090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007126544976573597,
+      "loss": 0.6297,
+      "step": 21100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007126120969280672,
+      "loss": 0.6384,
+      "step": 21110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007125696961987747,
+      "loss": 0.5636,
+      "step": 21120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007125272954694821,
+      "loss": 0.5653,
+      "step": 21130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007124848947401896,
+      "loss": 0.5446,
+      "step": 21140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000712442494010897,
+      "loss": 0.6408,
+      "step": 21150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007124000932816045,
+      "loss": 0.6568,
+      "step": 21160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000712357692552312,
+      "loss": 0.5491,
+      "step": 21170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007123152918230193,
+      "loss": 0.6168,
+      "step": 21180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007122728910937268,
+      "loss": 0.685,
+      "step": 21190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007122304903644344,
+      "loss": 0.5489,
+      "step": 21200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007121880896351417,
+      "loss": 0.5288,
+      "step": 21210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007121456889058492,
+      "loss": 0.6228,
+      "step": 21220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007121032881765566,
+      "loss": 0.6075,
+      "step": 21230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007120608874472641,
+      "loss": 0.6071,
+      "step": 21240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007120184867179716,
+      "loss": 0.6104,
+      "step": 21250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000711976085988679,
+      "loss": 0.6406,
+      "step": 21260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007119336852593864,
+      "loss": 0.6664,
+      "step": 21270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000711891284530094,
+      "loss": 0.6303,
+      "step": 21280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007118488838008014,
+      "loss": 0.5787,
+      "step": 21290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007118064830715088,
+      "loss": 0.5408,
+      "step": 21300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007117640823422163,
+      "loss": 0.6108,
+      "step": 21310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007117216816129238,
+      "loss": 0.5309,
+      "step": 21320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007116792808836312,
+      "loss": 0.6953,
+      "step": 21330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007116368801543387,
+      "loss": 0.6161,
+      "step": 21340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007115944794250461,
+      "loss": 0.6533,
+      "step": 21350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007115520786957536,
+      "loss": 0.715,
+      "step": 21360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007115096779664611,
+      "loss": 0.5497,
+      "step": 21370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007114672772371685,
+      "loss": 0.6655,
+      "step": 21380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007114248765078759,
+      "loss": 0.6049,
+      "step": 21390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007113824757785835,
+      "loss": 0.5993,
+      "step": 21400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007113400750492909,
+      "loss": 0.6398,
+      "step": 21410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007112976743199983,
+      "loss": 0.6148,
+      "step": 21420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007112552735907058,
+      "loss": 0.6681,
+      "step": 21430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007112128728614133,
+      "loss": 0.6382,
+      "step": 21440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007111704721321207,
+      "loss": 0.7087,
+      "step": 21450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007111280714028282,
+      "loss": 0.6605,
+      "step": 21460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007110856706735356,
+      "loss": 0.6312,
+      "step": 21470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007110432699442431,
+      "loss": 0.6083,
+      "step": 21480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007110008692149506,
+      "loss": 0.7642,
+      "step": 21490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000710958468485658,
+      "loss": 0.5539,
+      "step": 21500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007109160677563654,
+      "loss": 0.5903,
+      "step": 21510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007108736670270729,
+      "loss": 0.5887,
+      "step": 21520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007108312662977804,
+      "loss": 0.6,
+      "step": 21530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007107888655684878,
+      "loss": 0.6409,
+      "step": 21540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007107464648391952,
+      "loss": 0.5514,
+      "step": 21550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007107040641099028,
+      "loss": 0.6284,
+      "step": 21560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007106616633806102,
+      "loss": 0.5739,
+      "step": 21570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007106192626513176,
+      "loss": 0.6499,
+      "step": 21580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000710576861922025,
+      "loss": 0.692,
+      "step": 21590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007105344611927325,
+      "loss": 0.6065,
+      "step": 21600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00071049206046344,
+      "loss": 0.6687,
+      "step": 21610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007104496597341475,
+      "loss": 0.6152,
+      "step": 21620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007104072590048549,
+      "loss": 0.6808,
+      "step": 21630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007103648582755624,
+      "loss": 0.707,
+      "step": 21640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007103224575462699,
+      "loss": 0.567,
+      "step": 21650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007102800568169773,
+      "loss": 0.5303,
+      "step": 21660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007102376560876847,
+      "loss": 0.5269,
+      "step": 21670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007101952553583921,
+      "loss": 0.5847,
+      "step": 21680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007101528546290997,
+      "loss": 0.6181,
+      "step": 21690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007101104538998071,
+      "loss": 0.654,
+      "step": 21700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007100680531705145,
+      "loss": 0.6816,
+      "step": 21710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000710025652441222,
+      "loss": 0.651,
+      "step": 21720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007099832517119295,
+      "loss": 0.5812,
+      "step": 21730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000709940850982637,
+      "loss": 0.6924,
+      "step": 21740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007098984502533444,
+      "loss": 0.6462,
+      "step": 21750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007098560495240518,
+      "loss": 0.5935,
+      "step": 21760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007098136487947593,
+      "loss": 0.6244,
+      "step": 21770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007097712480654668,
+      "loss": 0.6925,
+      "step": 21780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007097288473361742,
+      "loss": 0.5978,
+      "step": 21790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007096864466068816,
+      "loss": 0.6391,
+      "step": 21800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007096440458775892,
+      "loss": 0.562,
+      "step": 21810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007096016451482966,
+      "loss": 0.619,
+      "step": 21820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000709559244419004,
+      "loss": 0.6623,
+      "step": 21830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007095168436897115,
+      "loss": 0.6397,
+      "step": 21840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000709474442960419,
+      "loss": 0.6841,
+      "step": 21850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007094320422311264,
+      "loss": 0.5645,
+      "step": 21860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007093896415018338,
+      "loss": 0.5963,
+      "step": 21870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007093472407725413,
+      "loss": 0.5978,
+      "step": 21880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007093048400432488,
+      "loss": 0.708,
+      "step": 21890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007092624393139562,
+      "loss": 0.5719,
+      "step": 21900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007092200385846637,
+      "loss": 0.6186,
+      "step": 21910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007091776378553711,
+      "loss": 0.5051,
+      "step": 21920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007091352371260786,
+      "loss": 0.6176,
+      "step": 21930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007090928363967861,
+      "loss": 0.6242,
+      "step": 21940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007090504356674935,
+      "loss": 0.668,
+      "step": 21950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007090080349382009,
+      "loss": 0.715,
+      "step": 21960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007089656342089085,
+      "loss": 0.584,
+      "step": 21970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007089232334796159,
+      "loss": 0.6717,
+      "step": 21980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007088808327503233,
+      "loss": 0.6739,
+      "step": 21990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007088384320210308,
+      "loss": 0.578,
+      "step": 22000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007087960312917383,
+      "loss": 0.597,
+      "step": 22010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007087536305624457,
+      "loss": 0.6072,
+      "step": 22020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007087112298331532,
+      "loss": 0.5684,
+      "step": 22030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007086688291038606,
+      "loss": 0.5653,
+      "step": 22040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007086264283745681,
+      "loss": 0.6884,
+      "step": 22050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007085840276452756,
+      "loss": 0.7782,
+      "step": 22060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000708541626915983,
+      "loss": 0.6606,
+      "step": 22070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007084992261866904,
+      "loss": 0.6001,
+      "step": 22080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000708456825457398,
+      "loss": 0.5788,
+      "step": 22090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007084144247281054,
+      "loss": 0.773,
+      "step": 22100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007083720239988128,
+      "loss": 0.6497,
+      "step": 22110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007083296232695203,
+      "loss": 0.6063,
+      "step": 22120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007082872225402277,
+      "loss": 0.6154,
+      "step": 22130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007082448218109352,
+      "loss": 0.5729,
+      "step": 22140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007082024210816427,
+      "loss": 0.6894,
+      "step": 22150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00070816002035235,
+      "loss": 0.6178,
+      "step": 22160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007081176196230576,
+      "loss": 0.629,
+      "step": 22170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000708075218893765,
+      "loss": 0.5572,
+      "step": 22180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007080328181644724,
+      "loss": 0.6512,
+      "step": 22190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007079904174351799,
+      "loss": 0.656,
+      "step": 22200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007079480167058873,
+      "loss": 0.6024,
+      "step": 22210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007079056159765948,
+      "loss": 0.6325,
+      "step": 22220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007078632152473023,
+      "loss": 0.6019,
+      "step": 22230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007078208145180097,
+      "loss": 0.6477,
+      "step": 22240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007077784137887172,
+      "loss": 0.5758,
+      "step": 22250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007077360130594247,
+      "loss": 0.6005,
+      "step": 22260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007076936123301321,
+      "loss": 0.6484,
+      "step": 22270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007076512116008395,
+      "loss": 0.6001,
+      "step": 22280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000707608810871547,
+      "loss": 0.6617,
+      "step": 22290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007075664101422545,
+      "loss": 0.6509,
+      "step": 22300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007075240094129619,
+      "loss": 0.6199,
+      "step": 22310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007074816086836694,
+      "loss": 0.5061,
+      "step": 22320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007074392079543768,
+      "loss": 0.5844,
+      "step": 22330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007073968072250843,
+      "loss": 0.6367,
+      "step": 22340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007073544064957918,
+      "loss": 0.5865,
+      "step": 22350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007073120057664992,
+      "loss": 0.6757,
+      "step": 22360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007072696050372067,
+      "loss": 0.6322,
+      "step": 22370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007072272043079142,
+      "loss": 0.5538,
+      "step": 22380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007071848035786216,
+      "loss": 0.5663,
+      "step": 22390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000707142402849329,
+      "loss": 0.5554,
+      "step": 22400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007071000021200365,
+      "loss": 0.5334,
+      "step": 22410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000707057601390744,
+      "loss": 0.6138,
+      "step": 22420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007070152006614514,
+      "loss": 0.6498,
+      "step": 22430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007069727999321589,
+      "loss": 0.637,
+      "step": 22440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007069303992028663,
+      "loss": 0.5299,
+      "step": 22450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007068879984735738,
+      "loss": 0.6525,
+      "step": 22460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007068455977442813,
+      "loss": 0.6105,
+      "step": 22470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007068031970149886,
+      "loss": 0.6262,
+      "step": 22480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007067607962856961,
+      "loss": 0.5891,
+      "step": 22490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007067183955564037,
+      "loss": 0.6563,
+      "step": 22500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000706675994827111,
+      "loss": 0.5711,
+      "step": 22510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007066335940978185,
+      "loss": 0.6365,
+      "step": 22520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000706591193368526,
+      "loss": 0.635,
+      "step": 22530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007065487926392334,
+      "loss": 0.6868,
+      "step": 22540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007065063919099409,
+      "loss": 0.597,
+      "step": 22550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007064639911806484,
+      "loss": 0.5961,
+      "step": 22560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007064215904513557,
+      "loss": 0.6529,
+      "step": 22570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007063791897220633,
+      "loss": 0.5869,
+      "step": 22580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007063367889927708,
+      "loss": 0.6421,
+      "step": 22590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007062943882634781,
+      "loss": 0.705,
+      "step": 22600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007062519875341856,
+      "loss": 0.6322,
+      "step": 22610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007062095868048932,
+      "loss": 0.5739,
+      "step": 22620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007061671860756005,
+      "loss": 0.6547,
+      "step": 22630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000706124785346308,
+      "loss": 0.6634,
+      "step": 22640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007060823846170154,
+      "loss": 0.5673,
+      "step": 22650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007060399838877229,
+      "loss": 0.536,
+      "step": 22660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007059975831584304,
+      "loss": 0.5347,
+      "step": 22670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007059551824291378,
+      "loss": 0.5288,
+      "step": 22680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007059127816998452,
+      "loss": 0.5419,
+      "step": 22690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007058703809705528,
+      "loss": 0.6644,
+      "step": 22700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007058279802412602,
+      "loss": 0.6728,
+      "step": 22710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007057855795119676,
+      "loss": 0.6428,
+      "step": 22720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007057431787826751,
+      "loss": 0.651,
+      "step": 22730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007057007780533825,
+      "loss": 0.6546,
+      "step": 22740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00070565837732409,
+      "loss": 0.6206,
+      "step": 22750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007056159765947975,
+      "loss": 0.5655,
+      "step": 22760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007055735758655048,
+      "loss": 0.6946,
+      "step": 22770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007055311751362124,
+      "loss": 0.6061,
+      "step": 22780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007054887744069199,
+      "loss": 0.6721,
+      "step": 22790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007054463736776272,
+      "loss": 0.5201,
+      "step": 22800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007054039729483347,
+      "loss": 0.649,
+      "step": 22810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007053615722190423,
+      "loss": 0.6368,
+      "step": 22820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007053191714897496,
+      "loss": 0.6428,
+      "step": 22830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007052767707604571,
+      "loss": 0.6092,
+      "step": 22840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007052343700311646,
+      "loss": 0.6464,
+      "step": 22850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000705191969301872,
+      "loss": 0.6384,
+      "step": 22860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007051495685725795,
+      "loss": 0.5484,
+      "step": 22870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000705107167843287,
+      "loss": 0.6678,
+      "step": 22880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007050647671139943,
+      "loss": 0.598,
+      "step": 22890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007050223663847019,
+      "loss": 0.5829,
+      "step": 22900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007049799656554094,
+      "loss": 0.5523,
+      "step": 22910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007049375649261167,
+      "loss": 0.7557,
+      "step": 22920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007048951641968242,
+      "loss": 0.6861,
+      "step": 22930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007048527634675317,
+      "loss": 0.6454,
+      "step": 22940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007048103627382391,
+      "loss": 0.6413,
+      "step": 22950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007047679620089466,
+      "loss": 0.709,
+      "step": 22960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007047255612796541,
+      "loss": 0.6126,
+      "step": 22970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007046831605503615,
+      "loss": 0.5864,
+      "step": 22980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000704640759821069,
+      "loss": 0.5941,
+      "step": 22990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007045983590917765,
+      "loss": 0.666,
+      "step": 23000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007045559583624838,
+      "loss": 0.5371,
+      "step": 23010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007045135576331913,
+      "loss": 0.5663,
+      "step": 23020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007044711569038989,
+      "loss": 0.6226,
+      "step": 23030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007044287561746062,
+      "loss": 0.6017,
+      "step": 23040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007043863554453137,
+      "loss": 0.6189,
+      "step": 23050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007043439547160212,
+      "loss": 0.6393,
+      "step": 23060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007043015539867286,
+      "loss": 0.6345,
+      "step": 23070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007042591532574361,
+      "loss": 0.6487,
+      "step": 23080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007042167525281434,
+      "loss": 0.5666,
+      "step": 23090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007041743517988509,
+      "loss": 0.6711,
+      "step": 23100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007041319510695585,
+      "loss": 0.5355,
+      "step": 23110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007040895503402658,
+      "loss": 0.6456,
+      "step": 23120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007040471496109733,
+      "loss": 0.6645,
+      "step": 23130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007040047488816808,
+      "loss": 0.7074,
+      "step": 23140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007039623481523882,
+      "loss": 0.6012,
+      "step": 23150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007039199474230957,
+      "loss": 0.7067,
+      "step": 23160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007038775466938032,
+      "loss": 0.6977,
+      "step": 23170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007038351459645105,
+      "loss": 0.6459,
+      "step": 23180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007037927452352181,
+      "loss": 0.6519,
+      "step": 23190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007037503445059256,
+      "loss": 0.6257,
+      "step": 23200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007037079437766329,
+      "loss": 0.6431,
+      "step": 23210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007036655430473404,
+      "loss": 0.6272,
+      "step": 23220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000703623142318048,
+      "loss": 0.653,
+      "step": 23230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007035807415887553,
+      "loss": 0.5919,
+      "step": 23240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007035383408594628,
+      "loss": 0.5866,
+      "step": 23250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007034959401301703,
+      "loss": 0.6435,
+      "step": 23260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007034535394008777,
+      "loss": 0.6517,
+      "step": 23270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007034111386715852,
+      "loss": 0.6048,
+      "step": 23280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007033687379422927,
+      "loss": 0.593,
+      "step": 23290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000703326337213,
+      "loss": 0.6295,
+      "step": 23300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007032839364837076,
+      "loss": 0.6414,
+      "step": 23310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007032415357544151,
+      "loss": 0.6102,
+      "step": 23320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007031991350251224,
+      "loss": 0.6823,
+      "step": 23330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007031567342958299,
+      "loss": 0.5635,
+      "step": 23340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007031143335665375,
+      "loss": 0.691,
+      "step": 23350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007030719328372448,
+      "loss": 0.6135,
+      "step": 23360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007030295321079523,
+      "loss": 0.6744,
+      "step": 23370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007029871313786597,
+      "loss": 0.5247,
+      "step": 23380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007029447306493672,
+      "loss": 0.6435,
+      "step": 23390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007029023299200747,
+      "loss": 0.6787,
+      "step": 23400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007028599291907821,
+      "loss": 0.6846,
+      "step": 23410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007028175284614895,
+      "loss": 0.7162,
+      "step": 23420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007027751277321971,
+      "loss": 0.6054,
+      "step": 23430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007027327270029045,
+      "loss": 0.652,
+      "step": 23440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007026903262736119,
+      "loss": 0.5605,
+      "step": 23450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007026479255443194,
+      "loss": 0.6684,
+      "step": 23460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007026055248150269,
+      "loss": 0.637,
+      "step": 23470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007025631240857343,
+      "loss": 0.5821,
+      "step": 23480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007025207233564418,
+      "loss": 0.5681,
+      "step": 23490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007024783226271491,
+      "loss": 0.5773,
+      "step": 23500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007024359218978567,
+      "loss": 0.5677,
+      "step": 23510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007023935211685642,
+      "loss": 0.6691,
+      "step": 23520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007023511204392716,
+      "loss": 0.6414,
+      "step": 23530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000702308719709979,
+      "loss": 0.6296,
+      "step": 23540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007022663189806865,
+      "loss": 0.4861,
+      "step": 23550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000702223918251394,
+      "loss": 0.5755,
+      "step": 23560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007021815175221014,
+      "loss": 0.6427,
+      "step": 23570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007021391167928089,
+      "loss": 0.527,
+      "step": 23580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007020967160635164,
+      "loss": 0.6524,
+      "step": 23590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007020543153342238,
+      "loss": 0.781,
+      "step": 23600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007020119146049313,
+      "loss": 0.5479,
+      "step": 23610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007019695138756386,
+      "loss": 0.5733,
+      "step": 23620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007019271131463461,
+      "loss": 0.6175,
+      "step": 23630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007018847124170537,
+      "loss": 0.5407,
+      "step": 23640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000701842311687761,
+      "loss": 0.646,
+      "step": 23650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007017999109584685,
+      "loss": 0.5768,
+      "step": 23660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000701757510229176,
+      "loss": 0.5123,
+      "step": 23670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007017151094998834,
+      "loss": 0.5601,
+      "step": 23680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007016727087705909,
+      "loss": 0.6888,
+      "step": 23690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007016303080412983,
+      "loss": 0.518,
+      "step": 23700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007015879073120057,
+      "loss": 0.4827,
+      "step": 23710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007015455065827133,
+      "loss": 0.6067,
+      "step": 23720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007015031058534207,
+      "loss": 0.5729,
+      "step": 23730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007014607051241281,
+      "loss": 0.5277,
+      "step": 23740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007014183043948356,
+      "loss": 0.6001,
+      "step": 23750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007013759036655431,
+      "loss": 0.5276,
+      "step": 23760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007013335029362505,
+      "loss": 0.6003,
+      "step": 23770
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000701291102206958,
+      "loss": 0.469,
+      "step": 23780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007012487014776654,
+      "loss": 0.6175,
+      "step": 23790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007012063007483729,
+      "loss": 0.6383,
+      "step": 23800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007011639000190804,
+      "loss": 0.6678,
+      "step": 23810
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007011214992897878,
+      "loss": 0.5253,
+      "step": 23820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007010790985604952,
+      "loss": 0.5248,
+      "step": 23830
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007010366978312028,
+      "loss": 0.5087,
+      "step": 23840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007009942971019102,
+      "loss": 0.5807,
+      "step": 23850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007009518963726176,
+      "loss": 0.5219,
+      "step": 23860
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007009094956433251,
+      "loss": 0.5977,
+      "step": 23870
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007008670949140326,
+      "loss": 0.5786,
+      "step": 23880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00070082469418474,
+      "loss": 0.5504,
+      "step": 23890
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007007822934554475,
+      "loss": 0.6085,
+      "step": 23900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007007398927261549,
+      "loss": 0.5364,
+      "step": 23910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007006974919968624,
+      "loss": 0.5677,
+      "step": 23920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007006550912675699,
+      "loss": 0.6672,
+      "step": 23930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007006126905382773,
+      "loss": 0.514,
+      "step": 23940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007005702898089847,
+      "loss": 0.4747,
+      "step": 23950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007005278890796923,
+      "loss": 0.5778,
+      "step": 23960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007004854883503997,
+      "loss": 0.5654,
+      "step": 23970
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007004430876211071,
+      "loss": 0.5595,
+      "step": 23980
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007004006868918146,
+      "loss": 0.618,
+      "step": 23990
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007003582861625221,
+      "loss": 0.6047,
+      "step": 24000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007003158854332295,
+      "loss": 0.5487,
+      "step": 24010
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007002734847039369,
+      "loss": 0.5975,
+      "step": 24020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007002310839746443,
+      "loss": 0.6538,
+      "step": 24030
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007001886832453519,
+      "loss": 0.5883,
+      "step": 24040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007001462825160593,
+      "loss": 0.5701,
+      "step": 24050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007001038817867667,
+      "loss": 0.7498,
+      "step": 24060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007000614810574742,
+      "loss": 0.5149,
+      "step": 24070
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007000190803281817,
+      "loss": 0.5168,
+      "step": 24080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006999766795988891,
+      "loss": 0.556,
+      "step": 24090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006999342788695966,
+      "loss": 0.5341,
+      "step": 24100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000699891878140304,
+      "loss": 0.5227,
+      "step": 24110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006998494774110115,
+      "loss": 0.556,
+      "step": 24120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000699807076681719,
+      "loss": 0.5245,
+      "step": 24130
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006997646759524264,
+      "loss": 0.635,
+      "step": 24140
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006997222752231338,
+      "loss": 0.535,
+      "step": 24150
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006996798744938413,
+      "loss": 0.5778,
+      "step": 24160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006996374737645488,
+      "loss": 0.5199,
+      "step": 24170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006995950730352562,
+      "loss": 0.6748,
+      "step": 24180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006995526723059637,
+      "loss": 0.5759,
+      "step": 24190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006995102715766712,
+      "loss": 0.5207,
+      "step": 24200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006994678708473786,
+      "loss": 0.6965,
+      "step": 24210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006994254701180861,
+      "loss": 0.5454,
+      "step": 24220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0006993830693887935,
+      "loss": 0.5924,
+      "step": 24230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006993406686595009,
+      "loss": 0.622,
+      "step": 24240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006992982679302085,
+      "loss": 0.6326,
+      "step": 24250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006992558672009159,
+      "loss": 0.6048,
+      "step": 24260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006992134664716233,
+      "loss": 0.5181,
+      "step": 24270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006991710657423308,
+      "loss": 0.6171,
+      "step": 24280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006991286650130383,
+      "loss": 0.5832,
+      "step": 24290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006990862642837457,
+      "loss": 0.5336,
+      "step": 24300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006990438635544531,
+      "loss": 0.5098,
+      "step": 24310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006990014628251606,
+      "loss": 0.5386,
+      "step": 24320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006989590620958681,
+      "loss": 0.5651,
+      "step": 24330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006989166613665755,
+      "loss": 0.5576,
+      "step": 24340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000698874260637283,
+      "loss": 0.577,
+      "step": 24350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006988318599079904,
+      "loss": 0.5504,
+      "step": 24360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006987894591786979,
+      "loss": 0.5514,
+      "step": 24370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006987470584494054,
+      "loss": 0.5736,
+      "step": 24380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006987046577201128,
+      "loss": 0.5235,
+      "step": 24390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006986622569908203,
+      "loss": 0.5624,
+      "step": 24400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006986198562615278,
+      "loss": 0.5762,
+      "step": 24410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006985774555322352,
+      "loss": 0.552,
+      "step": 24420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006985350548029426,
+      "loss": 0.5433,
+      "step": 24430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00069849265407365,
+      "loss": 0.6427,
+      "step": 24440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006984502533443576,
+      "loss": 0.5829,
+      "step": 24450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000698407852615065,
+      "loss": 0.546,
+      "step": 24460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0006983654518857725,
+      "loss": 0.6099,
+      "step": 24470
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006983230511564799,
+      "loss": 0.5575,
+      "step": 24480
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006982806504271874,
+      "loss": 0.5227,
+      "step": 24490
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006982382496978949,
+      "loss": 0.605,
+      "step": 24500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006981958489686023,
+      "loss": 0.5974,
+      "step": 24510
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006981534482393097,
+      "loss": 0.5444,
+      "step": 24520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006981110475100173,
+      "loss": 0.5775,
+      "step": 24530
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006980686467807247,
+      "loss": 0.5505,
+      "step": 24540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006980262460514321,
+      "loss": 0.5253,
+      "step": 24550
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006979838453221395,
+      "loss": 0.4977,
+      "step": 24560
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006979414445928471,
+      "loss": 0.5343,
+      "step": 24570
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006978990438635545,
+      "loss": 0.5587,
+      "step": 24580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006978566431342619,
+      "loss": 0.5814,
+      "step": 24590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006978142424049694,
+      "loss": 0.5183,
+      "step": 24600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006977718416756769,
+      "loss": 0.5714,
+      "step": 24610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006977294409463843,
+      "loss": 0.5093,
+      "step": 24620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006976870402170917,
+      "loss": 0.5629,
+      "step": 24630
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006976446394877992,
+      "loss": 0.6162,
+      "step": 24640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006976022387585067,
+      "loss": 0.5614,
+      "step": 24650
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006975598380292141,
+      "loss": 0.6527,
+      "step": 24660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006975174372999216,
+      "loss": 0.6788,
+      "step": 24670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000697475036570629,
+      "loss": 0.5523,
+      "step": 24680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006974326358413365,
+      "loss": 0.6047,
+      "step": 24690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000697390235112044,
+      "loss": 0.5696,
+      "step": 24700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006973478343827514,
+      "loss": 0.6413,
+      "step": 24710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006973054336534588,
+      "loss": 0.5421,
+      "step": 24720
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006972630329241664,
+      "loss": 0.5973,
+      "step": 24730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006972206321948738,
+      "loss": 0.6337,
+      "step": 24740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006971782314655812,
+      "loss": 0.5656,
+      "step": 24750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006971358307362887,
+      "loss": 0.5187,
+      "step": 24760
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006970934300069961,
+      "loss": 0.5453,
+      "step": 24770
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006970510292777036,
+      "loss": 0.5258,
+      "step": 24780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006970086285484111,
+      "loss": 0.5629,
+      "step": 24790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006969662278191185,
+      "loss": 0.5342,
+      "step": 24800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000696923827089826,
+      "loss": 0.6197,
+      "step": 24810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006968814263605335,
+      "loss": 0.6046,
+      "step": 24820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006968390256312409,
+      "loss": 0.4763,
+      "step": 24830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006967966249019483,
+      "loss": 0.6106,
+      "step": 24840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006967542241726559,
+      "loss": 0.6816,
+      "step": 24850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006967118234433633,
+      "loss": 0.5438,
+      "step": 24860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006966694227140707,
+      "loss": 0.6479,
+      "step": 24870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006966270219847782,
+      "loss": 0.5023,
+      "step": 24880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006965846212554856,
+      "loss": 0.5059,
+      "step": 24890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006965422205261931,
+      "loss": 0.5901,
+      "step": 24900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006964998197969006,
+      "loss": 0.6659,
+      "step": 24910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006964574190676079,
+      "loss": 0.6221,
+      "step": 24920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006964150183383155,
+      "loss": 0.6237,
+      "step": 24930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000696372617609023,
+      "loss": 0.6555,
+      "step": 24940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006963302168797303,
+      "loss": 0.7002,
+      "step": 24950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006962878161504378,
+      "loss": 0.649,
+      "step": 24960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006962454154211453,
+      "loss": 0.6351,
+      "step": 24970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006962030146918527,
+      "loss": 0.5112,
+      "step": 24980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006961606139625602,
+      "loss": 0.5888,
+      "step": 24990
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006961182132332677,
+      "loss": 0.6111,
+      "step": 25000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006960758125039751,
+      "loss": 0.5891,
+      "step": 25010
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006960334117746826,
+      "loss": 0.6024,
+      "step": 25020
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00069599101104539,
+      "loss": 0.5441,
+      "step": 25030
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006959486103160974,
+      "loss": 0.5834,
+      "step": 25040
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006959062095868049,
+      "loss": 0.5977,
+      "step": 25050
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006958638088575125,
+      "loss": 0.5637,
+      "step": 25060
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006958214081282198,
+      "loss": 0.6074,
+      "step": 25070
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006957790073989273,
+      "loss": 0.6314,
+      "step": 25080
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006957366066696347,
+      "loss": 0.6009,
+      "step": 25090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006956942059403422,
+      "loss": 0.6261,
+      "step": 25100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006956518052110497,
+      "loss": 0.6088,
+      "step": 25110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006956094044817571,
+      "loss": 0.529,
+      "step": 25120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006955670037524645,
+      "loss": 0.5781,
+      "step": 25130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006955246030231721,
+      "loss": 0.5641,
+      "step": 25140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006954822022938795,
+      "loss": 0.6374,
+      "step": 25150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006954398015645869,
+      "loss": 0.6331,
+      "step": 25160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006953974008352944,
+      "loss": 0.5078,
+      "step": 25170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006953550001060019,
+      "loss": 0.6404,
+      "step": 25180
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006953125993767093,
+      "loss": 0.5963,
+      "step": 25190
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006952701986474168,
+      "loss": 0.5921,
+      "step": 25200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006952277979181242,
+      "loss": 0.6606,
+      "step": 25210
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006951853971888317,
+      "loss": 0.6495,
+      "step": 25220
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006951429964595392,
+      "loss": 0.5645,
+      "step": 25230
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006951005957302465,
+      "loss": 0.5183,
+      "step": 25240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000695058195000954,
+      "loss": 0.5287,
+      "step": 25250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006950157942716616,
+      "loss": 0.5061,
+      "step": 25260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006949733935423689,
+      "loss": 0.6056,
+      "step": 25270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006949309928130764,
+      "loss": 0.6307,
+      "step": 25280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006948885920837839,
+      "loss": 0.5398,
+      "step": 25290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006948461913544913,
+      "loss": 0.5338,
+      "step": 25300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006948037906251988,
+      "loss": 0.5989,
+      "step": 25310
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006947613898959063,
+      "loss": 0.58,
+      "step": 25320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006947189891666136,
+      "loss": 0.6035,
+      "step": 25330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006946765884373212,
+      "loss": 0.5323,
+      "step": 25340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006946341877080287,
+      "loss": 0.5299,
+      "step": 25350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000694591786978736,
+      "loss": 0.5456,
+      "step": 25360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006945493862494435,
+      "loss": 0.5729,
+      "step": 25370
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006945069855201511,
+      "loss": 0.5174,
+      "step": 25380
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006944645847908584,
+      "loss": 0.6261,
+      "step": 25390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006944221840615659,
+      "loss": 0.7251,
+      "step": 25400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006943797833322734,
+      "loss": 0.5504,
+      "step": 25410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006943373826029808,
+      "loss": 0.5738,
+      "step": 25420
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006942949818736883,
+      "loss": 0.5357,
+      "step": 25430
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006942525811443958,
+      "loss": 0.6319,
+      "step": 25440
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006942101804151031,
+      "loss": 0.5715,
+      "step": 25450
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006941677796858107,
+      "loss": 0.6966,
+      "step": 25460
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006941253789565182,
+      "loss": 0.6355,
+      "step": 25470
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006940829782272255,
+      "loss": 0.5878,
+      "step": 25480
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000694040577497933,
+      "loss": 0.5854,
+      "step": 25490
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006939981767686404,
+      "loss": 0.552,
+      "step": 25500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006939557760393479,
+      "loss": 0.5348,
+      "step": 25510
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006939133753100554,
+      "loss": 0.5801,
+      "step": 25520
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006938709745807627,
+      "loss": 0.4255,
+      "step": 25530
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006938285738514703,
+      "loss": 0.535,
+      "step": 25540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006937861731221778,
+      "loss": 0.5846,
+      "step": 25550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006937437723928851,
+      "loss": 0.5823,
+      "step": 25560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006937013716635926,
+      "loss": 0.671,
+      "step": 25570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006936589709343001,
+      "loss": 0.574,
+      "step": 25580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006936165702050075,
+      "loss": 0.6175,
+      "step": 25590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000693574169475715,
+      "loss": 0.5127,
+      "step": 25600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006935317687464225,
+      "loss": 0.5691,
+      "step": 25610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006934893680171299,
+      "loss": 0.6279,
+      "step": 25620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006934469672878374,
+      "loss": 0.5667,
+      "step": 25630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006934045665585449,
+      "loss": 0.5849,
+      "step": 25640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006933621658292522,
+      "loss": 0.5885,
+      "step": 25650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006933197650999597,
+      "loss": 0.6158,
+      "step": 25660
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006932773643706673,
+      "loss": 0.5853,
+      "step": 25670
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006932349636413746,
+      "loss": 0.5683,
+      "step": 25680
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006931925629120821,
+      "loss": 0.6243,
+      "step": 25690
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006931501621827896,
+      "loss": 0.6842,
+      "step": 25700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000693107761453497,
+      "loss": 0.5601,
+      "step": 25710
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006930653607242045,
+      "loss": 0.5747,
+      "step": 25720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000693022959994912,
+      "loss": 0.4747,
+      "step": 25730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006929805592656193,
+      "loss": 0.6375,
+      "step": 25740
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006929381585363269,
+      "loss": 0.4729,
+      "step": 25750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006928957578070344,
+      "loss": 0.5623,
+      "step": 25760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006928533570777417,
+      "loss": 0.5665,
+      "step": 25770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006928109563484492,
+      "loss": 0.4765,
+      "step": 25780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006927685556191568,
+      "loss": 0.584,
+      "step": 25790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006927261548898641,
+      "loss": 0.548,
+      "step": 25800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006926837541605716,
+      "loss": 0.5884,
+      "step": 25810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006926413534312791,
+      "loss": 0.6063,
+      "step": 25820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006925989527019865,
+      "loss": 0.5335,
+      "step": 25830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000692556551972694,
+      "loss": 0.5116,
+      "step": 25840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006925141512434014,
+      "loss": 0.5986,
+      "step": 25850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006924717505141088,
+      "loss": 0.5328,
+      "step": 25860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006924293497848164,
+      "loss": 0.5848,
+      "step": 25870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006923869490555238,
+      "loss": 0.6491,
+      "step": 25880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006923445483262312,
+      "loss": 0.5726,
+      "step": 25890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006923021475969387,
+      "loss": 0.5517,
+      "step": 25900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006922597468676462,
+      "loss": 0.5889,
+      "step": 25910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006922173461383536,
+      "loss": 0.6953,
+      "step": 25920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006921749454090611,
+      "loss": 0.5754,
+      "step": 25930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006921325446797684,
+      "loss": 0.5994,
+      "step": 25940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000692090143950476,
+      "loss": 0.4993,
+      "step": 25950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006920477432211835,
+      "loss": 0.5768,
+      "step": 25960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006920053424918908,
+      "loss": 0.5483,
+      "step": 25970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006919629417625983,
+      "loss": 0.697,
+      "step": 25980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006919205410333059,
+      "loss": 0.6415,
+      "step": 25990
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006918781403040132,
+      "loss": 0.52,
+      "step": 26000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006918357395747207,
+      "loss": 0.6452,
+      "step": 26010
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006917933388454282,
+      "loss": 0.6071,
+      "step": 26020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006917509381161356,
+      "loss": 0.5752,
+      "step": 26030
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006917085373868431,
+      "loss": 0.5113,
+      "step": 26040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006916661366575506,
+      "loss": 0.6,
+      "step": 26050
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006916237359282579,
+      "loss": 0.5362,
+      "step": 26060
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006915813351989655,
+      "loss": 0.6591,
+      "step": 26070
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000691538934469673,
+      "loss": 0.5323,
+      "step": 26080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006914965337403803,
+      "loss": 0.5769,
+      "step": 26090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006914541330110878,
+      "loss": 0.5578,
+      "step": 26100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006914117322817953,
+      "loss": 0.5462,
+      "step": 26110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006913693315525027,
+      "loss": 0.6396,
+      "step": 26120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006913269308232102,
+      "loss": 0.6143,
+      "step": 26130
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006912845300939177,
+      "loss": 0.5494,
+      "step": 26140
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006912421293646251,
+      "loss": 0.6032,
+      "step": 26150
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006911997286353326,
+      "loss": 0.6202,
+      "step": 26160
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00069115732790604,
+      "loss": 0.6222,
+      "step": 26170
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006911149271767474,
+      "loss": 0.5147,
+      "step": 26180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006910725264474549,
+      "loss": 0.5999,
+      "step": 26190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006910301257181624,
+      "loss": 0.6171,
+      "step": 26200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006909877249888698,
+      "loss": 0.5012,
+      "step": 26210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006909453242595773,
+      "loss": 0.5372,
+      "step": 26220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006909029235302848,
+      "loss": 0.5925,
+      "step": 26230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006908605228009922,
+      "loss": 0.5717,
+      "step": 26240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006908181220716997,
+      "loss": 0.5678,
+      "step": 26250
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006907757213424071,
+      "loss": 0.5222,
+      "step": 26260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006907333206131145,
+      "loss": 0.6088,
+      "step": 26270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006906909198838221,
+      "loss": 0.5573,
+      "step": 26280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006906485191545295,
+      "loss": 0.6007,
+      "step": 26290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006906061184252369,
+      "loss": 0.606,
+      "step": 26300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006905637176959444,
+      "loss": 0.4931,
+      "step": 26310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006905213169666519,
+      "loss": 0.5421,
+      "step": 26320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006904789162373593,
+      "loss": 0.6568,
+      "step": 26330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006904365155080668,
+      "loss": 0.4719,
+      "step": 26340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006903941147787743,
+      "loss": 0.5117,
+      "step": 26350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006903517140494817,
+      "loss": 0.5725,
+      "step": 26360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006903093133201892,
+      "loss": 0.6667,
+      "step": 26370
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006902669125908965,
+      "loss": 0.6083,
+      "step": 26380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000690224511861604,
+      "loss": 0.5744,
+      "step": 26390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006901821111323116,
+      "loss": 0.5125,
+      "step": 26400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000690139710403019,
+      "loss": 0.565,
+      "step": 26410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006900973096737264,
+      "loss": 0.6076,
+      "step": 26420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006900549089444339,
+      "loss": 0.6309,
+      "step": 26430
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006900125082151414,
+      "loss": 0.5827,
+      "step": 26440
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006899701074858488,
+      "loss": 0.641,
+      "step": 26450
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006899277067565562,
+      "loss": 0.4923,
+      "step": 26460
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006898853060272636,
+      "loss": 0.4364,
+      "step": 26470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006898429052979712,
+      "loss": 0.5288,
+      "step": 26480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006898005045686786,
+      "loss": 0.519,
+      "step": 26490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000689758103839386,
+      "loss": 0.5937,
+      "step": 26500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006897157031100935,
+      "loss": 0.6305,
+      "step": 26510
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000689673302380801,
+      "loss": 0.649,
+      "step": 26520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006896309016515084,
+      "loss": 0.5124,
+      "step": 26530
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006895885009222159,
+      "loss": 0.4449,
+      "step": 26540
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006895461001929233,
+      "loss": 0.5322,
+      "step": 26550
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006895036994636308,
+      "loss": 0.5477,
+      "step": 26560
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006894612987343383,
+      "loss": 0.598,
+      "step": 26570
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006894188980050457,
+      "loss": 0.5701,
+      "step": 26580
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006893764972757531,
+      "loss": 0.5802,
+      "step": 26590
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006893340965464607,
+      "loss": 0.5165,
+      "step": 26600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006892916958171681,
+      "loss": 0.5221,
+      "step": 26610
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006892492950878755,
+      "loss": 0.6323,
+      "step": 26620
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000689206894358583,
+      "loss": 0.6157,
+      "step": 26630
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006891644936292905,
+      "loss": 0.5799,
+      "step": 26640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006891220928999979,
+      "loss": 0.5535,
+      "step": 26650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006890796921707054,
+      "loss": 0.7069,
+      "step": 26660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006890372914414128,
+      "loss": 0.6223,
+      "step": 26670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006889948907121203,
+      "loss": 0.4631,
+      "step": 26680
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006889524899828278,
+      "loss": 0.5508,
+      "step": 26690
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006889100892535352,
+      "loss": 0.6021,
+      "step": 26700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006888676885242426,
+      "loss": 0.479,
+      "step": 26710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006888252877949501,
+      "loss": 0.5727,
+      "step": 26720
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006887828870656576,
+      "loss": 0.5909,
+      "step": 26730
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000688740486336365,
+      "loss": 0.4967,
+      "step": 26740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006886980856070725,
+      "loss": 0.6219,
+      "step": 26750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00068865568487778,
+      "loss": 0.6883,
+      "step": 26760
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006886132841484874,
+      "loss": 0.6034,
+      "step": 26770
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006885708834191948,
+      "loss": 0.5622,
+      "step": 26780
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006885284826899023,
+      "loss": 0.5401,
+      "step": 26790
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006884860819606097,
+      "loss": 0.6162,
+      "step": 26800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006884436812313172,
+      "loss": 0.6447,
+      "step": 26810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006884012805020247,
+      "loss": 0.5589,
+      "step": 26820
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006883588797727321,
+      "loss": 0.6537,
+      "step": 26830
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006883164790434396,
+      "loss": 0.6269,
+      "step": 26840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006882740783141471,
+      "loss": 0.531,
+      "step": 26850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006882316775848545,
+      "loss": 0.5767,
+      "step": 26860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006881892768555619,
+      "loss": 0.6455,
+      "step": 26870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006881468761262695,
+      "loss": 0.5292,
+      "step": 26880
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006881044753969769,
+      "loss": 0.5397,
+      "step": 26890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006880620746676843,
+      "loss": 0.494,
+      "step": 26900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006880196739383917,
+      "loss": 0.4578,
+      "step": 26910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006879772732090992,
+      "loss": 0.5656,
+      "step": 26920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006879348724798067,
+      "loss": 0.5726,
+      "step": 26930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006878924717505141,
+      "loss": 0.612,
+      "step": 26940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006878500710212216,
+      "loss": 0.5796,
+      "step": 26950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006878076702919291,
+      "loss": 0.5246,
+      "step": 26960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006877652695626365,
+      "loss": 0.5426,
+      "step": 26970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000687722868833344,
+      "loss": 0.5193,
+      "step": 26980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006876804681040514,
+      "loss": 0.5662,
+      "step": 26990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006876380673747588,
+      "loss": 0.5151,
+      "step": 27000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006875956666454664,
+      "loss": 0.5513,
+      "step": 27010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006875532659161738,
+      "loss": 0.5754,
+      "step": 27020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006875108651868812,
+      "loss": 0.5476,
+      "step": 27030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006874684644575887,
+      "loss": 0.6118,
+      "step": 27040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006874260637282962,
+      "loss": 0.6001,
+      "step": 27050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006873836629990036,
+      "loss": 0.6224,
+      "step": 27060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000687341262269711,
+      "loss": 0.6493,
+      "step": 27070
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006872988615404185,
+      "loss": 0.5517,
+      "step": 27080
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000687256460811126,
+      "loss": 0.4907,
+      "step": 27090
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006872140600818334,
+      "loss": 0.5035,
+      "step": 27100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006871716593525409,
+      "loss": 0.5734,
+      "step": 27110
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006871292586232483,
+      "loss": 0.5609,
+      "step": 27120
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006870868578939558,
+      "loss": 0.5619,
+      "step": 27130
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006870444571646633,
+      "loss": 0.6514,
+      "step": 27140
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006870020564353707,
+      "loss": 0.5676,
+      "step": 27150
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006869596557060781,
+      "loss": 0.5087,
+      "step": 27160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006869172549767857,
+      "loss": 0.5278,
+      "step": 27170
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006868748542474931,
+      "loss": 0.5596,
+      "step": 27180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006868324535182005,
+      "loss": 0.4936,
+      "step": 27190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000686790052788908,
+      "loss": 0.5692,
+      "step": 27200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006867476520596155,
+      "loss": 0.6747,
+      "step": 27210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006867052513303229,
+      "loss": 0.6599,
+      "step": 27220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006866628506010304,
+      "loss": 0.6341,
+      "step": 27230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006866204498717378,
+      "loss": 0.5822,
+      "step": 27240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006865780491424453,
+      "loss": 0.5721,
+      "step": 27250
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006865356484131528,
+      "loss": 0.6001,
+      "step": 27260
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006864932476838602,
+      "loss": 0.6228,
+      "step": 27270
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006864508469545676,
+      "loss": 0.5955,
+      "step": 27280
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006864084462252752,
+      "loss": 0.4978,
+      "step": 27290
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006863660454959826,
+      "loss": 0.5665,
+      "step": 27300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00068632364476669,
+      "loss": 0.7129,
+      "step": 27310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006862812440373975,
+      "loss": 0.5458,
+      "step": 27320
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000686238843308105,
+      "loss": 0.5993,
+      "step": 27330
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006861964425788124,
+      "loss": 0.5737,
+      "step": 27340
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006861540418495199,
+      "loss": 0.6236,
+      "step": 27350
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006861116411202273,
+      "loss": 0.5938,
+      "step": 27360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006860692403909348,
+      "loss": 0.6408,
+      "step": 27370
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006860268396616423,
+      "loss": 0.5185,
+      "step": 27380
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006859844389323496,
+      "loss": 0.5327,
+      "step": 27390
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006859420382030571,
+      "loss": 0.6654,
+      "step": 27400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006858996374737647,
+      "loss": 0.5416,
+      "step": 27410
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000685857236744472,
+      "loss": 0.6141,
+      "step": 27420
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006858148360151795,
+      "loss": 0.5844,
+      "step": 27430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006857724352858869,
+      "loss": 0.6177,
+      "step": 27440
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006857300345565944,
+      "loss": 0.5986,
+      "step": 27450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006856876338273019,
+      "loss": 0.576,
+      "step": 27460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006856452330980093,
+      "loss": 0.5784,
+      "step": 27470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006856028323687167,
+      "loss": 0.6798,
+      "step": 27480
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006855604316394243,
+      "loss": 0.5377,
+      "step": 27490
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006855180309101317,
+      "loss": 0.5427,
+      "step": 27500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006854756301808391,
+      "loss": 0.6156,
+      "step": 27510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006854332294515466,
+      "loss": 0.5572,
+      "step": 27520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000685390828722254,
+      "loss": 0.6011,
+      "step": 27530
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006853484279929615,
+      "loss": 0.5765,
+      "step": 27540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000685306027263669,
+      "loss": 0.5492,
+      "step": 27550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006852636265343764,
+      "loss": 0.7331,
+      "step": 27560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006852212258050839,
+      "loss": 0.608,
+      "step": 27570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006851788250757914,
+      "loss": 0.5258,
+      "step": 27580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006851364243464988,
+      "loss": 0.5885,
+      "step": 27590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006850940236172062,
+      "loss": 0.5075,
+      "step": 27600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006850516228879137,
+      "loss": 0.5575,
+      "step": 27610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006850092221586212,
+      "loss": 0.5651,
+      "step": 27620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006849668214293286,
+      "loss": 0.5876,
+      "step": 27630
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006849244207000361,
+      "loss": 0.5443,
+      "step": 27640
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006848820199707435,
+      "loss": 0.7017,
+      "step": 27650
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000684839619241451,
+      "loss": 0.5877,
+      "step": 27660
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006847972185121585,
+      "loss": 0.5473,
+      "step": 27670
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006847548177828658,
+      "loss": 0.6034,
+      "step": 27680
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006847124170535733,
+      "loss": 0.5962,
+      "step": 27690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006846700163242809,
+      "loss": 0.5006,
+      "step": 27700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006846276155949882,
+      "loss": 0.5988,
+      "step": 27710
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006845852148656957,
+      "loss": 0.5941,
+      "step": 27720
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006845428141364032,
+      "loss": 0.5514,
+      "step": 27730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006845004134071106,
+      "loss": 0.5456,
+      "step": 27740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006844580126778181,
+      "loss": 0.6653,
+      "step": 27750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006844156119485256,
+      "loss": 0.634,
+      "step": 27760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006843732112192329,
+      "loss": 0.5701,
+      "step": 27770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006843308104899405,
+      "loss": 0.6192,
+      "step": 27780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000684288409760648,
+      "loss": 0.6494,
+      "step": 27790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006842460090313553,
+      "loss": 0.5872,
+      "step": 27800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006842036083020628,
+      "loss": 0.5553,
+      "step": 27810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006841612075727704,
+      "loss": 0.5927,
+      "step": 27820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006841188068434777,
+      "loss": 0.551,
+      "step": 27830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006840764061141852,
+      "loss": 0.6772,
+      "step": 27840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006840340053848926,
+      "loss": 0.6099,
+      "step": 27850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006839916046556001,
+      "loss": 0.5538,
+      "step": 27860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006839492039263076,
+      "loss": 0.7256,
+      "step": 27870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000683906803197015,
+      "loss": 0.6736,
+      "step": 27880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006838644024677224,
+      "loss": 0.5215,
+      "step": 27890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00068382200173843,
+      "loss": 0.5873,
+      "step": 27900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006837796010091375,
+      "loss": 0.546,
+      "step": 27910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006837372002798448,
+      "loss": 0.5603,
+      "step": 27920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006836947995505523,
+      "loss": 0.5531,
+      "step": 27930
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006836523988212599,
+      "loss": 0.6051,
+      "step": 27940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006836099980919672,
+      "loss": 0.5002,
+      "step": 27950
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006835675973626747,
+      "loss": 0.5597,
+      "step": 27960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006835251966333821,
+      "loss": 0.5233,
+      "step": 27970
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006834827959040896,
+      "loss": 0.5973,
+      "step": 27980
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006834403951747971,
+      "loss": 0.5774,
+      "step": 27990
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006833979944455044,
+      "loss": 0.6091,
+      "step": 28000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006833555937162119,
+      "loss": 0.5433,
+      "step": 28010
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006833131929869195,
+      "loss": 0.6489,
+      "step": 28020
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006832707922576268,
+      "loss": 0.568,
+      "step": 28030
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006832283915283343,
+      "loss": 0.6453,
+      "step": 28040
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006831859907990418,
+      "loss": 0.5244,
+      "step": 28050
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006831435900697492,
+      "loss": 0.5397,
+      "step": 28060
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006831011893404567,
+      "loss": 0.5682,
+      "step": 28070
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006830587886111642,
+      "loss": 0.5399,
+      "step": 28080
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006830163878818715,
+      "loss": 0.5871,
+      "step": 28090
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006829739871525791,
+      "loss": 0.5482,
+      "step": 28100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006829315864232866,
+      "loss": 0.5687,
+      "step": 28110
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006828891856939939,
+      "loss": 0.5135,
+      "step": 28120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006828467849647014,
+      "loss": 0.568,
+      "step": 28130
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006828043842354089,
+      "loss": 0.6056,
+      "step": 28140
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006827619835061163,
+      "loss": 0.5911,
+      "step": 28150
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006827195827768238,
+      "loss": 0.6205,
+      "step": 28160
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006826771820475313,
+      "loss": 0.5102,
+      "step": 28170
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006826347813182387,
+      "loss": 0.5685,
+      "step": 28180
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006825923805889462,
+      "loss": 0.545,
+      "step": 28190
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006825499798596537,
+      "loss": 0.5157,
+      "step": 28200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000682507579130361,
+      "loss": 0.6216,
+      "step": 28210
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006824651784010685,
+      "loss": 0.6402,
+      "step": 28220
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006824227776717761,
+      "loss": 0.5277,
+      "step": 28230
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006823803769424834,
+      "loss": 0.569,
+      "step": 28240
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006823379762131909,
+      "loss": 0.5537,
+      "step": 28250
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006822955754838984,
+      "loss": 0.4878,
+      "step": 28260
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006822531747546058,
+      "loss": 0.6577,
+      "step": 28270
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006822107740253133,
+      "loss": 0.6072,
+      "step": 28280
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006821683732960208,
+      "loss": 0.6501,
+      "step": 28290
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006821259725667281,
+      "loss": 0.5199,
+      "step": 28300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006820835718374357,
+      "loss": 0.5334,
+      "step": 28310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000682041171108143,
+      "loss": 0.5634,
+      "step": 28320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006819987703788505,
+      "loss": 0.5156,
+      "step": 28330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000681956369649558,
+      "loss": 0.5616,
+      "step": 28340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006819139689202654,
+      "loss": 0.5901,
+      "step": 28350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006818715681909729,
+      "loss": 0.5634,
+      "step": 28360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006818291674616804,
+      "loss": 0.6075,
+      "step": 28370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006817867667323878,
+      "loss": 0.5313,
+      "step": 28380
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006817443660030953,
+      "loss": 0.5094,
+      "step": 28390
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006817019652738028,
+      "loss": 0.591,
+      "step": 28400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006816595645445101,
+      "loss": 0.6013,
+      "step": 28410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006816171638152176,
+      "loss": 0.6337,
+      "step": 28420
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006815747630859252,
+      "loss": 0.5155,
+      "step": 28430
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006815323623566325,
+      "loss": 0.5974,
+      "step": 28440
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00068148996162734,
+      "loss": 0.5978,
+      "step": 28450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006814475608980475,
+      "loss": 0.4894,
+      "step": 28460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006814051601687549,
+      "loss": 0.5808,
+      "step": 28470
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006813627594394624,
+      "loss": 0.519,
+      "step": 28480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006813203587101699,
+      "loss": 0.5275,
+      "step": 28490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006812779579808772,
+      "loss": 0.6131,
+      "step": 28500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006812355572515848,
+      "loss": 0.5598,
+      "step": 28510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006811931565222923,
+      "loss": 0.5844,
+      "step": 28520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006811507557929996,
+      "loss": 0.6436,
+      "step": 28530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006811083550637071,
+      "loss": 0.6174,
+      "step": 28540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006810659543344147,
+      "loss": 0.6006,
+      "step": 28550
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000681023553605122,
+      "loss": 0.5411,
+      "step": 28560
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006809811528758295,
+      "loss": 0.6515,
+      "step": 28570
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000680938752146537,
+      "loss": 0.5411,
+      "step": 28580
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006808963514172444,
+      "loss": 0.5201,
+      "step": 28590
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006808539506879519,
+      "loss": 0.5766,
+      "step": 28600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006808115499586593,
+      "loss": 0.5849,
+      "step": 28610
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006807691492293667,
+      "loss": 0.5649,
+      "step": 28620
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006807267485000743,
+      "loss": 0.6626,
+      "step": 28630
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006806843477707817,
+      "loss": 0.6439,
+      "step": 28640
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006806419470414891,
+      "loss": 0.5206,
+      "step": 28650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006805995463121966,
+      "loss": 0.623,
+      "step": 28660
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006805571455829041,
+      "loss": 0.4991,
+      "step": 28670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006805147448536115,
+      "loss": 0.5428,
+      "step": 28680
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000680472344124319,
+      "loss": 0.5681,
+      "step": 28690
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006804299433950264,
+      "loss": 0.5937,
+      "step": 28700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006803875426657339,
+      "loss": 0.5214,
+      "step": 28710
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006803451419364414,
+      "loss": 0.5706,
+      "step": 28720
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006803027412071488,
+      "loss": 0.5806,
+      "step": 28730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006802603404778562,
+      "loss": 0.6901,
+      "step": 28740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006802179397485637,
+      "loss": 0.6234,
+      "step": 28750
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006801755390192712,
+      "loss": 0.5528,
+      "step": 28760
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006801331382899786,
+      "loss": 0.6514,
+      "step": 28770
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006800907375606861,
+      "loss": 0.6532,
+      "step": 28780
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006800483368313936,
+      "loss": 0.5345,
+      "step": 28790
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000680005936102101,
+      "loss": 0.5273,
+      "step": 28800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006799635353728085,
+      "loss": 0.5781,
+      "step": 28810
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006799211346435158,
+      "loss": 0.5454,
+      "step": 28820
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006798787339142233,
+      "loss": 0.5455,
+      "step": 28830
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006798363331849309,
+      "loss": 0.5811,
+      "step": 28840
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006797939324556382,
+      "loss": 0.6328,
+      "step": 28850
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006797515317263457,
+      "loss": 0.5401,
+      "step": 28860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006797091309970532,
+      "loss": 0.5971,
+      "step": 28870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006796667302677606,
+      "loss": 0.5965,
+      "step": 28880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006796243295384681,
+      "loss": 0.6427,
+      "step": 28890
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006795819288091756,
+      "loss": 0.6014,
+      "step": 28900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000679539528079883,
+      "loss": 0.5325,
+      "step": 28910
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006794971273505905,
+      "loss": 0.5478,
+      "step": 28920
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006794547266212979,
+      "loss": 0.5691,
+      "step": 28930
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006794123258920053,
+      "loss": 0.6676,
+      "step": 28940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006793699251627128,
+      "loss": 0.5983,
+      "step": 28950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006793275244334203,
+      "loss": 0.6161,
+      "step": 28960
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006792851237041277,
+      "loss": 0.6049,
+      "step": 28970
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006792427229748352,
+      "loss": 0.6736,
+      "step": 28980
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006792003222455427,
+      "loss": 0.5515,
+      "step": 28990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006791579215162501,
+      "loss": 0.456,
+      "step": 29000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006791155207869576,
+      "loss": 0.6256,
+      "step": 29010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000679073120057665,
+      "loss": 0.6029,
+      "step": 29020
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006790307193283724,
+      "loss": 0.4936,
+      "step": 29030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00067898831859908,
+      "loss": 0.6513,
+      "step": 29040
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006789459178697874,
+      "loss": 0.6708,
+      "step": 29050
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006789035171404948,
+      "loss": 0.6428,
+      "step": 29060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006788611164112023,
+      "loss": 0.6065,
+      "step": 29070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006788187156819098,
+      "loss": 0.704,
+      "step": 29080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006787763149526172,
+      "loss": 0.6592,
+      "step": 29090
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006787339142233247,
+      "loss": 0.5423,
+      "step": 29100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006786915134940321,
+      "loss": 0.6338,
+      "step": 29110
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006786491127647396,
+      "loss": 0.6561,
+      "step": 29120
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006786067120354471,
+      "loss": 0.4925,
+      "step": 29130
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006785643113061545,
+      "loss": 0.5727,
+      "step": 29140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006785219105768619,
+      "loss": 0.6433,
+      "step": 29150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006784795098475695,
+      "loss": 0.5899,
+      "step": 29160
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006784371091182769,
+      "loss": 0.5404,
+      "step": 29170
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006783947083889843,
+      "loss": 0.6078,
+      "step": 29180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006783523076596918,
+      "loss": 0.6414,
+      "step": 29190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006783099069303993,
+      "loss": 0.5501,
+      "step": 29200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006782675062011067,
+      "loss": 0.5376,
+      "step": 29210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006782251054718141,
+      "loss": 0.5393,
+      "step": 29220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006781827047425215,
+      "loss": 0.5841,
+      "step": 29230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006781403040132291,
+      "loss": 0.695,
+      "step": 29240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006780979032839365,
+      "loss": 0.659,
+      "step": 29250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000678055502554644,
+      "loss": 0.6178,
+      "step": 29260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006780131018253514,
+      "loss": 0.7016,
+      "step": 29270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006779707010960589,
+      "loss": 0.6255,
+      "step": 29280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006779283003667663,
+      "loss": 0.6202,
+      "step": 29290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006778858996374738,
+      "loss": 0.5437,
+      "step": 29300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006778434989081812,
+      "loss": 0.4701,
+      "step": 29310
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006778010981788887,
+      "loss": 0.6011,
+      "step": 29320
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006777586974495962,
+      "loss": 0.5492,
+      "step": 29330
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006777162967203036,
+      "loss": 0.6659,
+      "step": 29340
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000677673895991011,
+      "loss": 0.5414,
+      "step": 29350
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006776314952617186,
+      "loss": 0.6549,
+      "step": 29360
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000677589094532426,
+      "loss": 0.5967,
+      "step": 29370
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006775466938031334,
+      "loss": 0.5793,
+      "step": 29380
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006775042930738409,
+      "loss": 0.6028,
+      "step": 29390
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006774618923445484,
+      "loss": 0.4739,
+      "step": 29400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006774194916152558,
+      "loss": 0.5812,
+      "step": 29410
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006773770908859633,
+      "loss": 0.5955,
+      "step": 29420
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006773346901566707,
+      "loss": 0.6288,
+      "step": 29430
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006772922894273782,
+      "loss": 0.7082,
+      "step": 29440
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006772498886980857,
+      "loss": 0.5245,
+      "step": 29450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006772074879687931,
+      "loss": 0.7395,
+      "step": 29460
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006771650872395005,
+      "loss": 0.5862,
+      "step": 29470
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000677122686510208,
+      "loss": 0.5887,
+      "step": 29480
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006770802857809155,
+      "loss": 0.5407,
+      "step": 29490
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006770378850516229,
+      "loss": 0.6337,
+      "step": 29500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006769954843223304,
+      "loss": 0.6177,
+      "step": 29510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006769530835930379,
+      "loss": 0.6543,
+      "step": 29520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006769106828637453,
+      "loss": 0.5974,
+      "step": 29530
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006768682821344527,
+      "loss": 0.5944,
+      "step": 29540
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006768258814051602,
+      "loss": 0.5679,
+      "step": 29550
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006767834806758676,
+      "loss": 0.6102,
+      "step": 29560
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006767410799465751,
+      "loss": 0.5444,
+      "step": 29570
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006766986792172826,
+      "loss": 0.4687,
+      "step": 29580
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00067665627848799,
+      "loss": 0.5924,
+      "step": 29590
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006766138777586975,
+      "loss": 0.5215,
+      "step": 29600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000676571477029405,
+      "loss": 0.5816,
+      "step": 29610
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006765290763001124,
+      "loss": 0.5781,
+      "step": 29620
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006764866755708198,
+      "loss": 0.5977,
+      "step": 29630
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006764442748415273,
+      "loss": 0.4882,
+      "step": 29640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006764018741122348,
+      "loss": 0.6981,
+      "step": 29650
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006763594733829422,
+      "loss": 0.6863,
+      "step": 29660
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006763170726536497,
+      "loss": 0.541,
+      "step": 29670
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006762746719243571,
+      "loss": 0.498,
+      "step": 29680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006762322711950646,
+      "loss": 0.5225,
+      "step": 29690
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000676189870465772,
+      "loss": 0.5959,
+      "step": 29700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006761474697364795,
+      "loss": 0.5704,
+      "step": 29710
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006761050690071869,
+      "loss": 0.6006,
+      "step": 29720
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006760626682778945,
+      "loss": 0.6235,
+      "step": 29730
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006760202675486019,
+      "loss": 0.7029,
+      "step": 29740
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006759778668193093,
+      "loss": 0.5663,
+      "step": 29750
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006759354660900167,
+      "loss": 0.5109,
+      "step": 29760
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006758930653607243,
+      "loss": 0.5139,
+      "step": 29770
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006758506646314317,
+      "loss": 0.539,
+      "step": 29780
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006758082639021391,
+      "loss": 0.5511,
+      "step": 29790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006757658631728466,
+      "loss": 0.5445,
+      "step": 29800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006757234624435541,
+      "loss": 0.5365,
+      "step": 29810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006756810617142615,
+      "loss": 0.6377,
+      "step": 29820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006756386609849689,
+      "loss": 0.5776,
+      "step": 29830
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006755962602556764,
+      "loss": 0.5897,
+      "step": 29840
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000675553859526384,
+      "loss": 0.5625,
+      "step": 29850
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006755114587970913,
+      "loss": 0.5558,
+      "step": 29860
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006754690580677988,
+      "loss": 0.5291,
+      "step": 29870
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006754266573385062,
+      "loss": 0.5913,
+      "step": 29880
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006753842566092137,
+      "loss": 0.5776,
+      "step": 29890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006753418558799212,
+      "loss": 0.6106,
+      "step": 29900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006752994551506286,
+      "loss": 0.6628,
+      "step": 29910
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000675257054421336,
+      "loss": 0.583,
+      "step": 29920
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006752146536920436,
+      "loss": 0.5745,
+      "step": 29930
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000675172252962751,
+      "loss": 0.5825,
+      "step": 29940
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006751298522334584,
+      "loss": 0.5568,
+      "step": 29950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006750874515041659,
+      "loss": 0.5764,
+      "step": 29960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006750450507748734,
+      "loss": 0.5142,
+      "step": 29970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006750026500455808,
+      "loss": 0.4487,
+      "step": 29980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006749602493162883,
+      "loss": 0.6577,
+      "step": 29990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006749178485869957,
+      "loss": 0.5821,
+      "step": 30000
+    },
+    {
+      "epoch": 1.27,
+      "eval_loss": 0.6254510283470154,
+      "eval_runtime": 337.5949,
+      "eval_samples_per_second": 15.566,
+      "eval_steps_per_second": 3.892,
+      "step": 30000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006748754478577032,
+      "loss": 0.5532,
+      "step": 30010
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006748330471284107,
+      "loss": 0.5679,
+      "step": 30020
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006747906463991181,
+      "loss": 0.5665,
+      "step": 30030
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006747482456698255,
+      "loss": 0.5408,
+      "step": 30040
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006747058449405331,
+      "loss": 0.5008,
+      "step": 30050
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006746634442112405,
+      "loss": 0.6044,
+      "step": 30060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006746210434819479,
+      "loss": 0.5533,
+      "step": 30070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006745786427526554,
+      "loss": 0.5834,
+      "step": 30080
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006745362420233628,
+      "loss": 0.5195,
+      "step": 30090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006744938412940703,
+      "loss": 0.5288,
+      "step": 30100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006744514405647778,
+      "loss": 0.5713,
+      "step": 30110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006744090398354852,
+      "loss": 0.6587,
+      "step": 30120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006743666391061927,
+      "loss": 0.5719,
+      "step": 30130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006743242383769002,
+      "loss": 0.6385,
+      "step": 30140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006742818376476075,
+      "loss": 0.604,
+      "step": 30150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000674239436918315,
+      "loss": 0.5908,
+      "step": 30160
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006741970361890225,
+      "loss": 0.5629,
+      "step": 30170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006741546354597299,
+      "loss": 0.603,
+      "step": 30180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006741122347304374,
+      "loss": 0.4915,
+      "step": 30190
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006740698340011449,
+      "loss": 0.5905,
+      "step": 30200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006740274332718523,
+      "loss": 0.5249,
+      "step": 30210
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006739850325425598,
+      "loss": 0.6394,
+      "step": 30220
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006739426318132673,
+      "loss": 0.6733,
+      "step": 30230
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006739002310839746,
+      "loss": 0.5909,
+      "step": 30240
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006738578303546821,
+      "loss": 0.6294,
+      "step": 30250
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006738154296253897,
+      "loss": 0.6335,
+      "step": 30260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000673773028896097,
+      "loss": 0.605,
+      "step": 30270
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006737306281668045,
+      "loss": 0.5363,
+      "step": 30280
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006736882274375119,
+      "loss": 0.6013,
+      "step": 30290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006736458267082194,
+      "loss": 0.5636,
+      "step": 30300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006736034259789269,
+      "loss": 0.5715,
+      "step": 30310
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006735610252496343,
+      "loss": 0.4594,
+      "step": 30320
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006735186245203417,
+      "loss": 0.5405,
+      "step": 30330
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006734762237910493,
+      "loss": 0.5305,
+      "step": 30340
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006734338230617567,
+      "loss": 0.609,
+      "step": 30350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006733914223324641,
+      "loss": 0.5529,
+      "step": 30360
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006733490216031716,
+      "loss": 0.6016,
+      "step": 30370
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006733066208738791,
+      "loss": 0.6497,
+      "step": 30380
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006732642201445865,
+      "loss": 0.5799,
+      "step": 30390
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000673221819415294,
+      "loss": 0.6471,
+      "step": 30400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006731794186860014,
+      "loss": 0.6029,
+      "step": 30410
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006731370179567089,
+      "loss": 0.5474,
+      "step": 30420
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006730946172274164,
+      "loss": 0.5511,
+      "step": 30430
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006730522164981238,
+      "loss": 0.5881,
+      "step": 30440
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006730098157688312,
+      "loss": 0.5267,
+      "step": 30450
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006729674150395388,
+      "loss": 0.6721,
+      "step": 30460
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006729250143102461,
+      "loss": 0.5739,
+      "step": 30470
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006728826135809536,
+      "loss": 0.4932,
+      "step": 30480
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006728402128516611,
+      "loss": 0.6823,
+      "step": 30490
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006727978121223685,
+      "loss": 0.4928,
+      "step": 30500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000672755411393076,
+      "loss": 0.4665,
+      "step": 30510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006727130106637835,
+      "loss": 0.5201,
+      "step": 30520
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006726706099344908,
+      "loss": 0.578,
+      "step": 30530
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006726282092051984,
+      "loss": 0.6476,
+      "step": 30540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006725858084759059,
+      "loss": 0.5621,
+      "step": 30550
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006725434077466132,
+      "loss": 0.5384,
+      "step": 30560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006725010070173207,
+      "loss": 0.518,
+      "step": 30570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006724586062880283,
+      "loss": 0.7168,
+      "step": 30580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006724162055587356,
+      "loss": 0.5477,
+      "step": 30590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006723738048294431,
+      "loss": 0.5453,
+      "step": 30600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006723314041001506,
+      "loss": 0.5937,
+      "step": 30610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000672289003370858,
+      "loss": 0.6398,
+      "step": 30620
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006722466026415655,
+      "loss": 0.5308,
+      "step": 30630
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000672204201912273,
+      "loss": 0.5712,
+      "step": 30640
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006721618011829803,
+      "loss": 0.5579,
+      "step": 30650
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006721194004536879,
+      "loss": 0.5448,
+      "step": 30660
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006720769997243954,
+      "loss": 0.5702,
+      "step": 30670
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006720345989951027,
+      "loss": 0.6084,
+      "step": 30680
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006719921982658102,
+      "loss": 0.5092,
+      "step": 30690
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006719497975365176,
+      "loss": 0.5488,
+      "step": 30700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006719073968072251,
+      "loss": 0.6385,
+      "step": 30710
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006718649960779326,
+      "loss": 0.6713,
+      "step": 30720
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00067182259534864,
+      "loss": 0.5464,
+      "step": 30730
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006717801946193475,
+      "loss": 0.5958,
+      "step": 30740
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000671737793890055,
+      "loss": 0.6336,
+      "step": 30750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006716953931607623,
+      "loss": 0.6284,
+      "step": 30760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006716529924314698,
+      "loss": 0.6545,
+      "step": 30770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006716105917021773,
+      "loss": 0.5968,
+      "step": 30780
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006715681909728847,
+      "loss": 0.5287,
+      "step": 30790
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006715257902435922,
+      "loss": 0.617,
+      "step": 30800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006714833895142997,
+      "loss": 0.4703,
+      "step": 30810
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006714409887850071,
+      "loss": 0.586,
+      "step": 30820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006713985880557146,
+      "loss": 0.5227,
+      "step": 30830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006713561873264221,
+      "loss": 0.6134,
+      "step": 30840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006713137865971294,
+      "loss": 0.4928,
+      "step": 30850
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000671271385867837,
+      "loss": 0.6177,
+      "step": 30860
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006712289851385445,
+      "loss": 0.631,
+      "step": 30870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006711865844092518,
+      "loss": 0.6482,
+      "step": 30880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006711441836799593,
+      "loss": 0.5884,
+      "step": 30890
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006711017829506668,
+      "loss": 0.5557,
+      "step": 30900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006710593822213742,
+      "loss": 0.6637,
+      "step": 30910
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006710169814920817,
+      "loss": 0.5333,
+      "step": 30920
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006709745807627892,
+      "loss": 0.5776,
+      "step": 30930
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006709321800334966,
+      "loss": 0.5177,
+      "step": 30940
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006708897793042041,
+      "loss": 0.5666,
+      "step": 30950
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006708473785749116,
+      "loss": 0.5767,
+      "step": 30960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006708049778456189,
+      "loss": 0.6629,
+      "step": 30970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006707625771163264,
+      "loss": 0.5344,
+      "step": 30980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000670720176387034,
+      "loss": 0.6217,
+      "step": 30990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006706777756577413,
+      "loss": 0.5567,
+      "step": 31000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006706353749284488,
+      "loss": 0.5374,
+      "step": 31010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006705929741991563,
+      "loss": 0.6178,
+      "step": 31020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006705505734698637,
+      "loss": 0.5655,
+      "step": 31030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006705081727405712,
+      "loss": 0.6168,
+      "step": 31040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006704657720112787,
+      "loss": 0.69,
+      "step": 31050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000670423371281986,
+      "loss": 0.5561,
+      "step": 31060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006703809705526936,
+      "loss": 0.5657,
+      "step": 31070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000670338569823401,
+      "loss": 0.6699,
+      "step": 31080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006702961690941084,
+      "loss": 0.6052,
+      "step": 31090
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006702537683648159,
+      "loss": 0.5972,
+      "step": 31100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006702113676355234,
+      "loss": 0.531,
+      "step": 31110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006701689669062308,
+      "loss": 0.6316,
+      "step": 31120
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006701265661769383,
+      "loss": 0.5442,
+      "step": 31130
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006700841654476456,
+      "loss": 0.5981,
+      "step": 31140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006700417647183532,
+      "loss": 0.6147,
+      "step": 31150
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006699993639890607,
+      "loss": 0.5729,
+      "step": 31160
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000669956963259768,
+      "loss": 0.5123,
+      "step": 31170
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006699145625304755,
+      "loss": 0.6195,
+      "step": 31180
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006698721618011831,
+      "loss": 0.6114,
+      "step": 31190
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006698297610718904,
+      "loss": 0.5496,
+      "step": 31200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006697873603425979,
+      "loss": 0.584,
+      "step": 31210
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006697449596133054,
+      "loss": 0.6061,
+      "step": 31220
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006697025588840128,
+      "loss": 0.5134,
+      "step": 31230
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006696601581547203,
+      "loss": 0.6406,
+      "step": 31240
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006696177574254278,
+      "loss": 0.5839,
+      "step": 31250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006695753566961351,
+      "loss": 0.6163,
+      "step": 31260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006695329559668427,
+      "loss": 0.6628,
+      "step": 31270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006694905552375502,
+      "loss": 0.596,
+      "step": 31280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006694481545082575,
+      "loss": 0.5939,
+      "step": 31290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000669405753778965,
+      "loss": 0.513,
+      "step": 31300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006693633530496725,
+      "loss": 0.5646,
+      "step": 31310
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006693209523203799,
+      "loss": 0.517,
+      "step": 31320
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006692785515910874,
+      "loss": 0.6649,
+      "step": 31330
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006692361508617949,
+      "loss": 0.5832,
+      "step": 31340
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006691937501325023,
+      "loss": 0.637,
+      "step": 31350
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006691513494032098,
+      "loss": 0.5676,
+      "step": 31360
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006691089486739172,
+      "loss": 0.6194,
+      "step": 31370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006690665479446246,
+      "loss": 0.6133,
+      "step": 31380
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006690241472153322,
+      "loss": 0.5792,
+      "step": 31390
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006689817464860396,
+      "loss": 0.5685,
+      "step": 31400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000668939345756747,
+      "loss": 0.6092,
+      "step": 31410
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006688969450274545,
+      "loss": 0.5517,
+      "step": 31420
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000668854544298162,
+      "loss": 0.5929,
+      "step": 31430
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006688121435688694,
+      "loss": 0.5498,
+      "step": 31440
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006687697428395769,
+      "loss": 0.5424,
+      "step": 31450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006687273421102843,
+      "loss": 0.562,
+      "step": 31460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006686849413809918,
+      "loss": 0.5742,
+      "step": 31470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006686425406516993,
+      "loss": 0.5556,
+      "step": 31480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006686001399224067,
+      "loss": 0.636,
+      "step": 31490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006685577391931141,
+      "loss": 0.5775,
+      "step": 31500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006685153384638216,
+      "loss": 0.6124,
+      "step": 31510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006684729377345291,
+      "loss": 0.5638,
+      "step": 31520
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006684305370052365,
+      "loss": 0.5982,
+      "step": 31530
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000668388136275944,
+      "loss": 0.6329,
+      "step": 31540
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006683457355466515,
+      "loss": 0.6211,
+      "step": 31550
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006683033348173589,
+      "loss": 0.5607,
+      "step": 31560
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006682609340880664,
+      "loss": 0.5139,
+      "step": 31570
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006682185333587738,
+      "loss": 0.5795,
+      "step": 31580
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006681761326294812,
+      "loss": 0.5231,
+      "step": 31590
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006681337319001888,
+      "loss": 0.5728,
+      "step": 31600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006680913311708962,
+      "loss": 0.5798,
+      "step": 31610
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006680489304416036,
+      "loss": 0.6287,
+      "step": 31620
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006680065297123111,
+      "loss": 0.6851,
+      "step": 31630
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006679641289830186,
+      "loss": 0.5691,
+      "step": 31640
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000667921728253726,
+      "loss": 0.5494,
+      "step": 31650
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006678793275244335,
+      "loss": 0.6589,
+      "step": 31660
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006678369267951408,
+      "loss": 0.548,
+      "step": 31670
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006677945260658484,
+      "loss": 0.585,
+      "step": 31680
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006677521253365558,
+      "loss": 0.6812,
+      "step": 31690
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006677097246072632,
+      "loss": 0.5725,
+      "step": 31700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006676673238779707,
+      "loss": 0.5668,
+      "step": 31710
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006676249231486782,
+      "loss": 0.6998,
+      "step": 31720
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006675825224193856,
+      "loss": 0.4934,
+      "step": 31730
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006675401216900931,
+      "loss": 0.6001,
+      "step": 31740
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006674977209608005,
+      "loss": 0.5702,
+      "step": 31750
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000667455320231508,
+      "loss": 0.5654,
+      "step": 31760
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006674129195022155,
+      "loss": 0.6848,
+      "step": 31770
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006673705187729229,
+      "loss": 0.5054,
+      "step": 31780
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006673281180436303,
+      "loss": 0.6013,
+      "step": 31790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006672857173143379,
+      "loss": 0.5444,
+      "step": 31800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006672433165850453,
+      "loss": 0.5099,
+      "step": 31810
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006672009158557527,
+      "loss": 0.5926,
+      "step": 31820
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006671585151264602,
+      "loss": 0.6552,
+      "step": 31830
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006671161143971677,
+      "loss": 0.5806,
+      "step": 31840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006670737136678751,
+      "loss": 0.623,
+      "step": 31850
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006670313129385826,
+      "loss": 0.7061,
+      "step": 31860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00066698891220929,
+      "loss": 0.6776,
+      "step": 31870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006669465114799975,
+      "loss": 0.526,
+      "step": 31880
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000666904110750705,
+      "loss": 0.6113,
+      "step": 31890
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006668617100214124,
+      "loss": 0.5601,
+      "step": 31900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006668193092921198,
+      "loss": 0.5326,
+      "step": 31910
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006667769085628274,
+      "loss": 0.4983,
+      "step": 31920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006667345078335348,
+      "loss": 0.6145,
+      "step": 31930
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006666921071042422,
+      "loss": 0.5763,
+      "step": 31940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006666497063749497,
+      "loss": 0.5647,
+      "step": 31950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006666073056456572,
+      "loss": 0.5927,
+      "step": 31960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006665649049163646,
+      "loss": 0.6277,
+      "step": 31970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000666522504187072,
+      "loss": 0.6477,
+      "step": 31980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006664801034577795,
+      "loss": 0.5098,
+      "step": 31990
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000666437702728487,
+      "loss": 0.5947,
+      "step": 32000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006663953019991944,
+      "loss": 0.6288,
+      "step": 32010
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006663529012699019,
+      "loss": 0.6199,
+      "step": 32020
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006663105005406093,
+      "loss": 0.5647,
+      "step": 32030
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006662680998113168,
+      "loss": 0.6075,
+      "step": 32040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006662256990820243,
+      "loss": 0.5576,
+      "step": 32050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006661832983527317,
+      "loss": 0.5963,
+      "step": 32060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006661408976234391,
+      "loss": 0.6591,
+      "step": 32070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006660984968941467,
+      "loss": 0.5288,
+      "step": 32080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006660560961648541,
+      "loss": 0.6203,
+      "step": 32090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006660136954355615,
+      "loss": 0.5845,
+      "step": 32100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000665971294706269,
+      "loss": 0.6671,
+      "step": 32110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006659288939769764,
+      "loss": 0.7039,
+      "step": 32120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006658864932476839,
+      "loss": 0.6239,
+      "step": 32130
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006658440925183913,
+      "loss": 0.5408,
+      "step": 32140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006658016917890988,
+      "loss": 0.6777,
+      "step": 32150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006657592910598063,
+      "loss": 0.6263,
+      "step": 32160
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006657168903305137,
+      "loss": 0.5272,
+      "step": 32170
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006656744896012212,
+      "loss": 0.6719,
+      "step": 32180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006656320888719286,
+      "loss": 0.5613,
+      "step": 32190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000665589688142636,
+      "loss": 0.6035,
+      "step": 32200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006655472874133436,
+      "loss": 0.6037,
+      "step": 32210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000665504886684051,
+      "loss": 0.6595,
+      "step": 32220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006654624859547584,
+      "loss": 0.6295,
+      "step": 32230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006654200852254659,
+      "loss": 0.6422,
+      "step": 32240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006653776844961734,
+      "loss": 0.5268,
+      "step": 32250
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006653352837668808,
+      "loss": 0.5394,
+      "step": 32260
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006652928830375883,
+      "loss": 0.5376,
+      "step": 32270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006652504823082957,
+      "loss": 0.6102,
+      "step": 32280
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006652080815790032,
+      "loss": 0.6374,
+      "step": 32290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006651656808497106,
+      "loss": 0.5256,
+      "step": 32300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006651232801204181,
+      "loss": 0.5949,
+      "step": 32310
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006650808793911255,
+      "loss": 0.6553,
+      "step": 32320
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000665038478661833,
+      "loss": 0.4782,
+      "step": 32330
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006649960779325405,
+      "loss": 0.5803,
+      "step": 32340
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006649536772032479,
+      "loss": 0.532,
+      "step": 32350
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006649112764739554,
+      "loss": 0.6171,
+      "step": 32360
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006648688757446629,
+      "loss": 0.6078,
+      "step": 32370
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006648264750153703,
+      "loss": 0.6035,
+      "step": 32380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006647840742860777,
+      "loss": 0.5346,
+      "step": 32390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006647416735567852,
+      "loss": 0.6346,
+      "step": 32400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006646992728274927,
+      "loss": 0.5507,
+      "step": 32410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006646568720982001,
+      "loss": 0.5847,
+      "step": 32420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006646144713689076,
+      "loss": 0.7278,
+      "step": 32430
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000664572070639615,
+      "loss": 0.6262,
+      "step": 32440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006645296699103225,
+      "loss": 0.5835,
+      "step": 32450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00066448726918103,
+      "loss": 0.6199,
+      "step": 32460
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006644448684517374,
+      "loss": 0.6838,
+      "step": 32470
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006644024677224448,
+      "loss": 0.6497,
+      "step": 32480
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006643600669931524,
+      "loss": 0.5752,
+      "step": 32490
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006643176662638598,
+      "loss": 0.6802,
+      "step": 32500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006642752655345672,
+      "loss": 0.4919,
+      "step": 32510
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006642328648052747,
+      "loss": 0.5996,
+      "step": 32520
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006641904640759822,
+      "loss": 0.5516,
+      "step": 32530
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006641480633466896,
+      "loss": 0.6427,
+      "step": 32540
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000664105662617397,
+      "loss": 0.6397,
+      "step": 32550
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006640632618881045,
+      "loss": 0.5461,
+      "step": 32560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000664020861158812,
+      "loss": 0.6651,
+      "step": 32570
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006639784604295195,
+      "loss": 0.5433,
+      "step": 32580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006639360597002269,
+      "loss": 0.5176,
+      "step": 32590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006638936589709343,
+      "loss": 0.6062,
+      "step": 32600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006638512582416419,
+      "loss": 0.5838,
+      "step": 32610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006638088575123492,
+      "loss": 0.624,
+      "step": 32620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006637664567830567,
+      "loss": 0.5744,
+      "step": 32630
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006637240560537641,
+      "loss": 0.5272,
+      "step": 32640
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006636816553244716,
+      "loss": 0.6506,
+      "step": 32650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006636392545951791,
+      "loss": 0.5812,
+      "step": 32660
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006635968538658865,
+      "loss": 0.6021,
+      "step": 32670
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006635544531365939,
+      "loss": 0.7001,
+      "step": 32680
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006635120524073015,
+      "loss": 0.5969,
+      "step": 32690
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000663469651678009,
+      "loss": 0.6279,
+      "step": 32700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006634272509487163,
+      "loss": 0.6545,
+      "step": 32710
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006633848502194238,
+      "loss": 0.5827,
+      "step": 32720
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006633424494901312,
+      "loss": 0.551,
+      "step": 32730
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006633000487608387,
+      "loss": 0.4983,
+      "step": 32740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006632576480315462,
+      "loss": 0.6099,
+      "step": 32750
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006632152473022536,
+      "loss": 0.5467,
+      "step": 32760
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006631728465729611,
+      "loss": 0.5463,
+      "step": 32770
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006631304458436686,
+      "loss": 0.5434,
+      "step": 32780
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000663088045114376,
+      "loss": 0.669,
+      "step": 32790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006630456443850834,
+      "loss": 0.5524,
+      "step": 32800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006630032436557909,
+      "loss": 0.4795,
+      "step": 32810
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006629608429264984,
+      "loss": 0.5896,
+      "step": 32820
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006629184421972058,
+      "loss": 0.5922,
+      "step": 32830
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006628760414679133,
+      "loss": 0.5733,
+      "step": 32840
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006628336407386207,
+      "loss": 0.609,
+      "step": 32850
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006627912400093282,
+      "loss": 0.6344,
+      "step": 32860
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006627488392800357,
+      "loss": 0.6187,
+      "step": 32870
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006627064385507431,
+      "loss": 0.5814,
+      "step": 32880
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006626640378214506,
+      "loss": 0.6508,
+      "step": 32890
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006626216370921581,
+      "loss": 0.5435,
+      "step": 32900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006625792363628654,
+      "loss": 0.5739,
+      "step": 32910
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006625368356335729,
+      "loss": 0.6312,
+      "step": 32920
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006624944349042804,
+      "loss": 0.6776,
+      "step": 32930
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006624520341749878,
+      "loss": 0.5323,
+      "step": 32940
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006624096334456953,
+      "loss": 0.6112,
+      "step": 32950
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006623672327164028,
+      "loss": 0.6221,
+      "step": 32960
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006623248319871102,
+      "loss": 0.4679,
+      "step": 32970
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006622824312578177,
+      "loss": 0.6121,
+      "step": 32980
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006622400305285252,
+      "loss": 0.5472,
+      "step": 32990
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006621976297992325,
+      "loss": 0.5169,
+      "step": 33000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00066215522906994,
+      "loss": 0.5561,
+      "step": 33010
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006621128283406476,
+      "loss": 0.5663,
+      "step": 33020
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006620704276113549,
+      "loss": 0.678,
+      "step": 33030
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006620280268820624,
+      "loss": 0.5931,
+      "step": 33040
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006619856261527699,
+      "loss": 0.5811,
+      "step": 33050
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006619432254234773,
+      "loss": 0.6967,
+      "step": 33060
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006619008246941848,
+      "loss": 0.5198,
+      "step": 33070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006618584239648923,
+      "loss": 0.6195,
+      "step": 33080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006618160232355996,
+      "loss": 0.6173,
+      "step": 33090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006617736225063072,
+      "loss": 0.6146,
+      "step": 33100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006617312217770147,
+      "loss": 0.5114,
+      "step": 33110
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000661688821047722,
+      "loss": 0.6521,
+      "step": 33120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006616464203184295,
+      "loss": 0.4649,
+      "step": 33130
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000661604019589137,
+      "loss": 0.5698,
+      "step": 33140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006615616188598444,
+      "loss": 0.53,
+      "step": 33150
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006615192181305519,
+      "loss": 0.4872,
+      "step": 33160
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006614768174012593,
+      "loss": 0.5327,
+      "step": 33170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006614344166719668,
+      "loss": 0.5487,
+      "step": 33180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006613920159426743,
+      "loss": 0.6574,
+      "step": 33190
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006613496152133817,
+      "loss": 0.6345,
+      "step": 33200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006613072144840891,
+      "loss": 0.5783,
+      "step": 33210
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006612648137547967,
+      "loss": 0.6401,
+      "step": 33220
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000661222413025504,
+      "loss": 0.5805,
+      "step": 33230
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006611800122962115,
+      "loss": 0.5164,
+      "step": 33240
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000661137611566919,
+      "loss": 0.6184,
+      "step": 33250
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006610952108376264,
+      "loss": 0.6018,
+      "step": 33260
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006610528101083339,
+      "loss": 0.6152,
+      "step": 33270
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006610104093790414,
+      "loss": 0.5837,
+      "step": 33280
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006609680086497487,
+      "loss": 0.6302,
+      "step": 33290
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006609256079204563,
+      "loss": 0.5672,
+      "step": 33300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006608832071911638,
+      "loss": 0.5776,
+      "step": 33310
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006608408064618711,
+      "loss": 0.5595,
+      "step": 33320
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006607984057325786,
+      "loss": 0.6399,
+      "step": 33330
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006607560050032861,
+      "loss": 0.5944,
+      "step": 33340
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006607136042739935,
+      "loss": 0.5115,
+      "step": 33350
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000660671203544701,
+      "loss": 0.6287,
+      "step": 33360
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006606288028154085,
+      "loss": 0.5536,
+      "step": 33370
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006605864020861159,
+      "loss": 0.5395,
+      "step": 33380
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006605440013568234,
+      "loss": 0.5011,
+      "step": 33390
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006605016006275309,
+      "loss": 0.6054,
+      "step": 33400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006604591998982382,
+      "loss": 0.5526,
+      "step": 33410
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006604167991689458,
+      "loss": 0.6049,
+      "step": 33420
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006603743984396533,
+      "loss": 0.5933,
+      "step": 33430
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006603319977103606,
+      "loss": 0.5941,
+      "step": 33440
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006602895969810681,
+      "loss": 0.5708,
+      "step": 33450
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006602471962517756,
+      "loss": 0.6385,
+      "step": 33460
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000660204795522483,
+      "loss": 0.5586,
+      "step": 33470
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006601623947931905,
+      "loss": 0.5506,
+      "step": 33480
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000660119994063898,
+      "loss": 0.5724,
+      "step": 33490
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006600775933346054,
+      "loss": 0.6068,
+      "step": 33500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006600351926053129,
+      "loss": 0.5173,
+      "step": 33510
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006599927918760202,
+      "loss": 0.5249,
+      "step": 33520
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006599503911467277,
+      "loss": 0.5404,
+      "step": 33530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006599079904174352,
+      "loss": 0.6615,
+      "step": 33540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006598655896881426,
+      "loss": 0.5503,
+      "step": 33550
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006598231889588501,
+      "loss": 0.6833,
+      "step": 33560
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006597807882295576,
+      "loss": 0.6727,
+      "step": 33570
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000659738387500265,
+      "loss": 0.7116,
+      "step": 33580
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006596959867709725,
+      "loss": 0.5573,
+      "step": 33590
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00065965358604168,
+      "loss": 0.5533,
+      "step": 33600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006596111853123873,
+      "loss": 0.4972,
+      "step": 33610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006595687845830948,
+      "loss": 0.5929,
+      "step": 33620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006595263838538024,
+      "loss": 0.6794,
+      "step": 33630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006594839831245097,
+      "loss": 0.5427,
+      "step": 33640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006594415823952172,
+      "loss": 0.5489,
+      "step": 33650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006593991816659247,
+      "loss": 0.5622,
+      "step": 33660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006593567809366321,
+      "loss": 0.6172,
+      "step": 33670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006593143802073396,
+      "loss": 0.6136,
+      "step": 33680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006592719794780471,
+      "loss": 0.6229,
+      "step": 33690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006592295787487544,
+      "loss": 0.489,
+      "step": 33700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000659187178019462,
+      "loss": 0.5492,
+      "step": 33710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006591447772901695,
+      "loss": 0.5326,
+      "step": 33720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006591023765608768,
+      "loss": 0.5783,
+      "step": 33730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006590599758315843,
+      "loss": 0.5461,
+      "step": 33740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006590175751022919,
+      "loss": 0.6071,
+      "step": 33750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006589751743729992,
+      "loss": 0.5792,
+      "step": 33760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006589327736437067,
+      "loss": 0.5692,
+      "step": 33770
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006588903729144142,
+      "loss": 0.5619,
+      "step": 33780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006588479721851216,
+      "loss": 0.6287,
+      "step": 33790
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006588055714558291,
+      "loss": 0.5472,
+      "step": 33800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006587631707265366,
+      "loss": 0.6087,
+      "step": 33810
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006587207699972439,
+      "loss": 0.5904,
+      "step": 33820
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006586783692679515,
+      "loss": 0.5144,
+      "step": 33830
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006586359685386589,
+      "loss": 0.5891,
+      "step": 33840
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006585935678093663,
+      "loss": 0.5351,
+      "step": 33850
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006585511670800738,
+      "loss": 0.5895,
+      "step": 33860
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006585087663507813,
+      "loss": 0.6109,
+      "step": 33870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006584663656214887,
+      "loss": 0.5469,
+      "step": 33880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006584239648921962,
+      "loss": 0.6339,
+      "step": 33890
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006583815641629036,
+      "loss": 0.6026,
+      "step": 33900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006583391634336111,
+      "loss": 0.5529,
+      "step": 33910
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006582967627043186,
+      "loss": 0.6255,
+      "step": 33920
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000658254361975026,
+      "loss": 0.5655,
+      "step": 33930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006582119612457334,
+      "loss": 0.5899,
+      "step": 33940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000658169560516441,
+      "loss": 0.5772,
+      "step": 33950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006581271597871484,
+      "loss": 0.6523,
+      "step": 33960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006580847590578558,
+      "loss": 0.5286,
+      "step": 33970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006580423583285633,
+      "loss": 0.579,
+      "step": 33980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006579999575992708,
+      "loss": 0.5713,
+      "step": 33990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006579575568699782,
+      "loss": 0.6089,
+      "step": 34000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006579151561406857,
+      "loss": 0.6204,
+      "step": 34010
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000657872755411393,
+      "loss": 0.5458,
+      "step": 34020
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006578303546821006,
+      "loss": 0.591,
+      "step": 34030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006577879539528081,
+      "loss": 0.5084,
+      "step": 34040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006577455532235154,
+      "loss": 0.5959,
+      "step": 34050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006577031524942229,
+      "loss": 0.7334,
+      "step": 34060
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006576607517649304,
+      "loss": 0.6088,
+      "step": 34070
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006576183510356378,
+      "loss": 0.5507,
+      "step": 34080
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006575759503063453,
+      "loss": 0.6238,
+      "step": 34090
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006575335495770528,
+      "loss": 0.5001,
+      "step": 34100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006574911488477602,
+      "loss": 0.5522,
+      "step": 34110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006574487481184677,
+      "loss": 0.5583,
+      "step": 34120
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006574063473891751,
+      "loss": 0.5908,
+      "step": 34130
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006573639466598825,
+      "loss": 0.592,
+      "step": 34140
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00065732154593059,
+      "loss": 0.5751,
+      "step": 34150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006572791452012975,
+      "loss": 0.5885,
+      "step": 34160
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006572367444720049,
+      "loss": 0.4917,
+      "step": 34170
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006571943437427124,
+      "loss": 0.5177,
+      "step": 34180
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006571519430134199,
+      "loss": 0.5418,
+      "step": 34190
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006571095422841273,
+      "loss": 0.6658,
+      "step": 34200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006570671415548348,
+      "loss": 0.5959,
+      "step": 34210
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006570247408255422,
+      "loss": 0.5428,
+      "step": 34220
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006569823400962496,
+      "loss": 0.5095,
+      "step": 34230
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006569399393669572,
+      "loss": 0.4962,
+      "step": 34240
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006568975386376646,
+      "loss": 0.5608,
+      "step": 34250
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000656855137908372,
+      "loss": 0.5392,
+      "step": 34260
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006568127371790795,
+      "loss": 0.6249,
+      "step": 34270
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000656770336449787,
+      "loss": 0.5231,
+      "step": 34280
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006567279357204944,
+      "loss": 0.6333,
+      "step": 34290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006566855349912019,
+      "loss": 0.595,
+      "step": 34300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006566431342619093,
+      "loss": 0.5062,
+      "step": 34310
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006566007335326168,
+      "loss": 0.5551,
+      "step": 34320
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006565583328033243,
+      "loss": 0.6041,
+      "step": 34330
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006565159320740317,
+      "loss": 0.6024,
+      "step": 34340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006564735313447391,
+      "loss": 0.5634,
+      "step": 34350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006564311306154467,
+      "loss": 0.5655,
+      "step": 34360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006563887298861541,
+      "loss": 0.5576,
+      "step": 34370
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006563463291568615,
+      "loss": 0.6054,
+      "step": 34380
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000656303928427569,
+      "loss": 0.5784,
+      "step": 34390
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006562615276982765,
+      "loss": 0.6068,
+      "step": 34400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006562191269689839,
+      "loss": 0.6692,
+      "step": 34410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006561767262396914,
+      "loss": 0.5653,
+      "step": 34420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006561343255103987,
+      "loss": 0.6411,
+      "step": 34430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006560919247811063,
+      "loss": 0.6501,
+      "step": 34440
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006560495240518137,
+      "loss": 0.5875,
+      "step": 34450
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006560071233225211,
+      "loss": 0.5994,
+      "step": 34460
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006559647225932286,
+      "loss": 0.606,
+      "step": 34470
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006559223218639361,
+      "loss": 0.593,
+      "step": 34480
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006558799211346436,
+      "loss": 0.5689,
+      "step": 34490
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000655837520405351,
+      "loss": 0.4565,
+      "step": 34500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006557951196760584,
+      "loss": 0.6361,
+      "step": 34510
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000655752718946766,
+      "loss": 0.5607,
+      "step": 34520
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006557103182174734,
+      "loss": 0.5451,
+      "step": 34530
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006556679174881808,
+      "loss": 0.6119,
+      "step": 34540
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006556255167588882,
+      "loss": 0.5535,
+      "step": 34550
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006555831160295958,
+      "loss": 0.582,
+      "step": 34560
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006555407153003032,
+      "loss": 0.5903,
+      "step": 34570
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006554983145710106,
+      "loss": 0.699,
+      "step": 34580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006554559138417181,
+      "loss": 0.5786,
+      "step": 34590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006554135131124256,
+      "loss": 0.5899,
+      "step": 34600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000655371112383133,
+      "loss": 0.5541,
+      "step": 34610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006553287116538405,
+      "loss": 0.5996,
+      "step": 34620
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006552863109245479,
+      "loss": 0.5115,
+      "step": 34630
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006552439101952554,
+      "loss": 0.6655,
+      "step": 34640
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006552015094659629,
+      "loss": 0.6737,
+      "step": 34650
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006551591087366703,
+      "loss": 0.6093,
+      "step": 34660
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006551167080073777,
+      "loss": 0.5247,
+      "step": 34670
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006550743072780852,
+      "loss": 0.5155,
+      "step": 34680
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006550319065487927,
+      "loss": 0.575,
+      "step": 34690
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006549895058195001,
+      "loss": 0.5058,
+      "step": 34700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006549471050902076,
+      "loss": 0.5735,
+      "step": 34710
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006549047043609151,
+      "loss": 0.5428,
+      "step": 34720
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006548623036316225,
+      "loss": 0.4919,
+      "step": 34730
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00065481990290233,
+      "loss": 0.5371,
+      "step": 34740
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006547775021730374,
+      "loss": 0.6579,
+      "step": 34750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006547351014437448,
+      "loss": 0.6301,
+      "step": 34760
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006546927007144523,
+      "loss": 0.6454,
+      "step": 34770
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006546502999851598,
+      "loss": 0.6279,
+      "step": 34780
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006546078992558672,
+      "loss": 0.5818,
+      "step": 34790
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006545654985265747,
+      "loss": 0.6159,
+      "step": 34800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006545230977972822,
+      "loss": 0.5559,
+      "step": 34810
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006544806970679896,
+      "loss": 0.5666,
+      "step": 34820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000654438296338697,
+      "loss": 0.531,
+      "step": 34830
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006543958956094045,
+      "loss": 0.5311,
+      "step": 34840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000654353494880112,
+      "loss": 0.6246,
+      "step": 34850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006543110941508194,
+      "loss": 0.5468,
+      "step": 34860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006542686934215269,
+      "loss": 0.5602,
+      "step": 34870
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006542262926922343,
+      "loss": 0.6101,
+      "step": 34880
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006541838919629418,
+      "loss": 0.6997,
+      "step": 34890
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006541414912336493,
+      "loss": 0.5446,
+      "step": 34900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006540990905043567,
+      "loss": 0.657,
+      "step": 34910
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006540566897750642,
+      "loss": 0.6019,
+      "step": 34920
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006540142890457717,
+      "loss": 0.5414,
+      "step": 34930
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006539718883164791,
+      "loss": 0.5609,
+      "step": 34940
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006539294875871865,
+      "loss": 0.5155,
+      "step": 34950
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000653887086857894,
+      "loss": 0.645,
+      "step": 34960
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006538446861286015,
+      "loss": 0.6204,
+      "step": 34970
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006538022853993089,
+      "loss": 0.6152,
+      "step": 34980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006537598846700163,
+      "loss": 0.5709,
+      "step": 34990
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006537174839407238,
+      "loss": 0.5639,
+      "step": 35000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006536750832114313,
+      "loss": 0.6408,
+      "step": 35010
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006536326824821387,
+      "loss": 0.6894,
+      "step": 35020
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006535902817528462,
+      "loss": 0.6108,
+      "step": 35030
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006535478810235536,
+      "loss": 0.5345,
+      "step": 35040
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006535054802942611,
+      "loss": 0.5109,
+      "step": 35050
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006534630795649685,
+      "loss": 0.6259,
+      "step": 35060
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000653420678835676,
+      "loss": 0.5285,
+      "step": 35070
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006533782781063834,
+      "loss": 0.6326,
+      "step": 35080
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006533358773770909,
+      "loss": 0.5406,
+      "step": 35090
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006532934766477984,
+      "loss": 0.5619,
+      "step": 35100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006532510759185058,
+      "loss": 0.6635,
+      "step": 35110
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006532086751892132,
+      "loss": 0.5535,
+      "step": 35120
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006531662744599208,
+      "loss": 0.5916,
+      "step": 35130
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006531238737306282,
+      "loss": 0.5474,
+      "step": 35140
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006530814730013356,
+      "loss": 0.5703,
+      "step": 35150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006530390722720431,
+      "loss": 0.6214,
+      "step": 35160
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006529966715427506,
+      "loss": 0.5701,
+      "step": 35170
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000652954270813458,
+      "loss": 0.5929,
+      "step": 35180
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006529118700841655,
+      "loss": 0.5987,
+      "step": 35190
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006528694693548729,
+      "loss": 0.5978,
+      "step": 35200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006528270686255804,
+      "loss": 0.4837,
+      "step": 35210
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006527846678962879,
+      "loss": 0.4946,
+      "step": 35220
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006527422671669953,
+      "loss": 0.5093,
+      "step": 35230
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006526998664377027,
+      "loss": 0.6668,
+      "step": 35240
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006526574657084103,
+      "loss": 0.6552,
+      "step": 35250
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006526150649791177,
+      "loss": 0.5461,
+      "step": 35260
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006525726642498251,
+      "loss": 0.6849,
+      "step": 35270
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006525302635205326,
+      "loss": 0.6009,
+      "step": 35280
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00065248786279124,
+      "loss": 0.6449,
+      "step": 35290
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006524454620619475,
+      "loss": 0.6361,
+      "step": 35300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000652403061332655,
+      "loss": 0.4585,
+      "step": 35310
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006523606606033624,
+      "loss": 0.5605,
+      "step": 35320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006523182598740699,
+      "loss": 0.7019,
+      "step": 35330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006522758591447774,
+      "loss": 0.6024,
+      "step": 35340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006522334584154848,
+      "loss": 0.5308,
+      "step": 35350
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006521910576861922,
+      "loss": 0.5911,
+      "step": 35360
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006521486569568998,
+      "loss": 0.6132,
+      "step": 35370
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006521062562276071,
+      "loss": 0.5152,
+      "step": 35380
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006520638554983146,
+      "loss": 0.5995,
+      "step": 35390
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000652021454769022,
+      "loss": 0.6342,
+      "step": 35400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006519790540397295,
+      "loss": 0.6711,
+      "step": 35410
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000651936653310437,
+      "loss": 0.7022,
+      "step": 35420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006518942525811445,
+      "loss": 0.5101,
+      "step": 35430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006518518518518518,
+      "loss": 0.4465,
+      "step": 35440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006518094511225594,
+      "loss": 0.6372,
+      "step": 35450
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006517670503932669,
+      "loss": 0.5903,
+      "step": 35460
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006517246496639742,
+      "loss": 0.5248,
+      "step": 35470
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006516822489346817,
+      "loss": 0.6212,
+      "step": 35480
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006516398482053891,
+      "loss": 0.5091,
+      "step": 35490
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006515974474760966,
+      "loss": 0.5802,
+      "step": 35500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006515550467468041,
+      "loss": 0.5759,
+      "step": 35510
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006515126460175115,
+      "loss": 0.6052,
+      "step": 35520
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000651470245288219,
+      "loss": 0.5493,
+      "step": 35530
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006514278445589265,
+      "loss": 0.6218,
+      "step": 35540
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006513854438296339,
+      "loss": 0.6072,
+      "step": 35550
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006513430431003413,
+      "loss": 0.5502,
+      "step": 35560
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006513006423710488,
+      "loss": 0.5589,
+      "step": 35570
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006512582416417563,
+      "loss": 0.6258,
+      "step": 35580
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006512158409124637,
+      "loss": 0.6379,
+      "step": 35590
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006511734401831712,
+      "loss": 0.5588,
+      "step": 35600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006511310394538786,
+      "loss": 0.5298,
+      "step": 35610
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006510886387245861,
+      "loss": 0.667,
+      "step": 35620
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006510462379952936,
+      "loss": 0.5768,
+      "step": 35630
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000651003837266001,
+      "loss": 0.6077,
+      "step": 35640
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006509614365367084,
+      "loss": 0.55,
+      "step": 35650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000650919035807416,
+      "loss": 0.6988,
+      "step": 35660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006508766350781233,
+      "loss": 0.5932,
+      "step": 35670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006508342343488308,
+      "loss": 0.6253,
+      "step": 35680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006507918336195383,
+      "loss": 0.5712,
+      "step": 35690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006507494328902457,
+      "loss": 0.6432,
+      "step": 35700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006507070321609532,
+      "loss": 0.7242,
+      "step": 35710
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006506646314316607,
+      "loss": 0.628,
+      "step": 35720
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000650622230702368,
+      "loss": 0.5913,
+      "step": 35730
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006505798299730756,
+      "loss": 0.4931,
+      "step": 35740
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006505374292437831,
+      "loss": 0.6097,
+      "step": 35750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006504950285144904,
+      "loss": 0.5587,
+      "step": 35760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006504526277851979,
+      "loss": 0.5166,
+      "step": 35770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006504102270559055,
+      "loss": 0.4948,
+      "step": 35780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006503678263266128,
+      "loss": 0.5993,
+      "step": 35790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006503254255973203,
+      "loss": 0.5252,
+      "step": 35800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006502830248680278,
+      "loss": 0.6534,
+      "step": 35810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006502406241387352,
+      "loss": 0.5852,
+      "step": 35820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006501982234094427,
+      "loss": 0.631,
+      "step": 35830
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006501558226801502,
+      "loss": 0.5765,
+      "step": 35840
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006501134219508575,
+      "loss": 0.6836,
+      "step": 35850
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006500710212215651,
+      "loss": 0.5529,
+      "step": 35860
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006500286204922726,
+      "loss": 0.6735,
+      "step": 35870
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006499862197629799,
+      "loss": 0.5714,
+      "step": 35880
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006499438190336874,
+      "loss": 0.562,
+      "step": 35890
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000649901418304395,
+      "loss": 0.4197,
+      "step": 35900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006498590175751023,
+      "loss": 0.5941,
+      "step": 35910
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006498166168458098,
+      "loss": 0.6129,
+      "step": 35920
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006497742161165172,
+      "loss": 0.6396,
+      "step": 35930
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006497318153872247,
+      "loss": 0.6277,
+      "step": 35940
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006496894146579322,
+      "loss": 0.5461,
+      "step": 35950
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006496470139286397,
+      "loss": 0.5818,
+      "step": 35960
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000649604613199347,
+      "loss": 0.5532,
+      "step": 35970
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006495622124700546,
+      "loss": 0.5413,
+      "step": 35980
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006495198117407619,
+      "loss": 0.5896,
+      "step": 35990
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006494774110114694,
+      "loss": 0.6515,
+      "step": 36000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006494350102821769,
+      "loss": 0.547,
+      "step": 36010
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006493926095528843,
+      "loss": 0.549,
+      "step": 36020
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006493502088235918,
+      "loss": 0.5615,
+      "step": 36030
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006493078080942993,
+      "loss": 0.5395,
+      "step": 36040
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006492654073650066,
+      "loss": 0.7291,
+      "step": 36050
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006492230066357142,
+      "loss": 0.6571,
+      "step": 36060
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006491806059064217,
+      "loss": 0.64,
+      "step": 36070
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000649138205177129,
+      "loss": 0.5687,
+      "step": 36080
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006490958044478365,
+      "loss": 0.5841,
+      "step": 36090
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000649053403718544,
+      "loss": 0.6434,
+      "step": 36100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006490110029892514,
+      "loss": 0.6252,
+      "step": 36110
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006489686022599589,
+      "loss": 0.5223,
+      "step": 36120
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006489262015306664,
+      "loss": 0.6125,
+      "step": 36130
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006488838008013738,
+      "loss": 0.5846,
+      "step": 36140
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006488414000720813,
+      "loss": 0.561,
+      "step": 36150
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006487989993427888,
+      "loss": 0.5889,
+      "step": 36160
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006487565986134961,
+      "loss": 0.5742,
+      "step": 36170
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006487141978842036,
+      "loss": 0.642,
+      "step": 36180
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006486717971549112,
+      "loss": 0.5811,
+      "step": 36190
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006486293964256185,
+      "loss": 0.6441,
+      "step": 36200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000648586995696326,
+      "loss": 0.6345,
+      "step": 36210
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006485445949670335,
+      "loss": 0.6463,
+      "step": 36220
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006485021942377409,
+      "loss": 0.5278,
+      "step": 36230
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006484597935084484,
+      "loss": 0.5412,
+      "step": 36240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006484173927791559,
+      "loss": 0.5616,
+      "step": 36250
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006483749920498632,
+      "loss": 0.5041,
+      "step": 36260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006483325913205708,
+      "loss": 0.5646,
+      "step": 36270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006482901905912782,
+      "loss": 0.593,
+      "step": 36280
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006482477898619856,
+      "loss": 0.608,
+      "step": 36290
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006482053891326931,
+      "loss": 0.5979,
+      "step": 36300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006481629884034006,
+      "loss": 0.6315,
+      "step": 36310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000648120587674108,
+      "loss": 0.5372,
+      "step": 36320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006480781869448155,
+      "loss": 0.5394,
+      "step": 36330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006480357862155228,
+      "loss": 0.6276,
+      "step": 36340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006479933854862304,
+      "loss": 0.5604,
+      "step": 36350
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006479509847569379,
+      "loss": 0.7263,
+      "step": 36360
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006479085840276452,
+      "loss": 0.6035,
+      "step": 36370
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006478661832983527,
+      "loss": 0.5122,
+      "step": 36380
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006478237825690603,
+      "loss": 0.552,
+      "step": 36390
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006477813818397676,
+      "loss": 0.6376,
+      "step": 36400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006477389811104751,
+      "loss": 0.5471,
+      "step": 36410
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006476965803811826,
+      "loss": 0.5282,
+      "step": 36420
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00064765417965189,
+      "loss": 0.5356,
+      "step": 36430
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006476117789225975,
+      "loss": 0.5602,
+      "step": 36440
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000647569378193305,
+      "loss": 0.6486,
+      "step": 36450
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006475269774640123,
+      "loss": 0.6088,
+      "step": 36460
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006474845767347199,
+      "loss": 0.6552,
+      "step": 36470
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006474421760054274,
+      "loss": 0.4906,
+      "step": 36480
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006473997752761347,
+      "loss": 0.5401,
+      "step": 36490
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006473573745468422,
+      "loss": 0.6328,
+      "step": 36500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006473149738175498,
+      "loss": 0.5935,
+      "step": 36510
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006472725730882571,
+      "loss": 0.5979,
+      "step": 36520
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006472301723589646,
+      "loss": 0.6404,
+      "step": 36530
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006471877716296721,
+      "loss": 0.4864,
+      "step": 36540
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006471453709003795,
+      "loss": 0.5975,
+      "step": 36550
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000647102970171087,
+      "loss": 0.5982,
+      "step": 36560
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006470605694417945,
+      "loss": 0.6079,
+      "step": 36570
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006470181687125018,
+      "loss": 0.5871,
+      "step": 36580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006469757679832094,
+      "loss": 0.6382,
+      "step": 36590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006469333672539168,
+      "loss": 0.6443,
+      "step": 36600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006468909665246242,
+      "loss": 0.5298,
+      "step": 36610
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006468485657953317,
+      "loss": 0.5477,
+      "step": 36620
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006468061650660392,
+      "loss": 0.7045,
+      "step": 36630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006467637643367466,
+      "loss": 0.5301,
+      "step": 36640
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006467213636074541,
+      "loss": 0.7244,
+      "step": 36650
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006466789628781615,
+      "loss": 0.5855,
+      "step": 36660
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000646636562148869,
+      "loss": 0.6063,
+      "step": 36670
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006465941614195765,
+      "loss": 0.6183,
+      "step": 36680
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006465517606902839,
+      "loss": 0.6987,
+      "step": 36690
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006465093599609913,
+      "loss": 0.5941,
+      "step": 36700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006464669592316988,
+      "loss": 0.5842,
+      "step": 36710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006464245585024063,
+      "loss": 0.5101,
+      "step": 36720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006463821577731137,
+      "loss": 0.6404,
+      "step": 36730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006463397570438212,
+      "loss": 0.5867,
+      "step": 36740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006462973563145287,
+      "loss": 0.5186,
+      "step": 36750
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006462549555852361,
+      "loss": 0.5866,
+      "step": 36760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006462125548559436,
+      "loss": 0.6136,
+      "step": 36770
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000646170154126651,
+      "loss": 0.5117,
+      "step": 36780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006461277533973584,
+      "loss": 0.6658,
+      "step": 36790
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000646085352668066,
+      "loss": 0.5488,
+      "step": 36800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006460429519387734,
+      "loss": 0.6615,
+      "step": 36810
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006460005512094808,
+      "loss": 0.6085,
+      "step": 36820
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006459581504801883,
+      "loss": 0.5563,
+      "step": 36830
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006459157497508958,
+      "loss": 0.6335,
+      "step": 36840
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006458733490216032,
+      "loss": 0.5782,
+      "step": 36850
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006458309482923107,
+      "loss": 0.531,
+      "step": 36860
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006457885475630182,
+      "loss": 0.5525,
+      "step": 36870
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006457461468337256,
+      "loss": 0.5856,
+      "step": 36880
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006457037461044331,
+      "loss": 0.5965,
+      "step": 36890
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006456613453751404,
+      "loss": 0.6823,
+      "step": 36900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006456189446458479,
+      "loss": 0.5349,
+      "step": 36910
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006455765439165554,
+      "loss": 0.6979,
+      "step": 36920
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006455341431872628,
+      "loss": 0.5549,
+      "step": 36930
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006454917424579703,
+      "loss": 0.6653,
+      "step": 36940
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006454493417286778,
+      "loss": 0.5427,
+      "step": 36950
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006454069409993852,
+      "loss": 0.5699,
+      "step": 36960
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006453645402700927,
+      "loss": 0.6264,
+      "step": 36970
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006453221395408001,
+      "loss": 0.4644,
+      "step": 36980
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006452797388115075,
+      "loss": 0.5186,
+      "step": 36990
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006452373380822151,
+      "loss": 0.6045,
+      "step": 37000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006451949373529225,
+      "loss": 0.5357,
+      "step": 37010
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006451525366236299,
+      "loss": 0.661,
+      "step": 37020
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006451101358943374,
+      "loss": 0.4918,
+      "step": 37030
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006450677351650449,
+      "loss": 0.585,
+      "step": 37040
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006450253344357523,
+      "loss": 0.4625,
+      "step": 37050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006449829337064598,
+      "loss": 0.6295,
+      "step": 37060
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006449405329771672,
+      "loss": 0.5401,
+      "step": 37070
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006448981322478747,
+      "loss": 0.5094,
+      "step": 37080
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006448557315185822,
+      "loss": 0.6904,
+      "step": 37090
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006448133307892896,
+      "loss": 0.7571,
+      "step": 37100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000644770930059997,
+      "loss": 0.6165,
+      "step": 37110
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006447285293307046,
+      "loss": 0.5996,
+      "step": 37120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000644686128601412,
+      "loss": 0.5219,
+      "step": 37130
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006446437278721194,
+      "loss": 0.5385,
+      "step": 37140
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006446013271428269,
+      "loss": 0.5843,
+      "step": 37150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006445589264135344,
+      "loss": 0.4785,
+      "step": 37160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006445165256842418,
+      "loss": 0.5496,
+      "step": 37170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006444741249549493,
+      "loss": 0.5914,
+      "step": 37180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006444317242256567,
+      "loss": 0.6184,
+      "step": 37190
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006443893234963642,
+      "loss": 0.5834,
+      "step": 37200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006443469227670716,
+      "loss": 0.5847,
+      "step": 37210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006443045220377791,
+      "loss": 0.7343,
+      "step": 37220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006442621213084865,
+      "loss": 0.578,
+      "step": 37230
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000644219720579194,
+      "loss": 0.6081,
+      "step": 37240
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006441773198499015,
+      "loss": 0.6648,
+      "step": 37250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006441349191206089,
+      "loss": 0.6322,
+      "step": 37260
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006440925183913163,
+      "loss": 0.644,
+      "step": 37270
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006440501176620239,
+      "loss": 0.6735,
+      "step": 37280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006440077169327313,
+      "loss": 0.522,
+      "step": 37290
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006439653162034387,
+      "loss": 0.5467,
+      "step": 37300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006439229154741461,
+      "loss": 0.5468,
+      "step": 37310
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006438805147448536,
+      "loss": 0.7204,
+      "step": 37320
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006438381140155611,
+      "loss": 0.6305,
+      "step": 37330
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006437957132862685,
+      "loss": 0.6388,
+      "step": 37340
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000643753312556976,
+      "loss": 0.6814,
+      "step": 37350
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006437109118276835,
+      "loss": 0.5988,
+      "step": 37360
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000643668511098391,
+      "loss": 0.6133,
+      "step": 37370
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006436261103690984,
+      "loss": 0.625,
+      "step": 37380
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006435837096398058,
+      "loss": 0.7304,
+      "step": 37390
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006435413089105133,
+      "loss": 0.644,
+      "step": 37400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006434989081812208,
+      "loss": 0.5484,
+      "step": 37410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006434565074519282,
+      "loss": 0.5207,
+      "step": 37420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006434141067226356,
+      "loss": 0.6061,
+      "step": 37430
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006433717059933431,
+      "loss": 0.5283,
+      "step": 37440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006433293052640506,
+      "loss": 0.5624,
+      "step": 37450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000643286904534758,
+      "loss": 0.6262,
+      "step": 37460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006432445038054655,
+      "loss": 0.6179,
+      "step": 37470
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000643202103076173,
+      "loss": 0.5793,
+      "step": 37480
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006431597023468804,
+      "loss": 0.6772,
+      "step": 37490
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006431173016175879,
+      "loss": 0.6382,
+      "step": 37500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006430749008882953,
+      "loss": 0.6591,
+      "step": 37510
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006430325001590027,
+      "loss": 0.5974,
+      "step": 37520
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006429900994297102,
+      "loss": 0.5558,
+      "step": 37530
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006429476987004177,
+      "loss": 0.5412,
+      "step": 37540
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006429052979711251,
+      "loss": 0.5325,
+      "step": 37550
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006428628972418326,
+      "loss": 0.5871,
+      "step": 37560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006428204965125401,
+      "loss": 0.679,
+      "step": 37570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006427780957832475,
+      "loss": 0.6163,
+      "step": 37580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006427356950539549,
+      "loss": 0.5485,
+      "step": 37590
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006426932943246624,
+      "loss": 0.512,
+      "step": 37600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006426508935953699,
+      "loss": 0.6242,
+      "step": 37610
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006426084928660773,
+      "loss": 0.5117,
+      "step": 37620
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006425660921367848,
+      "loss": 0.5884,
+      "step": 37630
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006425236914074922,
+      "loss": 0.6016,
+      "step": 37640
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006424812906781997,
+      "loss": 0.6501,
+      "step": 37650
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006424388899489072,
+      "loss": 0.5921,
+      "step": 37660
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006423964892196146,
+      "loss": 0.5737,
+      "step": 37670
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000642354088490322,
+      "loss": 0.6607,
+      "step": 37680
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006423116877610296,
+      "loss": 0.5341,
+      "step": 37690
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000642269287031737,
+      "loss": 0.6577,
+      "step": 37700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006422268863024444,
+      "loss": 0.5865,
+      "step": 37710
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006421844855731519,
+      "loss": 0.4806,
+      "step": 37720
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006421420848438594,
+      "loss": 0.573,
+      "step": 37730
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006420996841145668,
+      "loss": 0.6048,
+      "step": 37740
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006420572833852743,
+      "loss": 0.5695,
+      "step": 37750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006420148826559817,
+      "loss": 0.582,
+      "step": 37760
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006419724819266892,
+      "loss": 0.5926,
+      "step": 37770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006419300811973967,
+      "loss": 0.5994,
+      "step": 37780
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006418876804681041,
+      "loss": 0.5516,
+      "step": 37790
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006418452797388115,
+      "loss": 0.5916,
+      "step": 37800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006418028790095191,
+      "loss": 0.6777,
+      "step": 37810
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006417604782802264,
+      "loss": 0.5148,
+      "step": 37820
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006417180775509339,
+      "loss": 0.6273,
+      "step": 37830
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006416756768216413,
+      "loss": 0.5604,
+      "step": 37840
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006416332760923488,
+      "loss": 0.624,
+      "step": 37850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006415908753630563,
+      "loss": 0.5423,
+      "step": 37860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006415484746337637,
+      "loss": 0.5332,
+      "step": 37870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006415060739044711,
+      "loss": 0.5563,
+      "step": 37880
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006414636731751787,
+      "loss": 0.4979,
+      "step": 37890
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006414212724458861,
+      "loss": 0.6359,
+      "step": 37900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006413788717165935,
+      "loss": 0.6516,
+      "step": 37910
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000641336470987301,
+      "loss": 0.5788,
+      "step": 37920
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006412940702580085,
+      "loss": 0.5746,
+      "step": 37930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006412516695287159,
+      "loss": 0.6057,
+      "step": 37940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006412092687994234,
+      "loss": 0.6027,
+      "step": 37950
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006411668680701308,
+      "loss": 0.533,
+      "step": 37960
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006411244673408383,
+      "loss": 0.522,
+      "step": 37970
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006410820666115458,
+      "loss": 0.5735,
+      "step": 37980
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006410396658822532,
+      "loss": 0.6731,
+      "step": 37990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006409972651529606,
+      "loss": 0.5508,
+      "step": 38000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006409548644236682,
+      "loss": 0.6606,
+      "step": 38010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006409124636943756,
+      "loss": 0.6639,
+      "step": 38020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000640870062965083,
+      "loss": 0.6482,
+      "step": 38030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006408276622357905,
+      "loss": 0.7,
+      "step": 38040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006407852615064979,
+      "loss": 0.6405,
+      "step": 38050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006407428607772054,
+      "loss": 0.58,
+      "step": 38060
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006407004600479129,
+      "loss": 0.6648,
+      "step": 38070
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006406580593186203,
+      "loss": 0.6335,
+      "step": 38080
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006406156585893278,
+      "loss": 0.6199,
+      "step": 38090
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006405732578600353,
+      "loss": 0.5507,
+      "step": 38100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006405308571307427,
+      "loss": 0.5812,
+      "step": 38110
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006404884564014501,
+      "loss": 0.5797,
+      "step": 38120
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006404460556721576,
+      "loss": 0.5626,
+      "step": 38130
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000640403654942865,
+      "loss": 0.5758,
+      "step": 38140
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006403612542135725,
+      "loss": 0.636,
+      "step": 38150
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00064031885348428,
+      "loss": 0.5421,
+      "step": 38160
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006402764527549874,
+      "loss": 0.6274,
+      "step": 38170
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006402340520256949,
+      "loss": 1.0423,
+      "step": 38180
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006401916512964024,
+      "loss": 0.5282,
+      "step": 38190
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006401492505671097,
+      "loss": 0.5791,
+      "step": 38200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006401068498378172,
+      "loss": 0.6264,
+      "step": 38210
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006400644491085248,
+      "loss": 0.6134,
+      "step": 38220
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006400220483792321,
+      "loss": 0.595,
+      "step": 38230
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006399796476499396,
+      "loss": 0.6046,
+      "step": 38240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000639937246920647,
+      "loss": 0.4572,
+      "step": 38250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006398948461913545,
+      "loss": 0.5985,
+      "step": 38260
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000639852445462062,
+      "loss": 0.462,
+      "step": 38270
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006398100447327695,
+      "loss": 0.5743,
+      "step": 38280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006397676440034768,
+      "loss": 0.5352,
+      "step": 38290
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006397252432741844,
+      "loss": 0.5709,
+      "step": 38300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006396828425448919,
+      "loss": 0.5659,
+      "step": 38310
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006396404418155992,
+      "loss": 0.6955,
+      "step": 38320
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006395980410863067,
+      "loss": 0.5623,
+      "step": 38330
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006395556403570143,
+      "loss": 0.593,
+      "step": 38340
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006395132396277216,
+      "loss": 0.5411,
+      "step": 38350
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006394708388984291,
+      "loss": 0.6041,
+      "step": 38360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006394284381691365,
+      "loss": 0.5911,
+      "step": 38370
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000639386037439844,
+      "loss": 0.6456,
+      "step": 38380
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006393436367105515,
+      "loss": 0.5982,
+      "step": 38390
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006393012359812589,
+      "loss": 0.567,
+      "step": 38400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006392588352519663,
+      "loss": 0.5701,
+      "step": 38410
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006392164345226739,
+      "loss": 0.6319,
+      "step": 38420
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006391740337933812,
+      "loss": 0.5555,
+      "step": 38430
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006391316330640887,
+      "loss": 0.5087,
+      "step": 38440
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006390892323347962,
+      "loss": 0.5948,
+      "step": 38450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006390468316055036,
+      "loss": 0.5538,
+      "step": 38460
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006390044308762111,
+      "loss": 0.503,
+      "step": 38470
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006389620301469186,
+      "loss": 0.5129,
+      "step": 38480
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006389196294176259,
+      "loss": 0.6116,
+      "step": 38490
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006388772286883335,
+      "loss": 0.5761,
+      "step": 38500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000638834827959041,
+      "loss": 0.5907,
+      "step": 38510
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006387924272297483,
+      "loss": 0.5914,
+      "step": 38520
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006387500265004558,
+      "loss": 0.5859,
+      "step": 38530
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006387076257711634,
+      "loss": 0.5369,
+      "step": 38540
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006386652250418707,
+      "loss": 0.6555,
+      "step": 38550
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006386228243125782,
+      "loss": 0.6267,
+      "step": 38560
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006385804235832857,
+      "loss": 0.6781,
+      "step": 38570
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006385380228539931,
+      "loss": 0.5366,
+      "step": 38580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006384956221247006,
+      "loss": 0.5004,
+      "step": 38590
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006384532213954081,
+      "loss": 0.5173,
+      "step": 38600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006384108206661154,
+      "loss": 0.5856,
+      "step": 38610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000638368419936823,
+      "loss": 0.5892,
+      "step": 38620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006383260192075305,
+      "loss": 0.5863,
+      "step": 38630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006382836184782378,
+      "loss": 0.7465,
+      "step": 38640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006382412177489453,
+      "loss": 0.5423,
+      "step": 38650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006381988170196528,
+      "loss": 0.5473,
+      "step": 38660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006381564162903602,
+      "loss": 0.6195,
+      "step": 38670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006381140155610677,
+      "loss": 0.5709,
+      "step": 38680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006380716148317752,
+      "loss": 0.5441,
+      "step": 38690
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006380292141024826,
+      "loss": 0.573,
+      "step": 38700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006379868133731901,
+      "loss": 0.6984,
+      "step": 38710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006379444126438976,
+      "loss": 0.497,
+      "step": 38720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006379020119146049,
+      "loss": 0.6734,
+      "step": 38730
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006378596111853124,
+      "loss": 0.6099,
+      "step": 38740
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006378172104560198,
+      "loss": 0.5798,
+      "step": 38750
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006377748097267273,
+      "loss": 0.5783,
+      "step": 38760
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006377324089974348,
+      "loss": 0.5854,
+      "step": 38770
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006376900082681422,
+      "loss": 0.5601,
+      "step": 38780
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006376476075388497,
+      "loss": 0.5597,
+      "step": 38790
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006376052068095572,
+      "loss": 0.643,
+      "step": 38800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006375628060802645,
+      "loss": 0.6061,
+      "step": 38810
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000637520405350972,
+      "loss": 0.614,
+      "step": 38820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006374780046216796,
+      "loss": 0.6285,
+      "step": 38830
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006374356038923869,
+      "loss": 0.5379,
+      "step": 38840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006373932031630944,
+      "loss": 0.6096,
+      "step": 38850
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006373508024338019,
+      "loss": 0.4671,
+      "step": 38860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006373084017045093,
+      "loss": 0.4919,
+      "step": 38870
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006372660009752168,
+      "loss": 0.5478,
+      "step": 38880
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006372236002459243,
+      "loss": 0.5716,
+      "step": 38890
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006371811995166317,
+      "loss": 0.6693,
+      "step": 38900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006371387987873392,
+      "loss": 0.6354,
+      "step": 38910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006370963980580467,
+      "loss": 0.4859,
+      "step": 38920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000637053997328754,
+      "loss": 0.6553,
+      "step": 38930
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006370115965994615,
+      "loss": 0.6254,
+      "step": 38940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006369691958701691,
+      "loss": 0.5227,
+      "step": 38950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006369267951408764,
+      "loss": 0.6704,
+      "step": 38960
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006368843944115839,
+      "loss": 0.539,
+      "step": 38970
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006368419936822914,
+      "loss": 0.5756,
+      "step": 38980
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006367995929529988,
+      "loss": 0.6459,
+      "step": 38990
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006367571922237063,
+      "loss": 0.574,
+      "step": 39000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006367147914944138,
+      "loss": 0.6317,
+      "step": 39010
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006366723907651211,
+      "loss": 0.5662,
+      "step": 39020
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006366299900358287,
+      "loss": 0.6288,
+      "step": 39030
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006365875893065361,
+      "loss": 0.6249,
+      "step": 39040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006365451885772435,
+      "loss": 0.5671,
+      "step": 39050
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000636502787847951,
+      "loss": 0.5595,
+      "step": 39060
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006364603871186585,
+      "loss": 0.5487,
+      "step": 39070
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006364179863893659,
+      "loss": 0.6679,
+      "step": 39080
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006363755856600734,
+      "loss": 0.669,
+      "step": 39090
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006363331849307808,
+      "loss": 0.6747,
+      "step": 39100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006362907842014883,
+      "loss": 0.5714,
+      "step": 39110
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006362483834721958,
+      "loss": 0.5654,
+      "step": 39120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006362059827429032,
+      "loss": 0.5714,
+      "step": 39130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006361635820136106,
+      "loss": 0.5507,
+      "step": 39140
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006361211812843182,
+      "loss": 0.6204,
+      "step": 39150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006360787805550256,
+      "loss": 0.5869,
+      "step": 39160
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000636036379825733,
+      "loss": 0.5017,
+      "step": 39170
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006359939790964405,
+      "loss": 0.6093,
+      "step": 39180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000635951578367148,
+      "loss": 0.6214,
+      "step": 39190
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006359091776378554,
+      "loss": 0.6087,
+      "step": 39200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006358667769085629,
+      "loss": 0.4818,
+      "step": 39210
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006358243761792702,
+      "loss": 0.5813,
+      "step": 39220
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006357819754499778,
+      "loss": 0.6087,
+      "step": 39230
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006357395747206853,
+      "loss": 0.6092,
+      "step": 39240
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006356971739913926,
+      "loss": 0.6679,
+      "step": 39250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006356547732621001,
+      "loss": 0.6649,
+      "step": 39260
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006356123725328076,
+      "loss": 0.5302,
+      "step": 39270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000635569971803515,
+      "loss": 0.5459,
+      "step": 39280
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006355275710742225,
+      "loss": 0.5934,
+      "step": 39290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00063548517034493,
+      "loss": 0.6161,
+      "step": 39300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006354427696156374,
+      "loss": 0.5832,
+      "step": 39310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006354003688863449,
+      "loss": 0.5874,
+      "step": 39320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006353579681570524,
+      "loss": 0.5814,
+      "step": 39330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006353155674277597,
+      "loss": 0.6918,
+      "step": 39340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006352731666984672,
+      "loss": 0.6171,
+      "step": 39350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006352307659691747,
+      "loss": 0.4692,
+      "step": 39360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006351883652398821,
+      "loss": 0.6155,
+      "step": 39370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006351459645105896,
+      "loss": 0.7185,
+      "step": 39380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006351035637812971,
+      "loss": 0.4758,
+      "step": 39390
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006350611630520045,
+      "loss": 0.5741,
+      "step": 39400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000635018762322712,
+      "loss": 0.6034,
+      "step": 39410
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006349763615934194,
+      "loss": 0.4682,
+      "step": 39420
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006349339608641269,
+      "loss": 0.6522,
+      "step": 39430
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006348915601348344,
+      "loss": 0.5565,
+      "step": 39440
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006348491594055418,
+      "loss": 0.603,
+      "step": 39450
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006348067586762492,
+      "loss": 0.5157,
+      "step": 39460
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006347643579469567,
+      "loss": 0.5789,
+      "step": 39470
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006347219572176642,
+      "loss": 0.541,
+      "step": 39480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006346795564883716,
+      "loss": 0.6552,
+      "step": 39490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006346371557590791,
+      "loss": 0.5832,
+      "step": 39500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006345947550297866,
+      "loss": 0.5883,
+      "step": 39510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000634552354300494,
+      "loss": 0.602,
+      "step": 39520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006345099535712015,
+      "loss": 0.5786,
+      "step": 39530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006344675528419089,
+      "loss": 0.5217,
+      "step": 39540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006344251521126163,
+      "loss": 0.5893,
+      "step": 39550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006343827513833239,
+      "loss": 0.5391,
+      "step": 39560
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006343403506540313,
+      "loss": 0.6379,
+      "step": 39570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006342979499247387,
+      "loss": 0.6073,
+      "step": 39580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006342555491954462,
+      "loss": 0.5449,
+      "step": 39590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006342131484661537,
+      "loss": 0.6176,
+      "step": 39600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006341707477368611,
+      "loss": 0.5854,
+      "step": 39610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006341283470075686,
+      "loss": 0.6026,
+      "step": 39620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000634085946278276,
+      "loss": 0.7012,
+      "step": 39630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006340435455489835,
+      "loss": 0.493,
+      "step": 39640
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000634001144819691,
+      "loss": 0.565,
+      "step": 39650
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006339587440903984,
+      "loss": 0.5109,
+      "step": 39660
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006339163433611058,
+      "loss": 0.5588,
+      "step": 39670
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006338739426318133,
+      "loss": 0.6806,
+      "step": 39680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006338315419025208,
+      "loss": 0.6096,
+      "step": 39690
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006337891411732282,
+      "loss": 0.6131,
+      "step": 39700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006337467404439356,
+      "loss": 0.5603,
+      "step": 39710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006337043397146432,
+      "loss": 0.6375,
+      "step": 39720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006336619389853506,
+      "loss": 0.6107,
+      "step": 39730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000633619538256058,
+      "loss": 0.5562,
+      "step": 39740
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006335771375267654,
+      "loss": 0.5609,
+      "step": 39750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000633534736797473,
+      "loss": 0.6522,
+      "step": 39760
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006334923360681804,
+      "loss": 0.5204,
+      "step": 39770
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006334499353388878,
+      "loss": 0.6555,
+      "step": 39780
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006334075346095953,
+      "loss": 0.5546,
+      "step": 39790
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006333651338803028,
+      "loss": 0.5227,
+      "step": 39800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006333227331510102,
+      "loss": 0.5434,
+      "step": 39810
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006332803324217177,
+      "loss": 0.5781,
+      "step": 39820
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006332379316924251,
+      "loss": 0.5375,
+      "step": 39830
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006331955309631326,
+      "loss": 0.5492,
+      "step": 39840
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006331531302338401,
+      "loss": 0.6527,
+      "step": 39850
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006331107295045475,
+      "loss": 0.6247,
+      "step": 39860
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006330683287752549,
+      "loss": 0.5187,
+      "step": 39870
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006330259280459625,
+      "loss": 0.5244,
+      "step": 39880
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006329835273166699,
+      "loss": 0.6077,
+      "step": 39890
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006329411265873773,
+      "loss": 0.5265,
+      "step": 39900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006328987258580848,
+      "loss": 0.574,
+      "step": 39910
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006328563251287923,
+      "loss": 0.5568,
+      "step": 39920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006328139243994997,
+      "loss": 0.7126,
+      "step": 39930
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006327715236702072,
+      "loss": 0.5745,
+      "step": 39940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006327291229409146,
+      "loss": 0.5836,
+      "step": 39950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006326867222116221,
+      "loss": 0.5586,
+      "step": 39960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006326443214823295,
+      "loss": 0.6035,
+      "step": 39970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.000632601920753037,
+      "loss": 0.5863,
+      "step": 39980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006325595200237444,
+      "loss": 0.5918,
+      "step": 39990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006325171192944519,
+      "loss": 0.585,
+      "step": 40000
+    },
+    {
+      "epoch": 1.69,
+      "eval_loss": 0.6182475090026855,
+      "eval_runtime": 337.5574,
+      "eval_samples_per_second": 15.568,
+      "eval_steps_per_second": 3.893,
+      "step": 40000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006324747185651594,
+      "loss": 0.5426,
+      "step": 40010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006324323178358668,
+      "loss": 0.6274,
+      "step": 40020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006323899171065742,
+      "loss": 0.5879,
+      "step": 40030
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006323475163772818,
+      "loss": 0.5892,
+      "step": 40040
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006323051156479892,
+      "loss": 0.5792,
+      "step": 40050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006322627149186966,
+      "loss": 0.6454,
+      "step": 40060
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006322203141894041,
+      "loss": 0.532,
+      "step": 40070
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006321779134601115,
+      "loss": 0.678,
+      "step": 40080
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000632135512730819,
+      "loss": 0.5556,
+      "step": 40090
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006320931120015265,
+      "loss": 0.5789,
+      "step": 40100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006320507112722339,
+      "loss": 0.7067,
+      "step": 40110
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006320083105429414,
+      "loss": 0.5701,
+      "step": 40120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006319659098136489,
+      "loss": 0.6824,
+      "step": 40130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006319235090843563,
+      "loss": 0.6875,
+      "step": 40140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006318811083550637,
+      "loss": 0.6386,
+      "step": 40150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006318387076257711,
+      "loss": 0.5898,
+      "step": 40160
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006317963068964787,
+      "loss": 0.5183,
+      "step": 40170
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006317539061671861,
+      "loss": 0.4507,
+      "step": 40180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006317115054378935,
+      "loss": 0.6666,
+      "step": 40190
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000631669104708601,
+      "loss": 0.508,
+      "step": 40200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006316267039793085,
+      "loss": 0.5691,
+      "step": 40210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000631584303250016,
+      "loss": 0.6355,
+      "step": 40220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006315419025207234,
+      "loss": 0.592,
+      "step": 40230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006314995017914308,
+      "loss": 0.6097,
+      "step": 40240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006314571010621383,
+      "loss": 0.5983,
+      "step": 40250
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006314147003328458,
+      "loss": 0.673,
+      "step": 40260
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006313722996035532,
+      "loss": 0.6025,
+      "step": 40270
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006313298988742606,
+      "loss": 0.6121,
+      "step": 40280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006312874981449681,
+      "loss": 0.6074,
+      "step": 40290
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006312450974156756,
+      "loss": 0.5227,
+      "step": 40300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000631202696686383,
+      "loss": 0.5474,
+      "step": 40310
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006311602959570904,
+      "loss": 0.4988,
+      "step": 40320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000631117895227798,
+      "loss": 0.6253,
+      "step": 40330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006310754944985054,
+      "loss": 0.551,
+      "step": 40340
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006310330937692128,
+      "loss": 0.5844,
+      "step": 40350
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006309906930399203,
+      "loss": 0.6082,
+      "step": 40360
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006309482923106278,
+      "loss": 0.5127,
+      "step": 40370
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006309058915813352,
+      "loss": 0.5563,
+      "step": 40380
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006308634908520427,
+      "loss": 0.536,
+      "step": 40390
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006308210901227501,
+      "loss": 0.4597,
+      "step": 40400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006307786893934576,
+      "loss": 0.5677,
+      "step": 40410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006307362886641651,
+      "loss": 0.5443,
+      "step": 40420
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006306938879348725,
+      "loss": 0.5492,
+      "step": 40430
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006306514872055799,
+      "loss": 0.4875,
+      "step": 40440
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006306090864762875,
+      "loss": 0.5447,
+      "step": 40450
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006305666857469949,
+      "loss": 0.5852,
+      "step": 40460
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006305242850177023,
+      "loss": 0.5139,
+      "step": 40470
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006304818842884098,
+      "loss": 0.6172,
+      "step": 40480
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006304394835591173,
+      "loss": 0.5326,
+      "step": 40490
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006303970828298247,
+      "loss": 0.6218,
+      "step": 40500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006303546821005322,
+      "loss": 0.5461,
+      "step": 40510
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006303122813712396,
+      "loss": 0.5841,
+      "step": 40520
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006302698806419471,
+      "loss": 0.5323,
+      "step": 40530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006302274799126546,
+      "loss": 0.5947,
+      "step": 40540
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000630185079183362,
+      "loss": 0.5795,
+      "step": 40550
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006301426784540694,
+      "loss": 0.5301,
+      "step": 40560
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000630100277724777,
+      "loss": 0.5347,
+      "step": 40570
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006300578769954843,
+      "loss": 0.5948,
+      "step": 40580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006300154762661918,
+      "loss": 0.5919,
+      "step": 40590
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006299730755368993,
+      "loss": 0.5644,
+      "step": 40600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006299306748076067,
+      "loss": 0.6178,
+      "step": 40610
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006298882740783142,
+      "loss": 0.6663,
+      "step": 40620
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006298458733490217,
+      "loss": 0.4909,
+      "step": 40630
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000629803472619729,
+      "loss": 0.4903,
+      "step": 40640
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006297610718904366,
+      "loss": 0.5632,
+      "step": 40650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000629718671161144,
+      "loss": 0.5755,
+      "step": 40660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006296762704318514,
+      "loss": 0.5456,
+      "step": 40670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006296338697025589,
+      "loss": 0.4702,
+      "step": 40680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006295914689732663,
+      "loss": 0.6645,
+      "step": 40690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006295490682439738,
+      "loss": 0.571,
+      "step": 40700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006295066675146813,
+      "loss": 0.6265,
+      "step": 40710
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006294642667853887,
+      "loss": 0.5373,
+      "step": 40720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006294218660560962,
+      "loss": 0.5615,
+      "step": 40730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006293794653268037,
+      "loss": 0.6148,
+      "step": 40740
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006293370645975111,
+      "loss": 0.5863,
+      "step": 40750
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006292946638682185,
+      "loss": 0.6081,
+      "step": 40760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000629252263138926,
+      "loss": 0.5869,
+      "step": 40770
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006292098624096335,
+      "loss": 0.5189,
+      "step": 40780
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006291674616803409,
+      "loss": 0.5895,
+      "step": 40790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006291250609510484,
+      "loss": 0.5804,
+      "step": 40800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006290826602217558,
+      "loss": 0.652,
+      "step": 40810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006290402594924633,
+      "loss": 0.6245,
+      "step": 40820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006289978587631708,
+      "loss": 0.5259,
+      "step": 40830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006289554580338782,
+      "loss": 0.6479,
+      "step": 40840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006289130573045856,
+      "loss": 0.6396,
+      "step": 40850
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006288706565752932,
+      "loss": 0.6402,
+      "step": 40860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006288282558460006,
+      "loss": 0.5821,
+      "step": 40870
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000628785855116708,
+      "loss": 0.5575,
+      "step": 40880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006287434543874155,
+      "loss": 0.5848,
+      "step": 40890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006287010536581229,
+      "loss": 0.5589,
+      "step": 40900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006286586529288304,
+      "loss": 0.5364,
+      "step": 40910
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006286162521995379,
+      "loss": 0.5688,
+      "step": 40920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006285738514702453,
+      "loss": 0.4957,
+      "step": 40930
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006285314507409528,
+      "loss": 0.6339,
+      "step": 40940
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006284890500116603,
+      "loss": 0.6041,
+      "step": 40950
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006284466492823676,
+      "loss": 0.6032,
+      "step": 40960
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006284042485530751,
+      "loss": 0.581,
+      "step": 40970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006283618478237827,
+      "loss": 0.6911,
+      "step": 40980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00062831944709449,
+      "loss": 0.6301,
+      "step": 40990
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006282770463651975,
+      "loss": 0.6087,
+      "step": 41000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000628234645635905,
+      "loss": 0.5661,
+      "step": 41010
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006281922449066124,
+      "loss": 0.6667,
+      "step": 41020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006281498441773199,
+      "loss": 0.655,
+      "step": 41030
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006281074434480274,
+      "loss": 0.5581,
+      "step": 41040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006280650427187347,
+      "loss": 0.6045,
+      "step": 41050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006280226419894423,
+      "loss": 0.5529,
+      "step": 41060
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006279802412601498,
+      "loss": 0.6128,
+      "step": 41070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006279378405308571,
+      "loss": 0.5946,
+      "step": 41080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006278954398015646,
+      "loss": 0.5492,
+      "step": 41090
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006278530390722722,
+      "loss": 0.5525,
+      "step": 41100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006278106383429795,
+      "loss": 0.522,
+      "step": 41110
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000627768237613687,
+      "loss": 0.6424,
+      "step": 41120
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006277258368843945,
+      "loss": 0.5707,
+      "step": 41130
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006276834361551019,
+      "loss": 0.6055,
+      "step": 41140
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006276410354258094,
+      "loss": 0.5524,
+      "step": 41150
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006275986346965169,
+      "loss": 0.6014,
+      "step": 41160
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006275562339672242,
+      "loss": 0.5136,
+      "step": 41170
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006275138332379318,
+      "loss": 0.5601,
+      "step": 41180
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006274714325086391,
+      "loss": 0.649,
+      "step": 41190
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006274290317793466,
+      "loss": 0.5739,
+      "step": 41200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006273866310500541,
+      "loss": 0.4601,
+      "step": 41210
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006273442303207615,
+      "loss": 0.6124,
+      "step": 41220
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000627301829591469,
+      "loss": 0.5127,
+      "step": 41230
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006272594288621765,
+      "loss": 0.6041,
+      "step": 41240
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006272170281328838,
+      "loss": 0.5507,
+      "step": 41250
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006271746274035914,
+      "loss": 0.6215,
+      "step": 41260
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006271322266742989,
+      "loss": 0.5771,
+      "step": 41270
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006270898259450062,
+      "loss": 0.5565,
+      "step": 41280
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006270474252157137,
+      "loss": 0.5613,
+      "step": 41290
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006270050244864212,
+      "loss": 0.5661,
+      "step": 41300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006269626237571286,
+      "loss": 0.5549,
+      "step": 41310
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006269202230278361,
+      "loss": 0.5698,
+      "step": 41320
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006268778222985436,
+      "loss": 0.6032,
+      "step": 41330
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000626835421569251,
+      "loss": 0.5016,
+      "step": 41340
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006267930208399585,
+      "loss": 0.65,
+      "step": 41350
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000626750620110666,
+      "loss": 0.5221,
+      "step": 41360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006267082193813733,
+      "loss": 0.595,
+      "step": 41370
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006266658186520809,
+      "loss": 0.4469,
+      "step": 41380
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006266234179227884,
+      "loss": 0.4944,
+      "step": 41390
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006265810171934957,
+      "loss": 0.5513,
+      "step": 41400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006265386164642032,
+      "loss": 0.6646,
+      "step": 41410
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006264962157349107,
+      "loss": 0.5601,
+      "step": 41420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006264538150056181,
+      "loss": 0.5495,
+      "step": 41430
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006264114142763256,
+      "loss": 0.5021,
+      "step": 41440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006263690135470331,
+      "loss": 0.589,
+      "step": 41450
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006263266128177405,
+      "loss": 0.6017,
+      "step": 41460
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000626284212088448,
+      "loss": 0.6146,
+      "step": 41470
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006262418113591555,
+      "loss": 0.6958,
+      "step": 41480
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006261994106298628,
+      "loss": 0.6282,
+      "step": 41490
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006261570099005703,
+      "loss": 0.5522,
+      "step": 41500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006261146091712778,
+      "loss": 0.6089,
+      "step": 41510
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006260722084419852,
+      "loss": 0.565,
+      "step": 41520
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006260298077126927,
+      "loss": 0.5942,
+      "step": 41530
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006259874069834002,
+      "loss": 0.5994,
+      "step": 41540
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006259450062541076,
+      "loss": 0.5144,
+      "step": 41550
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006259026055248151,
+      "loss": 0.5488,
+      "step": 41560
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006258602047955224,
+      "loss": 0.5206,
+      "step": 41570
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006258178040662299,
+      "loss": 0.672,
+      "step": 41580
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006257754033369375,
+      "loss": 0.5341,
+      "step": 41590
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006257330026076448,
+      "loss": 0.5835,
+      "step": 41600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006256906018783523,
+      "loss": 0.6827,
+      "step": 41610
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006256482011490598,
+      "loss": 0.5773,
+      "step": 41620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006256058004197672,
+      "loss": 0.563,
+      "step": 41630
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006255633996904747,
+      "loss": 0.4976,
+      "step": 41640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006255209989611822,
+      "loss": 0.534,
+      "step": 41650
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006254785982318895,
+      "loss": 0.5606,
+      "step": 41660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006254361975025971,
+      "loss": 0.5169,
+      "step": 41670
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006253937967733046,
+      "loss": 0.549,
+      "step": 41680
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006253513960440119,
+      "loss": 0.7435,
+      "step": 41690
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006253089953147194,
+      "loss": 0.5547,
+      "step": 41700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.000625266594585427,
+      "loss": 0.6174,
+      "step": 41710
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006252241938561343,
+      "loss": 0.5857,
+      "step": 41720
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006251817931268418,
+      "loss": 0.5375,
+      "step": 41730
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006251393923975493,
+      "loss": 0.6071,
+      "step": 41740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006250969916682567,
+      "loss": 0.5182,
+      "step": 41750
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006250545909389642,
+      "loss": 0.6664,
+      "step": 41760
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006250121902096717,
+      "loss": 0.6367,
+      "step": 41770
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.000624969789480379,
+      "loss": 0.5673,
+      "step": 41780
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006249273887510866,
+      "loss": 0.7235,
+      "step": 41790
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006248849880217941,
+      "loss": 0.5942,
+      "step": 41800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006248425872925014,
+      "loss": 0.641,
+      "step": 41810
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006248001865632089,
+      "loss": 0.6033,
+      "step": 41820
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006247577858339164,
+      "loss": 0.6109,
+      "step": 41830
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006247153851046238,
+      "loss": 0.6247,
+      "step": 41840
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006246729843753313,
+      "loss": 0.5922,
+      "step": 41850
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006246305836460387,
+      "loss": 0.6587,
+      "step": 41860
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006245881829167462,
+      "loss": 0.5594,
+      "step": 41870
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006245457821874537,
+      "loss": 0.7009,
+      "step": 41880
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006245033814581611,
+      "loss": 0.6113,
+      "step": 41890
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006244609807288685,
+      "loss": 0.4689,
+      "step": 41900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006244185799995761,
+      "loss": 0.4873,
+      "step": 41910
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006243761792702835,
+      "loss": 0.5575,
+      "step": 41920
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006243337785409909,
+      "loss": 0.5808,
+      "step": 41930
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006242913778116984,
+      "loss": 0.5909,
+      "step": 41940
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006242489770824059,
+      "loss": 0.5974,
+      "step": 41950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006242065763531133,
+      "loss": 0.6776,
+      "step": 41960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006241641756238208,
+      "loss": 0.5409,
+      "step": 41970
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006241217748945282,
+      "loss": 0.5944,
+      "step": 41980
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006240793741652357,
+      "loss": 0.5664,
+      "step": 41990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006240369734359432,
+      "loss": 0.621,
+      "step": 42000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006239945727066506,
+      "loss": 0.6263,
+      "step": 42010
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.000623952171977358,
+      "loss": 0.5701,
+      "step": 42020
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006239097712480655,
+      "loss": 0.5263,
+      "step": 42030
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.000623867370518773,
+      "loss": 0.5343,
+      "step": 42040
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006238249697894804,
+      "loss": 0.7297,
+      "step": 42050
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006237825690601879,
+      "loss": 0.6225,
+      "step": 42060
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006237401683308954,
+      "loss": 0.6182,
+      "step": 42070
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006236977676016028,
+      "loss": 0.6286,
+      "step": 42080
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006236553668723103,
+      "loss": 0.5204,
+      "step": 42090
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006236129661430176,
+      "loss": 0.5539,
+      "step": 42100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006235705654137251,
+      "loss": 0.6387,
+      "step": 42110
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006235281646844326,
+      "loss": 0.6853,
+      "step": 42120
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00062348576395514,
+      "loss": 0.5343,
+      "step": 42130
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006234433632258475,
+      "loss": 0.5768,
+      "step": 42140
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.000623400962496555,
+      "loss": 0.682,
+      "step": 42150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006233585617672624,
+      "loss": 0.6244,
+      "step": 42160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006233161610379699,
+      "loss": 0.5599,
+      "step": 42170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006232737603086773,
+      "loss": 0.5796,
+      "step": 42180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006232313595793847,
+      "loss": 0.5682,
+      "step": 42190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006231889588500923,
+      "loss": 0.5569,
+      "step": 42200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006231465581207997,
+      "loss": 0.5363,
+      "step": 42210
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006231041573915071,
+      "loss": 0.6217,
+      "step": 42220
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006230617566622146,
+      "loss": 0.5763,
+      "step": 42230
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006230193559329221,
+      "loss": 0.6068,
+      "step": 42240
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006229769552036295,
+      "loss": 0.5501,
+      "step": 42250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000622934554474337,
+      "loss": 0.6336,
+      "step": 42260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006228921537450444,
+      "loss": 0.5064,
+      "step": 42270
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006228497530157519,
+      "loss": 0.535,
+      "step": 42280
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006228073522864594,
+      "loss": 0.6124,
+      "step": 42290
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006227649515571668,
+      "loss": 0.5892,
+      "step": 42300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006227225508278742,
+      "loss": 0.5643,
+      "step": 42310
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006226801500985818,
+      "loss": 0.628,
+      "step": 42320
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006226377493692892,
+      "loss": 0.6752,
+      "step": 42330
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006225953486399966,
+      "loss": 0.6427,
+      "step": 42340
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006225529479107041,
+      "loss": 0.6457,
+      "step": 42350
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006225105471814116,
+      "loss": 0.6105,
+      "step": 42360
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000622468146452119,
+      "loss": 0.6257,
+      "step": 42370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006224257457228265,
+      "loss": 0.5828,
+      "step": 42380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006223833449935339,
+      "loss": 0.6686,
+      "step": 42390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006223409442642414,
+      "loss": 0.5731,
+      "step": 42400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006222985435349489,
+      "loss": 0.6268,
+      "step": 42410
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006222561428056563,
+      "loss": 0.5006,
+      "step": 42420
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006222137420763637,
+      "loss": 0.5582,
+      "step": 42430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006221713413470712,
+      "loss": 0.4704,
+      "step": 42440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006221289406177787,
+      "loss": 0.6218,
+      "step": 42450
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006220865398884861,
+      "loss": 0.6605,
+      "step": 42460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006220441391591935,
+      "loss": 0.5671,
+      "step": 42470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006220017384299011,
+      "loss": 0.6484,
+      "step": 42480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006219593377006085,
+      "loss": 0.503,
+      "step": 42490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006219169369713159,
+      "loss": 0.5798,
+      "step": 42500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006218745362420233,
+      "loss": 0.5666,
+      "step": 42510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006218321355127309,
+      "loss": 0.618,
+      "step": 42520
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006217897347834383,
+      "loss": 0.7225,
+      "step": 42530
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006217473340541457,
+      "loss": 0.5401,
+      "step": 42540
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006217049333248532,
+      "loss": 0.6292,
+      "step": 42550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006216625325955607,
+      "loss": 0.5774,
+      "step": 42560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006216201318662682,
+      "loss": 0.5775,
+      "step": 42570
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006215777311369756,
+      "loss": 0.6216,
+      "step": 42580
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000621535330407683,
+      "loss": 0.6002,
+      "step": 42590
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006214929296783906,
+      "loss": 0.6121,
+      "step": 42600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000621450528949098,
+      "loss": 0.6587,
+      "step": 42610
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006214081282198054,
+      "loss": 0.4774,
+      "step": 42620
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006213657274905128,
+      "loss": 0.5954,
+      "step": 42630
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006213233267612203,
+      "loss": 0.546,
+      "step": 42640
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006212809260319278,
+      "loss": 0.6274,
+      "step": 42650
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006212385253026352,
+      "loss": 0.549,
+      "step": 42660
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006211961245733427,
+      "loss": 0.5217,
+      "step": 42670
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006211537238440502,
+      "loss": 0.6003,
+      "step": 42680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006211113231147576,
+      "loss": 0.5325,
+      "step": 42690
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006210689223854651,
+      "loss": 0.5656,
+      "step": 42700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006210265216561725,
+      "loss": 0.5843,
+      "step": 42710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006209841209268799,
+      "loss": 0.5326,
+      "step": 42720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006209417201975874,
+      "loss": 0.6718,
+      "step": 42730
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006208993194682949,
+      "loss": 0.5997,
+      "step": 42740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006208569187390023,
+      "loss": 0.555,
+      "step": 42750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006208145180097098,
+      "loss": 0.5723,
+      "step": 42760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006207721172804173,
+      "loss": 0.7162,
+      "step": 42770
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006207297165511247,
+      "loss": 0.6034,
+      "step": 42780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006206873158218321,
+      "loss": 0.4909,
+      "step": 42790
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006206449150925396,
+      "loss": 0.6001,
+      "step": 42800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006206025143632471,
+      "loss": 0.5577,
+      "step": 42810
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006205601136339545,
+      "loss": 0.5302,
+      "step": 42820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000620517712904662,
+      "loss": 0.7032,
+      "step": 42830
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006204753121753694,
+      "loss": 0.6105,
+      "step": 42840
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006204329114460769,
+      "loss": 0.5439,
+      "step": 42850
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006203905107167844,
+      "loss": 0.6147,
+      "step": 42860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006203481099874918,
+      "loss": 0.6507,
+      "step": 42870
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006203057092581992,
+      "loss": 0.537,
+      "step": 42880
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006202633085289068,
+      "loss": 0.5432,
+      "step": 42890
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006202209077996142,
+      "loss": 0.565,
+      "step": 42900
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006201785070703216,
+      "loss": 0.5427,
+      "step": 42910
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006201361063410291,
+      "loss": 0.5936,
+      "step": 42920
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006200937056117366,
+      "loss": 0.64,
+      "step": 42930
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000620051304882444,
+      "loss": 0.5198,
+      "step": 42940
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006200089041531515,
+      "loss": 0.5847,
+      "step": 42950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006199665034238589,
+      "loss": 0.5382,
+      "step": 42960
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006199241026945664,
+      "loss": 0.5916,
+      "step": 42970
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006198817019652739,
+      "loss": 0.609,
+      "step": 42980
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006198393012359813,
+      "loss": 0.6113,
+      "step": 42990
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006197969005066887,
+      "loss": 0.6072,
+      "step": 43000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006197544997773963,
+      "loss": 0.6521,
+      "step": 43010
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006197120990481037,
+      "loss": 0.6341,
+      "step": 43020
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006196696983188111,
+      "loss": 0.5327,
+      "step": 43030
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006196272975895185,
+      "loss": 0.5515,
+      "step": 43040
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000619584896860226,
+      "loss": 0.536,
+      "step": 43050
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006195424961309335,
+      "loss": 0.6058,
+      "step": 43060
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000619500095401641,
+      "loss": 0.6576,
+      "step": 43070
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006194576946723483,
+      "loss": 0.6169,
+      "step": 43080
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006194152939430559,
+      "loss": 0.5565,
+      "step": 43090
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006193728932137633,
+      "loss": 0.5909,
+      "step": 43100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006193304924844707,
+      "loss": 0.6015,
+      "step": 43110
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006192880917551782,
+      "loss": 0.5745,
+      "step": 43120
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006192456910258857,
+      "loss": 0.6315,
+      "step": 43130
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006192032902965931,
+      "loss": 0.5972,
+      "step": 43140
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006191608895673006,
+      "loss": 0.5302,
+      "step": 43150
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000619118488838008,
+      "loss": 0.5799,
+      "step": 43160
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006190760881087155,
+      "loss": 0.5787,
+      "step": 43170
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000619033687379423,
+      "loss": 0.6096,
+      "step": 43180
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006189912866501304,
+      "loss": 0.6091,
+      "step": 43190
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006189488859208378,
+      "loss": 0.4776,
+      "step": 43200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006189064851915454,
+      "loss": 0.647,
+      "step": 43210
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006188640844622528,
+      "loss": 0.5232,
+      "step": 43220
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006188216837329602,
+      "loss": 0.5694,
+      "step": 43230
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006187792830036677,
+      "loss": 0.7137,
+      "step": 43240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006187368822743751,
+      "loss": 0.6293,
+      "step": 43250
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006186944815450826,
+      "loss": 0.5722,
+      "step": 43260
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006186520808157901,
+      "loss": 0.5468,
+      "step": 43270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006186096800864975,
+      "loss": 0.6743,
+      "step": 43280
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000618567279357205,
+      "loss": 0.5774,
+      "step": 43290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006185248786279125,
+      "loss": 0.665,
+      "step": 43300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006184824778986199,
+      "loss": 0.6315,
+      "step": 43310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006184400771693273,
+      "loss": 0.4686,
+      "step": 43320
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006183976764400348,
+      "loss": 0.6186,
+      "step": 43330
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006183552757107422,
+      "loss": 0.5621,
+      "step": 43340
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006183128749814497,
+      "loss": 0.5727,
+      "step": 43350
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006182704742521572,
+      "loss": 0.5476,
+      "step": 43360
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006182280735228646,
+      "loss": 0.5919,
+      "step": 43370
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006181856727935721,
+      "loss": 0.6658,
+      "step": 43380
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006181432720642796,
+      "loss": 0.5833,
+      "step": 43390
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006181008713349869,
+      "loss": 0.5359,
+      "step": 43400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006180584706056945,
+      "loss": 0.6815,
+      "step": 43410
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000618016069876402,
+      "loss": 0.6158,
+      "step": 43420
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006179736691471093,
+      "loss": 0.5931,
+      "step": 43430
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006179312684178168,
+      "loss": 0.5563,
+      "step": 43440
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006178888676885243,
+      "loss": 0.6955,
+      "step": 43450
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006178464669592317,
+      "loss": 0.5506,
+      "step": 43460
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006178040662299392,
+      "loss": 0.5916,
+      "step": 43470
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006177616655006467,
+      "loss": 0.5318,
+      "step": 43480
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006177192647713541,
+      "loss": 0.5677,
+      "step": 43490
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006176768640420616,
+      "loss": 0.5753,
+      "step": 43500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000617634463312769,
+      "loss": 0.4892,
+      "step": 43510
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006175920625834764,
+      "loss": 0.5645,
+      "step": 43520
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006175496618541839,
+      "loss": 0.6074,
+      "step": 43530
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006175072611248915,
+      "loss": 0.5384,
+      "step": 43540
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006174648603955988,
+      "loss": 0.5266,
+      "step": 43550
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006174224596663063,
+      "loss": 0.5903,
+      "step": 43560
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006173800589370137,
+      "loss": 0.6302,
+      "step": 43570
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006173376582077212,
+      "loss": 0.667,
+      "step": 43580
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006172952574784287,
+      "loss": 0.6498,
+      "step": 43590
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006172528567491361,
+      "loss": 0.5569,
+      "step": 43600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006172104560198435,
+      "loss": 0.5859,
+      "step": 43610
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006171680552905511,
+      "loss": 0.5957,
+      "step": 43620
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006171256545612585,
+      "loss": 0.5008,
+      "step": 43630
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006170832538319659,
+      "loss": 0.5327,
+      "step": 43640
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006170408531026734,
+      "loss": 0.7035,
+      "step": 43650
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006169984523733808,
+      "loss": 0.7324,
+      "step": 43660
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006169560516440883,
+      "loss": 0.6243,
+      "step": 43670
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006169136509147958,
+      "loss": 0.7245,
+      "step": 43680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006168712501855031,
+      "loss": 0.6368,
+      "step": 43690
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006168288494562107,
+      "loss": 0.5457,
+      "step": 43700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006167864487269182,
+      "loss": 0.5525,
+      "step": 43710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006167440479976255,
+      "loss": 0.5622,
+      "step": 43720
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000616701647268333,
+      "loss": 0.5707,
+      "step": 43730
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006166592465390406,
+      "loss": 0.6482,
+      "step": 43740
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006166168458097479,
+      "loss": 0.5924,
+      "step": 43750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006165744450804554,
+      "loss": 0.6356,
+      "step": 43760
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006165320443511629,
+      "loss": 0.7152,
+      "step": 43770
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006164896436218703,
+      "loss": 0.5268,
+      "step": 43780
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006164472428925778,
+      "loss": 0.5331,
+      "step": 43790
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006164048421632853,
+      "loss": 0.6067,
+      "step": 43800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006163624414339926,
+      "loss": 0.5553,
+      "step": 43810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006163200407047002,
+      "loss": 0.5467,
+      "step": 43820
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006162776399754077,
+      "loss": 0.4813,
+      "step": 43830
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000616235239246115,
+      "loss": 0.5494,
+      "step": 43840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006161928385168225,
+      "loss": 0.533,
+      "step": 43850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00061615043778753,
+      "loss": 0.6369,
+      "step": 43860
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006161080370582374,
+      "loss": 0.7125,
+      "step": 43870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006160656363289449,
+      "loss": 0.6046,
+      "step": 43880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006160232355996524,
+      "loss": 0.5251,
+      "step": 43890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006159808348703598,
+      "loss": 0.5718,
+      "step": 43900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006159384341410673,
+      "loss": 0.6478,
+      "step": 43910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006158960334117748,
+      "loss": 0.5402,
+      "step": 43920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006158536326824821,
+      "loss": 0.4754,
+      "step": 43930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006158112319531897,
+      "loss": 0.6246,
+      "step": 43940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006157688312238972,
+      "loss": 0.5531,
+      "step": 43950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006157264304946045,
+      "loss": 0.6153,
+      "step": 43960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000615684029765312,
+      "loss": 0.5588,
+      "step": 43970
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006156416290360194,
+      "loss": 0.5548,
+      "step": 43980
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006155992283067269,
+      "loss": 0.6271,
+      "step": 43990
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006155568275774344,
+      "loss": 0.5514,
+      "step": 44000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006155144268481417,
+      "loss": 0.5239,
+      "step": 44010
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006154720261188493,
+      "loss": 0.6776,
+      "step": 44020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006154296253895568,
+      "loss": 0.6317,
+      "step": 44030
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006153872246602641,
+      "loss": 0.6495,
+      "step": 44040
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006153448239309716,
+      "loss": 0.5355,
+      "step": 44050
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006153024232016791,
+      "loss": 0.5914,
+      "step": 44060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006152600224723865,
+      "loss": 0.5448,
+      "step": 44070
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000615217621743094,
+      "loss": 0.5764,
+      "step": 44080
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006151752210138015,
+      "loss": 0.5872,
+      "step": 44090
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006151328202845089,
+      "loss": 0.4692,
+      "step": 44100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006150904195552164,
+      "loss": 0.6471,
+      "step": 44110
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006150480188259239,
+      "loss": 0.5143,
+      "step": 44120
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006150056180966312,
+      "loss": 0.6618,
+      "step": 44130
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006149632173673387,
+      "loss": 0.501,
+      "step": 44140
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006149208166380463,
+      "loss": 0.6278,
+      "step": 44150
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006148784159087536,
+      "loss": 0.5065,
+      "step": 44160
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006148360151794611,
+      "loss": 0.6516,
+      "step": 44170
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006147936144501686,
+      "loss": 0.7077,
+      "step": 44180
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000614751213720876,
+      "loss": 0.6638,
+      "step": 44190
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006147088129915835,
+      "loss": 0.5935,
+      "step": 44200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000614666412262291,
+      "loss": 0.5317,
+      "step": 44210
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006146240115329983,
+      "loss": 0.591,
+      "step": 44220
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006145816108037059,
+      "loss": 0.5809,
+      "step": 44230
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006145392100744134,
+      "loss": 0.5758,
+      "step": 44240
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006144968093451207,
+      "loss": 0.6067,
+      "step": 44250
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006144544086158282,
+      "loss": 0.6172,
+      "step": 44260
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006144120078865357,
+      "loss": 0.6034,
+      "step": 44270
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006143696071572431,
+      "loss": 0.5916,
+      "step": 44280
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006143272064279506,
+      "loss": 0.523,
+      "step": 44290
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000614284805698658,
+      "loss": 0.5579,
+      "step": 44300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006142424049693655,
+      "loss": 0.6438,
+      "step": 44310
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000614200004240073,
+      "loss": 0.6097,
+      "step": 44320
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006141576035107804,
+      "loss": 0.5954,
+      "step": 44330
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006141152027814878,
+      "loss": 0.5392,
+      "step": 44340
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006140728020521954,
+      "loss": 0.6803,
+      "step": 44350
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006140304013229028,
+      "loss": 0.5811,
+      "step": 44360
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006139880005936102,
+      "loss": 0.6774,
+      "step": 44370
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006139455998643177,
+      "loss": 0.6091,
+      "step": 44380
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006139031991350252,
+      "loss": 0.4999,
+      "step": 44390
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006138607984057326,
+      "loss": 0.5748,
+      "step": 44400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006138183976764401,
+      "loss": 0.5628,
+      "step": 44410
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006137759969471474,
+      "loss": 0.5733,
+      "step": 44420
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000613733596217855,
+      "loss": 0.5116,
+      "step": 44430
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006136911954885625,
+      "loss": 0.596,
+      "step": 44440
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006136487947592698,
+      "loss": 0.6044,
+      "step": 44450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006136063940299773,
+      "loss": 0.6229,
+      "step": 44460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006135639933006849,
+      "loss": 0.652,
+      "step": 44470
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006135215925713922,
+      "loss": 0.5762,
+      "step": 44480
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006134791918420997,
+      "loss": 0.5425,
+      "step": 44490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006134367911128072,
+      "loss": 0.6174,
+      "step": 44500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006133943903835146,
+      "loss": 0.5959,
+      "step": 44510
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006133519896542221,
+      "loss": 0.558,
+      "step": 44520
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006133095889249296,
+      "loss": 0.571,
+      "step": 44530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006132671881956369,
+      "loss": 0.492,
+      "step": 44540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006132247874663445,
+      "loss": 0.7502,
+      "step": 44550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000613182386737052,
+      "loss": 0.6024,
+      "step": 44560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006131399860077593,
+      "loss": 0.5883,
+      "step": 44570
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006130975852784668,
+      "loss": 0.5787,
+      "step": 44580
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006130551845491743,
+      "loss": 0.5087,
+      "step": 44590
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006130127838198817,
+      "loss": 0.5823,
+      "step": 44600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006129703830905892,
+      "loss": 0.6235,
+      "step": 44610
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006129279823612966,
+      "loss": 0.5018,
+      "step": 44620
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006128855816320041,
+      "loss": 0.625,
+      "step": 44630
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006128431809027116,
+      "loss": 0.553,
+      "step": 44640
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000612800780173419,
+      "loss": 0.4955,
+      "step": 44650
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006127583794441264,
+      "loss": 0.5309,
+      "step": 44660
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006127159787148339,
+      "loss": 0.6306,
+      "step": 44670
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006126735779855414,
+      "loss": 0.5178,
+      "step": 44680
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006126311772562488,
+      "loss": 0.5506,
+      "step": 44690
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006125887765269563,
+      "loss": 0.5605,
+      "step": 44700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006125463757976638,
+      "loss": 0.5941,
+      "step": 44710
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006125039750683712,
+      "loss": 0.493,
+      "step": 44720
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006124615743390787,
+      "loss": 0.6731,
+      "step": 44730
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006124191736097861,
+      "loss": 0.5892,
+      "step": 44740
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006123767728804935,
+      "loss": 0.5891,
+      "step": 44750
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006123343721512011,
+      "loss": 0.5475,
+      "step": 44760
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006122919714219085,
+      "loss": 0.5666,
+      "step": 44770
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006122495706926159,
+      "loss": 0.5683,
+      "step": 44780
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006122071699633234,
+      "loss": 0.5549,
+      "step": 44790
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006121647692340309,
+      "loss": 0.4885,
+      "step": 44800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006121223685047383,
+      "loss": 0.5739,
+      "step": 44810
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006120799677754458,
+      "loss": 0.697,
+      "step": 44820
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006120375670461532,
+      "loss": 0.5882,
+      "step": 44830
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006119951663168607,
+      "loss": 0.6313,
+      "step": 44840
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006119527655875682,
+      "loss": 0.5344,
+      "step": 44850
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006119103648582756,
+      "loss": 0.54,
+      "step": 44860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000611867964128983,
+      "loss": 0.6136,
+      "step": 44870
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006118255633996905,
+      "loss": 0.6768,
+      "step": 44880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000611783162670398,
+      "loss": 0.5452,
+      "step": 44890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006117407619411054,
+      "loss": 0.5857,
+      "step": 44900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006116983612118129,
+      "loss": 0.653,
+      "step": 44910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006116559604825204,
+      "loss": 0.5315,
+      "step": 44920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006116135597532278,
+      "loss": 0.4933,
+      "step": 44930
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006115711590239352,
+      "loss": 0.7106,
+      "step": 44940
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006115287582946426,
+      "loss": 0.6066,
+      "step": 44950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006114863575653502,
+      "loss": 0.5959,
+      "step": 44960
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006114439568360576,
+      "loss": 0.5163,
+      "step": 44970
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000611401556106765,
+      "loss": 0.5967,
+      "step": 44980
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006113591553774725,
+      "loss": 0.5914,
+      "step": 44990
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00061131675464818,
+      "loss": 0.5842,
+      "step": 45000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006112743539188874,
+      "loss": 0.4903,
+      "step": 45010
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006112319531895949,
+      "loss": 0.5452,
+      "step": 45020
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006111895524603023,
+      "loss": 0.6404,
+      "step": 45030
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006111471517310098,
+      "loss": 0.5661,
+      "step": 45040
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006111047510017173,
+      "loss": 0.6001,
+      "step": 45050
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006110623502724247,
+      "loss": 0.5376,
+      "step": 45060
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006110199495431321,
+      "loss": 0.526,
+      "step": 45070
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006109775488138397,
+      "loss": 0.5801,
+      "step": 45080
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006109351480845471,
+      "loss": 0.5569,
+      "step": 45090
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006108927473552545,
+      "loss": 0.5212,
+      "step": 45100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000610850346625962,
+      "loss": 0.5371,
+      "step": 45110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006108079458966695,
+      "loss": 0.7621,
+      "step": 45120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006107655451673769,
+      "loss": 0.6012,
+      "step": 45130
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006107231444380844,
+      "loss": 0.6583,
+      "step": 45140
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006106807437087918,
+      "loss": 0.5542,
+      "step": 45150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006106383429794993,
+      "loss": 0.5806,
+      "step": 45160
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006105959422502068,
+      "loss": 0.6262,
+      "step": 45170
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006105535415209142,
+      "loss": 0.472,
+      "step": 45180
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006105111407916216,
+      "loss": 0.59,
+      "step": 45190
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006104687400623291,
+      "loss": 0.6389,
+      "step": 45200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006104263393330366,
+      "loss": 0.6566,
+      "step": 45210
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000610383938603744,
+      "loss": 0.6652,
+      "step": 45220
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006103415378744514,
+      "loss": 0.6456,
+      "step": 45230
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000610299137145159,
+      "loss": 0.5463,
+      "step": 45240
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006102567364158664,
+      "loss": 0.618,
+      "step": 45250
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006102143356865738,
+      "loss": 0.5365,
+      "step": 45260
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006101719349572813,
+      "loss": 0.6094,
+      "step": 45270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006101295342279887,
+      "loss": 0.5744,
+      "step": 45280
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006100871334986962,
+      "loss": 0.6096,
+      "step": 45290
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006100447327694037,
+      "loss": 0.7585,
+      "step": 45300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006100023320401111,
+      "loss": 0.5543,
+      "step": 45310
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006099599313108186,
+      "loss": 0.5147,
+      "step": 45320
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006099175305815261,
+      "loss": 0.6975,
+      "step": 45330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006098751298522335,
+      "loss": 0.6398,
+      "step": 45340
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006098327291229409,
+      "loss": 0.5749,
+      "step": 45350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006097903283936483,
+      "loss": 0.5769,
+      "step": 45360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006097479276643559,
+      "loss": 0.74,
+      "step": 45370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006097055269350633,
+      "loss": 0.5703,
+      "step": 45380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006096631262057707,
+      "loss": 0.5404,
+      "step": 45390
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006096207254764782,
+      "loss": 0.6391,
+      "step": 45400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006095783247471857,
+      "loss": 0.6135,
+      "step": 45410
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006095359240178931,
+      "loss": 0.6167,
+      "step": 45420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006094935232886006,
+      "loss": 0.603,
+      "step": 45430
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006094511225593081,
+      "loss": 0.6398,
+      "step": 45440
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006094087218300155,
+      "loss": 0.6446,
+      "step": 45450
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.000609366321100723,
+      "loss": 0.637,
+      "step": 45460
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006093239203714304,
+      "loss": 0.6142,
+      "step": 45470
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006092815196421378,
+      "loss": 0.5809,
+      "step": 45480
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006092391189128453,
+      "loss": 0.5511,
+      "step": 45490
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006091967181835528,
+      "loss": 0.5633,
+      "step": 45500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006091543174542602,
+      "loss": 0.5239,
+      "step": 45510
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006091119167249677,
+      "loss": 0.5826,
+      "step": 45520
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006090695159956752,
+      "loss": 0.6117,
+      "step": 45530
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006090271152663826,
+      "loss": 0.6223,
+      "step": 45540
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00060898471453709,
+      "loss": 0.6524,
+      "step": 45550
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006089423138077975,
+      "loss": 0.6316,
+      "step": 45560
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.000608899913078505,
+      "loss": 0.5877,
+      "step": 45570
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006088575123492124,
+      "loss": 0.5368,
+      "step": 45580
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006088151116199199,
+      "loss": 0.5482,
+      "step": 45590
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006087727108906273,
+      "loss": 0.5637,
+      "step": 45600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006087303101613348,
+      "loss": 0.6886,
+      "step": 45610
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006086879094320423,
+      "loss": 0.6657,
+      "step": 45620
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006086455087027497,
+      "loss": 0.5644,
+      "step": 45630
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006086031079734571,
+      "loss": 0.5414,
+      "step": 45640
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006085607072441647,
+      "loss": 0.6004,
+      "step": 45650
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006085183065148721,
+      "loss": 0.6727,
+      "step": 45660
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006084759057855795,
+      "loss": 0.5909,
+      "step": 45670
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.000608433505056287,
+      "loss": 0.68,
+      "step": 45680
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006083911043269945,
+      "loss": 0.6293,
+      "step": 45690
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006083487035977019,
+      "loss": 0.5979,
+      "step": 45700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006083063028684094,
+      "loss": 0.5941,
+      "step": 45710
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006082639021391168,
+      "loss": 0.5237,
+      "step": 45720
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006082215014098243,
+      "loss": 0.5852,
+      "step": 45730
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006081791006805318,
+      "loss": 0.4999,
+      "step": 45740
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006081366999512392,
+      "loss": 0.632,
+      "step": 45750
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006080942992219466,
+      "loss": 0.5527,
+      "step": 45760
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006080518984926542,
+      "loss": 0.5195,
+      "step": 45770
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006080094977633616,
+      "loss": 0.538,
+      "step": 45780
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000607967097034069,
+      "loss": 0.5172,
+      "step": 45790
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006079246963047765,
+      "loss": 0.6341,
+      "step": 45800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006078822955754839,
+      "loss": 0.6048,
+      "step": 45810
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006078398948461914,
+      "loss": 0.5927,
+      "step": 45820
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006077974941168989,
+      "loss": 0.5849,
+      "step": 45830
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006077550933876062,
+      "loss": 0.5208,
+      "step": 45840
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006077126926583138,
+      "loss": 0.5738,
+      "step": 45850
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006076702919290213,
+      "loss": 0.5654,
+      "step": 45860
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006076278911997286,
+      "loss": 0.6391,
+      "step": 45870
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006075854904704361,
+      "loss": 0.6384,
+      "step": 45880
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006075430897411437,
+      "loss": 0.6442,
+      "step": 45890
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000607500689011851,
+      "loss": 0.595,
+      "step": 45900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006074582882825585,
+      "loss": 0.5227,
+      "step": 45910
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000607415887553266,
+      "loss": 0.4981,
+      "step": 45920
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006073734868239734,
+      "loss": 0.6192,
+      "step": 45930
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006073310860946809,
+      "loss": 0.5633,
+      "step": 45940
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006072886853653883,
+      "loss": 0.611,
+      "step": 45950
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006072462846360957,
+      "loss": 0.6079,
+      "step": 45960
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006072038839068033,
+      "loss": 0.6901,
+      "step": 45970
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006071614831775107,
+      "loss": 0.582,
+      "step": 45980
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006071190824482181,
+      "loss": 0.6609,
+      "step": 45990
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006070766817189256,
+      "loss": 0.7101,
+      "step": 46000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000607034280989633,
+      "loss": 0.6436,
+      "step": 46010
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006069918802603405,
+      "loss": 0.5068,
+      "step": 46020
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000606949479531048,
+      "loss": 0.6014,
+      "step": 46030
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006069070788017554,
+      "loss": 0.6152,
+      "step": 46040
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006068646780724629,
+      "loss": 0.6296,
+      "step": 46050
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006068222773431704,
+      "loss": 0.599,
+      "step": 46060
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006067798766138778,
+      "loss": 0.5763,
+      "step": 46070
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006067374758845852,
+      "loss": 0.5048,
+      "step": 46080
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006066950751552927,
+      "loss": 0.6124,
+      "step": 46090
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006066526744260002,
+      "loss": 0.5962,
+      "step": 46100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006066102736967076,
+      "loss": 0.5553,
+      "step": 46110
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006065678729674151,
+      "loss": 0.6346,
+      "step": 46120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006065254722381225,
+      "loss": 0.5596,
+      "step": 46130
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00060648307150883,
+      "loss": 0.508,
+      "step": 46140
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006064406707795375,
+      "loss": 0.5891,
+      "step": 46150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006063982700502448,
+      "loss": 0.5864,
+      "step": 46160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006063558693209523,
+      "loss": 0.5487,
+      "step": 46170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006063134685916599,
+      "loss": 0.687,
+      "step": 46180
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006062710678623672,
+      "loss": 0.572,
+      "step": 46190
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006062286671330747,
+      "loss": 0.5618,
+      "step": 46200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006061862664037822,
+      "loss": 0.5253,
+      "step": 46210
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006061438656744896,
+      "loss": 0.5966,
+      "step": 46220
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006061014649451971,
+      "loss": 0.6183,
+      "step": 46230
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006060590642159046,
+      "loss": 0.6367,
+      "step": 46240
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006060166634866119,
+      "loss": 0.4946,
+      "step": 46250
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006059742627573195,
+      "loss": 0.597,
+      "step": 46260
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000605931862028027,
+      "loss": 0.545,
+      "step": 46270
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006058894612987343,
+      "loss": 0.5803,
+      "step": 46280
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006058470605694418,
+      "loss": 0.5621,
+      "step": 46290
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006058046598401494,
+      "loss": 0.5714,
+      "step": 46300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006057622591108567,
+      "loss": 0.6039,
+      "step": 46310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006057198583815642,
+      "loss": 0.6685,
+      "step": 46320
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006056774576522717,
+      "loss": 0.6506,
+      "step": 46330
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006056350569229791,
+      "loss": 0.5803,
+      "step": 46340
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006055926561936866,
+      "loss": 0.5297,
+      "step": 46350
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000605550255464394,
+      "loss": 0.5265,
+      "step": 46360
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006055078547351014,
+      "loss": 0.5858,
+      "step": 46370
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000605465454005809,
+      "loss": 0.6351,
+      "step": 46380
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006054230532765165,
+      "loss": 0.5843,
+      "step": 46390
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006053806525472238,
+      "loss": 0.6335,
+      "step": 46400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006053382518179313,
+      "loss": 0.6042,
+      "step": 46410
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006052958510886387,
+      "loss": 0.6457,
+      "step": 46420
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006052534503593462,
+      "loss": 0.4788,
+      "step": 46430
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006052110496300537,
+      "loss": 0.6689,
+      "step": 46440
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000605168648900761,
+      "loss": 0.5651,
+      "step": 46450
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006051262481714686,
+      "loss": 0.5655,
+      "step": 46460
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006050838474421761,
+      "loss": 0.5944,
+      "step": 46470
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006050414467128834,
+      "loss": 0.5554,
+      "step": 46480
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006049990459835909,
+      "loss": 0.5536,
+      "step": 46490
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006049566452542985,
+      "loss": 0.4895,
+      "step": 46500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006049142445250058,
+      "loss": 0.7098,
+      "step": 46510
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006048718437957133,
+      "loss": 0.6085,
+      "step": 46520
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006048294430664208,
+      "loss": 0.544,
+      "step": 46530
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006047870423371282,
+      "loss": 0.5347,
+      "step": 46540
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006047446416078357,
+      "loss": 0.536,
+      "step": 46550
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006047022408785432,
+      "loss": 0.6088,
+      "step": 46560
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006046598401492505,
+      "loss": 0.6528,
+      "step": 46570
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006046174394199581,
+      "loss": 0.5187,
+      "step": 46580
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006045750386906656,
+      "loss": 0.5515,
+      "step": 46590
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006045326379613729,
+      "loss": 0.5586,
+      "step": 46600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006044902372320804,
+      "loss": 0.5781,
+      "step": 46610
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006044478365027879,
+      "loss": 0.4801,
+      "step": 46620
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006044054357734953,
+      "loss": 0.5935,
+      "step": 46630
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006043630350442028,
+      "loss": 0.6245,
+      "step": 46640
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006043206343149103,
+      "loss": 0.6431,
+      "step": 46650
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006042782335856177,
+      "loss": 0.4822,
+      "step": 46660
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006042358328563252,
+      "loss": 0.568,
+      "step": 46670
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006041934321270327,
+      "loss": 0.6676,
+      "step": 46680
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00060415103139774,
+      "loss": 0.6664,
+      "step": 46690
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006041086306684475,
+      "loss": 0.6917,
+      "step": 46700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006040662299391551,
+      "loss": 0.5591,
+      "step": 46710
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006040238292098624,
+      "loss": 0.6007,
+      "step": 46720
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006039814284805699,
+      "loss": 0.6154,
+      "step": 46730
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006039390277512774,
+      "loss": 0.5721,
+      "step": 46740
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006038966270219848,
+      "loss": 0.5543,
+      "step": 46750
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006038542262926923,
+      "loss": 0.597,
+      "step": 46760
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006038118255633996,
+      "loss": 0.594,
+      "step": 46770
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006037694248341071,
+      "loss": 0.6628,
+      "step": 46780
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006037270241048147,
+      "loss": 0.6152,
+      "step": 46790
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000603684623375522,
+      "loss": 0.5721,
+      "step": 46800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006036422226462295,
+      "loss": 0.6436,
+      "step": 46810
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000603599821916937,
+      "loss": 0.5414,
+      "step": 46820
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006035574211876444,
+      "loss": 0.5524,
+      "step": 46830
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006035150204583519,
+      "loss": 0.5498,
+      "step": 46840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006034726197290594,
+      "loss": 0.6065,
+      "step": 46850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006034302189997667,
+      "loss": 0.5742,
+      "step": 46860
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006033878182704743,
+      "loss": 0.6128,
+      "step": 46870
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006033454175411818,
+      "loss": 0.5571,
+      "step": 46880
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006033030168118891,
+      "loss": 0.6009,
+      "step": 46890
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006032606160825966,
+      "loss": 0.552,
+      "step": 46900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006032182153533042,
+      "loss": 0.6414,
+      "step": 46910
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006031758146240115,
+      "loss": 0.5441,
+      "step": 46920
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000603133413894719,
+      "loss": 0.6564,
+      "step": 46930
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006030910131654265,
+      "loss": 0.5307,
+      "step": 46940
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006030486124361339,
+      "loss": 0.6455,
+      "step": 46950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006030062117068414,
+      "loss": 0.5962,
+      "step": 46960
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006029638109775489,
+      "loss": 0.6292,
+      "step": 46970
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006029214102482562,
+      "loss": 0.6203,
+      "step": 46980
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006028790095189638,
+      "loss": 0.5581,
+      "step": 46990
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006028366087896713,
+      "loss": 0.62,
+      "step": 47000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006027942080603786,
+      "loss": 0.7055,
+      "step": 47010
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006027518073310861,
+      "loss": 0.5764,
+      "step": 47020
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006027094066017936,
+      "loss": 0.6051,
+      "step": 47030
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000602667005872501,
+      "loss": 0.5505,
+      "step": 47040
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006026246051432085,
+      "loss": 0.5565,
+      "step": 47050
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006025822044139159,
+      "loss": 0.6092,
+      "step": 47060
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006025398036846234,
+      "loss": 0.5324,
+      "step": 47070
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006024974029553309,
+      "loss": 0.589,
+      "step": 47080
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006024550022260383,
+      "loss": 0.5959,
+      "step": 47090
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006024126014967457,
+      "loss": 0.5362,
+      "step": 47100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006023702007674533,
+      "loss": 0.6606,
+      "step": 47110
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006023278000381607,
+      "loss": 0.5745,
+      "step": 47120
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006022853993088681,
+      "loss": 0.4986,
+      "step": 47130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006022429985795756,
+      "loss": 0.5306,
+      "step": 47140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006022005978502831,
+      "loss": 0.5907,
+      "step": 47150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006021581971209905,
+      "loss": 0.6354,
+      "step": 47160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000602115796391698,
+      "loss": 0.5049,
+      "step": 47170
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006020733956624054,
+      "loss": 0.683,
+      "step": 47180
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006020309949331129,
+      "loss": 0.6151,
+      "step": 47190
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006019885942038204,
+      "loss": 0.5382,
+      "step": 47200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006019461934745278,
+      "loss": 0.6837,
+      "step": 47210
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006019037927452352,
+      "loss": 0.5442,
+      "step": 47220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006018613920159427,
+      "loss": 0.5025,
+      "step": 47230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006018189912866502,
+      "loss": 0.6,
+      "step": 47240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006017765905573576,
+      "loss": 0.6435,
+      "step": 47250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006017341898280651,
+      "loss": 0.4832,
+      "step": 47260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006016917890987726,
+      "loss": 0.6172,
+      "step": 47270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00060164938836948,
+      "loss": 0.6005,
+      "step": 47280
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006016069876401875,
+      "loss": 0.4877,
+      "step": 47290
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006015645869108948,
+      "loss": 0.4214,
+      "step": 47300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006015221861816023,
+      "loss": 0.4822,
+      "step": 47310
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006014797854523099,
+      "loss": 0.556,
+      "step": 47320
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006014373847230172,
+      "loss": 0.5535,
+      "step": 47330
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006013949839937247,
+      "loss": 0.5547,
+      "step": 47340
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006013525832644322,
+      "loss": 0.5328,
+      "step": 47350
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006013101825351396,
+      "loss": 0.5801,
+      "step": 47360
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006012677818058471,
+      "loss": 0.646,
+      "step": 47370
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006012253810765545,
+      "loss": 0.4888,
+      "step": 47380
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006011829803472619,
+      "loss": 0.5579,
+      "step": 47390
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006011405796179695,
+      "loss": 0.4464,
+      "step": 47400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006010981788886769,
+      "loss": 0.4954,
+      "step": 47410
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006010557781593843,
+      "loss": 0.5144,
+      "step": 47420
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006010133774300918,
+      "loss": 0.5419,
+      "step": 47430
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006009709767007993,
+      "loss": 0.4752,
+      "step": 47440
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006009285759715067,
+      "loss": 0.4848,
+      "step": 47450
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006008861752422142,
+      "loss": 0.6621,
+      "step": 47460
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006008437745129217,
+      "loss": 0.5249,
+      "step": 47470
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006008013737836291,
+      "loss": 0.5356,
+      "step": 47480
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006007589730543366,
+      "loss": 0.5444,
+      "step": 47490
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000600716572325044,
+      "loss": 0.5465,
+      "step": 47500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006006741715957514,
+      "loss": 0.5447,
+      "step": 47510
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000600631770866459,
+      "loss": 0.5371,
+      "step": 47520
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006005893701371664,
+      "loss": 0.5976,
+      "step": 47530
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006005469694078738,
+      "loss": 0.7002,
+      "step": 47540
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006005045686785813,
+      "loss": 0.5911,
+      "step": 47550
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006004621679492888,
+      "loss": 0.523,
+      "step": 47560
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006004197672199962,
+      "loss": 0.5444,
+      "step": 47570
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006003773664907037,
+      "loss": 0.5343,
+      "step": 47580
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006003349657614111,
+      "loss": 0.5129,
+      "step": 47590
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006002925650321186,
+      "loss": 0.5073,
+      "step": 47600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006002501643028261,
+      "loss": 0.4659,
+      "step": 47610
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006002077635735335,
+      "loss": 0.486,
+      "step": 47620
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006001653628442409,
+      "loss": 0.5903,
+      "step": 47630
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006001229621149484,
+      "loss": 0.5711,
+      "step": 47640
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006000805613856559,
+      "loss": 0.5755,
+      "step": 47650
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006000381606563633,
+      "loss": 0.5813,
+      "step": 47660
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005999957599270707,
+      "loss": 0.5408,
+      "step": 47670
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005999533591977783,
+      "loss": 0.5102,
+      "step": 47680
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005999109584684857,
+      "loss": 0.4828,
+      "step": 47690
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005998685577391931,
+      "loss": 0.6098,
+      "step": 47700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005998261570099006,
+      "loss": 0.4746,
+      "step": 47710
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005997837562806081,
+      "loss": 0.5494,
+      "step": 47720
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005997413555513155,
+      "loss": 0.4538,
+      "step": 47730
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000599698954822023,
+      "loss": 0.5321,
+      "step": 47740
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005996565540927304,
+      "loss": 0.4866,
+      "step": 47750
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005996141533634379,
+      "loss": 0.629,
+      "step": 47760
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005995717526341454,
+      "loss": 0.5901,
+      "step": 47770
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005995293519048528,
+      "loss": 0.529,
+      "step": 47780
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005994869511755602,
+      "loss": 0.6794,
+      "step": 47790
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005994445504462678,
+      "loss": 0.5362,
+      "step": 47800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005994021497169752,
+      "loss": 0.584,
+      "step": 47810
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005993597489876826,
+      "loss": 0.6118,
+      "step": 47820
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00059931734825839,
+      "loss": 0.547,
+      "step": 47830
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005992749475290975,
+      "loss": 0.5298,
+      "step": 47840
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000599232546799805,
+      "loss": 0.5472,
+      "step": 47850
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005991901460705124,
+      "loss": 0.585,
+      "step": 47860
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005991477453412199,
+      "loss": 0.5333,
+      "step": 47870
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005991053446119274,
+      "loss": 0.541,
+      "step": 47880
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005990629438826348,
+      "loss": 0.4582,
+      "step": 47890
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005990205431533423,
+      "loss": 0.5284,
+      "step": 47900
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005989781424240497,
+      "loss": 0.5652,
+      "step": 47910
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005989357416947572,
+      "loss": 0.5085,
+      "step": 47920
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005988933409654647,
+      "loss": 0.5001,
+      "step": 47930
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005988509402361721,
+      "loss": 0.5362,
+      "step": 47940
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005988085395068795,
+      "loss": 0.5168,
+      "step": 47950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000598766138777587,
+      "loss": 0.5041,
+      "step": 47960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005987237380482945,
+      "loss": 0.4952,
+      "step": 47970
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005986813373190019,
+      "loss": 0.5274,
+      "step": 47980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005986389365897093,
+      "loss": 0.5558,
+      "step": 47990
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005985965358604169,
+      "loss": 0.5881,
+      "step": 48000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005985541351311243,
+      "loss": 0.6355,
+      "step": 48010
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005985117344018317,
+      "loss": 0.5972,
+      "step": 48020
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005984693336725392,
+      "loss": 0.5286,
+      "step": 48030
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005984269329432466,
+      "loss": 0.5369,
+      "step": 48040
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005983845322139541,
+      "loss": 0.588,
+      "step": 48050
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005983421314846616,
+      "loss": 0.4542,
+      "step": 48060
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000598299730755369,
+      "loss": 0.5687,
+      "step": 48070
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005982573300260765,
+      "loss": 0.6087,
+      "step": 48080
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000598214929296784,
+      "loss": 0.4903,
+      "step": 48090
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005981725285674914,
+      "loss": 0.6037,
+      "step": 48100
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005981301278381988,
+      "loss": 0.4358,
+      "step": 48110
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005980877271089063,
+      "loss": 0.5041,
+      "step": 48120
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005980453263796138,
+      "loss": 0.492,
+      "step": 48130
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005980029256503212,
+      "loss": 0.5596,
+      "step": 48140
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005979605249210287,
+      "loss": 0.5638,
+      "step": 48150
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005979181241917361,
+      "loss": 0.5343,
+      "step": 48160
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005978757234624436,
+      "loss": 0.5111,
+      "step": 48170
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005978333227331511,
+      "loss": 0.5832,
+      "step": 48180
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005977909220038585,
+      "loss": 0.5941,
+      "step": 48190
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005977485212745659,
+      "loss": 0.5591,
+      "step": 48200
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005977061205452735,
+      "loss": 0.5082,
+      "step": 48210
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005976637198159809,
+      "loss": 0.6203,
+      "step": 48220
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005976213190866883,
+      "loss": 0.5419,
+      "step": 48230
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005975789183573957,
+      "loss": 0.6102,
+      "step": 48240
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005975365176281033,
+      "loss": 0.5592,
+      "step": 48250
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005974941168988107,
+      "loss": 0.5151,
+      "step": 48260
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005974517161695181,
+      "loss": 0.5853,
+      "step": 48270
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005974093154402255,
+      "loss": 0.5653,
+      "step": 48280
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005973669147109331,
+      "loss": 0.4786,
+      "step": 48290
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005973245139816405,
+      "loss": 0.5053,
+      "step": 48300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005972821132523479,
+      "loss": 0.539,
+      "step": 48310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005972397125230554,
+      "loss": 0.4795,
+      "step": 48320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.000597197311793763,
+      "loss": 0.535,
+      "step": 48330
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005971549110644703,
+      "loss": 0.5899,
+      "step": 48340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005971125103351778,
+      "loss": 0.4316,
+      "step": 48350
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005970701096058852,
+      "loss": 0.5189,
+      "step": 48360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005970277088765927,
+      "loss": 0.4826,
+      "step": 48370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005969853081473002,
+      "loss": 0.5916,
+      "step": 48380
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005969429074180076,
+      "loss": 0.6229,
+      "step": 48390
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.000596900506688715,
+      "loss": 0.479,
+      "step": 48400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005968581059594226,
+      "loss": 0.4506,
+      "step": 48410
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00059681570523013,
+      "loss": 0.5624,
+      "step": 48420
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005967733045008374,
+      "loss": 0.5045,
+      "step": 48430
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005967309037715449,
+      "loss": 0.5681,
+      "step": 48440
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005966885030422524,
+      "loss": 0.4812,
+      "step": 48450
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005966461023129598,
+      "loss": 0.5523,
+      "step": 48460
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005966037015836673,
+      "loss": 0.5555,
+      "step": 48470
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005965613008543747,
+      "loss": 0.6421,
+      "step": 48480
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005965189001250822,
+      "loss": 0.5489,
+      "step": 48490
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005964764993957897,
+      "loss": 0.5237,
+      "step": 48500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005964340986664971,
+      "loss": 0.5918,
+      "step": 48510
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005963916979372045,
+      "loss": 0.5362,
+      "step": 48520
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005963492972079121,
+      "loss": 0.6179,
+      "step": 48530
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005963068964786195,
+      "loss": 0.5668,
+      "step": 48540
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005962644957493269,
+      "loss": 0.5857,
+      "step": 48550
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005962220950200344,
+      "loss": 0.5496,
+      "step": 48560
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005961796942907418,
+      "loss": 0.603,
+      "step": 48570
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005961372935614493,
+      "loss": 0.5505,
+      "step": 48580
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005960948928321568,
+      "loss": 0.5916,
+      "step": 48590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005960524921028641,
+      "loss": 0.5834,
+      "step": 48600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005960100913735717,
+      "loss": 0.5816,
+      "step": 48610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005959676906442792,
+      "loss": 0.4856,
+      "step": 48620
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005959252899149865,
+      "loss": 0.5361,
+      "step": 48630
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000595882889185694,
+      "loss": 0.5317,
+      "step": 48640
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005958404884564015,
+      "loss": 0.5581,
+      "step": 48650
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005957980877271089,
+      "loss": 0.4612,
+      "step": 48660
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005957556869978164,
+      "loss": 0.5933,
+      "step": 48670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005957132862685239,
+      "loss": 0.5767,
+      "step": 48680
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005956708855392313,
+      "loss": 0.5853,
+      "step": 48690
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005956284848099388,
+      "loss": 0.5457,
+      "step": 48700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005955860840806463,
+      "loss": 0.5193,
+      "step": 48710
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005955436833513536,
+      "loss": 0.4677,
+      "step": 48720
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005955012826220611,
+      "loss": 0.5892,
+      "step": 48730
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005954588818927687,
+      "loss": 0.5073,
+      "step": 48740
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000595416481163476,
+      "loss": 0.4742,
+      "step": 48750
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005953740804341835,
+      "loss": 0.6194,
+      "step": 48760
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000595331679704891,
+      "loss": 0.5076,
+      "step": 48770
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005952892789755984,
+      "loss": 0.5721,
+      "step": 48780
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005952468782463059,
+      "loss": 0.5754,
+      "step": 48790
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005952044775170133,
+      "loss": 0.4978,
+      "step": 48800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005951620767877207,
+      "loss": 0.5219,
+      "step": 48810
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005951196760584283,
+      "loss": 0.5751,
+      "step": 48820
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005950772753291357,
+      "loss": 0.5683,
+      "step": 48830
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005950348745998431,
+      "loss": 0.5748,
+      "step": 48840
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005949924738705506,
+      "loss": 0.5972,
+      "step": 48850
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005949500731412581,
+      "loss": 0.5945,
+      "step": 48860
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005949076724119655,
+      "loss": 0.4958,
+      "step": 48870
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000594865271682673,
+      "loss": 0.4413,
+      "step": 48880
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005948228709533803,
+      "loss": 0.549,
+      "step": 48890
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005947804702240879,
+      "loss": 0.4655,
+      "step": 48900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005947380694947954,
+      "loss": 0.5296,
+      "step": 48910
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005946956687655027,
+      "loss": 0.6241,
+      "step": 48920
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005946532680362102,
+      "loss": 0.6697,
+      "step": 48930
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005946108673069178,
+      "loss": 0.5372,
+      "step": 48940
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005945684665776251,
+      "loss": 0.4896,
+      "step": 48950
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005945260658483326,
+      "loss": 0.4592,
+      "step": 48960
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005944836651190401,
+      "loss": 0.6048,
+      "step": 48970
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005944412643897475,
+      "loss": 0.5882,
+      "step": 48980
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000594398863660455,
+      "loss": 0.5275,
+      "step": 48990
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005943564629311625,
+      "loss": 0.5742,
+      "step": 49000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005943140622018698,
+      "loss": 0.4468,
+      "step": 49010
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005942716614725774,
+      "loss": 0.4737,
+      "step": 49020
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005942292607432849,
+      "loss": 0.4391,
+      "step": 49030
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005941868600139922,
+      "loss": 0.5189,
+      "step": 49040
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005941444592846997,
+      "loss": 0.5509,
+      "step": 49050
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005941020585554073,
+      "loss": 0.5452,
+      "step": 49060
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005940596578261146,
+      "loss": 0.6074,
+      "step": 49070
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005940172570968221,
+      "loss": 0.5489,
+      "step": 49080
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005939748563675296,
+      "loss": 0.6784,
+      "step": 49090
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000593932455638237,
+      "loss": 0.5531,
+      "step": 49100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005938900549089445,
+      "loss": 0.5241,
+      "step": 49110
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000593847654179652,
+      "loss": 0.5231,
+      "step": 49120
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005938052534503593,
+      "loss": 0.4759,
+      "step": 49130
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005937628527210669,
+      "loss": 0.518,
+      "step": 49140
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005937204519917744,
+      "loss": 0.5035,
+      "step": 49150
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005936780512624817,
+      "loss": 0.549,
+      "step": 49160
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005936356505331892,
+      "loss": 0.4949,
+      "step": 49170
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005935932498038967,
+      "loss": 0.5347,
+      "step": 49180
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005935508490746041,
+      "loss": 0.5604,
+      "step": 49190
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005935084483453116,
+      "loss": 0.5155,
+      "step": 49200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005934660476160189,
+      "loss": 0.4977,
+      "step": 49210
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005934236468867265,
+      "loss": 0.5182,
+      "step": 49220
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000593381246157434,
+      "loss": 0.4931,
+      "step": 49230
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005933388454281413,
+      "loss": 0.4678,
+      "step": 49240
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005932964446988488,
+      "loss": 0.5558,
+      "step": 49250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005932540439695563,
+      "loss": 0.4478,
+      "step": 49260
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005932116432402637,
+      "loss": 0.4686,
+      "step": 49270
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005931692425109712,
+      "loss": 0.5382,
+      "step": 49280
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005931268417816787,
+      "loss": 0.5689,
+      "step": 49290
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005930844410523861,
+      "loss": 0.5655,
+      "step": 49300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005930420403230936,
+      "loss": 0.5342,
+      "step": 49310
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005929996395938011,
+      "loss": 0.4755,
+      "step": 49320
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005929572388645084,
+      "loss": 0.5873,
+      "step": 49330
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005929148381352159,
+      "loss": 0.5306,
+      "step": 49340
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005928724374059235,
+      "loss": 0.4782,
+      "step": 49350
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005928300366766308,
+      "loss": 0.5349,
+      "step": 49360
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005927876359473383,
+      "loss": 0.5029,
+      "step": 49370
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005927452352180458,
+      "loss": 0.5368,
+      "step": 49380
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005927028344887532,
+      "loss": 0.5253,
+      "step": 49390
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005926604337594607,
+      "loss": 0.5828,
+      "step": 49400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005926180330301682,
+      "loss": 0.6064,
+      "step": 49410
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005925756323008756,
+      "loss": 0.4694,
+      "step": 49420
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005925332315715831,
+      "loss": 0.565,
+      "step": 49430
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005924908308422906,
+      "loss": 0.5276,
+      "step": 49440
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005924484301129979,
+      "loss": 0.5381,
+      "step": 49450
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005924060293837054,
+      "loss": 0.6455,
+      "step": 49460
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000592363628654413,
+      "loss": 0.507,
+      "step": 49470
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005923212279251203,
+      "loss": 0.5315,
+      "step": 49480
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005922788271958278,
+      "loss": 0.5264,
+      "step": 49490
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005922364264665353,
+      "loss": 0.5586,
+      "step": 49500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005921940257372427,
+      "loss": 0.5342,
+      "step": 49510
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005921516250079502,
+      "loss": 0.5159,
+      "step": 49520
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005921092242786576,
+      "loss": 0.6701,
+      "step": 49530
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000592066823549365,
+      "loss": 0.6391,
+      "step": 49540
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005920244228200726,
+      "loss": 0.5247,
+      "step": 49550
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00059198202209078,
+      "loss": 0.5875,
+      "step": 49560
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005919396213614874,
+      "loss": 0.618,
+      "step": 49570
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005918972206321949,
+      "loss": 0.6312,
+      "step": 49580
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005918548199029024,
+      "loss": 0.5089,
+      "step": 49590
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005918124191736098,
+      "loss": 0.632,
+      "step": 49600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005917700184443173,
+      "loss": 0.4787,
+      "step": 49610
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005917276177150246,
+      "loss": 0.4687,
+      "step": 49620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005916852169857322,
+      "loss": 0.5677,
+      "step": 49630
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005916428162564397,
+      "loss": 0.556,
+      "step": 49640
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000591600415527147,
+      "loss": 0.5876,
+      "step": 49650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005915580147978545,
+      "loss": 0.4774,
+      "step": 49660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005915156140685621,
+      "loss": 0.576,
+      "step": 49670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005914732133392694,
+      "loss": 0.5923,
+      "step": 49680
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005914308126099769,
+      "loss": 0.4974,
+      "step": 49690
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005913884118806844,
+      "loss": 0.5085,
+      "step": 49700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005913460111513918,
+      "loss": 0.5723,
+      "step": 49710
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005913036104220993,
+      "loss": 0.4923,
+      "step": 49720
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005912612096928068,
+      "loss": 0.5908,
+      "step": 49730
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005912188089635141,
+      "loss": 0.5213,
+      "step": 49740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005911764082342217,
+      "loss": 0.5104,
+      "step": 49750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005911340075049292,
+      "loss": 0.4723,
+      "step": 49760
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005910916067756365,
+      "loss": 0.4554,
+      "step": 49770
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.000591049206046344,
+      "loss": 0.6073,
+      "step": 49780
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005910068053170515,
+      "loss": 0.5642,
+      "step": 49790
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005909644045877589,
+      "loss": 0.5887,
+      "step": 49800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005909220038584664,
+      "loss": 0.5477,
+      "step": 49810
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005908796031291738,
+      "loss": 0.5425,
+      "step": 49820
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005908372023998813,
+      "loss": 0.615,
+      "step": 49830
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005907948016705888,
+      "loss": 0.6014,
+      "step": 49840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005907524009412962,
+      "loss": 0.419,
+      "step": 49850
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005907100002120036,
+      "loss": 0.5163,
+      "step": 49860
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005906675994827111,
+      "loss": 0.5545,
+      "step": 49870
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005906251987534186,
+      "loss": 0.547,
+      "step": 49880
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.000590582798024126,
+      "loss": 0.5646,
+      "step": 49890
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005905403972948335,
+      "loss": 0.5943,
+      "step": 49900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.000590497996565541,
+      "loss": 0.5962,
+      "step": 49910
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005904555958362484,
+      "loss": 0.5951,
+      "step": 49920
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005904131951069559,
+      "loss": 0.625,
+      "step": 49930
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005903707943776633,
+      "loss": 0.5408,
+      "step": 49940
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005903283936483708,
+      "loss": 0.5986,
+      "step": 49950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005902859929190783,
+      "loss": 0.5495,
+      "step": 49960
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005902435921897857,
+      "loss": 0.5875,
+      "step": 49970
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005902011914604931,
+      "loss": 0.5626,
+      "step": 49980
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005901587907312006,
+      "loss": 0.4972,
+      "step": 49990
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005901163900019081,
+      "loss": 0.5752,
+      "step": 50000
+    },
+    {
+      "epoch": 2.11,
+      "eval_loss": 0.6209712624549866,
+      "eval_runtime": 337.6425,
+      "eval_samples_per_second": 15.564,
+      "eval_steps_per_second": 3.892,
+      "step": 50000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005900739892726155,
+      "loss": 0.6667,
+      "step": 50010
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.000590031588543323,
+      "loss": 0.4868,
+      "step": 50020
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005899891878140305,
+      "loss": 0.4711,
+      "step": 50030
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005899467870847379,
+      "loss": 0.7268,
+      "step": 50040
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005899043863554454,
+      "loss": 0.556,
+      "step": 50050
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005898619856261528,
+      "loss": 0.6284,
+      "step": 50060
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005898195848968602,
+      "loss": 0.6613,
+      "step": 50070
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005897771841675678,
+      "loss": 0.5903,
+      "step": 50080
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005897347834382752,
+      "loss": 0.5359,
+      "step": 50090
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005896923827089826,
+      "loss": 0.5745,
+      "step": 50100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005896499819796901,
+      "loss": 0.5602,
+      "step": 50110
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005896075812503976,
+      "loss": 0.5169,
+      "step": 50120
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.000589565180521105,
+      "loss": 0.5398,
+      "step": 50130
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005895227797918124,
+      "loss": 0.4734,
+      "step": 50140
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005894803790625198,
+      "loss": 0.52,
+      "step": 50150
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005894379783332274,
+      "loss": 0.566,
+      "step": 50160
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005893955776039348,
+      "loss": 0.5178,
+      "step": 50170
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005893531768746422,
+      "loss": 0.5283,
+      "step": 50180
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005893107761453497,
+      "loss": 0.556,
+      "step": 50190
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005892683754160572,
+      "loss": 0.6206,
+      "step": 50200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005892259746867646,
+      "loss": 0.5014,
+      "step": 50210
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005891835739574721,
+      "loss": 0.5912,
+      "step": 50220
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005891411732281795,
+      "loss": 0.5519,
+      "step": 50230
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.000589098772498887,
+      "loss": 0.6274,
+      "step": 50240
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005890563717695945,
+      "loss": 0.5774,
+      "step": 50250
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005890139710403019,
+      "loss": 0.5574,
+      "step": 50260
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005889715703110093,
+      "loss": 0.5891,
+      "step": 50270
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005889291695817169,
+      "loss": 0.5965,
+      "step": 50280
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005888867688524243,
+      "loss": 0.5582,
+      "step": 50290
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005888443681231317,
+      "loss": 0.609,
+      "step": 50300
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005888019673938392,
+      "loss": 0.5217,
+      "step": 50310
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005887595666645467,
+      "loss": 0.6075,
+      "step": 50320
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005887171659352541,
+      "loss": 0.5001,
+      "step": 50330
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005886747652059616,
+      "loss": 0.5654,
+      "step": 50340
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000588632364476669,
+      "loss": 0.5833,
+      "step": 50350
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005885899637473765,
+      "loss": 0.5751,
+      "step": 50360
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000588547563018084,
+      "loss": 0.4976,
+      "step": 50370
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005885051622887914,
+      "loss": 0.5851,
+      "step": 50380
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005884627615594988,
+      "loss": 0.5588,
+      "step": 50390
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005884203608302064,
+      "loss": 0.5136,
+      "step": 50400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005883779601009138,
+      "loss": 0.4693,
+      "step": 50410
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005883355593716212,
+      "loss": 0.546,
+      "step": 50420
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005882931586423286,
+      "loss": 0.5373,
+      "step": 50430
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005882507579130362,
+      "loss": 0.5914,
+      "step": 50440
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005882083571837436,
+      "loss": 0.5186,
+      "step": 50450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000588165956454451,
+      "loss": 0.5005,
+      "step": 50460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005881235557251585,
+      "loss": 0.4518,
+      "step": 50470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000588081154995866,
+      "loss": 0.5228,
+      "step": 50480
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005880387542665734,
+      "loss": 0.6138,
+      "step": 50490
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005879963535372809,
+      "loss": 0.5852,
+      "step": 50500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005879539528079883,
+      "loss": 0.505,
+      "step": 50510
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005879115520786958,
+      "loss": 0.5563,
+      "step": 50520
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005878691513494033,
+      "loss": 0.6413,
+      "step": 50530
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005878267506201107,
+      "loss": 0.5169,
+      "step": 50540
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005877843498908181,
+      "loss": 0.6066,
+      "step": 50550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005877419491615257,
+      "loss": 0.6438,
+      "step": 50560
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005876995484322331,
+      "loss": 0.6042,
+      "step": 50570
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005876571477029405,
+      "loss": 0.6053,
+      "step": 50580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.000587614746973648,
+      "loss": 0.5252,
+      "step": 50590
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005875723462443554,
+      "loss": 0.4536,
+      "step": 50600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005875299455150629,
+      "loss": 0.6157,
+      "step": 50610
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005874875447857704,
+      "loss": 0.5278,
+      "step": 50620
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005874451440564778,
+      "loss": 0.6014,
+      "step": 50630
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005874027433271853,
+      "loss": 0.4818,
+      "step": 50640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005873603425978928,
+      "loss": 0.5893,
+      "step": 50650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005873179418686002,
+      "loss": 0.5342,
+      "step": 50660
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005872755411393076,
+      "loss": 0.4604,
+      "step": 50670
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.000587233140410015,
+      "loss": 0.5309,
+      "step": 50680
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005871907396807226,
+      "loss": 0.5289,
+      "step": 50690
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00058714833895143,
+      "loss": 0.5739,
+      "step": 50700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005871059382221374,
+      "loss": 0.5249,
+      "step": 50710
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005870635374928449,
+      "loss": 0.5144,
+      "step": 50720
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005870211367635524,
+      "loss": 0.6828,
+      "step": 50730
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005869787360342598,
+      "loss": 0.6959,
+      "step": 50740
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005869363353049672,
+      "loss": 0.5785,
+      "step": 50750
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005868939345756747,
+      "loss": 0.5282,
+      "step": 50760
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005868515338463822,
+      "loss": 0.65,
+      "step": 50770
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005868091331170896,
+      "loss": 0.4919,
+      "step": 50780
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005867667323877971,
+      "loss": 0.5421,
+      "step": 50790
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005867243316585045,
+      "loss": 0.6123,
+      "step": 50800
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.000586681930929212,
+      "loss": 0.4924,
+      "step": 50810
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005866395301999195,
+      "loss": 0.5152,
+      "step": 50820
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005865971294706269,
+      "loss": 0.5472,
+      "step": 50830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005865547287413343,
+      "loss": 0.5255,
+      "step": 50840
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005865123280120419,
+      "loss": 0.4997,
+      "step": 50850
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005864699272827493,
+      "loss": 0.5561,
+      "step": 50860
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005864275265534567,
+      "loss": 0.5544,
+      "step": 50870
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005863851258241642,
+      "loss": 0.5322,
+      "step": 50880
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005863427250948717,
+      "loss": 0.5739,
+      "step": 50890
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005863003243655791,
+      "loss": 0.5599,
+      "step": 50900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005862579236362866,
+      "loss": 0.5736,
+      "step": 50910
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.000586215522906994,
+      "loss": 0.5404,
+      "step": 50920
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005861731221777015,
+      "loss": 0.471,
+      "step": 50930
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.000586130721448409,
+      "loss": 0.433,
+      "step": 50940
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005860883207191164,
+      "loss": 0.7258,
+      "step": 50950
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005860459199898238,
+      "loss": 0.601,
+      "step": 50960
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005860035192605314,
+      "loss": 0.6164,
+      "step": 50970
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005859611185312388,
+      "loss": 0.4885,
+      "step": 50980
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005859187178019462,
+      "loss": 0.6148,
+      "step": 50990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005858763170726537,
+      "loss": 0.5312,
+      "step": 51000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005858339163433612,
+      "loss": 0.5341,
+      "step": 51010
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005857915156140686,
+      "loss": 0.5922,
+      "step": 51020
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005857491148847761,
+      "loss": 0.562,
+      "step": 51030
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005857067141554834,
+      "loss": 0.4545,
+      "step": 51040
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.000585664313426191,
+      "loss": 0.6225,
+      "step": 51050
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005856219126968985,
+      "loss": 0.5042,
+      "step": 51060
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005855795119676058,
+      "loss": 0.5098,
+      "step": 51070
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005855371112383133,
+      "loss": 0.5776,
+      "step": 51080
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005854947105090209,
+      "loss": 0.5457,
+      "step": 51090
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005854523097797282,
+      "loss": 0.4899,
+      "step": 51100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005854099090504357,
+      "loss": 0.4404,
+      "step": 51110
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005853675083211431,
+      "loss": 0.5087,
+      "step": 51120
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005853251075918506,
+      "loss": 0.5631,
+      "step": 51130
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005852827068625581,
+      "loss": 0.559,
+      "step": 51140
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005852403061332655,
+      "loss": 0.5756,
+      "step": 51150
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005851979054039729,
+      "loss": 0.5653,
+      "step": 51160
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005851555046746805,
+      "loss": 0.5888,
+      "step": 51170
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.000585113103945388,
+      "loss": 0.5178,
+      "step": 51180
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005850707032160953,
+      "loss": 0.4952,
+      "step": 51190
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005850283024868028,
+      "loss": 0.4933,
+      "step": 51200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005849859017575102,
+      "loss": 0.5579,
+      "step": 51210
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005849435010282177,
+      "loss": 0.5654,
+      "step": 51220
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005849011002989252,
+      "loss": 0.5135,
+      "step": 51230
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005848586995696326,
+      "loss": 0.5242,
+      "step": 51240
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005848162988403401,
+      "loss": 0.4766,
+      "step": 51250
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005847738981110476,
+      "loss": 0.4537,
+      "step": 51260
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.000584731497381755,
+      "loss": 0.5505,
+      "step": 51270
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005846890966524624,
+      "loss": 0.5223,
+      "step": 51280
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005846466959231699,
+      "loss": 0.4849,
+      "step": 51290
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005846042951938774,
+      "loss": 0.4304,
+      "step": 51300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005845618944645848,
+      "loss": 0.5402,
+      "step": 51310
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005845194937352923,
+      "loss": 0.5595,
+      "step": 51320
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005844770930059997,
+      "loss": 0.6415,
+      "step": 51330
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005844346922767072,
+      "loss": 0.6371,
+      "step": 51340
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005843922915474147,
+      "loss": 0.6055,
+      "step": 51350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.000584349890818122,
+      "loss": 0.5656,
+      "step": 51360
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005843074900888295,
+      "loss": 0.4312,
+      "step": 51370
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005842650893595371,
+      "loss": 0.6341,
+      "step": 51380
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005842226886302444,
+      "loss": 0.5313,
+      "step": 51390
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005841802879009519,
+      "loss": 0.6417,
+      "step": 51400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005841378871716594,
+      "loss": 0.4814,
+      "step": 51410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005840954864423668,
+      "loss": 0.4796,
+      "step": 51420
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005840530857130743,
+      "loss": 0.6121,
+      "step": 51430
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005840106849837818,
+      "loss": 0.6449,
+      "step": 51440
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005839682842544892,
+      "loss": 0.4985,
+      "step": 51450
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005839258835251967,
+      "loss": 0.68,
+      "step": 51460
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005838834827959042,
+      "loss": 0.5731,
+      "step": 51470
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005838410820666115,
+      "loss": 0.5508,
+      "step": 51480
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.000583798681337319,
+      "loss": 0.5573,
+      "step": 51490
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005837562806080266,
+      "loss": 0.4559,
+      "step": 51500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005837138798787339,
+      "loss": 0.5225,
+      "step": 51510
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005836714791494414,
+      "loss": 0.5508,
+      "step": 51520
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005836290784201489,
+      "loss": 0.5588,
+      "step": 51530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005835866776908563,
+      "loss": 0.5248,
+      "step": 51540
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005835442769615638,
+      "loss": 0.5573,
+      "step": 51550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005835018762322713,
+      "loss": 0.5023,
+      "step": 51560
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005834594755029786,
+      "loss": 0.5796,
+      "step": 51570
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005834170747736862,
+      "loss": 0.5753,
+      "step": 51580
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005833746740443937,
+      "loss": 0.5658,
+      "step": 51590
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.000583332273315101,
+      "loss": 0.5473,
+      "step": 51600
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005832898725858085,
+      "loss": 0.6368,
+      "step": 51610
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.000583247471856516,
+      "loss": 0.5308,
+      "step": 51620
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005832050711272234,
+      "loss": 0.4662,
+      "step": 51630
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005831626703979309,
+      "loss": 0.4769,
+      "step": 51640
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005831202696686382,
+      "loss": 0.5762,
+      "step": 51650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005830778689393458,
+      "loss": 0.5619,
+      "step": 51660
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005830354682100533,
+      "loss": 0.4849,
+      "step": 51670
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005829930674807606,
+      "loss": 0.573,
+      "step": 51680
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005829506667514681,
+      "loss": 0.5153,
+      "step": 51690
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005829082660221757,
+      "loss": 0.5465,
+      "step": 51700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.000582865865292883,
+      "loss": 0.5136,
+      "step": 51710
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005828234645635905,
+      "loss": 0.5402,
+      "step": 51720
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.000582781063834298,
+      "loss": 0.5092,
+      "step": 51730
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005827386631050054,
+      "loss": 0.6069,
+      "step": 51740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005826962623757129,
+      "loss": 0.5257,
+      "step": 51750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005826538616464204,
+      "loss": 0.4903,
+      "step": 51760
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005826114609171277,
+      "loss": 0.5139,
+      "step": 51770
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005825690601878353,
+      "loss": 0.531,
+      "step": 51780
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005825266594585428,
+      "loss": 0.5382,
+      "step": 51790
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005824842587292501,
+      "loss": 0.5554,
+      "step": 51800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005824418579999576,
+      "loss": 0.5095,
+      "step": 51810
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005823994572706651,
+      "loss": 0.5218,
+      "step": 51820
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005823570565413725,
+      "loss": 0.6137,
+      "step": 51830
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00058231465581208,
+      "loss": 0.5847,
+      "step": 51840
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005822722550827875,
+      "loss": 0.438,
+      "step": 51850
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005822298543534949,
+      "loss": 0.4941,
+      "step": 51860
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005821874536242024,
+      "loss": 0.5508,
+      "step": 51870
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005821450528949099,
+      "loss": 0.5063,
+      "step": 51880
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005821026521656172,
+      "loss": 0.5294,
+      "step": 51890
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005820602514363247,
+      "loss": 0.6089,
+      "step": 51900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005820178507070323,
+      "loss": 0.5715,
+      "step": 51910
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005819754499777396,
+      "loss": 0.5184,
+      "step": 51920
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005819330492484471,
+      "loss": 0.4328,
+      "step": 51930
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005818906485191546,
+      "loss": 0.5033,
+      "step": 51940
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.000581848247789862,
+      "loss": 0.5208,
+      "step": 51950
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005818058470605695,
+      "loss": 0.6041,
+      "step": 51960
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005817634463312768,
+      "loss": 0.5607,
+      "step": 51970
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005817210456019844,
+      "loss": 0.5245,
+      "step": 51980
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005816786448726919,
+      "loss": 0.4966,
+      "step": 51990
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005816362441433992,
+      "loss": 0.6623,
+      "step": 52000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005815938434141067,
+      "loss": 0.6049,
+      "step": 52010
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005815514426848142,
+      "loss": 0.5516,
+      "step": 52020
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005815090419555216,
+      "loss": 0.578,
+      "step": 52030
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005814666412262291,
+      "loss": 0.5082,
+      "step": 52040
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005814242404969366,
+      "loss": 0.6198,
+      "step": 52050
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.000581381839767644,
+      "loss": 0.5347,
+      "step": 52060
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005813394390383515,
+      "loss": 0.4767,
+      "step": 52070
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.000581297038309059,
+      "loss": 0.6466,
+      "step": 52080
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005812546375797663,
+      "loss": 0.5615,
+      "step": 52090
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005812122368504738,
+      "loss": 0.4641,
+      "step": 52100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005811698361211814,
+      "loss": 0.648,
+      "step": 52110
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005811274353918887,
+      "loss": 0.6095,
+      "step": 52120
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005810850346625962,
+      "loss": 0.5041,
+      "step": 52130
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005810426339333037,
+      "loss": 0.5076,
+      "step": 52140
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005810002332040111,
+      "loss": 0.6962,
+      "step": 52150
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005809578324747186,
+      "loss": 0.4983,
+      "step": 52160
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005809154317454261,
+      "loss": 0.5715,
+      "step": 52170
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005808730310161334,
+      "loss": 0.6177,
+      "step": 52180
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.000580830630286841,
+      "loss": 0.6148,
+      "step": 52190
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005807882295575485,
+      "loss": 0.5601,
+      "step": 52200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005807458288282558,
+      "loss": 0.6229,
+      "step": 52210
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005807034280989633,
+      "loss": 0.5258,
+      "step": 52220
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005806610273696709,
+      "loss": 0.6037,
+      "step": 52230
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005806186266403782,
+      "loss": 0.4773,
+      "step": 52240
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005805762259110857,
+      "loss": 0.5346,
+      "step": 52250
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005805338251817931,
+      "loss": 0.5886,
+      "step": 52260
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005804914244525006,
+      "loss": 0.6226,
+      "step": 52270
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005804490237232081,
+      "loss": 0.5693,
+      "step": 52280
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005804066229939155,
+      "loss": 0.5343,
+      "step": 52290
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005803642222646229,
+      "loss": 0.5677,
+      "step": 52300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005803218215353305,
+      "loss": 0.4871,
+      "step": 52310
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005802794208060379,
+      "loss": 0.4957,
+      "step": 52320
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005802370200767453,
+      "loss": 0.579,
+      "step": 52330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005801946193474528,
+      "loss": 0.5916,
+      "step": 52340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005801522186181603,
+      "loss": 0.5858,
+      "step": 52350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005801098178888677,
+      "loss": 0.5902,
+      "step": 52360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005800674171595752,
+      "loss": 0.5167,
+      "step": 52370
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005800250164302826,
+      "loss": 0.5289,
+      "step": 52380
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005799826157009901,
+      "loss": 0.5725,
+      "step": 52390
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005799402149716976,
+      "loss": 0.528,
+      "step": 52400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.000579897814242405,
+      "loss": 0.6092,
+      "step": 52410
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005798554135131124,
+      "loss": 0.5771,
+      "step": 52420
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00057981301278382,
+      "loss": 0.6738,
+      "step": 52430
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005797706120545274,
+      "loss": 0.5114,
+      "step": 52440
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005797282113252348,
+      "loss": 0.4938,
+      "step": 52450
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005796858105959423,
+      "loss": 0.5146,
+      "step": 52460
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005796434098666498,
+      "loss": 0.5492,
+      "step": 52470
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005796010091373572,
+      "loss": 0.5664,
+      "step": 52480
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005795586084080647,
+      "loss": 0.5716,
+      "step": 52490
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.000579516207678772,
+      "loss": 0.5549,
+      "step": 52500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005794738069494796,
+      "loss": 0.6391,
+      "step": 52510
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005794314062201871,
+      "loss": 0.5824,
+      "step": 52520
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005793890054908944,
+      "loss": 0.6416,
+      "step": 52530
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005793466047616019,
+      "loss": 0.6526,
+      "step": 52540
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005793042040323094,
+      "loss": 0.5576,
+      "step": 52550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005792618033030168,
+      "loss": 0.5571,
+      "step": 52560
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005792194025737243,
+      "loss": 0.6332,
+      "step": 52570
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005791770018444317,
+      "loss": 0.5914,
+      "step": 52580
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005791346011151392,
+      "loss": 0.4489,
+      "step": 52590
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005790922003858467,
+      "loss": 0.56,
+      "step": 52600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005790497996565541,
+      "loss": 0.532,
+      "step": 52610
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005790073989272615,
+      "loss": 0.5385,
+      "step": 52620
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.000578964998197969,
+      "loss": 0.6132,
+      "step": 52630
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005789225974686765,
+      "loss": 0.5939,
+      "step": 52640
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005788801967393839,
+      "loss": 0.5292,
+      "step": 52650
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005788377960100914,
+      "loss": 0.4504,
+      "step": 52660
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005787953952807989,
+      "loss": 0.5257,
+      "step": 52670
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005787529945515063,
+      "loss": 0.6144,
+      "step": 52680
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005787105938222138,
+      "loss": 0.5127,
+      "step": 52690
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005786681930929212,
+      "loss": 0.4926,
+      "step": 52700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005786257923636286,
+      "loss": 0.5985,
+      "step": 52710
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005785833916343362,
+      "loss": 0.5277,
+      "step": 52720
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005785409909050436,
+      "loss": 0.5498,
+      "step": 52730
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.000578498590175751,
+      "loss": 0.5169,
+      "step": 52740
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005784561894464585,
+      "loss": 0.5062,
+      "step": 52750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.000578413788717166,
+      "loss": 0.5106,
+      "step": 52760
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005783713879878734,
+      "loss": 0.5832,
+      "step": 52770
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005783289872585809,
+      "loss": 0.4333,
+      "step": 52780
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005782865865292883,
+      "loss": 0.5144,
+      "step": 52790
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005782441857999958,
+      "loss": 0.5038,
+      "step": 52800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005782017850707033,
+      "loss": 0.5306,
+      "step": 52810
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005781593843414107,
+      "loss": 0.5725,
+      "step": 52820
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005781169836121181,
+      "loss": 0.588,
+      "step": 52830
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005780745828828257,
+      "loss": 0.6158,
+      "step": 52840
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005780321821535331,
+      "loss": 0.5921,
+      "step": 52850
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005779897814242405,
+      "loss": 0.4961,
+      "step": 52860
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005779473806949479,
+      "loss": 0.5434,
+      "step": 52870
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005779049799656555,
+      "loss": 0.5944,
+      "step": 52880
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005778625792363629,
+      "loss": 0.6043,
+      "step": 52890
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005778201785070703,
+      "loss": 0.5002,
+      "step": 52900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005777777777777778,
+      "loss": 0.5842,
+      "step": 52910
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005777353770484853,
+      "loss": 0.568,
+      "step": 52920
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005776929763191927,
+      "loss": 0.5699,
+      "step": 52930
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005776505755899002,
+      "loss": 0.5143,
+      "step": 52940
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005776081748606076,
+      "loss": 0.5317,
+      "step": 52950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005775657741313151,
+      "loss": 0.5807,
+      "step": 52960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005775233734020226,
+      "loss": 0.5213,
+      "step": 52970
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00057748097267273,
+      "loss": 0.4169,
+      "step": 52980
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005774385719434374,
+      "loss": 0.5842,
+      "step": 52990
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.000577396171214145,
+      "loss": 0.631,
+      "step": 53000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005773537704848524,
+      "loss": 0.5452,
+      "step": 53010
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005773113697555598,
+      "loss": 0.5233,
+      "step": 53020
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005772689690262672,
+      "loss": 0.5356,
+      "step": 53030
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005772265682969748,
+      "loss": 0.584,
+      "step": 53040
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005771841675676822,
+      "loss": 0.5422,
+      "step": 53050
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005771417668383896,
+      "loss": 0.6475,
+      "step": 53060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005770993661090971,
+      "loss": 0.5606,
+      "step": 53070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005770569653798046,
+      "loss": 0.5272,
+      "step": 53080
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000577014564650512,
+      "loss": 0.632,
+      "step": 53090
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005769721639212195,
+      "loss": 0.5727,
+      "step": 53100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005769297631919269,
+      "loss": 0.6079,
+      "step": 53110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005768873624626344,
+      "loss": 0.6643,
+      "step": 53120
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005768449617333419,
+      "loss": 0.616,
+      "step": 53130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005768025610040493,
+      "loss": 0.497,
+      "step": 53140
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005767601602747567,
+      "loss": 0.4824,
+      "step": 53150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005767177595454642,
+      "loss": 0.5245,
+      "step": 53160
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005766753588161717,
+      "loss": 0.4854,
+      "step": 53170
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005766329580868791,
+      "loss": 0.6134,
+      "step": 53180
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005765905573575865,
+      "loss": 0.5409,
+      "step": 53190
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005765481566282941,
+      "loss": 0.5154,
+      "step": 53200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005765057558990015,
+      "loss": 0.5471,
+      "step": 53210
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005764633551697089,
+      "loss": 0.4887,
+      "step": 53220
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005764209544404164,
+      "loss": 0.562,
+      "step": 53230
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005763785537111238,
+      "loss": 0.5026,
+      "step": 53240
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005763361529818313,
+      "loss": 0.5624,
+      "step": 53250
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005762937522525388,
+      "loss": 0.5862,
+      "step": 53260
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005762513515232462,
+      "loss": 0.5449,
+      "step": 53270
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005762089507939537,
+      "loss": 0.485,
+      "step": 53280
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005761665500646612,
+      "loss": 0.65,
+      "step": 53290
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005761241493353686,
+      "loss": 0.6087,
+      "step": 53300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000576081748606076,
+      "loss": 0.6243,
+      "step": 53310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005760393478767835,
+      "loss": 0.5997,
+      "step": 53320
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000575996947147491,
+      "loss": 0.5928,
+      "step": 53330
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005759545464181984,
+      "loss": 0.6114,
+      "step": 53340
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005759121456889059,
+      "loss": 0.6406,
+      "step": 53350
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005758697449596133,
+      "loss": 0.6383,
+      "step": 53360
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005758273442303208,
+      "loss": 0.5361,
+      "step": 53370
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005757849435010283,
+      "loss": 0.752,
+      "step": 53380
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005757425427717357,
+      "loss": 0.5543,
+      "step": 53390
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005757001420424431,
+      "loss": 0.5028,
+      "step": 53400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005756577413131507,
+      "loss": 0.5278,
+      "step": 53410
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005756153405838581,
+      "loss": 0.5393,
+      "step": 53420
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005755729398545655,
+      "loss": 0.4457,
+      "step": 53430
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000575530539125273,
+      "loss": 0.5571,
+      "step": 53440
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005754881383959805,
+      "loss": 0.5333,
+      "step": 53450
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005754457376666879,
+      "loss": 0.5601,
+      "step": 53460
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005754033369373953,
+      "loss": 0.5598,
+      "step": 53470
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005753609362081028,
+      "loss": 0.6427,
+      "step": 53480
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005753185354788103,
+      "loss": 0.4862,
+      "step": 53490
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005752761347495177,
+      "loss": 0.5087,
+      "step": 53500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005752337340202251,
+      "loss": 0.7063,
+      "step": 53510
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005751913332909326,
+      "loss": 0.5382,
+      "step": 53520
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005751489325616402,
+      "loss": 0.5725,
+      "step": 53530
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005751065318323475,
+      "loss": 0.5567,
+      "step": 53540
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000575064131103055,
+      "loss": 0.5427,
+      "step": 53550
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005750217303737624,
+      "loss": 0.5625,
+      "step": 53560
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005749793296444699,
+      "loss": 0.6265,
+      "step": 53570
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005749369289151774,
+      "loss": 0.4798,
+      "step": 53580
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005748945281858848,
+      "loss": 0.5168,
+      "step": 53590
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005748521274565922,
+      "loss": 0.6712,
+      "step": 53600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005748097267272998,
+      "loss": 0.6749,
+      "step": 53610
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005747673259980072,
+      "loss": 0.6262,
+      "step": 53620
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005747249252687146,
+      "loss": 0.5397,
+      "step": 53630
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005746825245394221,
+      "loss": 0.487,
+      "step": 53640
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005746401238101296,
+      "loss": 0.5717,
+      "step": 53650
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000574597723080837,
+      "loss": 0.4783,
+      "step": 53660
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005745553223515445,
+      "loss": 0.602,
+      "step": 53670
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005745129216222519,
+      "loss": 0.5467,
+      "step": 53680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005744705208929594,
+      "loss": 0.4987,
+      "step": 53690
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005744281201636669,
+      "loss": 0.4431,
+      "step": 53700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005743857194343743,
+      "loss": 0.4971,
+      "step": 53710
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005743433187050817,
+      "loss": 0.5843,
+      "step": 53720
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005743009179757893,
+      "loss": 0.548,
+      "step": 53730
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005742585172464967,
+      "loss": 0.5955,
+      "step": 53740
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005742161165172041,
+      "loss": 0.5533,
+      "step": 53750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005741737157879116,
+      "loss": 0.4899,
+      "step": 53760
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000574131315058619,
+      "loss": 0.5893,
+      "step": 53770
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005740889143293265,
+      "loss": 0.583,
+      "step": 53780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000574046513600034,
+      "loss": 0.5442,
+      "step": 53790
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005740041128707413,
+      "loss": 0.5675,
+      "step": 53800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005739617121414489,
+      "loss": 0.5417,
+      "step": 53810
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005739193114121564,
+      "loss": 0.4939,
+      "step": 53820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005738769106828637,
+      "loss": 0.5553,
+      "step": 53830
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005738345099535712,
+      "loss": 0.5126,
+      "step": 53840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005737921092242787,
+      "loss": 0.5282,
+      "step": 53850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005737497084949861,
+      "loss": 0.4711,
+      "step": 53860
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005737073077656936,
+      "loss": 0.5113,
+      "step": 53870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000573664907036401,
+      "loss": 0.5746,
+      "step": 53880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005736225063071085,
+      "loss": 0.566,
+      "step": 53890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000573580105577816,
+      "loss": 0.433,
+      "step": 53900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005735377048485235,
+      "loss": 0.5095,
+      "step": 53910
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005734953041192308,
+      "loss": 0.4827,
+      "step": 53920
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005734529033899384,
+      "loss": 0.5784,
+      "step": 53930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005734105026606459,
+      "loss": 0.5683,
+      "step": 53940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005733681019313532,
+      "loss": 0.5308,
+      "step": 53950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005733257012020607,
+      "loss": 0.5355,
+      "step": 53960
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005732833004727681,
+      "loss": 0.5426,
+      "step": 53970
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005732408997434756,
+      "loss": 0.5047,
+      "step": 53980
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005731984990141831,
+      "loss": 0.4786,
+      "step": 53990
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005731560982848905,
+      "loss": 0.5134,
+      "step": 54000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000573113697555598,
+      "loss": 0.4157,
+      "step": 54010
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005730712968263055,
+      "loss": 0.5882,
+      "step": 54020
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000573028896097013,
+      "loss": 0.5142,
+      "step": 54030
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005729864953677203,
+      "loss": 0.5637,
+      "step": 54040
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005729440946384278,
+      "loss": 0.6372,
+      "step": 54050
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005729016939091353,
+      "loss": 0.5748,
+      "step": 54060
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005728592931798427,
+      "loss": 0.4832,
+      "step": 54070
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005728168924505502,
+      "loss": 0.5169,
+      "step": 54080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005727744917212576,
+      "loss": 0.5686,
+      "step": 54090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005727320909919651,
+      "loss": 0.4788,
+      "step": 54100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005726896902626726,
+      "loss": 0.5093,
+      "step": 54110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005726472895333799,
+      "loss": 0.5584,
+      "step": 54120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005726048888040874,
+      "loss": 0.5349,
+      "step": 54130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.000572562488074795,
+      "loss": 0.5658,
+      "step": 54140
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005725200873455023,
+      "loss": 0.5673,
+      "step": 54150
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005724776866162098,
+      "loss": 0.5128,
+      "step": 54160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005724352858869173,
+      "loss": 0.5241,
+      "step": 54170
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005723928851576247,
+      "loss": 0.5904,
+      "step": 54180
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005723504844283322,
+      "loss": 0.5631,
+      "step": 54190
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005723080836990397,
+      "loss": 0.5808,
+      "step": 54200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.000572265682969747,
+      "loss": 0.5334,
+      "step": 54210
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005722232822404546,
+      "loss": 0.5958,
+      "step": 54220
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005721808815111621,
+      "loss": 0.5251,
+      "step": 54230
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005721384807818694,
+      "loss": 0.4831,
+      "step": 54240
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005720960800525769,
+      "loss": 0.5635,
+      "step": 54250
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005720536793232845,
+      "loss": 0.5387,
+      "step": 54260
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005720112785939918,
+      "loss": 0.6658,
+      "step": 54270
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005719688778646993,
+      "loss": 0.5417,
+      "step": 54280
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005719264771354068,
+      "loss": 0.7009,
+      "step": 54290
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005718840764061142,
+      "loss": 0.6154,
+      "step": 54300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005718416756768217,
+      "loss": 0.6059,
+      "step": 54310
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005717992749475292,
+      "loss": 0.606,
+      "step": 54320
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005717568742182365,
+      "loss": 0.5046,
+      "step": 54330
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005717144734889441,
+      "loss": 0.6098,
+      "step": 54340
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005716720727596516,
+      "loss": 0.4544,
+      "step": 54350
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005716296720303589,
+      "loss": 0.618,
+      "step": 54360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005715872713010664,
+      "loss": 0.5662,
+      "step": 54370
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005715448705717739,
+      "loss": 0.5281,
+      "step": 54380
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005715024698424813,
+      "loss": 0.5062,
+      "step": 54390
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005714600691131888,
+      "loss": 0.5025,
+      "step": 54400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005714176683838961,
+      "loss": 0.597,
+      "step": 54410
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005713752676546037,
+      "loss": 0.4972,
+      "step": 54420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005713328669253112,
+      "loss": 0.5571,
+      "step": 54430
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005712904661960185,
+      "loss": 0.5924,
+      "step": 54440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.000571248065466726,
+      "loss": 0.5336,
+      "step": 54450
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005712056647374336,
+      "loss": 0.5648,
+      "step": 54460
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005711632640081409,
+      "loss": 0.5268,
+      "step": 54470
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005711208632788484,
+      "loss": 0.633,
+      "step": 54480
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005710784625495559,
+      "loss": 0.5307,
+      "step": 54490
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005710360618202633,
+      "loss": 0.5918,
+      "step": 54500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005709936610909708,
+      "loss": 0.4573,
+      "step": 54510
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005709512603616783,
+      "loss": 0.5096,
+      "step": 54520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005709088596323856,
+      "loss": 0.5082,
+      "step": 54530
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005708664589030932,
+      "loss": 0.511,
+      "step": 54540
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005708240581738007,
+      "loss": 0.5666,
+      "step": 54550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000570781657444508,
+      "loss": 0.6517,
+      "step": 54560
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005707392567152155,
+      "loss": 0.5056,
+      "step": 54570
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000570696855985923,
+      "loss": 0.5474,
+      "step": 54580
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005706544552566304,
+      "loss": 0.606,
+      "step": 54590
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005706120545273379,
+      "loss": 0.5325,
+      "step": 54600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005705696537980454,
+      "loss": 0.55,
+      "step": 54610
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005705272530687528,
+      "loss": 0.4731,
+      "step": 54620
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005704848523394603,
+      "loss": 0.4946,
+      "step": 54630
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005704424516101678,
+      "loss": 0.6169,
+      "step": 54640
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005704000508808751,
+      "loss": 0.5656,
+      "step": 54650
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005703576501515826,
+      "loss": 0.4863,
+      "step": 54660
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005703152494222902,
+      "loss": 0.5154,
+      "step": 54670
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005702728486929975,
+      "loss": 0.5406,
+      "step": 54680
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000570230447963705,
+      "loss": 0.5439,
+      "step": 54690
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005701880472344125,
+      "loss": 0.4997,
+      "step": 54700
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005701456465051199,
+      "loss": 0.5168,
+      "step": 54710
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005701032457758274,
+      "loss": 0.547,
+      "step": 54720
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005700608450465348,
+      "loss": 0.582,
+      "step": 54730
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005700184443172422,
+      "loss": 0.6018,
+      "step": 54740
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005699760435879498,
+      "loss": 0.6147,
+      "step": 54750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005699336428586572,
+      "loss": 0.4906,
+      "step": 54760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005698912421293646,
+      "loss": 0.5004,
+      "step": 54770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005698488414000721,
+      "loss": 0.5135,
+      "step": 54780
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005698064406707796,
+      "loss": 0.5753,
+      "step": 54790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000569764039941487,
+      "loss": 0.5334,
+      "step": 54800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005697216392121945,
+      "loss": 0.5567,
+      "step": 54810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005696792384829018,
+      "loss": 0.4885,
+      "step": 54820
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005696368377536094,
+      "loss": 0.4899,
+      "step": 54830
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005695944370243169,
+      "loss": 0.4705,
+      "step": 54840
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005695520362950242,
+      "loss": 0.6041,
+      "step": 54850
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005695096355657317,
+      "loss": 0.6187,
+      "step": 54860
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005694672348364393,
+      "loss": 0.5327,
+      "step": 54870
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005694248341071466,
+      "loss": 0.5895,
+      "step": 54880
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005693824333778541,
+      "loss": 0.5707,
+      "step": 54890
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005693400326485616,
+      "loss": 0.554,
+      "step": 54900
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000569297631919269,
+      "loss": 0.5775,
+      "step": 54910
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005692552311899765,
+      "loss": 0.5496,
+      "step": 54920
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000569212830460684,
+      "loss": 0.5226,
+      "step": 54930
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005691704297313913,
+      "loss": 0.546,
+      "step": 54940
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005691280290020989,
+      "loss": 0.5302,
+      "step": 54950
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005690856282728064,
+      "loss": 0.5663,
+      "step": 54960
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005690432275435137,
+      "loss": 0.6079,
+      "step": 54970
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005690008268142212,
+      "loss": 0.5562,
+      "step": 54980
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005689584260849288,
+      "loss": 0.667,
+      "step": 54990
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005689160253556361,
+      "loss": 0.5346,
+      "step": 55000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005688736246263436,
+      "loss": 0.5153,
+      "step": 55010
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.000568831223897051,
+      "loss": 0.5897,
+      "step": 55020
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005687888231677585,
+      "loss": 0.5558,
+      "step": 55030
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.000568746422438466,
+      "loss": 0.65,
+      "step": 55040
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005687040217091734,
+      "loss": 0.5143,
+      "step": 55050
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005686616209798808,
+      "loss": 0.6068,
+      "step": 55060
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005686192202505884,
+      "loss": 0.5345,
+      "step": 55070
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005685768195212958,
+      "loss": 0.5821,
+      "step": 55080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005685344187920032,
+      "loss": 0.5377,
+      "step": 55090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005684920180627107,
+      "loss": 0.5496,
+      "step": 55100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005684496173334182,
+      "loss": 0.6047,
+      "step": 55110
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005684072166041256,
+      "loss": 0.544,
+      "step": 55120
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005683648158748331,
+      "loss": 0.5076,
+      "step": 55130
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005683224151455405,
+      "loss": 0.4631,
+      "step": 55140
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.000568280014416248,
+      "loss": 0.5981,
+      "step": 55150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005682376136869555,
+      "loss": 0.556,
+      "step": 55160
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005681952129576629,
+      "loss": 0.5135,
+      "step": 55170
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005681528122283703,
+      "loss": 0.58,
+      "step": 55180
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005681104114990778,
+      "loss": 0.6407,
+      "step": 55190
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005680680107697853,
+      "loss": 0.5997,
+      "step": 55200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005680256100404927,
+      "loss": 0.5152,
+      "step": 55210
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005679832093112002,
+      "loss": 0.5646,
+      "step": 55220
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005679408085819077,
+      "loss": 0.5291,
+      "step": 55230
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005678984078526151,
+      "loss": 0.6682,
+      "step": 55240
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005678560071233226,
+      "loss": 0.5439,
+      "step": 55250
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00056781360639403,
+      "loss": 0.5261,
+      "step": 55260
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005677712056647374,
+      "loss": 0.5792,
+      "step": 55270
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.000567728804935445,
+      "loss": 0.5517,
+      "step": 55280
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005676864042061524,
+      "loss": 0.5743,
+      "step": 55290
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005676440034768598,
+      "loss": 0.5488,
+      "step": 55300
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005676016027475673,
+      "loss": 0.4516,
+      "step": 55310
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005675592020182748,
+      "loss": 0.5266,
+      "step": 55320
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005675168012889822,
+      "loss": 0.5802,
+      "step": 55330
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005674744005596896,
+      "loss": 0.4998,
+      "step": 55340
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.000567431999830397,
+      "loss": 0.6404,
+      "step": 55350
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005673895991011046,
+      "loss": 0.6038,
+      "step": 55360
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.000567347198371812,
+      "loss": 0.5937,
+      "step": 55370
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005673047976425194,
+      "loss": 0.589,
+      "step": 55380
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005672623969132269,
+      "loss": 0.5304,
+      "step": 55390
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005672199961839344,
+      "loss": 0.5797,
+      "step": 55400
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005671775954546418,
+      "loss": 0.5344,
+      "step": 55410
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005671351947253493,
+      "loss": 0.5789,
+      "step": 55420
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005670927939960568,
+      "loss": 0.6419,
+      "step": 55430
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005670503932667642,
+      "loss": 0.5965,
+      "step": 55440
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005670079925374717,
+      "loss": 0.5232,
+      "step": 55450
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005669655918081791,
+      "loss": 0.6434,
+      "step": 55460
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005669231910788865,
+      "loss": 0.6434,
+      "step": 55470
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005668807903495941,
+      "loss": 0.5476,
+      "step": 55480
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005668383896203015,
+      "loss": 0.5635,
+      "step": 55490
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005667959888910089,
+      "loss": 0.4776,
+      "step": 55500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005667535881617164,
+      "loss": 0.5336,
+      "step": 55510
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005667111874324239,
+      "loss": 0.7213,
+      "step": 55520
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005666687867031313,
+      "loss": 0.5582,
+      "step": 55530
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005666263859738388,
+      "loss": 0.5497,
+      "step": 55540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005665839852445462,
+      "loss": 0.5166,
+      "step": 55550
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005665415845152537,
+      "loss": 0.5266,
+      "step": 55560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005664991837859612,
+      "loss": 0.5959,
+      "step": 55570
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005664567830566686,
+      "loss": 0.5151,
+      "step": 55580
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000566414382327376,
+      "loss": 0.6289,
+      "step": 55590
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005663719815980836,
+      "loss": 0.6132,
+      "step": 55600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000566329580868791,
+      "loss": 0.7453,
+      "step": 55610
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005662871801394984,
+      "loss": 0.5524,
+      "step": 55620
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005662447794102058,
+      "loss": 0.5405,
+      "step": 55630
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005662023786809134,
+      "loss": 0.5286,
+      "step": 55640
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005661599779516208,
+      "loss": 0.5308,
+      "step": 55650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005661175772223282,
+      "loss": 0.4727,
+      "step": 55660
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005660751764930357,
+      "loss": 0.7251,
+      "step": 55670
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005660327757637432,
+      "loss": 0.5711,
+      "step": 55680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005659903750344506,
+      "loss": 0.5241,
+      "step": 55690
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005659479743051581,
+      "loss": 0.5588,
+      "step": 55700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005659055735758655,
+      "loss": 0.5789,
+      "step": 55710
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000565863172846573,
+      "loss": 0.6088,
+      "step": 55720
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005658207721172805,
+      "loss": 0.588,
+      "step": 55730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005657783713879879,
+      "loss": 0.5436,
+      "step": 55740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005657359706586953,
+      "loss": 0.5744,
+      "step": 55750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005656935699294029,
+      "loss": 0.5552,
+      "step": 55760
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005656511692001103,
+      "loss": 0.4999,
+      "step": 55770
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005656087684708177,
+      "loss": 0.5099,
+      "step": 55780
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005655663677415252,
+      "loss": 0.6363,
+      "step": 55790
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005655239670122326,
+      "loss": 0.5673,
+      "step": 55800
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005654815662829401,
+      "loss": 0.5138,
+      "step": 55810
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005654391655536476,
+      "loss": 0.6276,
+      "step": 55820
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000565396764824355,
+      "loss": 0.5727,
+      "step": 55830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005653543640950625,
+      "loss": 0.4723,
+      "step": 55840
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00056531196336577,
+      "loss": 0.5745,
+      "step": 55850
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005652695626364774,
+      "loss": 0.5802,
+      "step": 55860
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005652271619071848,
+      "loss": 0.5624,
+      "step": 55870
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005651847611778922,
+      "loss": 0.5577,
+      "step": 55880
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005651423604485998,
+      "loss": 0.5549,
+      "step": 55890
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005650999597193072,
+      "loss": 0.61,
+      "step": 55900
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005650575589900146,
+      "loss": 0.5965,
+      "step": 55910
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005650151582607221,
+      "loss": 0.4937,
+      "step": 55920
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005649727575314296,
+      "loss": 0.6106,
+      "step": 55930
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.000564930356802137,
+      "loss": 0.6339,
+      "step": 55940
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005648879560728444,
+      "loss": 0.5432,
+      "step": 55950
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.000564845555343552,
+      "loss": 0.5575,
+      "step": 55960
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005648031546142594,
+      "loss": 0.5714,
+      "step": 55970
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005647607538849668,
+      "loss": 0.6011,
+      "step": 55980
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005647183531556743,
+      "loss": 0.5142,
+      "step": 55990
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005646759524263817,
+      "loss": 0.601,
+      "step": 56000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005646335516970892,
+      "loss": 0.62,
+      "step": 56010
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005645911509677967,
+      "loss": 0.5306,
+      "step": 56020
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005645487502385041,
+      "loss": 0.547,
+      "step": 56030
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005645063495092116,
+      "loss": 0.5176,
+      "step": 56040
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005644639487799191,
+      "loss": 0.4862,
+      "step": 56050
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005644215480506265,
+      "loss": 0.626,
+      "step": 56060
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005643791473213339,
+      "loss": 0.5088,
+      "step": 56070
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005643367465920414,
+      "loss": 0.5485,
+      "step": 56080
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005642943458627489,
+      "loss": 0.5697,
+      "step": 56090
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005642519451334563,
+      "loss": 0.4896,
+      "step": 56100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005642095444041638,
+      "loss": 0.5423,
+      "step": 56110
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005641671436748712,
+      "loss": 0.5308,
+      "step": 56120
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005641247429455787,
+      "loss": 0.5562,
+      "step": 56130
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005640823422162862,
+      "loss": 0.5598,
+      "step": 56140
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005640399414869936,
+      "loss": 0.5479,
+      "step": 56150
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.000563997540757701,
+      "loss": 0.5503,
+      "step": 56160
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005639551400284086,
+      "loss": 0.5565,
+      "step": 56170
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.000563912739299116,
+      "loss": 0.4933,
+      "step": 56180
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005638703385698234,
+      "loss": 0.5328,
+      "step": 56190
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005638279378405309,
+      "loss": 0.6017,
+      "step": 56200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005637855371112384,
+      "loss": 0.52,
+      "step": 56210
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005637431363819458,
+      "loss": 0.5333,
+      "step": 56220
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005637007356526533,
+      "loss": 0.626,
+      "step": 56230
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005636583349233606,
+      "loss": 0.6135,
+      "step": 56240
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005636159341940682,
+      "loss": 0.6053,
+      "step": 56250
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005635735334647757,
+      "loss": 0.5556,
+      "step": 56260
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.000563531132735483,
+      "loss": 0.4902,
+      "step": 56270
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005634887320061905,
+      "loss": 0.5244,
+      "step": 56280
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005634463312768981,
+      "loss": 0.4365,
+      "step": 56290
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005634039305476054,
+      "loss": 0.554,
+      "step": 56300
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005633615298183129,
+      "loss": 0.5877,
+      "step": 56310
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005633191290890203,
+      "loss": 0.5552,
+      "step": 56320
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005632767283597278,
+      "loss": 0.5941,
+      "step": 56330
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005632343276304353,
+      "loss": 0.5789,
+      "step": 56340
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005631919269011427,
+      "loss": 0.5444,
+      "step": 56350
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005631495261718501,
+      "loss": 0.6905,
+      "step": 56360
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005631071254425577,
+      "loss": 0.5926,
+      "step": 56370
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005630647247132651,
+      "loss": 0.52,
+      "step": 56380
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005630223239839725,
+      "loss": 0.4549,
+      "step": 56390
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00056297992325468,
+      "loss": 0.5238,
+      "step": 56400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005629375225253874,
+      "loss": 0.5092,
+      "step": 56410
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005628951217960949,
+      "loss": 0.6304,
+      "step": 56420
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005628527210668024,
+      "loss": 0.4976,
+      "step": 56430
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005628103203375098,
+      "loss": 0.5495,
+      "step": 56440
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005627679196082173,
+      "loss": 0.5275,
+      "step": 56450
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005627255188789248,
+      "loss": 0.5516,
+      "step": 56460
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005626831181496322,
+      "loss": 0.4995,
+      "step": 56470
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005626407174203396,
+      "loss": 0.589,
+      "step": 56480
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005625983166910472,
+      "loss": 0.5293,
+      "step": 56490
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005625559159617546,
+      "loss": 0.4986,
+      "step": 56500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.000562513515232462,
+      "loss": 0.5304,
+      "step": 56510
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005624711145031695,
+      "loss": 0.5625,
+      "step": 56520
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005624287137738769,
+      "loss": 0.5416,
+      "step": 56530
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005623863130445844,
+      "loss": 0.5932,
+      "step": 56540
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005623439123152919,
+      "loss": 0.5493,
+      "step": 56550
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005623015115859992,
+      "loss": 0.5432,
+      "step": 56560
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005622591108567068,
+      "loss": 0.5103,
+      "step": 56570
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005622167101274143,
+      "loss": 0.5168,
+      "step": 56580
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005621743093981216,
+      "loss": 0.5633,
+      "step": 56590
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005621319086688291,
+      "loss": 0.5274,
+      "step": 56600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005620895079395366,
+      "loss": 0.4679,
+      "step": 56610
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.000562047107210244,
+      "loss": 0.5114,
+      "step": 56620
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005620047064809515,
+      "loss": 0.5912,
+      "step": 56630
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.000561962305751659,
+      "loss": 0.553,
+      "step": 56640
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005619199050223664,
+      "loss": 0.4378,
+      "step": 56650
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005618775042930739,
+      "loss": 0.4732,
+      "step": 56660
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005618351035637814,
+      "loss": 0.5626,
+      "step": 56670
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005617927028344887,
+      "loss": 0.5785,
+      "step": 56680
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005617503021051962,
+      "loss": 0.5616,
+      "step": 56690
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005617079013759038,
+      "loss": 0.6061,
+      "step": 56700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005616655006466111,
+      "loss": 0.589,
+      "step": 56710
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005616230999173186,
+      "loss": 0.5287,
+      "step": 56720
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.000561580699188026,
+      "loss": 0.5607,
+      "step": 56730
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005615382984587335,
+      "loss": 0.5583,
+      "step": 56740
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.000561495897729441,
+      "loss": 0.5545,
+      "step": 56750
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005614534970001485,
+      "loss": 0.6284,
+      "step": 56760
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005614110962708558,
+      "loss": 0.5135,
+      "step": 56770
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005613686955415634,
+      "loss": 0.5386,
+      "step": 56780
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005613262948122709,
+      "loss": 0.5378,
+      "step": 56790
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005612838940829782,
+      "loss": 0.5683,
+      "step": 56800
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005612414933536857,
+      "loss": 0.5048,
+      "step": 56810
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005611990926243933,
+      "loss": 0.5823,
+      "step": 56820
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005611566918951006,
+      "loss": 0.5877,
+      "step": 56830
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005611142911658081,
+      "loss": 0.4757,
+      "step": 56840
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005610718904365155,
+      "loss": 0.5431,
+      "step": 56850
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.000561029489707223,
+      "loss": 0.6014,
+      "step": 56860
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005609870889779305,
+      "loss": 0.5819,
+      "step": 56870
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005609446882486378,
+      "loss": 0.4981,
+      "step": 56880
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005609022875193453,
+      "loss": 0.5791,
+      "step": 56890
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005608598867900529,
+      "loss": 0.5418,
+      "step": 56900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005608174860607602,
+      "loss": 0.6301,
+      "step": 56910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005607750853314677,
+      "loss": 0.5327,
+      "step": 56920
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005607326846021752,
+      "loss": 0.5229,
+      "step": 56930
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005606902838728826,
+      "loss": 0.4874,
+      "step": 56940
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005606478831435901,
+      "loss": 0.6183,
+      "step": 56950
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005606054824142976,
+      "loss": 0.6354,
+      "step": 56960
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005605630816850049,
+      "loss": 0.6076,
+      "step": 56970
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005605206809557125,
+      "loss": 0.559,
+      "step": 56980
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00056047828022642,
+      "loss": 0.4822,
+      "step": 56990
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005604358794971273,
+      "loss": 0.5019,
+      "step": 57000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005603934787678348,
+      "loss": 0.526,
+      "step": 57010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005603510780385424,
+      "loss": 0.5834,
+      "step": 57020
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005603086773092497,
+      "loss": 0.517,
+      "step": 57030
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005602662765799572,
+      "loss": 0.5447,
+      "step": 57040
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005602238758506647,
+      "loss": 0.5892,
+      "step": 57050
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005601814751213721,
+      "loss": 0.6021,
+      "step": 57060
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005601390743920796,
+      "loss": 0.597,
+      "step": 57070
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005600966736627871,
+      "loss": 0.6265,
+      "step": 57080
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005600542729334944,
+      "loss": 0.6016,
+      "step": 57090
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.000560011872204202,
+      "loss": 0.612,
+      "step": 57100
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005599694714749095,
+      "loss": 0.4695,
+      "step": 57110
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005599270707456168,
+      "loss": 0.5754,
+      "step": 57120
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005598846700163243,
+      "loss": 0.5522,
+      "step": 57130
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005598422692870318,
+      "loss": 0.5973,
+      "step": 57140
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005597998685577392,
+      "loss": 0.5335,
+      "step": 57150
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005597574678284467,
+      "loss": 0.5432,
+      "step": 57160
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.000559715067099154,
+      "loss": 0.5078,
+      "step": 57170
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005596726663698616,
+      "loss": 0.5523,
+      "step": 57180
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005596302656405691,
+      "loss": 0.6013,
+      "step": 57190
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005595878649112765,
+      "loss": 0.5683,
+      "step": 57200
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005595454641819839,
+      "loss": 0.5164,
+      "step": 57210
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005595030634526914,
+      "loss": 0.6015,
+      "step": 57220
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005594606627233989,
+      "loss": 0.4152,
+      "step": 57230
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005594182619941063,
+      "loss": 0.5756,
+      "step": 57240
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005593758612648138,
+      "loss": 0.6152,
+      "step": 57250
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005593334605355213,
+      "loss": 0.5326,
+      "step": 57260
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005592910598062287,
+      "loss": 0.517,
+      "step": 57270
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005592486590769362,
+      "loss": 0.6328,
+      "step": 57280
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005592062583476435,
+      "loss": 0.6094,
+      "step": 57290
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.000559163857618351,
+      "loss": 0.5494,
+      "step": 57300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005591214568890586,
+      "loss": 0.5149,
+      "step": 57310
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005590790561597659,
+      "loss": 0.5713,
+      "step": 57320
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005590366554304734,
+      "loss": 0.5139,
+      "step": 57330
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005589942547011809,
+      "loss": 0.5915,
+      "step": 57340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005589518539718883,
+      "loss": 0.5676,
+      "step": 57350
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005589094532425958,
+      "loss": 0.5523,
+      "step": 57360
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005588670525133033,
+      "loss": 0.5143,
+      "step": 57370
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005588246517840106,
+      "loss": 0.5109,
+      "step": 57380
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005587822510547182,
+      "loss": 0.652,
+      "step": 57390
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005587398503254257,
+      "loss": 0.6057,
+      "step": 57400
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.000558697449596133,
+      "loss": 0.5545,
+      "step": 57410
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005586550488668405,
+      "loss": 0.515,
+      "step": 57420
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005586126481375481,
+      "loss": 0.5767,
+      "step": 57430
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005585702474082554,
+      "loss": 0.6014,
+      "step": 57440
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005585278466789629,
+      "loss": 0.571,
+      "step": 57450
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005584854459496704,
+      "loss": 0.5803,
+      "step": 57460
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005584430452203778,
+      "loss": 0.5304,
+      "step": 57470
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005584006444910853,
+      "loss": 0.5906,
+      "step": 57480
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005583582437617927,
+      "loss": 0.7207,
+      "step": 57490
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005583158430325001,
+      "loss": 0.5742,
+      "step": 57500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005582734423032077,
+      "loss": 0.4658,
+      "step": 57510
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005582310415739151,
+      "loss": 0.5384,
+      "step": 57520
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005581886408446225,
+      "loss": 0.5018,
+      "step": 57530
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00055814624011533,
+      "loss": 0.5508,
+      "step": 57540
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005581038393860375,
+      "loss": 0.5737,
+      "step": 57550
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005580614386567449,
+      "loss": 0.5859,
+      "step": 57560
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005580190379274524,
+      "loss": 0.5289,
+      "step": 57570
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005579766371981598,
+      "loss": 0.5235,
+      "step": 57580
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005579342364688673,
+      "loss": 0.5173,
+      "step": 57590
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005578918357395748,
+      "loss": 0.5387,
+      "step": 57600
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005578494350102822,
+      "loss": 0.5929,
+      "step": 57610
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005578070342809896,
+      "loss": 0.5984,
+      "step": 57620
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005577646335516972,
+      "loss": 0.5678,
+      "step": 57630
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005577222328224046,
+      "loss": 0.6249,
+      "step": 57640
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.000557679832093112,
+      "loss": 0.5558,
+      "step": 57650
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005576374313638195,
+      "loss": 0.5805,
+      "step": 57660
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.000557595030634527,
+      "loss": 0.5677,
+      "step": 57670
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005575526299052344,
+      "loss": 0.5469,
+      "step": 57680
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005575102291759419,
+      "loss": 0.604,
+      "step": 57690
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005574678284466492,
+      "loss": 0.5407,
+      "step": 57700
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005574254277173568,
+      "loss": 0.5309,
+      "step": 57710
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005573830269880643,
+      "loss": 0.5719,
+      "step": 57720
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005573406262587716,
+      "loss": 0.5927,
+      "step": 57730
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005572982255294791,
+      "loss": 0.5026,
+      "step": 57740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005572558248001866,
+      "loss": 0.5562,
+      "step": 57750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.000557213424070894,
+      "loss": 0.5561,
+      "step": 57760
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005571710233416015,
+      "loss": 0.4991,
+      "step": 57770
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005571286226123089,
+      "loss": 0.5123,
+      "step": 57780
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005570862218830164,
+      "loss": 0.5004,
+      "step": 57790
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005570438211537239,
+      "loss": 0.4923,
+      "step": 57800
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005570014204244313,
+      "loss": 0.5622,
+      "step": 57810
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005569590196951387,
+      "loss": 0.5475,
+      "step": 57820
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005569166189658462,
+      "loss": 0.559,
+      "step": 57830
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005568742182365537,
+      "loss": 0.5514,
+      "step": 57840
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005568318175072611,
+      "loss": 0.6223,
+      "step": 57850
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005567894167779686,
+      "loss": 0.5041,
+      "step": 57860
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005567470160486761,
+      "loss": 0.6847,
+      "step": 57870
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005567046153193835,
+      "loss": 0.5,
+      "step": 57880
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.000556662214590091,
+      "loss": 0.6563,
+      "step": 57890
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005566198138607984,
+      "loss": 0.5911,
+      "step": 57900
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005565774131315058,
+      "loss": 0.5629,
+      "step": 57910
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005565350124022134,
+      "loss": 0.6458,
+      "step": 57920
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005564926116729208,
+      "loss": 0.6508,
+      "step": 57930
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005564502109436282,
+      "loss": 0.5123,
+      "step": 57940
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005564078102143357,
+      "loss": 0.5002,
+      "step": 57950
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005563654094850432,
+      "loss": 0.5616,
+      "step": 57960
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005563230087557506,
+      "loss": 0.498,
+      "step": 57970
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005562806080264581,
+      "loss": 0.486,
+      "step": 57980
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005562382072971656,
+      "loss": 0.5186,
+      "step": 57990
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.000556195806567873,
+      "loss": 0.5373,
+      "step": 58000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005561534058385805,
+      "loss": 0.5718,
+      "step": 58010
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005561110051092879,
+      "loss": 0.5649,
+      "step": 58020
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005560686043799953,
+      "loss": 0.5045,
+      "step": 58030
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005560262036507029,
+      "loss": 0.5473,
+      "step": 58040
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005559838029214103,
+      "loss": 0.559,
+      "step": 58050
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005559414021921177,
+      "loss": 0.4806,
+      "step": 58060
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005558990014628252,
+      "loss": 0.5223,
+      "step": 58070
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005558566007335327,
+      "loss": 0.5027,
+      "step": 58080
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005558142000042401,
+      "loss": 0.5001,
+      "step": 58090
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005557717992749475,
+      "loss": 0.6493,
+      "step": 58100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000555729398545655,
+      "loss": 0.6827,
+      "step": 58110
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005556869978163625,
+      "loss": 0.549,
+      "step": 58120
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005556445970870699,
+      "loss": 0.5848,
+      "step": 58130
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005556021963577774,
+      "loss": 0.6631,
+      "step": 58140
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005555597956284848,
+      "loss": 0.4865,
+      "step": 58150
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005555173948991923,
+      "loss": 0.5454,
+      "step": 58160
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005554749941698998,
+      "loss": 0.6111,
+      "step": 58170
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005554325934406072,
+      "loss": 0.5929,
+      "step": 58180
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005553901927113146,
+      "loss": 0.518,
+      "step": 58190
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005553477919820222,
+      "loss": 0.5622,
+      "step": 58200
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005553053912527296,
+      "loss": 0.6257,
+      "step": 58210
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000555262990523437,
+      "loss": 0.5763,
+      "step": 58220
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005552205897941444,
+      "loss": 0.5826,
+      "step": 58230
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000555178189064852,
+      "loss": 0.5492,
+      "step": 58240
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005551357883355594,
+      "loss": 0.6057,
+      "step": 58250
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005550933876062668,
+      "loss": 0.5343,
+      "step": 58260
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005550509868769743,
+      "loss": 0.5451,
+      "step": 58270
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005550085861476818,
+      "loss": 0.4964,
+      "step": 58280
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005549661854183892,
+      "loss": 0.5515,
+      "step": 58290
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005549237846890967,
+      "loss": 0.5157,
+      "step": 58300
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005548813839598041,
+      "loss": 0.5318,
+      "step": 58310
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005548389832305116,
+      "loss": 0.489,
+      "step": 58320
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005547965825012191,
+      "loss": 0.4614,
+      "step": 58330
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005547541817719265,
+      "loss": 0.6319,
+      "step": 58340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005547117810426339,
+      "loss": 0.622,
+      "step": 58350
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005546693803133414,
+      "loss": 0.5256,
+      "step": 58360
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005546269795840489,
+      "loss": 0.5038,
+      "step": 58370
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005545845788547563,
+      "loss": 0.5473,
+      "step": 58380
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005545421781254637,
+      "loss": 0.5692,
+      "step": 58390
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005544997773961713,
+      "loss": 0.536,
+      "step": 58400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005544573766668787,
+      "loss": 0.5273,
+      "step": 58410
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005544149759375861,
+      "loss": 0.5627,
+      "step": 58420
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005543725752082936,
+      "loss": 0.4799,
+      "step": 58430
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005543301744790011,
+      "loss": 0.6208,
+      "step": 58440
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005542877737497085,
+      "loss": 0.5923,
+      "step": 58450
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.000554245373020416,
+      "loss": 0.492,
+      "step": 58460
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005542029722911234,
+      "loss": 0.6026,
+      "step": 58470
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005541605715618309,
+      "loss": 0.5129,
+      "step": 58480
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005541181708325384,
+      "loss": 0.5782,
+      "step": 58490
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005540757701032458,
+      "loss": 0.5227,
+      "step": 58500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005540333693739532,
+      "loss": 0.5224,
+      "step": 58510
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005539909686446608,
+      "loss": 0.534,
+      "step": 58520
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005539485679153682,
+      "loss": 0.5077,
+      "step": 58530
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005539061671860756,
+      "loss": 0.6322,
+      "step": 58540
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005538637664567831,
+      "loss": 0.5487,
+      "step": 58550
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005538213657274905,
+      "loss": 0.5015,
+      "step": 58560
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000553778964998198,
+      "loss": 0.6675,
+      "step": 58570
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005537365642689055,
+      "loss": 0.4788,
+      "step": 58580
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005536941635396129,
+      "loss": 0.4767,
+      "step": 58590
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005536517628103204,
+      "loss": 0.5272,
+      "step": 58600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005536093620810279,
+      "loss": 0.5105,
+      "step": 58610
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005535669613517353,
+      "loss": 0.5572,
+      "step": 58620
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005535245606224427,
+      "loss": 0.5376,
+      "step": 58630
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005534821598931501,
+      "loss": 0.5343,
+      "step": 58640
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005534397591638577,
+      "loss": 0.6082,
+      "step": 58650
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005533973584345651,
+      "loss": 0.5494,
+      "step": 58660
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005533549577052726,
+      "loss": 0.512,
+      "step": 58670
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00055331255697598,
+      "loss": 0.583,
+      "step": 58680
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005532701562466875,
+      "loss": 0.6124,
+      "step": 58690
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000553227755517395,
+      "loss": 0.6056,
+      "step": 58700
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005531853547881023,
+      "loss": 0.5013,
+      "step": 58710
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005531429540588098,
+      "loss": 0.5512,
+      "step": 58720
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005531005533295174,
+      "loss": 0.6035,
+      "step": 58730
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005530581526002247,
+      "loss": 0.5119,
+      "step": 58740
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005530157518709322,
+      "loss": 0.5042,
+      "step": 58750
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005529733511416396,
+      "loss": 0.5363,
+      "step": 58760
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005529309504123471,
+      "loss": 0.4638,
+      "step": 58770
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005528885496830546,
+      "loss": 0.6105,
+      "step": 58780
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.000552846148953762,
+      "loss": 0.4961,
+      "step": 58790
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005528037482244694,
+      "loss": 0.5692,
+      "step": 58800
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.000552761347495177,
+      "loss": 0.5878,
+      "step": 58810
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005527189467658844,
+      "loss": 0.5867,
+      "step": 58820
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005526765460365918,
+      "loss": 0.5736,
+      "step": 58830
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005526341453072993,
+      "loss": 0.677,
+      "step": 58840
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005525917445780068,
+      "loss": 0.5693,
+      "step": 58850
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005525493438487142,
+      "loss": 0.6159,
+      "step": 58860
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005525069431194217,
+      "loss": 0.5254,
+      "step": 58870
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005524645423901291,
+      "loss": 0.5877,
+      "step": 58880
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005524221416608366,
+      "loss": 0.4555,
+      "step": 58890
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005523797409315441,
+      "loss": 0.6234,
+      "step": 58900
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005523373402022515,
+      "loss": 0.4738,
+      "step": 58910
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005522949394729589,
+      "loss": 0.4954,
+      "step": 58920
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005522525387436665,
+      "loss": 0.5612,
+      "step": 58930
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005522101380143739,
+      "loss": 0.6385,
+      "step": 58940
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005521677372850813,
+      "loss": 0.5604,
+      "step": 58950
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005521253365557888,
+      "loss": 0.6475,
+      "step": 58960
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005520829358264963,
+      "loss": 0.5871,
+      "step": 58970
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005520405350972037,
+      "loss": 0.6235,
+      "step": 58980
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005519981343679112,
+      "loss": 0.5908,
+      "step": 58990
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005519557336386186,
+      "loss": 0.5087,
+      "step": 59000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005519133329093261,
+      "loss": 0.5674,
+      "step": 59010
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005518709321800336,
+      "loss": 0.6272,
+      "step": 59020
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005518285314507409,
+      "loss": 0.59,
+      "step": 59030
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005517861307214484,
+      "loss": 0.5812,
+      "step": 59040
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.000551743729992156,
+      "loss": 0.5477,
+      "step": 59050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005517013292628633,
+      "loss": 0.5762,
+      "step": 59060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005516589285335708,
+      "loss": 0.5223,
+      "step": 59070
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005516165278042783,
+      "loss": 0.5378,
+      "step": 59080
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005515741270749857,
+      "loss": 0.5626,
+      "step": 59090
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005515317263456932,
+      "loss": 0.596,
+      "step": 59100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005514893256164007,
+      "loss": 0.6499,
+      "step": 59110
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.000551446924887108,
+      "loss": 0.6724,
+      "step": 59120
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005514045241578156,
+      "loss": 0.6093,
+      "step": 59130
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005513621234285231,
+      "loss": 0.574,
+      "step": 59140
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005513197226992304,
+      "loss": 0.5883,
+      "step": 59150
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005512773219699379,
+      "loss": 0.5662,
+      "step": 59160
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005512349212406453,
+      "loss": 0.5895,
+      "step": 59170
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005511925205113528,
+      "loss": 0.5683,
+      "step": 59180
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005511501197820603,
+      "loss": 0.5199,
+      "step": 59190
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005511077190527677,
+      "loss": 0.5616,
+      "step": 59200
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005510653183234752,
+      "loss": 0.5707,
+      "step": 59210
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005510229175941827,
+      "loss": 0.5316,
+      "step": 59220
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005509805168648901,
+      "loss": 0.5373,
+      "step": 59230
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005509381161355975,
+      "loss": 0.5941,
+      "step": 59240
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.000550895715406305,
+      "loss": 0.5494,
+      "step": 59250
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005508533146770125,
+      "loss": 0.5411,
+      "step": 59260
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005508109139477199,
+      "loss": 0.5132,
+      "step": 59270
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005507685132184274,
+      "loss": 0.6374,
+      "step": 59280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005507261124891348,
+      "loss": 0.592,
+      "step": 59290
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005506837117598423,
+      "loss": 0.5795,
+      "step": 59300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005506413110305498,
+      "loss": 0.5276,
+      "step": 59310
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005505989103012571,
+      "loss": 0.5402,
+      "step": 59320
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005505565095719646,
+      "loss": 0.5217,
+      "step": 59330
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005505141088426722,
+      "loss": 0.667,
+      "step": 59340
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005504717081133795,
+      "loss": 0.6444,
+      "step": 59350
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.000550429307384087,
+      "loss": 0.5273,
+      "step": 59360
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005503869066547945,
+      "loss": 0.5754,
+      "step": 59370
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005503445059255019,
+      "loss": 0.5935,
+      "step": 59380
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005503021051962094,
+      "loss": 0.5677,
+      "step": 59390
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005502597044669169,
+      "loss": 0.5145,
+      "step": 59400
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005502173037376242,
+      "loss": 0.5718,
+      "step": 59410
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005501749030083318,
+      "loss": 0.4751,
+      "step": 59420
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005501325022790393,
+      "loss": 0.5026,
+      "step": 59430
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005500901015497466,
+      "loss": 0.4761,
+      "step": 59440
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005500477008204541,
+      "loss": 0.4975,
+      "step": 59450
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005500053000911617,
+      "loss": 0.5866,
+      "step": 59460
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.000549962899361869,
+      "loss": 0.4963,
+      "step": 59470
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005499204986325765,
+      "loss": 0.534,
+      "step": 59480
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.000549878097903284,
+      "loss": 0.579,
+      "step": 59490
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005498356971739914,
+      "loss": 0.5819,
+      "step": 59500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005497932964446989,
+      "loss": 0.5844,
+      "step": 59510
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005497508957154064,
+      "loss": 0.5217,
+      "step": 59520
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005497084949861137,
+      "loss": 0.5529,
+      "step": 59530
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005496660942568213,
+      "loss": 0.6351,
+      "step": 59540
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005496236935275288,
+      "loss": 0.4911,
+      "step": 59550
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005495812927982361,
+      "loss": 0.5049,
+      "step": 59560
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005495388920689436,
+      "loss": 0.5857,
+      "step": 59570
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005494964913396512,
+      "loss": 0.5859,
+      "step": 59580
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005494540906103585,
+      "loss": 0.5133,
+      "step": 59590
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.000549411689881066,
+      "loss": 0.556,
+      "step": 59600
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005493692891517735,
+      "loss": 0.5246,
+      "step": 59610
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005493268884224809,
+      "loss": 0.6271,
+      "step": 59620
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005492844876931884,
+      "loss": 0.5647,
+      "step": 59630
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005492420869638957,
+      "loss": 0.6235,
+      "step": 59640
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005491996862346032,
+      "loss": 0.5607,
+      "step": 59650
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005491572855053108,
+      "loss": 0.5519,
+      "step": 59660
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005491148847760181,
+      "loss": 0.4775,
+      "step": 59670
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005490724840467256,
+      "loss": 0.531,
+      "step": 59680
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005490300833174331,
+      "loss": 0.5525,
+      "step": 59690
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005489876825881405,
+      "loss": 0.4457,
+      "step": 59700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.000548945281858848,
+      "loss": 0.5814,
+      "step": 59710
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005489028811295555,
+      "loss": 0.6167,
+      "step": 59720
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005488604804002628,
+      "loss": 0.5607,
+      "step": 59730
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005488180796709704,
+      "loss": 0.5838,
+      "step": 59740
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005487756789416779,
+      "loss": 0.567,
+      "step": 59750
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005487332782123852,
+      "loss": 0.5789,
+      "step": 59760
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005486908774830927,
+      "loss": 0.5177,
+      "step": 59770
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005486484767538002,
+      "loss": 0.6863,
+      "step": 59780
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005486060760245076,
+      "loss": 0.5987,
+      "step": 59790
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005485636752952151,
+      "loss": 0.6308,
+      "step": 59800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005485212745659226,
+      "loss": 0.484,
+      "step": 59810
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00054847887383663,
+      "loss": 0.495,
+      "step": 59820
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005484364731073375,
+      "loss": 0.5858,
+      "step": 59830
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.000548394072378045,
+      "loss": 0.5158,
+      "step": 59840
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005483516716487523,
+      "loss": 0.5626,
+      "step": 59850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005483092709194598,
+      "loss": 0.6279,
+      "step": 59860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005482668701901674,
+      "loss": 0.5963,
+      "step": 59870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005482244694608747,
+      "loss": 0.722,
+      "step": 59880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005481820687315822,
+      "loss": 0.5392,
+      "step": 59890
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005481396680022897,
+      "loss": 0.5556,
+      "step": 59900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005480972672729971,
+      "loss": 0.5845,
+      "step": 59910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005480548665437046,
+      "loss": 0.6377,
+      "step": 59920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.000548012465814412,
+      "loss": 0.6205,
+      "step": 59930
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005479700650851195,
+      "loss": 0.4524,
+      "step": 59940
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.000547927664355827,
+      "loss": 0.4696,
+      "step": 59950
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005478852636265344,
+      "loss": 0.5068,
+      "step": 59960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005478428628972418,
+      "loss": 0.5286,
+      "step": 59970
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005478004621679493,
+      "loss": 0.5013,
+      "step": 59980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005477580614386568,
+      "loss": 0.576,
+      "step": 59990
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005477156607093642,
+      "loss": 0.597,
+      "step": 60000
+    },
+    {
+      "epoch": 2.54,
+      "eval_loss": 0.614812970161438,
+      "eval_runtime": 337.7266,
+      "eval_samples_per_second": 15.56,
+      "eval_steps_per_second": 3.891,
+      "step": 60000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005476732599800717,
+      "loss": 0.5573,
+      "step": 60010
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005476308592507792,
+      "loss": 0.5024,
+      "step": 60020
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005475884585214866,
+      "loss": 0.5397,
+      "step": 60030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005475460577921941,
+      "loss": 0.5716,
+      "step": 60040
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005475036570629014,
+      "loss": 0.5294,
+      "step": 60050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005474612563336089,
+      "loss": 0.5466,
+      "step": 60060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005474188556043165,
+      "loss": 0.5732,
+      "step": 60070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005473764548750238,
+      "loss": 0.578,
+      "step": 60080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005473340541457313,
+      "loss": 0.515,
+      "step": 60090
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005472916534164388,
+      "loss": 0.5342,
+      "step": 60100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005472492526871462,
+      "loss": 0.5687,
+      "step": 60110
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005472068519578537,
+      "loss": 0.5727,
+      "step": 60120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005471644512285612,
+      "loss": 0.5957,
+      "step": 60130
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005471220504992685,
+      "loss": 0.5407,
+      "step": 60140
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005470796497699761,
+      "loss": 0.6212,
+      "step": 60150
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005470372490406836,
+      "loss": 0.4535,
+      "step": 60160
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005469948483113909,
+      "loss": 0.5337,
+      "step": 60170
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005469524475820984,
+      "loss": 0.5265,
+      "step": 60180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.000546910046852806,
+      "loss": 0.5844,
+      "step": 60190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005468676461235133,
+      "loss": 0.4787,
+      "step": 60200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005468252453942208,
+      "loss": 0.5991,
+      "step": 60210
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005467828446649283,
+      "loss": 0.5654,
+      "step": 60220
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005467404439356357,
+      "loss": 0.526,
+      "step": 60230
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005466980432063432,
+      "loss": 0.5136,
+      "step": 60240
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005466556424770506,
+      "loss": 0.5956,
+      "step": 60250
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.000546613241747758,
+      "loss": 0.5169,
+      "step": 60260
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005465708410184656,
+      "loss": 0.5977,
+      "step": 60270
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.000546528440289173,
+      "loss": 0.5924,
+      "step": 60280
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005464860395598804,
+      "loss": 0.5212,
+      "step": 60290
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005464436388305879,
+      "loss": 0.5982,
+      "step": 60300
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005464012381012954,
+      "loss": 0.6596,
+      "step": 60310
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005463588373720028,
+      "loss": 0.4693,
+      "step": 60320
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005463164366427103,
+      "loss": 0.4414,
+      "step": 60330
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005462740359134177,
+      "loss": 0.5957,
+      "step": 60340
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005462316351841252,
+      "loss": 0.6169,
+      "step": 60350
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005461892344548327,
+      "loss": 0.49,
+      "step": 60360
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005461468337255401,
+      "loss": 0.5926,
+      "step": 60370
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005461044329962475,
+      "loss": 0.5905,
+      "step": 60380
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.000546062032266955,
+      "loss": 0.4116,
+      "step": 60390
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005460196315376625,
+      "loss": 0.5826,
+      "step": 60400
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005459772308083699,
+      "loss": 0.5376,
+      "step": 60410
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005459348300790774,
+      "loss": 0.4114,
+      "step": 60420
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005458924293497849,
+      "loss": 0.5173,
+      "step": 60430
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005458500286204923,
+      "loss": 0.504,
+      "step": 60440
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005458076278911998,
+      "loss": 0.4993,
+      "step": 60450
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005457652271619072,
+      "loss": 0.5309,
+      "step": 60460
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005457228264326147,
+      "loss": 0.4482,
+      "step": 60470
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005456804257033222,
+      "loss": 0.5491,
+      "step": 60480
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005456380249740296,
+      "loss": 0.5161,
+      "step": 60490
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000545595624244737,
+      "loss": 0.5631,
+      "step": 60500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005455532235154445,
+      "loss": 0.5399,
+      "step": 60510
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000545510822786152,
+      "loss": 0.5324,
+      "step": 60520
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005454684220568594,
+      "loss": 0.6212,
+      "step": 60530
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005454260213275668,
+      "loss": 0.4993,
+      "step": 60540
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005453836205982744,
+      "loss": 0.5176,
+      "step": 60550
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005453412198689818,
+      "loss": 0.6088,
+      "step": 60560
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005452988191396892,
+      "loss": 0.5554,
+      "step": 60570
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005452564184103966,
+      "loss": 0.5203,
+      "step": 60580
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005452140176811041,
+      "loss": 0.5476,
+      "step": 60590
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005451716169518116,
+      "loss": 0.5418,
+      "step": 60600
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000545129216222519,
+      "loss": 0.509,
+      "step": 60610
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005450868154932265,
+      "loss": 0.5772,
+      "step": 60620
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000545044414763934,
+      "loss": 0.5575,
+      "step": 60630
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005450020140346414,
+      "loss": 0.5033,
+      "step": 60640
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005449596133053489,
+      "loss": 0.6112,
+      "step": 60650
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005449172125760563,
+      "loss": 0.5542,
+      "step": 60660
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005448748118467637,
+      "loss": 0.5451,
+      "step": 60670
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005448324111174713,
+      "loss": 0.4488,
+      "step": 60680
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005447900103881787,
+      "loss": 0.5016,
+      "step": 60690
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005447476096588861,
+      "loss": 0.4687,
+      "step": 60700
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005447052089295936,
+      "loss": 0.5622,
+      "step": 60710
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005446628082003011,
+      "loss": 0.514,
+      "step": 60720
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005446204074710085,
+      "loss": 0.6491,
+      "step": 60730
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.000544578006741716,
+      "loss": 0.6118,
+      "step": 60740
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005445356060124234,
+      "loss": 0.5092,
+      "step": 60750
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005444932052831309,
+      "loss": 0.5162,
+      "step": 60760
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005444508045538384,
+      "loss": 0.4835,
+      "step": 60770
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005444084038245458,
+      "loss": 0.5225,
+      "step": 60780
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005443660030952532,
+      "loss": 0.5072,
+      "step": 60790
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005443236023659608,
+      "loss": 0.5549,
+      "step": 60800
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005442812016366682,
+      "loss": 0.5776,
+      "step": 60810
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005442388009073756,
+      "loss": 0.5537,
+      "step": 60820
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005441964001780831,
+      "loss": 0.4869,
+      "step": 60830
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005441539994487906,
+      "loss": 0.6895,
+      "step": 60840
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.000544111598719498,
+      "loss": 0.5598,
+      "step": 60850
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005440691979902054,
+      "loss": 0.4915,
+      "step": 60860
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005440267972609129,
+      "loss": 0.505,
+      "step": 60870
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005439843965316204,
+      "loss": 0.6275,
+      "step": 60880
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005439419958023278,
+      "loss": 0.4553,
+      "step": 60890
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005438995950730353,
+      "loss": 0.5269,
+      "step": 60900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005438571943437427,
+      "loss": 0.6661,
+      "step": 60910
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005438147936144502,
+      "loss": 0.5319,
+      "step": 60920
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005437723928851577,
+      "loss": 0.5353,
+      "step": 60930
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005437299921558651,
+      "loss": 0.5864,
+      "step": 60940
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005436875914265725,
+      "loss": 0.678,
+      "step": 60950
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005436451906972801,
+      "loss": 0.4873,
+      "step": 60960
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005436027899679875,
+      "loss": 0.5471,
+      "step": 60970
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005435603892386949,
+      "loss": 0.5628,
+      "step": 60980
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005435179885094024,
+      "loss": 0.5451,
+      "step": 60990
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005434755877801099,
+      "loss": 0.5602,
+      "step": 61000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005434331870508173,
+      "loss": 0.5452,
+      "step": 61010
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005433907863215248,
+      "loss": 0.4742,
+      "step": 61020
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005433483855922322,
+      "loss": 0.5716,
+      "step": 61030
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005433059848629397,
+      "loss": 0.5707,
+      "step": 61040
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005432635841336472,
+      "loss": 0.468,
+      "step": 61050
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005432211834043546,
+      "loss": 0.5106,
+      "step": 61060
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000543178782675062,
+      "loss": 0.5111,
+      "step": 61070
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005431363819457696,
+      "loss": 0.4967,
+      "step": 61080
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000543093981216477,
+      "loss": 0.4863,
+      "step": 61090
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005430515804871844,
+      "loss": 0.6062,
+      "step": 61100
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005430091797578918,
+      "loss": 0.5237,
+      "step": 61110
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005429667790285993,
+      "loss": 0.4689,
+      "step": 61120
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005429243782993068,
+      "loss": 0.5734,
+      "step": 61130
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005428819775700142,
+      "loss": 0.4921,
+      "step": 61140
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005428395768407217,
+      "loss": 0.5534,
+      "step": 61150
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005427971761114292,
+      "loss": 0.6353,
+      "step": 61160
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005427547753821366,
+      "loss": 0.645,
+      "step": 61170
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000542712374652844,
+      "loss": 0.6017,
+      "step": 61180
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005426699739235515,
+      "loss": 0.5586,
+      "step": 61190
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005426275731942589,
+      "loss": 0.5983,
+      "step": 61200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005425851724649664,
+      "loss": 0.5314,
+      "step": 61210
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005425427717356739,
+      "loss": 0.5566,
+      "step": 61220
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005425003710063813,
+      "loss": 0.5726,
+      "step": 61230
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005424579702770888,
+      "loss": 0.5729,
+      "step": 61240
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005424155695477963,
+      "loss": 0.6439,
+      "step": 61250
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005423731688185037,
+      "loss": 0.4948,
+      "step": 61260
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005423307680892111,
+      "loss": 0.6874,
+      "step": 61270
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005422883673599186,
+      "loss": 0.5299,
+      "step": 61280
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005422459666306261,
+      "loss": 0.4886,
+      "step": 61290
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005422035659013335,
+      "loss": 0.6245,
+      "step": 61300
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000542161165172041,
+      "loss": 0.5564,
+      "step": 61310
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005421187644427484,
+      "loss": 0.6703,
+      "step": 61320
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005420763637134559,
+      "loss": 0.6186,
+      "step": 61330
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005420339629841634,
+      "loss": 0.6104,
+      "step": 61340
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005419915622548708,
+      "loss": 0.5586,
+      "step": 61350
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005419491615255782,
+      "loss": 0.5316,
+      "step": 61360
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005419067607962858,
+      "loss": 0.4221,
+      "step": 61370
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005418643600669932,
+      "loss": 0.5421,
+      "step": 61380
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005418219593377006,
+      "loss": 0.5428,
+      "step": 61390
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005417795586084081,
+      "loss": 0.5993,
+      "step": 61400
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005417371578791156,
+      "loss": 0.5323,
+      "step": 61410
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000541694757149823,
+      "loss": 0.5041,
+      "step": 61420
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005416523564205305,
+      "loss": 0.5253,
+      "step": 61430
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005416099556912379,
+      "loss": 0.5677,
+      "step": 61440
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005415675549619454,
+      "loss": 0.641,
+      "step": 61450
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005415251542326529,
+      "loss": 0.6027,
+      "step": 61460
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005414827535033602,
+      "loss": 0.5493,
+      "step": 61470
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005414403527740677,
+      "loss": 0.4953,
+      "step": 61480
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005413979520447753,
+      "loss": 0.5974,
+      "step": 61490
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005413555513154826,
+      "loss": 0.5783,
+      "step": 61500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005413131505861901,
+      "loss": 0.558,
+      "step": 61510
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005412707498568975,
+      "loss": 0.5972,
+      "step": 61520
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000541228349127605,
+      "loss": 0.5828,
+      "step": 61530
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005411859483983125,
+      "loss": 0.5844,
+      "step": 61540
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00054114354766902,
+      "loss": 0.5492,
+      "step": 61550
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005411011469397273,
+      "loss": 0.5726,
+      "step": 61560
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005410587462104349,
+      "loss": 0.6693,
+      "step": 61570
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005410163454811423,
+      "loss": 0.5503,
+      "step": 61580
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005409739447518497,
+      "loss": 0.5151,
+      "step": 61590
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005409315440225572,
+      "loss": 0.5346,
+      "step": 61600
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005408891432932648,
+      "loss": 0.4459,
+      "step": 61610
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005408467425639721,
+      "loss": 0.4908,
+      "step": 61620
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005408043418346796,
+      "loss": 0.5772,
+      "step": 61630
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000540761941105387,
+      "loss": 0.5853,
+      "step": 61640
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005407195403760945,
+      "loss": 0.5015,
+      "step": 61650
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000540677139646802,
+      "loss": 0.5526,
+      "step": 61660
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005406347389175094,
+      "loss": 0.5633,
+      "step": 61670
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005405923381882168,
+      "loss": 0.5037,
+      "step": 61680
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005405499374589244,
+      "loss": 0.5331,
+      "step": 61690
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005405075367296318,
+      "loss": 0.5116,
+      "step": 61700
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005404651360003392,
+      "loss": 0.5214,
+      "step": 61710
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005404227352710467,
+      "loss": 0.5925,
+      "step": 61720
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005403803345417541,
+      "loss": 0.5107,
+      "step": 61730
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005403379338124616,
+      "loss": 0.6,
+      "step": 61740
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005402955330831691,
+      "loss": 0.5692,
+      "step": 61750
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005402531323538765,
+      "loss": 0.6205,
+      "step": 61760
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000540210731624584,
+      "loss": 0.551,
+      "step": 61770
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005401683308952915,
+      "loss": 0.6074,
+      "step": 61780
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005401259301659988,
+      "loss": 0.5455,
+      "step": 61790
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005400835294367063,
+      "loss": 0.5138,
+      "step": 61800
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005400411287074138,
+      "loss": 0.498,
+      "step": 61810
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005399987279781212,
+      "loss": 0.6401,
+      "step": 61820
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005399563272488287,
+      "loss": 0.5126,
+      "step": 61830
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005399139265195362,
+      "loss": 0.5412,
+      "step": 61840
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005398715257902436,
+      "loss": 0.5083,
+      "step": 61850
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005398291250609511,
+      "loss": 0.5974,
+      "step": 61860
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005397867243316586,
+      "loss": 0.628,
+      "step": 61870
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005397443236023659,
+      "loss": 0.5436,
+      "step": 61880
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005397019228730734,
+      "loss": 0.5576,
+      "step": 61890
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000539659522143781,
+      "loss": 0.5335,
+      "step": 61900
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005396171214144883,
+      "loss": 0.6635,
+      "step": 61910
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005395747206851958,
+      "loss": 0.5945,
+      "step": 61920
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005395323199559033,
+      "loss": 0.5894,
+      "step": 61930
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005394899192266107,
+      "loss": 0.5478,
+      "step": 61940
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005394475184973182,
+      "loss": 0.5898,
+      "step": 61950
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005394051177680257,
+      "loss": 0.4998,
+      "step": 61960
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005393627170387331,
+      "loss": 0.5604,
+      "step": 61970
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005393203163094406,
+      "loss": 0.5079,
+      "step": 61980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005392779155801481,
+      "loss": 0.624,
+      "step": 61990
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005392355148508554,
+      "loss": 0.6419,
+      "step": 62000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005391931141215629,
+      "loss": 0.5636,
+      "step": 62010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005391507133922705,
+      "loss": 0.5604,
+      "step": 62020
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005391083126629778,
+      "loss": 0.4566,
+      "step": 62030
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005390659119336853,
+      "loss": 0.6616,
+      "step": 62040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005390235112043927,
+      "loss": 0.5795,
+      "step": 62050
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005389811104751002,
+      "loss": 0.6493,
+      "step": 62060
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005389387097458077,
+      "loss": 0.4821,
+      "step": 62070
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000538896309016515,
+      "loss": 0.5762,
+      "step": 62080
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005388539082872225,
+      "loss": 0.5914,
+      "step": 62090
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005388115075579301,
+      "loss": 0.4733,
+      "step": 62100
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005387691068286374,
+      "loss": 0.5435,
+      "step": 62110
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005387267060993449,
+      "loss": 0.5099,
+      "step": 62120
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005386843053700524,
+      "loss": 0.514,
+      "step": 62130
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005386419046407598,
+      "loss": 0.6153,
+      "step": 62140
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005385995039114673,
+      "loss": 0.557,
+      "step": 62150
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005385571031821748,
+      "loss": 0.5452,
+      "step": 62160
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005385147024528821,
+      "loss": 0.6296,
+      "step": 62170
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005384723017235897,
+      "loss": 0.5771,
+      "step": 62180
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005384299009942972,
+      "loss": 0.522,
+      "step": 62190
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005383875002650045,
+      "loss": 0.5393,
+      "step": 62200
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000538345099535712,
+      "loss": 0.6076,
+      "step": 62210
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005383026988064196,
+      "loss": 0.6673,
+      "step": 62220
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005382602980771269,
+      "loss": 0.6234,
+      "step": 62230
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005382178973478344,
+      "loss": 0.5682,
+      "step": 62240
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005381754966185419,
+      "loss": 0.585,
+      "step": 62250
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005381330958892493,
+      "loss": 0.5313,
+      "step": 62260
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005380906951599568,
+      "loss": 0.5163,
+      "step": 62270
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005380482944306643,
+      "loss": 0.5368,
+      "step": 62280
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005380058937013716,
+      "loss": 0.5773,
+      "step": 62290
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005379634929720792,
+      "loss": 0.4948,
+      "step": 62300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005379210922427867,
+      "loss": 0.5364,
+      "step": 62310
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.000537878691513494,
+      "loss": 0.5809,
+      "step": 62320
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005378362907842015,
+      "loss": 0.6497,
+      "step": 62330
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.000537793890054909,
+      "loss": 0.5266,
+      "step": 62340
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005377514893256164,
+      "loss": 0.5228,
+      "step": 62350
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005377090885963239,
+      "loss": 0.5735,
+      "step": 62360
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005376666878670314,
+      "loss": 0.5673,
+      "step": 62370
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005376242871377388,
+      "loss": 0.5818,
+      "step": 62380
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005375818864084463,
+      "loss": 0.5311,
+      "step": 62390
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005375394856791537,
+      "loss": 0.568,
+      "step": 62400
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005374970849498611,
+      "loss": 0.5736,
+      "step": 62410
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005374546842205686,
+      "loss": 0.4852,
+      "step": 62420
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.000537412283491276,
+      "loss": 0.5895,
+      "step": 62430
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005373698827619835,
+      "loss": 0.4906,
+      "step": 62440
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.000537327482032691,
+      "loss": 0.505,
+      "step": 62450
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005372850813033985,
+      "loss": 0.5143,
+      "step": 62460
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005372426805741059,
+      "loss": 0.5496,
+      "step": 62470
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005372002798448134,
+      "loss": 0.5921,
+      "step": 62480
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005371578791155207,
+      "loss": 0.5165,
+      "step": 62490
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005371154783862283,
+      "loss": 0.568,
+      "step": 62500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005370730776569358,
+      "loss": 0.613,
+      "step": 62510
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005370306769276431,
+      "loss": 0.5791,
+      "step": 62520
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005369882761983506,
+      "loss": 0.5979,
+      "step": 62530
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005369458754690581,
+      "loss": 0.5731,
+      "step": 62540
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005369034747397655,
+      "loss": 0.5789,
+      "step": 62550
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.000536861074010473,
+      "loss": 0.5927,
+      "step": 62560
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005368186732811805,
+      "loss": 0.6206,
+      "step": 62570
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005367762725518879,
+      "loss": 0.4981,
+      "step": 62580
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005367338718225954,
+      "loss": 0.4971,
+      "step": 62590
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005366914710933029,
+      "loss": 0.5686,
+      "step": 62600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005366490703640102,
+      "loss": 0.5007,
+      "step": 62610
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005366066696347177,
+      "loss": 0.6311,
+      "step": 62620
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005365642689054253,
+      "loss": 0.5435,
+      "step": 62630
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005365218681761326,
+      "loss": 0.5352,
+      "step": 62640
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005364794674468401,
+      "loss": 0.5109,
+      "step": 62650
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005364370667175476,
+      "loss": 0.5277,
+      "step": 62660
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.000536394665988255,
+      "loss": 0.6674,
+      "step": 62670
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005363522652589625,
+      "loss": 0.4522,
+      "step": 62680
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005363098645296699,
+      "loss": 0.5409,
+      "step": 62690
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005362674638003773,
+      "loss": 0.4913,
+      "step": 62700
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005362250630710849,
+      "loss": 0.6608,
+      "step": 62710
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005361826623417923,
+      "loss": 0.4964,
+      "step": 62720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005361402616124997,
+      "loss": 0.5443,
+      "step": 62730
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005360978608832072,
+      "loss": 0.627,
+      "step": 62740
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005360554601539147,
+      "loss": 0.5027,
+      "step": 62750
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005360130594246221,
+      "loss": 0.553,
+      "step": 62760
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005359706586953296,
+      "loss": 0.612,
+      "step": 62770
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.000535928257966037,
+      "loss": 0.4994,
+      "step": 62780
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005358858572367445,
+      "loss": 0.4563,
+      "step": 62790
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000535843456507452,
+      "loss": 0.513,
+      "step": 62800
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005358010557781594,
+      "loss": 0.5919,
+      "step": 62810
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005357586550488668,
+      "loss": 0.478,
+      "step": 62820
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005357162543195744,
+      "loss": 0.5127,
+      "step": 62830
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005356738535902818,
+      "loss": 0.5484,
+      "step": 62840
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005356314528609892,
+      "loss": 0.6286,
+      "step": 62850
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005355890521316967,
+      "loss": 0.5532,
+      "step": 62860
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005355466514024042,
+      "loss": 0.5017,
+      "step": 62870
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005355042506731116,
+      "loss": 0.5736,
+      "step": 62880
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005354618499438191,
+      "loss": 0.5613,
+      "step": 62890
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005354194492145264,
+      "loss": 0.5709,
+      "step": 62900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000535377048485234,
+      "loss": 0.6232,
+      "step": 62910
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005353346477559415,
+      "loss": 0.4514,
+      "step": 62920
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005352922470266488,
+      "loss": 0.5809,
+      "step": 62930
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005352498462973563,
+      "loss": 0.5765,
+      "step": 62940
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005352074455680639,
+      "loss": 0.5277,
+      "step": 62950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005351650448387712,
+      "loss": 0.7146,
+      "step": 62960
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005351226441094787,
+      "loss": 0.5914,
+      "step": 62970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005350802433801862,
+      "loss": 0.5379,
+      "step": 62980
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005350378426508936,
+      "loss": 0.6106,
+      "step": 62990
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005349954419216011,
+      "loss": 0.4814,
+      "step": 63000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005349530411923085,
+      "loss": 0.4736,
+      "step": 63010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005349106404630159,
+      "loss": 0.6082,
+      "step": 63020
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005348682397337235,
+      "loss": 0.5278,
+      "step": 63030
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005348258390044309,
+      "loss": 0.5325,
+      "step": 63040
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005347834382751383,
+      "loss": 0.4819,
+      "step": 63050
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005347410375458458,
+      "loss": 0.491,
+      "step": 63060
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005346986368165533,
+      "loss": 0.6142,
+      "step": 63070
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005346562360872607,
+      "loss": 0.4459,
+      "step": 63080
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005346138353579682,
+      "loss": 0.6237,
+      "step": 63090
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005345714346286756,
+      "loss": 0.5468,
+      "step": 63100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005345290338993831,
+      "loss": 0.5413,
+      "step": 63110
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005344866331700906,
+      "loss": 0.5509,
+      "step": 63120
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.000534444232440798,
+      "loss": 0.5504,
+      "step": 63130
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005344018317115054,
+      "loss": 0.5245,
+      "step": 63140
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005343594309822129,
+      "loss": 0.4605,
+      "step": 63150
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005343170302529204,
+      "loss": 0.5651,
+      "step": 63160
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005342746295236278,
+      "loss": 0.5003,
+      "step": 63170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005342322287943353,
+      "loss": 0.5763,
+      "step": 63180
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005341898280650428,
+      "loss": 0.5379,
+      "step": 63190
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005341474273357502,
+      "loss": 0.5683,
+      "step": 63200
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005341050266064577,
+      "loss": 0.6308,
+      "step": 63210
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005340626258771651,
+      "loss": 0.6991,
+      "step": 63220
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005340202251478725,
+      "loss": 0.5527,
+      "step": 63230
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005339778244185801,
+      "loss": 0.5219,
+      "step": 63240
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005339354236892875,
+      "loss": 0.486,
+      "step": 63250
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005338930229599949,
+      "loss": 0.4883,
+      "step": 63260
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005338506222307024,
+      "loss": 0.6231,
+      "step": 63270
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005338082215014099,
+      "loss": 0.4818,
+      "step": 63280
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005337658207721173,
+      "loss": 0.502,
+      "step": 63290
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005337234200428247,
+      "loss": 0.4635,
+      "step": 63300
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005336810193135322,
+      "loss": 0.6835,
+      "step": 63310
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005336386185842397,
+      "loss": 0.5488,
+      "step": 63320
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005335962178549471,
+      "loss": 0.4601,
+      "step": 63330
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005335538171256546,
+      "loss": 0.5,
+      "step": 63340
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000533511416396362,
+      "loss": 0.5787,
+      "step": 63350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005334690156670695,
+      "loss": 0.5159,
+      "step": 63360
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000533426614937777,
+      "loss": 0.6051,
+      "step": 63370
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005333842142084844,
+      "loss": 0.5425,
+      "step": 63380
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005333418134791918,
+      "loss": 0.4451,
+      "step": 63390
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005332994127498994,
+      "loss": 0.6021,
+      "step": 63400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005332570120206068,
+      "loss": 0.5112,
+      "step": 63410
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005332146112913142,
+      "loss": 0.5167,
+      "step": 63420
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005331722105620216,
+      "loss": 0.5179,
+      "step": 63430
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005331298098327292,
+      "loss": 0.5562,
+      "step": 63440
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005330874091034366,
+      "loss": 0.4454,
+      "step": 63450
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000533045008374144,
+      "loss": 0.6274,
+      "step": 63460
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005330026076448515,
+      "loss": 0.542,
+      "step": 63470
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000532960206915559,
+      "loss": 0.4905,
+      "step": 63480
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005329178061862664,
+      "loss": 0.7345,
+      "step": 63490
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005328754054569739,
+      "loss": 0.519,
+      "step": 63500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005328330047276813,
+      "loss": 0.5283,
+      "step": 63510
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005327906039983888,
+      "loss": 0.506,
+      "step": 63520
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005327482032690963,
+      "loss": 0.6433,
+      "step": 63530
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005327058025398037,
+      "loss": 0.5913,
+      "step": 63540
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005326634018105111,
+      "loss": 0.5734,
+      "step": 63550
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005326210010812187,
+      "loss": 0.5057,
+      "step": 63560
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005325786003519261,
+      "loss": 0.5879,
+      "step": 63570
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005325361996226335,
+      "loss": 0.5597,
+      "step": 63580
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000532493798893341,
+      "loss": 0.4856,
+      "step": 63590
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005324513981640485,
+      "loss": 0.5882,
+      "step": 63600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005324089974347559,
+      "loss": 0.5134,
+      "step": 63610
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005323665967054633,
+      "loss": 0.6733,
+      "step": 63620
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005323241959761708,
+      "loss": 0.5876,
+      "step": 63630
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005322817952468783,
+      "loss": 0.6421,
+      "step": 63640
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005322393945175857,
+      "loss": 0.4651,
+      "step": 63650
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005321969937882932,
+      "loss": 0.5966,
+      "step": 63660
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005321545930590006,
+      "loss": 0.4412,
+      "step": 63670
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005321121923297081,
+      "loss": 0.5575,
+      "step": 63680
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005320697916004156,
+      "loss": 0.5714,
+      "step": 63690
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000532027390871123,
+      "loss": 0.6065,
+      "step": 63700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005319849901418304,
+      "loss": 0.5747,
+      "step": 63710
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000531942589412538,
+      "loss": 0.5839,
+      "step": 63720
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005319001886832454,
+      "loss": 0.6187,
+      "step": 63730
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005318577879539528,
+      "loss": 0.6066,
+      "step": 63740
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005318153872246603,
+      "loss": 0.5528,
+      "step": 63750
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005317729864953677,
+      "loss": 0.5183,
+      "step": 63760
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005317305857660752,
+      "loss": 0.5429,
+      "step": 63770
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005316881850367827,
+      "loss": 0.5855,
+      "step": 63780
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005316457843074901,
+      "loss": 0.6049,
+      "step": 63790
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005316033835781976,
+      "loss": 0.5627,
+      "step": 63800
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005315609828489051,
+      "loss": 0.5055,
+      "step": 63810
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005315185821196125,
+      "loss": 0.5854,
+      "step": 63820
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005314761813903199,
+      "loss": 0.5973,
+      "step": 63830
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005314337806610274,
+      "loss": 0.5174,
+      "step": 63840
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005313913799317349,
+      "loss": 0.6365,
+      "step": 63850
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005313489792024423,
+      "loss": 0.5359,
+      "step": 63860
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005313065784731498,
+      "loss": 0.6015,
+      "step": 63870
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005312641777438572,
+      "loss": 0.4582,
+      "step": 63880
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005312217770145647,
+      "loss": 0.4968,
+      "step": 63890
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005311793762852722,
+      "loss": 0.6134,
+      "step": 63900
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005311369755559796,
+      "loss": 0.5145,
+      "step": 63910
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000531094574826687,
+      "loss": 0.5066,
+      "step": 63920
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005310521740973946,
+      "loss": 0.5612,
+      "step": 63930
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005310097733681019,
+      "loss": 0.4812,
+      "step": 63940
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005309673726388094,
+      "loss": 0.5963,
+      "step": 63950
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005309249719095168,
+      "loss": 0.6209,
+      "step": 63960
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005308825711802243,
+      "loss": 0.577,
+      "step": 63970
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005308401704509318,
+      "loss": 0.5613,
+      "step": 63980
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005307977697216392,
+      "loss": 0.5283,
+      "step": 63990
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005307553689923467,
+      "loss": 0.592,
+      "step": 64000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005307129682630542,
+      "loss": 0.4972,
+      "step": 64010
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005306705675337616,
+      "loss": 0.501,
+      "step": 64020
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000530628166804469,
+      "loss": 0.6224,
+      "step": 64030
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005305857660751765,
+      "loss": 0.641,
+      "step": 64040
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000530543365345884,
+      "loss": 0.5952,
+      "step": 64050
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005305009646165914,
+      "loss": 0.6351,
+      "step": 64060
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005304585638872989,
+      "loss": 0.641,
+      "step": 64070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005304161631580063,
+      "loss": 0.5535,
+      "step": 64080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005303737624287138,
+      "loss": 0.4981,
+      "step": 64090
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005303313616994213,
+      "loss": 0.5695,
+      "step": 64100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005302889609701287,
+      "loss": 0.5256,
+      "step": 64110
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005302465602408361,
+      "loss": 0.565,
+      "step": 64120
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005302041595115437,
+      "loss": 0.5556,
+      "step": 64130
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005301617587822511,
+      "loss": 0.489,
+      "step": 64140
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005301193580529585,
+      "loss": 0.5536,
+      "step": 64150
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000530076957323666,
+      "loss": 0.6385,
+      "step": 64160
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005300345565943735,
+      "loss": 0.5813,
+      "step": 64170
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005299921558650809,
+      "loss": 0.559,
+      "step": 64180
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005299497551357884,
+      "loss": 0.5044,
+      "step": 64190
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005299073544064958,
+      "loss": 0.5908,
+      "step": 64200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005298649536772033,
+      "loss": 0.5873,
+      "step": 64210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005298225529479108,
+      "loss": 0.5167,
+      "step": 64220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005297801522186181,
+      "loss": 0.5785,
+      "step": 64230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005297377514893256,
+      "loss": 0.6001,
+      "step": 64240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005296953507600332,
+      "loss": 0.5348,
+      "step": 64250
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005296529500307405,
+      "loss": 0.5322,
+      "step": 64260
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.000529610549301448,
+      "loss": 0.6614,
+      "step": 64270
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005295681485721555,
+      "loss": 0.5957,
+      "step": 64280
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005295257478428629,
+      "loss": 0.5569,
+      "step": 64290
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005294833471135704,
+      "loss": 0.4545,
+      "step": 64300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005294409463842779,
+      "loss": 0.5741,
+      "step": 64310
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005293985456549852,
+      "loss": 0.5769,
+      "step": 64320
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005293561449256928,
+      "loss": 0.5855,
+      "step": 64330
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005293137441964003,
+      "loss": 0.5467,
+      "step": 64340
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005292713434671076,
+      "loss": 0.5051,
+      "step": 64350
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005292289427378151,
+      "loss": 0.5,
+      "step": 64360
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005291865420085225,
+      "loss": 0.6119,
+      "step": 64370
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00052914414127923,
+      "loss": 0.5282,
+      "step": 64380
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005291017405499375,
+      "loss": 0.6177,
+      "step": 64390
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.000529059339820645,
+      "loss": 0.5095,
+      "step": 64400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005290169390913524,
+      "loss": 0.5487,
+      "step": 64410
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005289745383620599,
+      "loss": 0.59,
+      "step": 64420
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005289321376327673,
+      "loss": 0.5499,
+      "step": 64430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005288897369034747,
+      "loss": 0.5957,
+      "step": 64440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005288473361741823,
+      "loss": 0.6216,
+      "step": 64450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005288049354448897,
+      "loss": 0.5405,
+      "step": 64460
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005287625347155971,
+      "loss": 0.5275,
+      "step": 64470
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005287201339863046,
+      "loss": 0.5763,
+      "step": 64480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000528677733257012,
+      "loss": 0.6097,
+      "step": 64490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005286353325277195,
+      "loss": 0.4259,
+      "step": 64500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000528592931798427,
+      "loss": 0.5273,
+      "step": 64510
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005285505310691344,
+      "loss": 0.5317,
+      "step": 64520
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005285081303398419,
+      "loss": 0.591,
+      "step": 64530
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005284657296105494,
+      "loss": 0.5253,
+      "step": 64540
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005284233288812567,
+      "loss": 0.618,
+      "step": 64550
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005283809281519642,
+      "loss": 0.638,
+      "step": 64560
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005283385274226717,
+      "loss": 0.5409,
+      "step": 64570
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005282961266933791,
+      "loss": 0.5381,
+      "step": 64580
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005282537259640866,
+      "loss": 0.5577,
+      "step": 64590
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005282113252347941,
+      "loss": 0.527,
+      "step": 64600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005281689245055015,
+      "loss": 0.6125,
+      "step": 64610
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000528126523776209,
+      "loss": 0.4781,
+      "step": 64620
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005280841230469165,
+      "loss": 0.5359,
+      "step": 64630
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005280417223176238,
+      "loss": 0.5531,
+      "step": 64640
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005279993215883313,
+      "loss": 0.5424,
+      "step": 64650
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005279569208590389,
+      "loss": 0.5875,
+      "step": 64660
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005279145201297462,
+      "loss": 0.6403,
+      "step": 64670
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005278721194004537,
+      "loss": 0.543,
+      "step": 64680
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005278297186711612,
+      "loss": 0.5608,
+      "step": 64690
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005277873179418686,
+      "loss": 0.4619,
+      "step": 64700
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005277449172125761,
+      "loss": 0.543,
+      "step": 64710
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005277025164832836,
+      "loss": 0.586,
+      "step": 64720
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005276601157539909,
+      "loss": 0.5554,
+      "step": 64730
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005276177150246985,
+      "loss": 0.6111,
+      "step": 64740
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000527575314295406,
+      "loss": 0.5757,
+      "step": 64750
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005275329135661133,
+      "loss": 0.5989,
+      "step": 64760
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005274905128368208,
+      "loss": 0.6962,
+      "step": 64770
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005274481121075284,
+      "loss": 0.5257,
+      "step": 64780
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005274057113782357,
+      "loss": 0.6447,
+      "step": 64790
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005273633106489432,
+      "loss": 0.4914,
+      "step": 64800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005273209099196507,
+      "loss": 0.535,
+      "step": 64810
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005272785091903581,
+      "loss": 0.4964,
+      "step": 64820
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005272361084610656,
+      "loss": 0.6069,
+      "step": 64830
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005271937077317729,
+      "loss": 0.5717,
+      "step": 64840
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005271513070024804,
+      "loss": 0.6049,
+      "step": 64850
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000527108906273188,
+      "loss": 0.4927,
+      "step": 64860
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005270665055438953,
+      "loss": 0.5566,
+      "step": 64870
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005270241048146028,
+      "loss": 0.6662,
+      "step": 64880
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005269817040853103,
+      "loss": 0.5116,
+      "step": 64890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005269393033560177,
+      "loss": 0.5188,
+      "step": 64900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005268969026267252,
+      "loss": 0.4953,
+      "step": 64910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005268545018974327,
+      "loss": 0.6639,
+      "step": 64920
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00052681210116814,
+      "loss": 0.573,
+      "step": 64930
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005267697004388476,
+      "loss": 0.6061,
+      "step": 64940
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005267272997095551,
+      "loss": 0.6073,
+      "step": 64950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005266848989802624,
+      "loss": 0.5409,
+      "step": 64960
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005266424982509699,
+      "loss": 0.6675,
+      "step": 64970
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005266000975216775,
+      "loss": 0.5805,
+      "step": 64980
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005265576967923848,
+      "loss": 0.5722,
+      "step": 64990
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005265152960630923,
+      "loss": 0.5811,
+      "step": 65000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005264728953337998,
+      "loss": 0.5526,
+      "step": 65010
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005264304946045072,
+      "loss": 0.5199,
+      "step": 65020
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005263880938752147,
+      "loss": 0.5603,
+      "step": 65030
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005263456931459222,
+      "loss": 0.5314,
+      "step": 65040
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005263032924166295,
+      "loss": 0.5364,
+      "step": 65050
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005262608916873371,
+      "loss": 0.519,
+      "step": 65060
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005262184909580446,
+      "loss": 0.5056,
+      "step": 65070
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005261760902287519,
+      "loss": 0.5263,
+      "step": 65080
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005261336894994594,
+      "loss": 0.5893,
+      "step": 65090
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005260912887701669,
+      "loss": 0.5459,
+      "step": 65100
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005260488880408743,
+      "loss": 0.549,
+      "step": 65110
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005260064873115818,
+      "loss": 0.5111,
+      "step": 65120
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005259640865822893,
+      "loss": 0.4562,
+      "step": 65130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005259216858529967,
+      "loss": 0.5349,
+      "step": 65140
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005258792851237042,
+      "loss": 0.5737,
+      "step": 65150
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005258368843944116,
+      "loss": 0.5936,
+      "step": 65160
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000525794483665119,
+      "loss": 0.5425,
+      "step": 65170
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005257520829358265,
+      "loss": 0.533,
+      "step": 65180
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000525709682206534,
+      "loss": 0.5776,
+      "step": 65190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005256672814772414,
+      "loss": 0.5567,
+      "step": 65200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005256248807479489,
+      "loss": 0.5869,
+      "step": 65210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005255824800186564,
+      "loss": 0.5645,
+      "step": 65220
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005255400792893638,
+      "loss": 0.5315,
+      "step": 65230
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005254976785600713,
+      "loss": 0.5717,
+      "step": 65240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005254552778307786,
+      "loss": 0.5497,
+      "step": 65250
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005254128771014861,
+      "loss": 0.6518,
+      "step": 65260
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005253704763721937,
+      "loss": 0.5005,
+      "step": 65270
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000525328075642901,
+      "loss": 0.5506,
+      "step": 65280
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005252856749136085,
+      "loss": 0.5421,
+      "step": 65290
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000525243274184316,
+      "loss": 0.4679,
+      "step": 65300
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005252008734550235,
+      "loss": 0.5018,
+      "step": 65310
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005251584727257309,
+      "loss": 0.6565,
+      "step": 65320
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005251160719964384,
+      "loss": 0.5195,
+      "step": 65330
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005250736712671457,
+      "loss": 0.4982,
+      "step": 65340
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005250312705378533,
+      "loss": 0.5682,
+      "step": 65350
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005249888698085608,
+      "loss": 0.6093,
+      "step": 65360
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005249464690792681,
+      "loss": 0.6474,
+      "step": 65370
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005249040683499756,
+      "loss": 0.5667,
+      "step": 65380
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005248616676206832,
+      "loss": 0.6015,
+      "step": 65390
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005248192668913905,
+      "loss": 0.4893,
+      "step": 65400
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.000524776866162098,
+      "loss": 0.535,
+      "step": 65410
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005247344654328055,
+      "loss": 0.6234,
+      "step": 65420
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005246920647035129,
+      "loss": 0.4796,
+      "step": 65430
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005246496639742204,
+      "loss": 0.6031,
+      "step": 65440
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005246072632449278,
+      "loss": 0.5497,
+      "step": 65450
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005245648625156352,
+      "loss": 0.459,
+      "step": 65460
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005245224617863428,
+      "loss": 0.5264,
+      "step": 65470
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005244800610570502,
+      "loss": 0.6038,
+      "step": 65480
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005244376603277576,
+      "loss": 0.5438,
+      "step": 65490
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005243952595984651,
+      "loss": 0.6504,
+      "step": 65500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005243528588691726,
+      "loss": 0.5372,
+      "step": 65510
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00052431045813988,
+      "loss": 0.6083,
+      "step": 65520
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005242680574105875,
+      "loss": 0.5688,
+      "step": 65530
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005242256566812949,
+      "loss": 0.5366,
+      "step": 65540
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005241832559520024,
+      "loss": 0.5622,
+      "step": 65550
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005241408552227099,
+      "loss": 0.4758,
+      "step": 65560
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005240984544934173,
+      "loss": 0.6,
+      "step": 65570
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005240560537641247,
+      "loss": 0.4609,
+      "step": 65580
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005240136530348323,
+      "loss": 0.5552,
+      "step": 65590
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005239712523055397,
+      "loss": 0.5454,
+      "step": 65600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005239288515762471,
+      "loss": 0.5821,
+      "step": 65610
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005238864508469546,
+      "loss": 0.6406,
+      "step": 65620
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005238440501176621,
+      "loss": 0.54,
+      "step": 65630
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005238016493883695,
+      "loss": 0.5965,
+      "step": 65640
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.000523759248659077,
+      "loss": 0.5455,
+      "step": 65650
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005237168479297844,
+      "loss": 0.6076,
+      "step": 65660
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005236744472004919,
+      "loss": 0.5878,
+      "step": 65670
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005236320464711994,
+      "loss": 0.5691,
+      "step": 65680
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005235896457419068,
+      "loss": 0.461,
+      "step": 65690
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005235472450126142,
+      "loss": 0.4622,
+      "step": 65700
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005235048442833217,
+      "loss": 0.5654,
+      "step": 65710
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005234624435540292,
+      "loss": 0.6422,
+      "step": 65720
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005234200428247366,
+      "loss": 0.583,
+      "step": 65730
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005233776420954441,
+      "loss": 0.5096,
+      "step": 65740
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005233352413661516,
+      "loss": 0.5181,
+      "step": 65750
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.000523292840636859,
+      "loss": 0.4637,
+      "step": 65760
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005232504399075664,
+      "loss": 0.5984,
+      "step": 65770
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005232080391782738,
+      "loss": 0.5771,
+      "step": 65780
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005231656384489813,
+      "loss": 0.5944,
+      "step": 65790
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005231232377196888,
+      "loss": 0.7141,
+      "step": 65800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005230808369903962,
+      "loss": 0.5445,
+      "step": 65810
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005230384362611037,
+      "loss": 0.5941,
+      "step": 65820
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005229960355318112,
+      "loss": 0.6274,
+      "step": 65830
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005229536348025186,
+      "loss": 0.5122,
+      "step": 65840
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005229112340732261,
+      "loss": 0.4749,
+      "step": 65850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005228688333439335,
+      "loss": 0.5781,
+      "step": 65860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005228264326146409,
+      "loss": 0.6577,
+      "step": 65870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005227840318853485,
+      "loss": 0.6542,
+      "step": 65880
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005227416311560559,
+      "loss": 0.4606,
+      "step": 65890
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005226992304267633,
+      "loss": 0.5439,
+      "step": 65900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005226568296974708,
+      "loss": 0.5499,
+      "step": 65910
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005226144289681783,
+      "loss": 0.6014,
+      "step": 65920
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005225720282388857,
+      "loss": 0.5778,
+      "step": 65930
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005225296275095932,
+      "loss": 0.5052,
+      "step": 65940
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005224872267803006,
+      "loss": 0.5712,
+      "step": 65950
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005224448260510081,
+      "loss": 0.5753,
+      "step": 65960
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005224024253217156,
+      "loss": 0.5409,
+      "step": 65970
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000522360024592423,
+      "loss": 0.4796,
+      "step": 65980
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005223176238631304,
+      "loss": 0.5941,
+      "step": 65990
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000522275223133838,
+      "loss": 0.6004,
+      "step": 66000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005222328224045454,
+      "loss": 0.5999,
+      "step": 66010
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005221904216752528,
+      "loss": 0.4649,
+      "step": 66020
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005221480209459603,
+      "loss": 0.6109,
+      "step": 66030
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005221056202166678,
+      "loss": 0.4958,
+      "step": 66040
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005220632194873752,
+      "loss": 0.5757,
+      "step": 66050
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005220208187580827,
+      "loss": 0.556,
+      "step": 66060
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005219784180287901,
+      "loss": 0.6853,
+      "step": 66070
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005219360172994976,
+      "loss": 0.5885,
+      "step": 66080
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000521893616570205,
+      "loss": 0.5715,
+      "step": 66090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005218512158409125,
+      "loss": 0.6034,
+      "step": 66100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005218088151116199,
+      "loss": 0.4779,
+      "step": 66110
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005217664143823274,
+      "loss": 0.5997,
+      "step": 66120
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005217240136530349,
+      "loss": 0.5409,
+      "step": 66130
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005216816129237423,
+      "loss": 0.5031,
+      "step": 66140
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005216392121944497,
+      "loss": 0.5848,
+      "step": 66150
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005215968114651573,
+      "loss": 0.5144,
+      "step": 66160
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005215544107358647,
+      "loss": 0.5486,
+      "step": 66170
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005215120100065721,
+      "loss": 0.5871,
+      "step": 66180
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005214696092772796,
+      "loss": 0.561,
+      "step": 66190
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005214272085479871,
+      "loss": 0.5522,
+      "step": 66200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005213848078186945,
+      "loss": 0.5706,
+      "step": 66210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000521342407089402,
+      "loss": 0.5238,
+      "step": 66220
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005213000063601094,
+      "loss": 0.5392,
+      "step": 66230
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005212576056308169,
+      "loss": 0.5042,
+      "step": 66240
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005212152049015244,
+      "loss": 0.6747,
+      "step": 66250
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005211728041722318,
+      "loss": 0.4271,
+      "step": 66260
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005211304034429392,
+      "loss": 0.6523,
+      "step": 66270
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005210880027136468,
+      "loss": 0.5325,
+      "step": 66280
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005210456019843542,
+      "loss": 0.6221,
+      "step": 66290
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005210032012550616,
+      "loss": 0.5425,
+      "step": 66300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000520960800525769,
+      "loss": 0.6491,
+      "step": 66310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005209183997964765,
+      "loss": 0.658,
+      "step": 66320
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000520875999067184,
+      "loss": 0.6298,
+      "step": 66330
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005208335983378914,
+      "loss": 0.6005,
+      "step": 66340
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005207911976085989,
+      "loss": 0.5716,
+      "step": 66350
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005207487968793064,
+      "loss": 0.6544,
+      "step": 66360
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005207063961500138,
+      "loss": 0.6155,
+      "step": 66370
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005206639954207212,
+      "loss": 0.5099,
+      "step": 66380
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005206215946914287,
+      "loss": 0.5366,
+      "step": 66390
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005205791939621361,
+      "loss": 0.4547,
+      "step": 66400
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005205367932328436,
+      "loss": 0.5339,
+      "step": 66410
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005204943925035511,
+      "loss": 0.5762,
+      "step": 66420
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005204519917742585,
+      "loss": 0.4792,
+      "step": 66430
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000520409591044966,
+      "loss": 0.4539,
+      "step": 66440
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005203671903156735,
+      "loss": 0.5326,
+      "step": 66450
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005203247895863809,
+      "loss": 0.534,
+      "step": 66460
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005202823888570883,
+      "loss": 0.6447,
+      "step": 66470
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005202399881277959,
+      "loss": 0.5231,
+      "step": 66480
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005201975873985033,
+      "loss": 0.5308,
+      "step": 66490
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005201551866692107,
+      "loss": 0.6164,
+      "step": 66500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005201127859399182,
+      "loss": 0.6078,
+      "step": 66510
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005200703852106256,
+      "loss": 0.5353,
+      "step": 66520
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005200279844813331,
+      "loss": 0.5898,
+      "step": 66530
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005199855837520406,
+      "loss": 0.5199,
+      "step": 66540
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000519943183022748,
+      "loss": 0.5812,
+      "step": 66550
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005199007822934555,
+      "loss": 0.5647,
+      "step": 66560
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000519858381564163,
+      "loss": 0.563,
+      "step": 66570
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005198159808348704,
+      "loss": 0.4807,
+      "step": 66580
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005197735801055778,
+      "loss": 0.4918,
+      "step": 66590
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005197311793762853,
+      "loss": 0.5777,
+      "step": 66600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005196887786469928,
+      "loss": 0.6448,
+      "step": 66610
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005196463779177002,
+      "loss": 0.5003,
+      "step": 66620
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005196039771884077,
+      "loss": 0.6408,
+      "step": 66630
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005195615764591151,
+      "loss": 0.5625,
+      "step": 66640
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005195191757298226,
+      "loss": 0.4986,
+      "step": 66650
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005194767750005301,
+      "loss": 0.5285,
+      "step": 66660
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005194343742712375,
+      "loss": 0.6886,
+      "step": 66670
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005193919735419449,
+      "loss": 0.6146,
+      "step": 66680
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005193495728126525,
+      "loss": 0.5014,
+      "step": 66690
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005193071720833598,
+      "loss": 0.5533,
+      "step": 66700
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005192647713540673,
+      "loss": 0.5377,
+      "step": 66710
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005192223706247748,
+      "loss": 0.5072,
+      "step": 66720
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005191799698954822,
+      "loss": 0.646,
+      "step": 66730
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005191375691661897,
+      "loss": 0.606,
+      "step": 66740
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005190951684368972,
+      "loss": 0.5916,
+      "step": 66750
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005190527677076045,
+      "loss": 0.5245,
+      "step": 66760
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005190103669783121,
+      "loss": 0.5324,
+      "step": 66770
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005189679662490196,
+      "loss": 0.5351,
+      "step": 66780
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005189255655197269,
+      "loss": 0.6992,
+      "step": 66790
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005188831647904344,
+      "loss": 0.526,
+      "step": 66800
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000518840764061142,
+      "loss": 0.4734,
+      "step": 66810
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005187983633318493,
+      "loss": 0.5802,
+      "step": 66820
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005187559626025568,
+      "loss": 0.5292,
+      "step": 66830
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005187135618732642,
+      "loss": 0.562,
+      "step": 66840
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005186711611439717,
+      "loss": 0.5512,
+      "step": 66850
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005186287604146792,
+      "loss": 0.6065,
+      "step": 66860
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005185863596853866,
+      "loss": 0.6211,
+      "step": 66870
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000518543958956094,
+      "loss": 0.5817,
+      "step": 66880
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005185015582268016,
+      "loss": 0.5376,
+      "step": 66890
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000518459157497509,
+      "loss": 0.6077,
+      "step": 66900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005184167567682164,
+      "loss": 0.5089,
+      "step": 66910
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005183743560389239,
+      "loss": 0.6108,
+      "step": 66920
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005183319553096313,
+      "loss": 0.5093,
+      "step": 66930
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005182895545803388,
+      "loss": 0.5961,
+      "step": 66940
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005182471538510463,
+      "loss": 0.5747,
+      "step": 66950
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005182047531217537,
+      "loss": 0.6478,
+      "step": 66960
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005181623523924612,
+      "loss": 0.4955,
+      "step": 66970
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005181199516631687,
+      "loss": 0.6236,
+      "step": 66980
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000518077550933876,
+      "loss": 0.456,
+      "step": 66990
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005180351502045835,
+      "loss": 0.5693,
+      "step": 67000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005179927494752911,
+      "loss": 0.4763,
+      "step": 67010
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005179503487459984,
+      "loss": 0.552,
+      "step": 67020
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005179079480167059,
+      "loss": 0.5903,
+      "step": 67030
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005178655472874134,
+      "loss": 0.5042,
+      "step": 67040
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005178231465581208,
+      "loss": 0.5885,
+      "step": 67050
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005177807458288283,
+      "loss": 0.5115,
+      "step": 67060
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005177383450995358,
+      "loss": 0.6079,
+      "step": 67070
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005176959443702431,
+      "loss": 0.5623,
+      "step": 67080
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005176535436409507,
+      "loss": 0.4898,
+      "step": 67090
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005176111429116582,
+      "loss": 0.5075,
+      "step": 67100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005175687421823655,
+      "loss": 0.5514,
+      "step": 67110
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000517526341453073,
+      "loss": 0.5684,
+      "step": 67120
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005174839407237805,
+      "loss": 0.5448,
+      "step": 67130
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005174415399944879,
+      "loss": 0.5326,
+      "step": 67140
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005173991392651954,
+      "loss": 0.5841,
+      "step": 67150
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005173567385359029,
+      "loss": 0.6066,
+      "step": 67160
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005173143378066103,
+      "loss": 0.5663,
+      "step": 67170
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005172719370773178,
+      "loss": 0.5904,
+      "step": 67180
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005172295363480253,
+      "loss": 0.4561,
+      "step": 67190
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005171871356187326,
+      "loss": 0.509,
+      "step": 67200
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005171447348894401,
+      "loss": 0.5424,
+      "step": 67210
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005171023341601477,
+      "loss": 0.6063,
+      "step": 67220
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000517059933430855,
+      "loss": 0.5457,
+      "step": 67230
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005170175327015625,
+      "loss": 0.5951,
+      "step": 67240
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00051697513197227,
+      "loss": 0.5199,
+      "step": 67250
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005169327312429774,
+      "loss": 0.624,
+      "step": 67260
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005168903305136849,
+      "loss": 0.5208,
+      "step": 67270
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005168479297843923,
+      "loss": 0.5034,
+      "step": 67280
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005168055290550997,
+      "loss": 0.5193,
+      "step": 67290
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005167631283258073,
+      "loss": 0.5758,
+      "step": 67300
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005167207275965146,
+      "loss": 0.4725,
+      "step": 67310
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005166783268672221,
+      "loss": 0.5471,
+      "step": 67320
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005166359261379296,
+      "loss": 0.5486,
+      "step": 67330
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000516593525408637,
+      "loss": 0.5003,
+      "step": 67340
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005165511246793445,
+      "loss": 0.6004,
+      "step": 67350
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000516508723950052,
+      "loss": 0.5643,
+      "step": 67360
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005164663232207593,
+      "loss": 0.5329,
+      "step": 67370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005164239224914669,
+      "loss": 0.6411,
+      "step": 67380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005163815217621744,
+      "loss": 0.5444,
+      "step": 67390
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005163391210328817,
+      "loss": 0.5373,
+      "step": 67400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005162967203035892,
+      "loss": 0.5599,
+      "step": 67410
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005162543195742968,
+      "loss": 0.6055,
+      "step": 67420
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005162119188450041,
+      "loss": 0.4969,
+      "step": 67430
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005161695181157116,
+      "loss": 0.5403,
+      "step": 67440
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005161271173864191,
+      "loss": 0.5353,
+      "step": 67450
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005160847166571265,
+      "loss": 0.5192,
+      "step": 67460
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000516042315927834,
+      "loss": 0.6583,
+      "step": 67470
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005159999151985415,
+      "loss": 0.5567,
+      "step": 67480
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005159575144692488,
+      "loss": 0.5318,
+      "step": 67490
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005159151137399564,
+      "loss": 0.5619,
+      "step": 67500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005158727130106639,
+      "loss": 0.4847,
+      "step": 67510
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005158303122813712,
+      "loss": 0.5692,
+      "step": 67520
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005157879115520787,
+      "loss": 0.5499,
+      "step": 67530
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005157455108227863,
+      "loss": 0.4291,
+      "step": 67540
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005157031100934936,
+      "loss": 0.6016,
+      "step": 67550
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005156607093642011,
+      "loss": 0.5498,
+      "step": 67560
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005156183086349086,
+      "loss": 0.5082,
+      "step": 67570
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.000515575907905616,
+      "loss": 0.6166,
+      "step": 67580
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005155335071763235,
+      "loss": 0.5829,
+      "step": 67590
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005154911064470309,
+      "loss": 0.5501,
+      "step": 67600
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005154487057177383,
+      "loss": 0.5623,
+      "step": 67610
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005154063049884459,
+      "loss": 0.4963,
+      "step": 67620
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005153639042591533,
+      "loss": 0.4252,
+      "step": 67630
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005153215035298607,
+      "loss": 0.4916,
+      "step": 67640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005152791028005682,
+      "loss": 0.6202,
+      "step": 67650
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005152367020712757,
+      "loss": 0.4733,
+      "step": 67660
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005151943013419831,
+      "loss": 0.5469,
+      "step": 67670
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005151519006126906,
+      "loss": 0.5301,
+      "step": 67680
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005151094998833979,
+      "loss": 0.5802,
+      "step": 67690
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005150670991541055,
+      "loss": 0.5928,
+      "step": 67700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.000515024698424813,
+      "loss": 0.5171,
+      "step": 67710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005149822976955203,
+      "loss": 0.5104,
+      "step": 67720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005149398969662278,
+      "loss": 0.4601,
+      "step": 67730
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005148974962369353,
+      "loss": 0.5395,
+      "step": 67740
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005148550955076427,
+      "loss": 0.5057,
+      "step": 67750
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005148126947783502,
+      "loss": 0.5907,
+      "step": 67760
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005147702940490577,
+      "loss": 0.5525,
+      "step": 67770
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005147278933197651,
+      "loss": 0.5272,
+      "step": 67780
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005146854925904726,
+      "loss": 0.6282,
+      "step": 67790
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005146430918611801,
+      "loss": 0.5243,
+      "step": 67800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005146006911318874,
+      "loss": 0.5145,
+      "step": 67810
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005145582904025949,
+      "loss": 0.5837,
+      "step": 67820
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005145158896733025,
+      "loss": 0.5913,
+      "step": 67830
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005144734889440098,
+      "loss": 0.5579,
+      "step": 67840
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005144310882147173,
+      "loss": 0.5843,
+      "step": 67850
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005143886874854248,
+      "loss": 0.6117,
+      "step": 67860
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005143462867561322,
+      "loss": 0.5865,
+      "step": 67870
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005143038860268397,
+      "loss": 0.6213,
+      "step": 67880
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005142614852975472,
+      "loss": 0.5766,
+      "step": 67890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005142190845682545,
+      "loss": 0.5452,
+      "step": 67900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005141766838389621,
+      "loss": 0.4795,
+      "step": 67910
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005141342831096695,
+      "loss": 0.5963,
+      "step": 67920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005140918823803769,
+      "loss": 0.6336,
+      "step": 67930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005140494816510844,
+      "loss": 0.5375,
+      "step": 67940
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005140070809217919,
+      "loss": 0.4748,
+      "step": 67950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005139646801924993,
+      "loss": 0.4954,
+      "step": 67960
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005139222794632068,
+      "loss": 0.5391,
+      "step": 67970
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005138798787339143,
+      "loss": 0.5234,
+      "step": 67980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005138374780046217,
+      "loss": 0.5253,
+      "step": 67990
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005137950772753292,
+      "loss": 0.5545,
+      "step": 68000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005137526765460366,
+      "loss": 0.5537,
+      "step": 68010
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.000513710275816744,
+      "loss": 0.5281,
+      "step": 68020
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005136678750874516,
+      "loss": 0.5674,
+      "step": 68030
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.000513625474358159,
+      "loss": 0.4665,
+      "step": 68040
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005135830736288664,
+      "loss": 0.5962,
+      "step": 68050
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005135406728995739,
+      "loss": 0.539,
+      "step": 68060
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005134982721702814,
+      "loss": 0.4893,
+      "step": 68070
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005134558714409888,
+      "loss": 0.6401,
+      "step": 68080
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005134134707116963,
+      "loss": 0.6025,
+      "step": 68090
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005133710699824036,
+      "loss": 0.5638,
+      "step": 68100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005133286692531112,
+      "loss": 0.6303,
+      "step": 68110
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005132862685238187,
+      "loss": 0.4704,
+      "step": 68120
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.000513243867794526,
+      "loss": 0.5518,
+      "step": 68130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005132014670652335,
+      "loss": 0.5557,
+      "step": 68140
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005131590663359411,
+      "loss": 0.5744,
+      "step": 68150
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005131166656066484,
+      "loss": 0.5317,
+      "step": 68160
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005130742648773559,
+      "loss": 0.5243,
+      "step": 68170
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005130318641480634,
+      "loss": 0.6434,
+      "step": 68180
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005129894634187709,
+      "loss": 0.6225,
+      "step": 68190
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005129470626894783,
+      "loss": 0.4556,
+      "step": 68200
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005129046619601858,
+      "loss": 0.553,
+      "step": 68210
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005128622612308931,
+      "loss": 0.5699,
+      "step": 68220
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005128198605016007,
+      "loss": 0.5654,
+      "step": 68230
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005127774597723081,
+      "loss": 0.57,
+      "step": 68240
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005127350590430155,
+      "loss": 0.5919,
+      "step": 68250
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000512692658313723,
+      "loss": 0.5484,
+      "step": 68260
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005126502575844305,
+      "loss": 0.5569,
+      "step": 68270
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005126078568551379,
+      "loss": 0.5497,
+      "step": 68280
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005125654561258454,
+      "loss": 0.5142,
+      "step": 68290
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005125230553965528,
+      "loss": 0.535,
+      "step": 68300
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005124806546672603,
+      "loss": 0.6414,
+      "step": 68310
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005124382539379678,
+      "loss": 0.5935,
+      "step": 68320
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005123958532086752,
+      "loss": 0.5595,
+      "step": 68330
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005123534524793826,
+      "loss": 0.5338,
+      "step": 68340
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005123110517500901,
+      "loss": 0.6319,
+      "step": 68350
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005122686510207976,
+      "loss": 0.592,
+      "step": 68360
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000512226250291505,
+      "loss": 0.5381,
+      "step": 68370
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005121838495622125,
+      "loss": 0.5511,
+      "step": 68380
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00051214144883292,
+      "loss": 0.6523,
+      "step": 68390
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005120990481036274,
+      "loss": 0.5386,
+      "step": 68400
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005120566473743349,
+      "loss": 0.5265,
+      "step": 68410
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005120142466450423,
+      "loss": 0.5652,
+      "step": 68420
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005119718459157497,
+      "loss": 0.5222,
+      "step": 68430
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005119294451864573,
+      "loss": 0.5012,
+      "step": 68440
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005118870444571647,
+      "loss": 0.6505,
+      "step": 68450
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005118446437278721,
+      "loss": 0.6113,
+      "step": 68460
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005118022429985796,
+      "loss": 0.5286,
+      "step": 68470
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005117598422692871,
+      "loss": 0.5892,
+      "step": 68480
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005117174415399945,
+      "loss": 0.5588,
+      "step": 68490
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.000511675040810702,
+      "loss": 0.6046,
+      "step": 68500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005116326400814095,
+      "loss": 0.5091,
+      "step": 68510
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005115902393521169,
+      "loss": 0.5746,
+      "step": 68520
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005115478386228243,
+      "loss": 0.6003,
+      "step": 68530
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005115054378935318,
+      "loss": 0.5846,
+      "step": 68540
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005114630371642392,
+      "loss": 0.4922,
+      "step": 68550
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005114206364349467,
+      "loss": 0.6708,
+      "step": 68560
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005113782357056542,
+      "loss": 0.5909,
+      "step": 68570
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005113358349763616,
+      "loss": 0.5713,
+      "step": 68580
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005112934342470691,
+      "loss": 0.5949,
+      "step": 68590
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005112510335177766,
+      "loss": 0.6228,
+      "step": 68600
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.000511208632788484,
+      "loss": 0.4823,
+      "step": 68610
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005111662320591914,
+      "loss": 0.5546,
+      "step": 68620
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005111238313298988,
+      "loss": 0.5948,
+      "step": 68630
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005110814306006064,
+      "loss": 0.6143,
+      "step": 68640
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005110390298713138,
+      "loss": 0.4605,
+      "step": 68650
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005109966291420212,
+      "loss": 0.5536,
+      "step": 68660
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005109542284127287,
+      "loss": 0.6364,
+      "step": 68670
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005109118276834362,
+      "loss": 0.4947,
+      "step": 68680
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005108694269541436,
+      "loss": 0.6201,
+      "step": 68690
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005108270262248511,
+      "loss": 0.4569,
+      "step": 68700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005107846254955585,
+      "loss": 0.5128,
+      "step": 68710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.000510742224766266,
+      "loss": 0.4574,
+      "step": 68720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005106998240369735,
+      "loss": 0.5286,
+      "step": 68730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005106574233076809,
+      "loss": 0.6162,
+      "step": 68740
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005106150225783883,
+      "loss": 0.5695,
+      "step": 68750
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005105726218490959,
+      "loss": 0.5295,
+      "step": 68760
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005105302211198033,
+      "loss": 0.5546,
+      "step": 68770
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005104878203905107,
+      "loss": 0.5193,
+      "step": 68780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005104454196612182,
+      "loss": 0.5785,
+      "step": 68790
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005104030189319257,
+      "loss": 0.5567,
+      "step": 68800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005103606182026331,
+      "loss": 0.5293,
+      "step": 68810
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005103182174733406,
+      "loss": 0.6194,
+      "step": 68820
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.000510275816744048,
+      "loss": 0.5798,
+      "step": 68830
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005102334160147555,
+      "loss": 0.4691,
+      "step": 68840
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005101910152854629,
+      "loss": 0.5981,
+      "step": 68850
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005101486145561704,
+      "loss": 0.5088,
+      "step": 68860
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005101062138268778,
+      "loss": 0.4538,
+      "step": 68870
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005100638130975853,
+      "loss": 0.5106,
+      "step": 68880
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005100214123682928,
+      "loss": 0.616,
+      "step": 68890
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005099790116390002,
+      "loss": 0.6236,
+      "step": 68900
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005099366109097076,
+      "loss": 0.5243,
+      "step": 68910
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005098942101804152,
+      "loss": 0.4851,
+      "step": 68920
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005098518094511226,
+      "loss": 0.6389,
+      "step": 68930
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00050980940872183,
+      "loss": 0.6198,
+      "step": 68940
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005097670079925375,
+      "loss": 0.6395,
+      "step": 68950
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.000509724607263245,
+      "loss": 0.5233,
+      "step": 68960
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005096822065339524,
+      "loss": 0.5446,
+      "step": 68970
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005096398058046599,
+      "loss": 0.608,
+      "step": 68980
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005095974050753673,
+      "loss": 0.4683,
+      "step": 68990
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005095550043460748,
+      "loss": 0.5657,
+      "step": 69000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005095126036167823,
+      "loss": 0.5576,
+      "step": 69010
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005094702028874897,
+      "loss": 0.6137,
+      "step": 69020
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005094278021581971,
+      "loss": 0.5629,
+      "step": 69030
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005093854014289047,
+      "loss": 0.5643,
+      "step": 69040
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005093430006996121,
+      "loss": 0.6008,
+      "step": 69050
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005093005999703195,
+      "loss": 0.565,
+      "step": 69060
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.000509258199241027,
+      "loss": 0.5376,
+      "step": 69070
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005092157985117344,
+      "loss": 0.6281,
+      "step": 69080
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005091733977824419,
+      "loss": 0.5168,
+      "step": 69090
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005091309970531494,
+      "loss": 0.5619,
+      "step": 69100
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005090885963238568,
+      "loss": 0.6353,
+      "step": 69110
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005090461955945643,
+      "loss": 0.5902,
+      "step": 69120
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005090037948652718,
+      "loss": 0.6062,
+      "step": 69130
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005089613941359791,
+      "loss": 0.5282,
+      "step": 69140
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005089189934066866,
+      "loss": 0.5772,
+      "step": 69150
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.000508876592677394,
+      "loss": 0.569,
+      "step": 69160
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005088341919481015,
+      "loss": 0.4475,
+      "step": 69170
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000508791791218809,
+      "loss": 0.6095,
+      "step": 69180
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005087493904895164,
+      "loss": 0.5579,
+      "step": 69190
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005087069897602239,
+      "loss": 0.6069,
+      "step": 69200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005086645890309314,
+      "loss": 0.5828,
+      "step": 69210
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005086221883016388,
+      "loss": 0.5043,
+      "step": 69220
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005085797875723462,
+      "loss": 0.4889,
+      "step": 69230
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005085373868430537,
+      "loss": 0.5201,
+      "step": 69240
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005084949861137612,
+      "loss": 0.476,
+      "step": 69250
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005084525853844686,
+      "loss": 0.7554,
+      "step": 69260
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005084101846551761,
+      "loss": 0.5318,
+      "step": 69270
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005083677839258835,
+      "loss": 0.5711,
+      "step": 69280
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000508325383196591,
+      "loss": 0.6282,
+      "step": 69290
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005082829824672985,
+      "loss": 0.5627,
+      "step": 69300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005082405817380059,
+      "loss": 0.5577,
+      "step": 69310
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005081981810087133,
+      "loss": 0.564,
+      "step": 69320
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005081557802794209,
+      "loss": 0.5482,
+      "step": 69330
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005081133795501283,
+      "loss": 0.5424,
+      "step": 69340
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005080709788208357,
+      "loss": 0.6234,
+      "step": 69350
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005080285780915432,
+      "loss": 0.5924,
+      "step": 69360
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005079861773622507,
+      "loss": 0.5763,
+      "step": 69370
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005079437766329581,
+      "loss": 0.5542,
+      "step": 69380
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005079013759036656,
+      "loss": 0.5198,
+      "step": 69390
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000507858975174373,
+      "loss": 0.5589,
+      "step": 69400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005078165744450805,
+      "loss": 0.521,
+      "step": 69410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.000507774173715788,
+      "loss": 0.5909,
+      "step": 69420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005077317729864954,
+      "loss": 0.6408,
+      "step": 69430
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005076893722572028,
+      "loss": 0.543,
+      "step": 69440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005076469715279104,
+      "loss": 0.5368,
+      "step": 69450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005076045707986177,
+      "loss": 0.5345,
+      "step": 69460
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005075621700693252,
+      "loss": 0.547,
+      "step": 69470
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005075197693400327,
+      "loss": 0.6726,
+      "step": 69480
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005074773686107401,
+      "loss": 0.5795,
+      "step": 69490
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005074349678814476,
+      "loss": 0.5378,
+      "step": 69500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005073925671521551,
+      "loss": 0.5856,
+      "step": 69510
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005073501664228624,
+      "loss": 0.5097,
+      "step": 69520
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00050730776569357,
+      "loss": 0.5611,
+      "step": 69530
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005072653649642775,
+      "loss": 0.5751,
+      "step": 69540
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005072229642349848,
+      "loss": 0.6031,
+      "step": 69550
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005071805635056923,
+      "loss": 0.4923,
+      "step": 69560
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005071381627763999,
+      "loss": 0.5571,
+      "step": 69570
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005070957620471072,
+      "loss": 0.5097,
+      "step": 69580
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005070533613178147,
+      "loss": 0.6107,
+      "step": 69590
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005070109605885221,
+      "loss": 0.556,
+      "step": 69600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005069685598592296,
+      "loss": 0.5025,
+      "step": 69610
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005069261591299371,
+      "loss": 0.5669,
+      "step": 69620
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005068837584006445,
+      "loss": 0.645,
+      "step": 69630
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005068413576713519,
+      "loss": 0.5799,
+      "step": 69640
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005067989569420595,
+      "loss": 0.5412,
+      "step": 69650
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.000506756556212767,
+      "loss": 0.5652,
+      "step": 69660
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005067141554834743,
+      "loss": 0.586,
+      "step": 69670
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005066717547541818,
+      "loss": 0.6181,
+      "step": 69680
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005066293540248892,
+      "loss": 0.5531,
+      "step": 69690
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005065869532955967,
+      "loss": 0.5551,
+      "step": 69700
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005065445525663042,
+      "loss": 0.586,
+      "step": 69710
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005065021518370116,
+      "loss": 0.5598,
+      "step": 69720
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005064597511077191,
+      "loss": 0.575,
+      "step": 69730
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005064173503784266,
+      "loss": 0.4883,
+      "step": 69740
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005063749496491339,
+      "loss": 0.593,
+      "step": 69750
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005063325489198414,
+      "loss": 0.527,
+      "step": 69760
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005062901481905489,
+      "loss": 0.5661,
+      "step": 69770
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005062477474612563,
+      "loss": 0.522,
+      "step": 69780
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005062053467319638,
+      "loss": 0.5918,
+      "step": 69790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005061629460026713,
+      "loss": 0.5937,
+      "step": 69800
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005061205452733787,
+      "loss": 0.4222,
+      "step": 69810
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005060781445440862,
+      "loss": 0.5006,
+      "step": 69820
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005060357438147937,
+      "loss": 0.5573,
+      "step": 69830
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.000505993343085501,
+      "loss": 0.5724,
+      "step": 69840
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005059509423562085,
+      "loss": 0.4367,
+      "step": 69850
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005059085416269161,
+      "loss": 0.5215,
+      "step": 69860
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005058661408976234,
+      "loss": 0.4401,
+      "step": 69870
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005058237401683309,
+      "loss": 0.4991,
+      "step": 69880
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005057813394390384,
+      "loss": 0.5107,
+      "step": 69890
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005057389387097458,
+      "loss": 0.5603,
+      "step": 69900
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005056965379804533,
+      "loss": 0.5796,
+      "step": 69910
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005056541372511608,
+      "loss": 0.4495,
+      "step": 69920
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005056117365218681,
+      "loss": 0.6021,
+      "step": 69930
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005055693357925757,
+      "loss": 0.5264,
+      "step": 69940
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005055269350632832,
+      "loss": 0.6888,
+      "step": 69950
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005054845343339905,
+      "loss": 0.5547,
+      "step": 69960
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.000505442133604698,
+      "loss": 0.5121,
+      "step": 69970
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005053997328754056,
+      "loss": 0.582,
+      "step": 69980
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005053573321461129,
+      "loss": 0.5956,
+      "step": 69990
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005053149314168204,
+      "loss": 0.5408,
+      "step": 70000
+    },
+    {
+      "epoch": 2.96,
+      "eval_loss": 0.6053586602210999,
+      "eval_runtime": 337.729,
+      "eval_samples_per_second": 15.56,
+      "eval_steps_per_second": 3.891,
+      "step": 70000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005052725306875279,
+      "loss": 0.5125,
+      "step": 70010
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005052301299582353,
+      "loss": 0.5051,
+      "step": 70020
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005051877292289428,
+      "loss": 0.5172,
+      "step": 70030
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005051453284996503,
+      "loss": 0.5293,
+      "step": 70040
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005051029277703576,
+      "loss": 0.6299,
+      "step": 70050
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005050605270410652,
+      "loss": 0.5143,
+      "step": 70060
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005050181263117725,
+      "loss": 0.5827,
+      "step": 70070
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00050497572558248,
+      "loss": 0.5883,
+      "step": 70080
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005049333248531875,
+      "loss": 0.5293,
+      "step": 70090
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.000504890924123895,
+      "loss": 0.5464,
+      "step": 70100
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005048485233946024,
+      "loss": 0.6002,
+      "step": 70110
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005048061226653099,
+      "loss": 0.7026,
+      "step": 70120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005047637219360172,
+      "loss": 0.4571,
+      "step": 70130
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005047213212067248,
+      "loss": 0.6722,
+      "step": 70140
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005046789204774323,
+      "loss": 0.6154,
+      "step": 70150
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005046365197481396,
+      "loss": 0.5586,
+      "step": 70160
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005045941190188471,
+      "loss": 0.5023,
+      "step": 70170
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005045517182895547,
+      "loss": 0.5799,
+      "step": 70180
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.000504509317560262,
+      "loss": 0.5724,
+      "step": 70190
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005044669168309695,
+      "loss": 0.514,
+      "step": 70200
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.000504424516101677,
+      "loss": 0.5404,
+      "step": 70210
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005043821153723844,
+      "loss": 0.5795,
+      "step": 70220
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005043397146430919,
+      "loss": 0.6498,
+      "step": 70230
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005042973139137994,
+      "loss": 0.5169,
+      "step": 70240
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005042549131845067,
+      "loss": 0.6216,
+      "step": 70250
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005042125124552143,
+      "loss": 0.5882,
+      "step": 70260
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005041701117259218,
+      "loss": 0.5458,
+      "step": 70270
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005041277109966291,
+      "loss": 0.5344,
+      "step": 70280
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005040853102673366,
+      "loss": 0.5116,
+      "step": 70290
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005040429095380441,
+      "loss": 0.6302,
+      "step": 70300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005040005088087515,
+      "loss": 0.5824,
+      "step": 70310
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.000503958108079459,
+      "loss": 0.6912,
+      "step": 70320
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005039157073501665,
+      "loss": 0.4882,
+      "step": 70330
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005038733066208739,
+      "loss": 0.4924,
+      "step": 70340
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005038309058915814,
+      "loss": 0.4844,
+      "step": 70350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005037885051622889,
+      "loss": 0.5262,
+      "step": 70360
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005037461044329962,
+      "loss": 0.6292,
+      "step": 70370
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005037037037037037,
+      "loss": 0.5322,
+      "step": 70380
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005036613029744112,
+      "loss": 0.5653,
+      "step": 70390
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005036189022451186,
+      "loss": 0.6209,
+      "step": 70400
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005035765015158261,
+      "loss": 0.5085,
+      "step": 70410
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005035341007865336,
+      "loss": 0.6068,
+      "step": 70420
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.000503491700057241,
+      "loss": 0.5698,
+      "step": 70430
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005034492993279485,
+      "loss": 0.502,
+      "step": 70440
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005034068985986559,
+      "loss": 0.5874,
+      "step": 70450
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005033644978693633,
+      "loss": 0.5782,
+      "step": 70460
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005033220971400709,
+      "loss": 0.5394,
+      "step": 70470
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005032796964107783,
+      "loss": 0.5976,
+      "step": 70480
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005032372956814857,
+      "loss": 0.5624,
+      "step": 70490
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005031948949521932,
+      "loss": 0.4806,
+      "step": 70500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005031524942229007,
+      "loss": 0.5768,
+      "step": 70510
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005031100934936081,
+      "loss": 0.6531,
+      "step": 70520
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005030676927643156,
+      "loss": 0.5966,
+      "step": 70530
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.000503025292035023,
+      "loss": 0.5612,
+      "step": 70540
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005029828913057305,
+      "loss": 0.5579,
+      "step": 70550
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.000502940490576438,
+      "loss": 0.5137,
+      "step": 70560
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005028980898471453,
+      "loss": 0.633,
+      "step": 70570
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005028556891178528,
+      "loss": 0.6308,
+      "step": 70580
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005028132883885604,
+      "loss": 0.6313,
+      "step": 70590
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005027708876592677,
+      "loss": 0.5713,
+      "step": 70600
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005027284869299752,
+      "loss": 0.6351,
+      "step": 70610
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005026860862006827,
+      "loss": 0.5522,
+      "step": 70620
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005026436854713901,
+      "loss": 0.4856,
+      "step": 70630
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005026012847420976,
+      "loss": 0.5759,
+      "step": 70640
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005025588840128051,
+      "loss": 0.5238,
+      "step": 70650
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005025164832835124,
+      "loss": 0.5602,
+      "step": 70660
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00050247408255422,
+      "loss": 0.5609,
+      "step": 70670
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005024316818249274,
+      "loss": 0.591,
+      "step": 70680
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005023892810956348,
+      "loss": 0.5414,
+      "step": 70690
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005023468803663423,
+      "loss": 0.6054,
+      "step": 70700
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005023044796370498,
+      "loss": 0.6063,
+      "step": 70710
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005022620789077572,
+      "loss": 0.5206,
+      "step": 70720
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005022196781784647,
+      "loss": 0.6796,
+      "step": 70730
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005021772774491721,
+      "loss": 0.5158,
+      "step": 70740
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005021348767198796,
+      "loss": 0.5476,
+      "step": 70750
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005020924759905871,
+      "loss": 0.5877,
+      "step": 70760
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005020500752612945,
+      "loss": 0.4583,
+      "step": 70770
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005020076745320019,
+      "loss": 0.5495,
+      "step": 70780
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005019652738027095,
+      "loss": 0.5862,
+      "step": 70790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005019228730734169,
+      "loss": 0.4068,
+      "step": 70800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005018804723441243,
+      "loss": 0.5669,
+      "step": 70810
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005018380716148318,
+      "loss": 0.6162,
+      "step": 70820
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005017956708855393,
+      "loss": 0.5455,
+      "step": 70830
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005017532701562467,
+      "loss": 0.5478,
+      "step": 70840
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005017108694269542,
+      "loss": 0.448,
+      "step": 70850
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005016684686976616,
+      "loss": 0.5944,
+      "step": 70860
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005016260679683691,
+      "loss": 0.5631,
+      "step": 70870
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005015836672390766,
+      "loss": 0.5651,
+      "step": 70880
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000501541266509784,
+      "loss": 0.4939,
+      "step": 70890
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005014988657804914,
+      "loss": 0.6093,
+      "step": 70900
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005014564650511989,
+      "loss": 0.4765,
+      "step": 70910
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005014140643219064,
+      "loss": 0.596,
+      "step": 70920
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005013716635926138,
+      "loss": 0.5989,
+      "step": 70930
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005013292628633213,
+      "loss": 0.5595,
+      "step": 70940
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005012868621340288,
+      "loss": 0.6006,
+      "step": 70950
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005012444614047362,
+      "loss": 0.5156,
+      "step": 70960
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005012020606754437,
+      "loss": 0.4382,
+      "step": 70970
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000501159659946151,
+      "loss": 0.5785,
+      "step": 70980
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005011172592168586,
+      "loss": 0.5198,
+      "step": 70990
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000501074858487566,
+      "loss": 0.4833,
+      "step": 71000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005010324577582734,
+      "loss": 0.5431,
+      "step": 71010
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005009900570289809,
+      "loss": 0.4276,
+      "step": 71020
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005009476562996884,
+      "loss": 0.519,
+      "step": 71030
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005009052555703958,
+      "loss": 0.4612,
+      "step": 71040
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005008628548411033,
+      "loss": 0.4552,
+      "step": 71050
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005008204541118107,
+      "loss": 0.5332,
+      "step": 71060
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005007780533825182,
+      "loss": 0.4765,
+      "step": 71070
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005007356526532257,
+      "loss": 0.5225,
+      "step": 71080
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005006932519239331,
+      "loss": 0.5441,
+      "step": 71090
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005006508511946405,
+      "loss": 0.4793,
+      "step": 71100
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.000500608450465348,
+      "loss": 0.4695,
+      "step": 71110
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005005660497360555,
+      "loss": 0.5591,
+      "step": 71120
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005005236490067629,
+      "loss": 0.4518,
+      "step": 71130
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005004812482774704,
+      "loss": 0.4745,
+      "step": 71140
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005004388475481779,
+      "loss": 0.5148,
+      "step": 71150
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005003964468188853,
+      "loss": 0.4996,
+      "step": 71160
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005003540460895928,
+      "loss": 0.4106,
+      "step": 71170
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005003116453603002,
+      "loss": 0.5095,
+      "step": 71180
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005002692446310076,
+      "loss": 0.4849,
+      "step": 71190
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005002268439017152,
+      "loss": 0.4828,
+      "step": 71200
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005001844431724226,
+      "loss": 0.518,
+      "step": 71210
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00050014204244313,
+      "loss": 0.4613,
+      "step": 71220
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005000996417138375,
+      "loss": 0.5248,
+      "step": 71230
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.000500057240984545,
+      "loss": 0.4476,
+      "step": 71240
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005000148402552524,
+      "loss": 0.4928,
+      "step": 71250
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0004999724395259599,
+      "loss": 0.5271,
+      "step": 71260
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0004999300387966673,
+      "loss": 0.5018,
+      "step": 71270
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0004998876380673748,
+      "loss": 0.4535,
+      "step": 71280
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0004998452373380822,
+      "loss": 0.4703,
+      "step": 71290
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004998028366087897,
+      "loss": 0.5359,
+      "step": 71300
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004997604358794971,
+      "loss": 0.5612,
+      "step": 71310
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004997180351502046,
+      "loss": 0.5504,
+      "step": 71320
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004996756344209121,
+      "loss": 0.533,
+      "step": 71330
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004996332336916195,
+      "loss": 0.4805,
+      "step": 71340
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004995908329623269,
+      "loss": 0.4741,
+      "step": 71350
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004995484322330345,
+      "loss": 0.4634,
+      "step": 71360
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004995060315037419,
+      "loss": 0.4991,
+      "step": 71370
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004994636307744493,
+      "loss": 0.5552,
+      "step": 71380
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004994212300451568,
+      "loss": 0.5193,
+      "step": 71390
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004993788293158643,
+      "loss": 0.5232,
+      "step": 71400
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004993364285865717,
+      "loss": 0.5304,
+      "step": 71410
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004992940278572792,
+      "loss": 0.6105,
+      "step": 71420
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004992516271279866,
+      "loss": 0.5924,
+      "step": 71430
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004992092263986941,
+      "loss": 0.5286,
+      "step": 71440
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004991668256694016,
+      "loss": 0.5514,
+      "step": 71450
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000499124424940109,
+      "loss": 0.4388,
+      "step": 71460
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004990820242108164,
+      "loss": 0.4027,
+      "step": 71470
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000499039623481524,
+      "loss": 0.5186,
+      "step": 71480
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004989972227522314,
+      "loss": 0.487,
+      "step": 71490
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004989548220229388,
+      "loss": 0.4596,
+      "step": 71500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004989124212936462,
+      "loss": 0.4839,
+      "step": 71510
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004988700205643538,
+      "loss": 0.558,
+      "step": 71520
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004988276198350612,
+      "loss": 0.5125,
+      "step": 71530
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004987852191057686,
+      "loss": 0.515,
+      "step": 71540
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004987428183764761,
+      "loss": 0.5388,
+      "step": 71550
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004987004176471836,
+      "loss": 0.5043,
+      "step": 71560
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000498658016917891,
+      "loss": 0.5008,
+      "step": 71570
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004986156161885985,
+      "loss": 0.507,
+      "step": 71580
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004985732154593059,
+      "loss": 0.5735,
+      "step": 71590
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004985308147300134,
+      "loss": 0.4633,
+      "step": 71600
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004984884140007208,
+      "loss": 0.5328,
+      "step": 71610
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004984460132714283,
+      "loss": 0.5205,
+      "step": 71620
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004984036125421357,
+      "loss": 0.5621,
+      "step": 71630
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004983612118128432,
+      "loss": 0.5216,
+      "step": 71640
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004983188110835507,
+      "loss": 0.5304,
+      "step": 71650
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004982764103542581,
+      "loss": 0.4998,
+      "step": 71660
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004982340096249655,
+      "loss": 0.5101,
+      "step": 71670
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004981916088956731,
+      "loss": 0.5165,
+      "step": 71680
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004981492081663805,
+      "loss": 0.592,
+      "step": 71690
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004981068074370879,
+      "loss": 0.5027,
+      "step": 71700
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004980644067077954,
+      "loss": 0.4887,
+      "step": 71710
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004980220059785028,
+      "loss": 0.5141,
+      "step": 71720
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004979796052492103,
+      "loss": 0.4989,
+      "step": 71730
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004979372045199178,
+      "loss": 0.5166,
+      "step": 71740
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004978948037906252,
+      "loss": 0.4397,
+      "step": 71750
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004978524030613327,
+      "loss": 0.5845,
+      "step": 71760
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004978100023320402,
+      "loss": 0.5411,
+      "step": 71770
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004977676016027476,
+      "loss": 0.4676,
+      "step": 71780
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.000497725200873455,
+      "loss": 0.4935,
+      "step": 71790
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004976828001441625,
+      "loss": 0.5317,
+      "step": 71800
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00049764039941487,
+      "loss": 0.5395,
+      "step": 71810
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004975979986855774,
+      "loss": 0.4697,
+      "step": 71820
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004975555979562849,
+      "loss": 0.4977,
+      "step": 71830
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004975131972269923,
+      "loss": 0.588,
+      "step": 71840
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004974707964976998,
+      "loss": 0.6394,
+      "step": 71850
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004974283957684073,
+      "loss": 0.5246,
+      "step": 71860
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004973859950391147,
+      "loss": 0.5227,
+      "step": 71870
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004973435943098221,
+      "loss": 0.5361,
+      "step": 71880
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004973011935805297,
+      "loss": 0.4055,
+      "step": 71890
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.000497258792851237,
+      "loss": 0.5008,
+      "step": 71900
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004972163921219445,
+      "loss": 0.535,
+      "step": 71910
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.000497173991392652,
+      "loss": 0.5468,
+      "step": 71920
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004971315906633594,
+      "loss": 0.5504,
+      "step": 71930
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004970891899340669,
+      "loss": 0.5302,
+      "step": 71940
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004970467892047744,
+      "loss": 0.5478,
+      "step": 71950
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004970043884754817,
+      "loss": 0.514,
+      "step": 71960
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004969619877461893,
+      "loss": 0.4169,
+      "step": 71970
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004969195870168968,
+      "loss": 0.5202,
+      "step": 71980
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004968771862876041,
+      "loss": 0.5228,
+      "step": 71990
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004968347855583116,
+      "loss": 0.5436,
+      "step": 72000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004967923848290192,
+      "loss": 0.54,
+      "step": 72010
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004967499840997265,
+      "loss": 0.461,
+      "step": 72020
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.000496707583370434,
+      "loss": 0.4488,
+      "step": 72030
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004966651826411414,
+      "loss": 0.5339,
+      "step": 72040
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004966227819118489,
+      "loss": 0.4892,
+      "step": 72050
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004965803811825564,
+      "loss": 0.5274,
+      "step": 72060
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004965379804532638,
+      "loss": 0.5486,
+      "step": 72070
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004964955797239712,
+      "loss": 0.5255,
+      "step": 72080
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004964531789946788,
+      "loss": 0.6218,
+      "step": 72090
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004964107782653862,
+      "loss": 0.5992,
+      "step": 72100
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004963683775360936,
+      "loss": 0.5378,
+      "step": 72110
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004963259768068011,
+      "loss": 0.611,
+      "step": 72120
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004962835760775086,
+      "loss": 0.4515,
+      "step": 72130
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.000496241175348216,
+      "loss": 0.5273,
+      "step": 72140
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004961987746189235,
+      "loss": 0.4744,
+      "step": 72150
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004961563738896309,
+      "loss": 0.5466,
+      "step": 72160
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004961139731603384,
+      "loss": 0.5004,
+      "step": 72170
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004960715724310459,
+      "loss": 0.5751,
+      "step": 72180
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004960291717017533,
+      "loss": 0.4606,
+      "step": 72190
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004959867709724607,
+      "loss": 0.4958,
+      "step": 72200
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004959443702431683,
+      "loss": 0.4351,
+      "step": 72210
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004959019695138756,
+      "loss": 0.5332,
+      "step": 72220
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004958595687845831,
+      "loss": 0.5246,
+      "step": 72230
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004958171680552906,
+      "loss": 0.5564,
+      "step": 72240
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000495774767325998,
+      "loss": 0.4552,
+      "step": 72250
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004957323665967055,
+      "loss": 0.5377,
+      "step": 72260
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000495689965867413,
+      "loss": 0.4634,
+      "step": 72270
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004956475651381203,
+      "loss": 0.5248,
+      "step": 72280
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004956051644088279,
+      "loss": 0.5232,
+      "step": 72290
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004955627636795354,
+      "loss": 0.5317,
+      "step": 72300
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004955203629502427,
+      "loss": 0.5773,
+      "step": 72310
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004954779622209502,
+      "loss": 0.494,
+      "step": 72320
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004954355614916577,
+      "loss": 0.5459,
+      "step": 72330
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004953931607623651,
+      "loss": 0.5463,
+      "step": 72340
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004953507600330726,
+      "loss": 0.4873,
+      "step": 72350
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004953083593037801,
+      "loss": 0.4884,
+      "step": 72360
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004952659585744875,
+      "loss": 0.536,
+      "step": 72370
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000495223557845195,
+      "loss": 0.5463,
+      "step": 72380
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004951811571159025,
+      "loss": 0.4813,
+      "step": 72390
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004951387563866098,
+      "loss": 0.6031,
+      "step": 72400
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004950963556573173,
+      "loss": 0.4803,
+      "step": 72410
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004950539549280249,
+      "loss": 0.506,
+      "step": 72420
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004950115541987322,
+      "loss": 0.4757,
+      "step": 72430
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004949691534694397,
+      "loss": 0.4803,
+      "step": 72440
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004949267527401471,
+      "loss": 0.5146,
+      "step": 72450
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004948843520108546,
+      "loss": 0.5381,
+      "step": 72460
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004948419512815621,
+      "loss": 0.5242,
+      "step": 72470
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004947995505522695,
+      "loss": 0.5278,
+      "step": 72480
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.000494757149822977,
+      "loss": 0.5766,
+      "step": 72490
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004947147490936845,
+      "loss": 0.5403,
+      "step": 72500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.000494672348364392,
+      "loss": 0.4668,
+      "step": 72510
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004946299476350993,
+      "loss": 0.541,
+      "step": 72520
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004945875469058068,
+      "loss": 0.5967,
+      "step": 72530
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004945451461765142,
+      "loss": 0.5478,
+      "step": 72540
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004945027454472217,
+      "loss": 0.5164,
+      "step": 72550
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004944603447179292,
+      "loss": 0.5291,
+      "step": 72560
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004944179439886366,
+      "loss": 0.5039,
+      "step": 72570
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004943755432593441,
+      "loss": 0.4855,
+      "step": 72580
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004943331425300516,
+      "loss": 0.4835,
+      "step": 72590
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004942907418007589,
+      "loss": 0.6097,
+      "step": 72600
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004942483410714664,
+      "loss": 0.5133,
+      "step": 72610
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.000494205940342174,
+      "loss": 0.5457,
+      "step": 72620
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004941635396128813,
+      "loss": 0.5608,
+      "step": 72630
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004941211388835888,
+      "loss": 0.5323,
+      "step": 72640
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004940787381542963,
+      "loss": 0.5183,
+      "step": 72650
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004940363374250037,
+      "loss": 0.5324,
+      "step": 72660
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004939939366957112,
+      "loss": 0.5807,
+      "step": 72670
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004939515359664187,
+      "loss": 0.4917,
+      "step": 72680
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.000493909135237126,
+      "loss": 0.5183,
+      "step": 72690
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004938667345078336,
+      "loss": 0.5434,
+      "step": 72700
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004938243337785411,
+      "loss": 0.5461,
+      "step": 72710
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004937819330492484,
+      "loss": 0.5218,
+      "step": 72720
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004937395323199559,
+      "loss": 0.5265,
+      "step": 72730
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004936971315906635,
+      "loss": 0.5529,
+      "step": 72740
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004936547308613708,
+      "loss": 0.4382,
+      "step": 72750
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004936123301320783,
+      "loss": 0.4791,
+      "step": 72760
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004935699294027858,
+      "loss": 0.4422,
+      "step": 72770
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004935275286734932,
+      "loss": 0.5278,
+      "step": 72780
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004934851279442007,
+      "loss": 0.5135,
+      "step": 72790
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004934427272149082,
+      "loss": 0.5586,
+      "step": 72800
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004934003264856155,
+      "loss": 0.5076,
+      "step": 72810
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004933579257563231,
+      "loss": 0.4554,
+      "step": 72820
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004933155250270305,
+      "loss": 0.5352,
+      "step": 72830
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004932731242977379,
+      "loss": 0.5573,
+      "step": 72840
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004932307235684454,
+      "loss": 0.4671,
+      "step": 72850
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004931883228391529,
+      "loss": 0.4761,
+      "step": 72860
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004931459221098603,
+      "loss": 0.5503,
+      "step": 72870
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004931035213805678,
+      "loss": 0.4144,
+      "step": 72880
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004930611206512751,
+      "loss": 0.4416,
+      "step": 72890
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004930187199219827,
+      "loss": 0.494,
+      "step": 72900
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004929763191926902,
+      "loss": 0.4925,
+      "step": 72910
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004929339184633975,
+      "loss": 0.4544,
+      "step": 72920
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000492891517734105,
+      "loss": 0.5669,
+      "step": 72930
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004928491170048125,
+      "loss": 0.5084,
+      "step": 72940
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00049280671627552,
+      "loss": 0.5562,
+      "step": 72950
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004927643155462274,
+      "loss": 0.5731,
+      "step": 72960
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004927219148169349,
+      "loss": 0.5005,
+      "step": 72970
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004926795140876423,
+      "loss": 0.563,
+      "step": 72980
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004926371133583498,
+      "loss": 0.5228,
+      "step": 72990
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004925947126290573,
+      "loss": 0.4966,
+      "step": 73000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004925523118997646,
+      "loss": 0.6024,
+      "step": 73010
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004925099111704722,
+      "loss": 0.4786,
+      "step": 73020
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004924675104411797,
+      "loss": 0.4879,
+      "step": 73030
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000492425109711887,
+      "loss": 0.4488,
+      "step": 73040
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004923827089825945,
+      "loss": 0.4996,
+      "step": 73050
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000492340308253302,
+      "loss": 0.4363,
+      "step": 73060
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004922979075240094,
+      "loss": 0.5704,
+      "step": 73070
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004922555067947169,
+      "loss": 0.6253,
+      "step": 73080
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004922131060654244,
+      "loss": 0.5385,
+      "step": 73090
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004921707053361318,
+      "loss": 0.5292,
+      "step": 73100
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004921283046068393,
+      "loss": 0.5467,
+      "step": 73110
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004920859038775468,
+      "loss": 0.5959,
+      "step": 73120
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004920435031482541,
+      "loss": 0.4598,
+      "step": 73130
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004920011024189616,
+      "loss": 0.5015,
+      "step": 73140
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004919587016896691,
+      "loss": 0.5917,
+      "step": 73150
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004919163009603765,
+      "loss": 0.5123,
+      "step": 73160
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000491873900231084,
+      "loss": 0.5387,
+      "step": 73170
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004918314995017915,
+      "loss": 0.5092,
+      "step": 73180
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004917890987724989,
+      "loss": 0.5036,
+      "step": 73190
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004917466980432064,
+      "loss": 0.4785,
+      "step": 73200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004917042973139138,
+      "loss": 0.4546,
+      "step": 73210
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004916618965846212,
+      "loss": 0.5189,
+      "step": 73220
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004916194958553288,
+      "loss": 0.4696,
+      "step": 73230
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004915770951260362,
+      "loss": 0.4645,
+      "step": 73240
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004915346943967436,
+      "loss": 0.501,
+      "step": 73250
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004914922936674511,
+      "loss": 0.4955,
+      "step": 73260
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004914498929381586,
+      "loss": 0.4628,
+      "step": 73270
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000491407492208866,
+      "loss": 0.4787,
+      "step": 73280
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004913650914795735,
+      "loss": 0.4952,
+      "step": 73290
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004913226907502808,
+      "loss": 0.4324,
+      "step": 73300
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004912802900209884,
+      "loss": 0.5118,
+      "step": 73310
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004912378892916959,
+      "loss": 0.5121,
+      "step": 73320
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004911954885624033,
+      "loss": 0.5002,
+      "step": 73330
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004911530878331107,
+      "loss": 0.4463,
+      "step": 73340
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004911106871038183,
+      "loss": 0.4805,
+      "step": 73350
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004910682863745257,
+      "loss": 0.5899,
+      "step": 73360
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004910258856452331,
+      "loss": 0.5429,
+      "step": 73370
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004909834849159406,
+      "loss": 0.4607,
+      "step": 73380
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000490941084186648,
+      "loss": 0.5488,
+      "step": 73390
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004908986834573555,
+      "loss": 0.4573,
+      "step": 73400
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000490856282728063,
+      "loss": 0.4594,
+      "step": 73410
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004908138819987703,
+      "loss": 0.5379,
+      "step": 73420
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004907714812694779,
+      "loss": 0.5829,
+      "step": 73430
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004907290805401853,
+      "loss": 0.4431,
+      "step": 73440
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004906866798108927,
+      "loss": 0.4935,
+      "step": 73450
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004906442790816002,
+      "loss": 0.5333,
+      "step": 73460
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004906018783523077,
+      "loss": 0.4704,
+      "step": 73470
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004905594776230151,
+      "loss": 0.5163,
+      "step": 73480
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004905170768937226,
+      "loss": 0.4549,
+      "step": 73490
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00049047467616443,
+      "loss": 0.4823,
+      "step": 73500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004904322754351375,
+      "loss": 0.5041,
+      "step": 73510
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000490389874705845,
+      "loss": 0.5117,
+      "step": 73520
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004903474739765524,
+      "loss": 0.4273,
+      "step": 73530
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004903050732472598,
+      "loss": 0.5989,
+      "step": 73540
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004902626725179674,
+      "loss": 0.4742,
+      "step": 73550
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004902202717886748,
+      "loss": 0.5185,
+      "step": 73560
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004901778710593822,
+      "loss": 0.5201,
+      "step": 73570
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004901354703300897,
+      "loss": 0.5478,
+      "step": 73580
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004900930696007972,
+      "loss": 0.4528,
+      "step": 73590
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004900506688715046,
+      "loss": 0.5149,
+      "step": 73600
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004900082681422121,
+      "loss": 0.5134,
+      "step": 73610
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004899658674129195,
+      "loss": 0.5406,
+      "step": 73620
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000489923466683627,
+      "loss": 0.5407,
+      "step": 73630
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004898810659543345,
+      "loss": 0.6125,
+      "step": 73640
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004898386652250419,
+      "loss": 0.4722,
+      "step": 73650
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004897962644957493,
+      "loss": 0.471,
+      "step": 73660
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004897538637664568,
+      "loss": 0.4476,
+      "step": 73670
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004897114630371643,
+      "loss": 0.5193,
+      "step": 73680
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004896690623078717,
+      "loss": 0.5127,
+      "step": 73690
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004896266615785792,
+      "loss": 0.4781,
+      "step": 73700
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004895842608492867,
+      "loss": 0.4825,
+      "step": 73710
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004895418601199941,
+      "loss": 0.5058,
+      "step": 73720
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004894994593907016,
+      "loss": 0.5268,
+      "step": 73730
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.000489457058661409,
+      "loss": 0.5138,
+      "step": 73740
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004894146579321164,
+      "loss": 0.4676,
+      "step": 73750
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004893722572028239,
+      "loss": 0.4773,
+      "step": 73760
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004893298564735314,
+      "loss": 0.4538,
+      "step": 73770
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004892874557442388,
+      "loss": 0.4822,
+      "step": 73780
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004892450550149463,
+      "loss": 0.4474,
+      "step": 73790
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004892026542856538,
+      "loss": 0.4985,
+      "step": 73800
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004891602535563612,
+      "loss": 0.521,
+      "step": 73810
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004891178528270686,
+      "loss": 0.5127,
+      "step": 73820
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.000489075452097776,
+      "loss": 0.4665,
+      "step": 73830
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004890330513684836,
+      "loss": 0.5128,
+      "step": 73840
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.000488990650639191,
+      "loss": 0.5272,
+      "step": 73850
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004889482499098984,
+      "loss": 0.423,
+      "step": 73860
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004889058491806059,
+      "loss": 0.6059,
+      "step": 73870
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004888634484513134,
+      "loss": 0.5059,
+      "step": 73880
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004888210477220208,
+      "loss": 0.4635,
+      "step": 73890
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004887786469927283,
+      "loss": 0.4782,
+      "step": 73900
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004887362462634357,
+      "loss": 0.5022,
+      "step": 73910
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004886938455341432,
+      "loss": 0.4395,
+      "step": 73920
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004886514448048507,
+      "loss": 0.4682,
+      "step": 73930
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004886090440755581,
+      "loss": 0.5286,
+      "step": 73940
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004885666433462655,
+      "loss": 0.5875,
+      "step": 73950
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004885242426169731,
+      "loss": 0.5407,
+      "step": 73960
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004884818418876805,
+      "loss": 0.6945,
+      "step": 73970
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004884394411583879,
+      "loss": 0.4727,
+      "step": 73980
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004883970404290954,
+      "loss": 0.4344,
+      "step": 73990
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004883546396998029,
+      "loss": 0.556,
+      "step": 74000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004883122389705103,
+      "loss": 0.582,
+      "step": 74010
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004882698382412178,
+      "loss": 0.5813,
+      "step": 74020
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004882274375119252,
+      "loss": 0.5199,
+      "step": 74030
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004881850367826327,
+      "loss": 0.5599,
+      "step": 74040
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004881426360533401,
+      "loss": 0.6021,
+      "step": 74050
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00048810023532404757,
+      "loss": 0.5208,
+      "step": 74060
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004880578345947551,
+      "loss": 0.456,
+      "step": 74070
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00048801543386546245,
+      "loss": 0.496,
+      "step": 74080
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00048797303313616997,
+      "loss": 0.5465,
+      "step": 74090
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00048793063240687743,
+      "loss": 0.5174,
+      "step": 74100
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00048788823167758485,
+      "loss": 0.5012,
+      "step": 74110
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004878458309482923,
+      "loss": 0.5082,
+      "step": 74120
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00048780343021899983,
+      "loss": 0.5289,
+      "step": 74130
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004877610294897072,
+      "loss": 0.5438,
+      "step": 74140
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004877186287604147,
+      "loss": 0.529,
+      "step": 74150
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004876762280311222,
+      "loss": 0.5028,
+      "step": 74160
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004876338273018296,
+      "loss": 0.5482,
+      "step": 74170
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00048759142657253706,
+      "loss": 0.4759,
+      "step": 74180
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004875490258432446,
+      "loss": 0.5872,
+      "step": 74190
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00048750662511395194,
+      "loss": 0.5212,
+      "step": 74200
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00048746422438465946,
+      "loss": 0.4995,
+      "step": 74210
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00048742182365536693,
+      "loss": 0.543,
+      "step": 74220
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00048737942292607434,
+      "loss": 0.5652,
+      "step": 74230
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004873370221967818,
+      "loss": 0.4894,
+      "step": 74240
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004872946214674893,
+      "loss": 0.4258,
+      "step": 74250
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004872522207381967,
+      "loss": 0.5535,
+      "step": 74260
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004872098200089042,
+      "loss": 0.5712,
+      "step": 74270
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004871674192796117,
+      "loss": 0.5439,
+      "step": 74280
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004871250185503191,
+      "loss": 0.5796,
+      "step": 74290
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00048708261782102655,
+      "loss": 0.4422,
+      "step": 74300
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.000487040217091734,
+      "loss": 0.5219,
+      "step": 74310
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00048699781636244143,
+      "loss": 0.5136,
+      "step": 74320
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004869554156331489,
+      "loss": 0.5618,
+      "step": 74330
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004869130149038564,
+      "loss": 0.5964,
+      "step": 74340
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00048687061417456383,
+      "loss": 0.4471,
+      "step": 74350
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004868282134452713,
+      "loss": 0.5655,
+      "step": 74360
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004867858127159787,
+      "loss": 0.5107,
+      "step": 74370
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004867434119866862,
+      "loss": 0.5547,
+      "step": 74380
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048670101125739365,
+      "loss": 0.5588,
+      "step": 74390
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048665861052810106,
+      "loss": 0.5385,
+      "step": 74400
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004866162097988086,
+      "loss": 0.4729,
+      "step": 74410
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048657380906951605,
+      "loss": 0.472,
+      "step": 74420
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048653140834022346,
+      "loss": 0.4596,
+      "step": 74430
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004864890076109309,
+      "loss": 0.517,
+      "step": 74440
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004864466068816384,
+      "loss": 0.5981,
+      "step": 74450
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004864042061523458,
+      "loss": 0.5713,
+      "step": 74460
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048636180542305327,
+      "loss": 0.4909,
+      "step": 74470
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004863194046937608,
+      "loss": 0.5372,
+      "step": 74480
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004862770039644682,
+      "loss": 0.528,
+      "step": 74490
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048623460323517567,
+      "loss": 0.4746,
+      "step": 74500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048619220250588314,
+      "loss": 0.5438,
+      "step": 74510
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048614980177659055,
+      "loss": 0.5697,
+      "step": 74520
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.000486107401047298,
+      "loss": 0.4827,
+      "step": 74530
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048606500031800554,
+      "loss": 0.5983,
+      "step": 74540
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004860225995887129,
+      "loss": 0.6071,
+      "step": 74550
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004859801988594204,
+      "loss": 0.5053,
+      "step": 74560
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004859377981301279,
+      "loss": 0.6286,
+      "step": 74570
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004858953974008353,
+      "loss": 0.4561,
+      "step": 74580
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048585299667154276,
+      "loss": 0.5321,
+      "step": 74590
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004858105959422503,
+      "loss": 0.5497,
+      "step": 74600
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048576819521295764,
+      "loss": 0.4808,
+      "step": 74610
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048572579448366516,
+      "loss": 0.5256,
+      "step": 74620
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048568339375437263,
+      "loss": 0.535,
+      "step": 74630
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048564099302508004,
+      "loss": 0.4878,
+      "step": 74640
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004855985922957875,
+      "loss": 0.5045,
+      "step": 74650
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048555619156649503,
+      "loss": 0.5911,
+      "step": 74660
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004855137908372024,
+      "loss": 0.4426,
+      "step": 74670
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004854713901079099,
+      "loss": 0.4799,
+      "step": 74680
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048542898937861727,
+      "loss": 0.5388,
+      "step": 74690
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004853865886493248,
+      "loss": 0.4935,
+      "step": 74700
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048534418792003226,
+      "loss": 0.4264,
+      "step": 74710
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048530178719073967,
+      "loss": 0.5304,
+      "step": 74720
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048525938646144714,
+      "loss": 0.4217,
+      "step": 74730
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048521698573215466,
+      "loss": 0.5422,
+      "step": 74740
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.000485174585002862,
+      "loss": 0.4844,
+      "step": 74750
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048513218427356954,
+      "loss": 0.4889,
+      "step": 74760
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.000485089783544277,
+      "loss": 0.4848,
+      "step": 74770
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004850473828149844,
+      "loss": 0.5355,
+      "step": 74780
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004850049820856919,
+      "loss": 0.5225,
+      "step": 74790
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004849625813563994,
+      "loss": 0.4961,
+      "step": 74800
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048492018062710676,
+      "loss": 0.5388,
+      "step": 74810
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004848777798978143,
+      "loss": 0.4639,
+      "step": 74820
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048483537916852175,
+      "loss": 0.5121,
+      "step": 74830
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048479297843922916,
+      "loss": 0.518,
+      "step": 74840
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048475057770993663,
+      "loss": 0.4298,
+      "step": 74850
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048470817698064415,
+      "loss": 0.4589,
+      "step": 74860
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004846657762513515,
+      "loss": 0.4803,
+      "step": 74870
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048462337552205903,
+      "loss": 0.4676,
+      "step": 74880
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004845809747927665,
+      "loss": 0.5189,
+      "step": 74890
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004845385740634739,
+      "loss": 0.4613,
+      "step": 74900
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004844961733341814,
+      "loss": 0.4898,
+      "step": 74910
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048445377260488884,
+      "loss": 0.4845,
+      "step": 74920
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048441137187559626,
+      "loss": 0.4895,
+      "step": 74930
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004843689711463038,
+      "loss": 0.5075,
+      "step": 74940
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048432657041701124,
+      "loss": 0.4986,
+      "step": 74950
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048428416968771866,
+      "loss": 0.5302,
+      "step": 74960
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004842417689584261,
+      "loss": 0.539,
+      "step": 74970
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048419936822913354,
+      "loss": 0.6087,
+      "step": 74980
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.000484156967499841,
+      "loss": 0.5667,
+      "step": 74990
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048411456677054847,
+      "loss": 0.5784,
+      "step": 75000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004840721660412559,
+      "loss": 0.486,
+      "step": 75010
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004840297653119634,
+      "loss": 0.5271,
+      "step": 75020
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048398736458267087,
+      "loss": 0.4607,
+      "step": 75030
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004839449638533783,
+      "loss": 0.5144,
+      "step": 75040
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048390256312408575,
+      "loss": 0.4709,
+      "step": 75050
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004838601623947932,
+      "loss": 0.5076,
+      "step": 75060
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048381776166550063,
+      "loss": 0.476,
+      "step": 75070
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004837753609362081,
+      "loss": 0.4717,
+      "step": 75080
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004837329602069156,
+      "loss": 0.472,
+      "step": 75090
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048369055947762303,
+      "loss": 0.5083,
+      "step": 75100
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004836481587483305,
+      "loss": 0.4591,
+      "step": 75110
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048360575801903796,
+      "loss": 0.5677,
+      "step": 75120
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004835633572897454,
+      "loss": 0.5859,
+      "step": 75130
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048352095656045284,
+      "loss": 0.553,
+      "step": 75140
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048347855583116036,
+      "loss": 0.5696,
+      "step": 75150
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004834361551018678,
+      "loss": 0.5231,
+      "step": 75160
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048339375437257524,
+      "loss": 0.5804,
+      "step": 75170
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004833513536432827,
+      "loss": 0.5019,
+      "step": 75180
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004833089529139901,
+      "loss": 0.5516,
+      "step": 75190
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004832665521846976,
+      "loss": 0.528,
+      "step": 75200
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004832241514554051,
+      "loss": 0.4757,
+      "step": 75210
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048318175072611247,
+      "loss": 0.5318,
+      "step": 75220
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048313934999682,
+      "loss": 0.5559,
+      "step": 75230
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048309694926752746,
+      "loss": 0.633,
+      "step": 75240
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048305454853823487,
+      "loss": 0.5745,
+      "step": 75250
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048301214780894233,
+      "loss": 0.4618,
+      "step": 75260
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048296974707964986,
+      "loss": 0.4105,
+      "step": 75270
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004829273463503572,
+      "loss": 0.552,
+      "step": 75280
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048288494562106474,
+      "loss": 0.63,
+      "step": 75290
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004828425448917721,
+      "loss": 0.5927,
+      "step": 75300
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004828001441624796,
+      "loss": 0.5353,
+      "step": 75310
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004827577434331871,
+      "loss": 0.4562,
+      "step": 75320
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004827153427038945,
+      "loss": 0.4668,
+      "step": 75330
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048267294197460196,
+      "loss": 0.4971,
+      "step": 75340
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004826305412453095,
+      "loss": 0.5424,
+      "step": 75350
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048258814051601684,
+      "loss": 0.5036,
+      "step": 75360
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048254573978672436,
+      "loss": 0.53,
+      "step": 75370
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048250333905743183,
+      "loss": 0.5401,
+      "step": 75380
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048246093832813924,
+      "loss": 0.5064,
+      "step": 75390
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004824185375988467,
+      "loss": 0.5401,
+      "step": 75400
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048237613686955423,
+      "loss": 0.5111,
+      "step": 75410
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004823337361402616,
+      "loss": 0.543,
+      "step": 75420
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004822913354109691,
+      "loss": 0.5015,
+      "step": 75430
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004822489346816766,
+      "loss": 0.4696,
+      "step": 75440
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.000482206533952384,
+      "loss": 0.5638,
+      "step": 75450
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048216413322309145,
+      "loss": 0.5414,
+      "step": 75460
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.000482121732493799,
+      "loss": 0.4752,
+      "step": 75470
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048207933176450633,
+      "loss": 0.5112,
+      "step": 75480
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048203693103521385,
+      "loss": 0.5937,
+      "step": 75490
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004819945303059213,
+      "loss": 0.5636,
+      "step": 75500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048195212957662873,
+      "loss": 0.4162,
+      "step": 75510
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004819097288473362,
+      "loss": 0.5701,
+      "step": 75520
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048186732811804367,
+      "loss": 0.5312,
+      "step": 75530
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004818249273887511,
+      "loss": 0.4948,
+      "step": 75540
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004817825266594586,
+      "loss": 0.5552,
+      "step": 75550
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048174012593016607,
+      "loss": 0.5866,
+      "step": 75560
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004816977252008735,
+      "loss": 0.5222,
+      "step": 75570
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048165532447158095,
+      "loss": 0.5167,
+      "step": 75580
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048161292374228836,
+      "loss": 0.5507,
+      "step": 75590
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004815705230129958,
+      "loss": 0.5202,
+      "step": 75600
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048152812228370335,
+      "loss": 0.4746,
+      "step": 75610
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004814857215544107,
+      "loss": 0.5439,
+      "step": 75620
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004814433208251182,
+      "loss": 0.5818,
+      "step": 75630
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004814009200958257,
+      "loss": 0.4889,
+      "step": 75640
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004813585193665331,
+      "loss": 0.549,
+      "step": 75650
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048131611863724057,
+      "loss": 0.4992,
+      "step": 75660
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048127371790794804,
+      "loss": 0.4875,
+      "step": 75670
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048123131717865545,
+      "loss": 0.54,
+      "step": 75680
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.000481188916449363,
+      "loss": 0.5764,
+      "step": 75690
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048114651572007044,
+      "loss": 0.4574,
+      "step": 75700
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048110411499077785,
+      "loss": 0.4398,
+      "step": 75710
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004810617142614853,
+      "loss": 0.5258,
+      "step": 75720
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004810193135321928,
+      "loss": 0.5225,
+      "step": 75730
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004809769128029002,
+      "loss": 0.5353,
+      "step": 75740
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048093451207360767,
+      "loss": 0.4473,
+      "step": 75750
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004808921113443152,
+      "loss": 0.4724,
+      "step": 75760
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004808497106150226,
+      "loss": 0.5771,
+      "step": 75770
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048080730988573007,
+      "loss": 0.585,
+      "step": 75780
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048076490915643753,
+      "loss": 0.6031,
+      "step": 75790
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048072250842714495,
+      "loss": 0.6068,
+      "step": 75800
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004806801076978524,
+      "loss": 0.5503,
+      "step": 75810
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048063770696855993,
+      "loss": 0.4862,
+      "step": 75820
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004805953062392673,
+      "loss": 0.5818,
+      "step": 75830
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004805529055099748,
+      "loss": 0.4202,
+      "step": 75840
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004805105047806823,
+      "loss": 0.5186,
+      "step": 75850
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004804681040513897,
+      "loss": 0.498,
+      "step": 75860
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048042570332209716,
+      "loss": 0.6796,
+      "step": 75870
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004803833025928047,
+      "loss": 0.5047,
+      "step": 75880
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048034090186351204,
+      "loss": 0.5118,
+      "step": 75890
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048029850113421956,
+      "loss": 0.5146,
+      "step": 75900
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048025610040492697,
+      "loss": 0.5572,
+      "step": 75910
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048021369967563444,
+      "loss": 0.4548,
+      "step": 75920
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004801712989463419,
+      "loss": 0.375,
+      "step": 75930
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004801288982170493,
+      "loss": 0.4809,
+      "step": 75940
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004800864974877568,
+      "loss": 0.5234,
+      "step": 75950
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004800440967584643,
+      "loss": 0.433,
+      "step": 75960
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048000169602917166,
+      "loss": 0.5337,
+      "step": 75970
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004799592952998792,
+      "loss": 0.5185,
+      "step": 75980
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00047991689457058665,
+      "loss": 0.5264,
+      "step": 75990
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00047987449384129406,
+      "loss": 0.51,
+      "step": 76000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00047983209311200153,
+      "loss": 0.4856,
+      "step": 76010
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00047978969238270905,
+      "loss": 0.515,
+      "step": 76020
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004797472916534164,
+      "loss": 0.5146,
+      "step": 76030
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00047970489092412393,
+      "loss": 0.5528,
+      "step": 76040
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004796624901948314,
+      "loss": 0.528,
+      "step": 76050
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004796200894655388,
+      "loss": 0.476,
+      "step": 76060
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004795776887362463,
+      "loss": 0.5561,
+      "step": 76070
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004795352880069538,
+      "loss": 0.538,
+      "step": 76080
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00047949288727766116,
+      "loss": 0.4592,
+      "step": 76090
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004794504865483687,
+      "loss": 0.4291,
+      "step": 76100
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00047940808581907614,
+      "loss": 0.5123,
+      "step": 76110
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00047936568508978356,
+      "loss": 0.5434,
+      "step": 76120
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.000479323284360491,
+      "loss": 0.4747,
+      "step": 76130
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00047928088363119854,
+      "loss": 0.4645,
+      "step": 76140
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004792384829019059,
+      "loss": 0.5325,
+      "step": 76150
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004791960821726134,
+      "loss": 0.6031,
+      "step": 76160
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004791536814433209,
+      "loss": 0.4842,
+      "step": 76170
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004791112807140283,
+      "loss": 0.4916,
+      "step": 76180
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00047906887998473577,
+      "loss": 0.6312,
+      "step": 76190
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004790264792554432,
+      "loss": 0.4754,
+      "step": 76200
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00047898407852615065,
+      "loss": 0.4842,
+      "step": 76210
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00047894167779685817,
+      "loss": 0.4404,
+      "step": 76220
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00047889927706756553,
+      "loss": 0.4689,
+      "step": 76230
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00047885687633827305,
+      "loss": 0.4999,
+      "step": 76240
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004788144756089805,
+      "loss": 0.4727,
+      "step": 76250
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00047877207487968793,
+      "loss": 0.4853,
+      "step": 76260
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004787296741503954,
+      "loss": 0.528,
+      "step": 76270
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047868727342110286,
+      "loss": 0.5291,
+      "step": 76280
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004786448726918103,
+      "loss": 0.5253,
+      "step": 76290
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004786024719625178,
+      "loss": 0.412,
+      "step": 76300
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047856007123322526,
+      "loss": 0.5087,
+      "step": 76310
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004785176705039327,
+      "loss": 0.4515,
+      "step": 76320
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047847526977464014,
+      "loss": 0.5723,
+      "step": 76330
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004784328690453476,
+      "loss": 0.601,
+      "step": 76340
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.000478390468316055,
+      "loss": 0.5171,
+      "step": 76350
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047834806758676254,
+      "loss": 0.4414,
+      "step": 76360
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047830566685747,
+      "loss": 0.5525,
+      "step": 76370
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004782632661281774,
+      "loss": 0.5213,
+      "step": 76380
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004782208653988849,
+      "loss": 0.5814,
+      "step": 76390
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047817846466959236,
+      "loss": 0.4739,
+      "step": 76400
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047813606394029977,
+      "loss": 0.5884,
+      "step": 76410
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047809366321100724,
+      "loss": 0.5032,
+      "step": 76420
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047805126248171476,
+      "loss": 0.4679,
+      "step": 76430
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047800886175242217,
+      "loss": 0.3853,
+      "step": 76440
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047796646102312964,
+      "loss": 0.5355,
+      "step": 76450
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004779240602938371,
+      "loss": 0.5662,
+      "step": 76460
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004778816595645445,
+      "loss": 0.4942,
+      "step": 76470
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.000477839258835252,
+      "loss": 0.4424,
+      "step": 76480
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004777968581059595,
+      "loss": 0.4922,
+      "step": 76490
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047775445737666686,
+      "loss": 0.5414,
+      "step": 76500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004777120566473744,
+      "loss": 0.5683,
+      "step": 76510
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004776696559180818,
+      "loss": 0.5588,
+      "step": 76520
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047762725518878926,
+      "loss": 0.4974,
+      "step": 76530
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047758485445949673,
+      "loss": 0.5487,
+      "step": 76540
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047754245373020414,
+      "loss": 0.421,
+      "step": 76550
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004775000530009116,
+      "loss": 0.5916,
+      "step": 76560
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047745765227161913,
+      "loss": 0.4629,
+      "step": 76570
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004774152515423265,
+      "loss": 0.5329,
+      "step": 76580
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000477372850813034,
+      "loss": 0.5355,
+      "step": 76590
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004773304500837415,
+      "loss": 0.572,
+      "step": 76600
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004772880493544489,
+      "loss": 0.5435,
+      "step": 76610
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047724564862515635,
+      "loss": 0.5154,
+      "step": 76620
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004772032478958639,
+      "loss": 0.5073,
+      "step": 76630
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047716084716657123,
+      "loss": 0.509,
+      "step": 76640
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047711844643727875,
+      "loss": 0.5813,
+      "step": 76650
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004770760457079862,
+      "loss": 0.5699,
+      "step": 76660
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047703364497869363,
+      "loss": 0.5242,
+      "step": 76670
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004769912442494011,
+      "loss": 0.5007,
+      "step": 76680
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004769488435201086,
+      "loss": 0.4746,
+      "step": 76690
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000476906442790816,
+      "loss": 0.43,
+      "step": 76700
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004768640420615235,
+      "loss": 0.5566,
+      "step": 76710
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047682164133223097,
+      "loss": 0.5325,
+      "step": 76720
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004767792406029384,
+      "loss": 0.4324,
+      "step": 76730
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047673683987364585,
+      "loss": 0.4257,
+      "step": 76740
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047669443914435337,
+      "loss": 0.5847,
+      "step": 76750
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004766520384150607,
+      "loss": 0.5264,
+      "step": 76760
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047660963768576825,
+      "loss": 0.4188,
+      "step": 76770
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004765672369564757,
+      "loss": 0.5866,
+      "step": 76780
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047652483622718313,
+      "loss": 0.5252,
+      "step": 76790
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004764824354978906,
+      "loss": 0.5626,
+      "step": 76800
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047644003476859806,
+      "loss": 0.5342,
+      "step": 76810
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004763976340393055,
+      "loss": 0.4544,
+      "step": 76820
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.000476355233310013,
+      "loss": 0.5763,
+      "step": 76830
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047631283258072035,
+      "loss": 0.5174,
+      "step": 76840
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004762704318514279,
+      "loss": 0.4455,
+      "step": 76850
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047622803112213534,
+      "loss": 0.4155,
+      "step": 76860
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047618563039284275,
+      "loss": 0.5356,
+      "step": 76870
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004761432296635502,
+      "loss": 0.5383,
+      "step": 76880
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047610082893425774,
+      "loss": 0.6553,
+      "step": 76890
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004760584282049651,
+      "loss": 0.4864,
+      "step": 76900
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004760160274756726,
+      "loss": 0.509,
+      "step": 76910
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004759736267463801,
+      "loss": 0.4601,
+      "step": 76920
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004759312260170875,
+      "loss": 0.5248,
+      "step": 76930
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047588882528779497,
+      "loss": 0.5178,
+      "step": 76940
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047584642455850243,
+      "loss": 0.6241,
+      "step": 76950
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047580402382920985,
+      "loss": 0.5301,
+      "step": 76960
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047576162309991737,
+      "loss": 0.4639,
+      "step": 76970
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047571922237062483,
+      "loss": 0.4565,
+      "step": 76980
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047567682164133225,
+      "loss": 0.4754,
+      "step": 76990
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004756344209120397,
+      "loss": 0.515,
+      "step": 77000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004755920201827472,
+      "loss": 0.4733,
+      "step": 77010
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004755496194534546,
+      "loss": 0.5727,
+      "step": 77020
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047550721872416206,
+      "loss": 0.471,
+      "step": 77030
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004754648179948696,
+      "loss": 0.5049,
+      "step": 77040
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.000475422417265577,
+      "loss": 0.5597,
+      "step": 77050
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047538001653628446,
+      "loss": 0.579,
+      "step": 77060
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004753376158069919,
+      "loss": 0.4981,
+      "step": 77070
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047529521507769934,
+      "loss": 0.458,
+      "step": 77080
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004752528143484068,
+      "loss": 0.4263,
+      "step": 77090
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004752104136191143,
+      "loss": 0.519,
+      "step": 77100
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047516801288982174,
+      "loss": 0.4756,
+      "step": 77110
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004751256121605292,
+      "loss": 0.6127,
+      "step": 77120
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004750832114312366,
+      "loss": 0.5791,
+      "step": 77130
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004750408107019441,
+      "loss": 0.4921,
+      "step": 77140
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047499840997265155,
+      "loss": 0.5414,
+      "step": 77150
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047495600924335896,
+      "loss": 0.4507,
+      "step": 77160
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047491360851406643,
+      "loss": 0.4666,
+      "step": 77170
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047487120778477395,
+      "loss": 0.4686,
+      "step": 77180
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047482880705548137,
+      "loss": 0.589,
+      "step": 77190
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047478640632618883,
+      "loss": 0.4556,
+      "step": 77200
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004747440055968963,
+      "loss": 0.4177,
+      "step": 77210
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004747016048676037,
+      "loss": 0.526,
+      "step": 77220
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004746592041383112,
+      "loss": 0.5222,
+      "step": 77230
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004746168034090187,
+      "loss": 0.5068,
+      "step": 77240
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047457440267972606,
+      "loss": 0.4607,
+      "step": 77250
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004745320019504336,
+      "loss": 0.5532,
+      "step": 77260
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047448960122114105,
+      "loss": 0.5266,
+      "step": 77270
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047444720049184846,
+      "loss": 0.57,
+      "step": 77280
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004744047997625559,
+      "loss": 0.4789,
+      "step": 77290
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047436239903326345,
+      "loss": 0.445,
+      "step": 77300
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004743199983039708,
+      "loss": 0.6136,
+      "step": 77310
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004742775975746783,
+      "loss": 0.46,
+      "step": 77320
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004742351968453858,
+      "loss": 0.5109,
+      "step": 77330
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004741927961160932,
+      "loss": 0.5145,
+      "step": 77340
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047415039538680067,
+      "loss": 0.5738,
+      "step": 77350
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004741079946575082,
+      "loss": 0.487,
+      "step": 77360
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047406559392821555,
+      "loss": 0.587,
+      "step": 77370
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047402319319892307,
+      "loss": 0.4403,
+      "step": 77380
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047398079246963054,
+      "loss": 0.5683,
+      "step": 77390
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047393839174033795,
+      "loss": 0.441,
+      "step": 77400
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004738959910110454,
+      "loss": 0.5224,
+      "step": 77410
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047385359028175294,
+      "loss": 0.5017,
+      "step": 77420
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004738111895524603,
+      "loss": 0.487,
+      "step": 77430
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004737687888231678,
+      "loss": 0.5515,
+      "step": 77440
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004737263880938752,
+      "loss": 0.4546,
+      "step": 77450
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004736839873645827,
+      "loss": 0.5731,
+      "step": 77460
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047364158663529016,
+      "loss": 0.5447,
+      "step": 77470
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004735991859059976,
+      "loss": 0.4933,
+      "step": 77480
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047355678517670504,
+      "loss": 0.5239,
+      "step": 77490
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047351438444741256,
+      "loss": 0.4848,
+      "step": 77500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004734719837181199,
+      "loss": 0.5548,
+      "step": 77510
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047342958298882744,
+      "loss": 0.5022,
+      "step": 77520
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004733871822595349,
+      "loss": 0.513,
+      "step": 77530
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004733447815302423,
+      "loss": 0.4325,
+      "step": 77540
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004733023808009498,
+      "loss": 0.5866,
+      "step": 77550
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047325998007165726,
+      "loss": 0.4742,
+      "step": 77560
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047321757934236467,
+      "loss": 0.5336,
+      "step": 77570
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004731751786130722,
+      "loss": 0.4936,
+      "step": 77580
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047313277788377966,
+      "loss": 0.6111,
+      "step": 77590
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047309037715448707,
+      "loss": 0.5541,
+      "step": 77600
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047304797642519454,
+      "loss": 0.5036,
+      "step": 77610
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.000473005575695902,
+      "loss": 0.5936,
+      "step": 77620
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004729631749666094,
+      "loss": 0.4725,
+      "step": 77630
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047292077423731694,
+      "loss": 0.4948,
+      "step": 77640
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004728783735080244,
+      "loss": 0.5842,
+      "step": 77650
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004728359727787318,
+      "loss": 0.57,
+      "step": 77660
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004727935720494393,
+      "loss": 0.4679,
+      "step": 77670
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047275117132014675,
+      "loss": 0.6122,
+      "step": 77680
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047270877059085416,
+      "loss": 0.5156,
+      "step": 77690
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047266636986156163,
+      "loss": 0.5896,
+      "step": 77700
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047262396913226915,
+      "loss": 0.5119,
+      "step": 77710
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047258156840297656,
+      "loss": 0.5107,
+      "step": 77720
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047253916767368403,
+      "loss": 0.4566,
+      "step": 77730
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047249676694439144,
+      "loss": 0.618,
+      "step": 77740
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004724543662150989,
+      "loss": 0.5318,
+      "step": 77750
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004724119654858064,
+      "loss": 0.4899,
+      "step": 77760
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004723695647565138,
+      "loss": 0.5115,
+      "step": 77770
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047232716402722126,
+      "loss": 0.5572,
+      "step": 77780
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004722847632979288,
+      "loss": 0.556,
+      "step": 77790
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004722423625686362,
+      "loss": 0.5518,
+      "step": 77800
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047219996183934366,
+      "loss": 0.4718,
+      "step": 77810
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004721575611100511,
+      "loss": 0.5253,
+      "step": 77820
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047211516038075853,
+      "loss": 0.5141,
+      "step": 77830
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.000472072759651466,
+      "loss": 0.5777,
+      "step": 77840
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004720303589221735,
+      "loss": 0.5581,
+      "step": 77850
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004719879581928809,
+      "loss": 0.5388,
+      "step": 77860
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004719455574635884,
+      "loss": 0.5617,
+      "step": 77870
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047190315673429587,
+      "loss": 0.5255,
+      "step": 77880
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004718607560050033,
+      "loss": 0.5058,
+      "step": 77890
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047181835527571075,
+      "loss": 0.4948,
+      "step": 77900
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047177595454641827,
+      "loss": 0.494,
+      "step": 77910
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047173355381712563,
+      "loss": 0.5289,
+      "step": 77920
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047169115308783315,
+      "loss": 0.4526,
+      "step": 77930
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004716487523585406,
+      "loss": 0.512,
+      "step": 77940
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047160635162924803,
+      "loss": 0.4611,
+      "step": 77950
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004715639508999555,
+      "loss": 0.5745,
+      "step": 77960
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.000471521550170663,
+      "loss": 0.6043,
+      "step": 77970
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004714791494413704,
+      "loss": 0.45,
+      "step": 77980
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004714367487120779,
+      "loss": 0.6834,
+      "step": 77990
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047139434798278536,
+      "loss": 0.4226,
+      "step": 78000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004713519472534928,
+      "loss": 0.5423,
+      "step": 78010
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047130954652420024,
+      "loss": 0.5776,
+      "step": 78020
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047126714579490776,
+      "loss": 0.4903,
+      "step": 78030
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004712247450656151,
+      "loss": 0.5277,
+      "step": 78040
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047118234433632264,
+      "loss": 0.4924,
+      "step": 78050
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047113994360703,
+      "loss": 0.5796,
+      "step": 78060
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004710975428777375,
+      "loss": 0.5438,
+      "step": 78070
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.000471055142148445,
+      "loss": 0.6223,
+      "step": 78080
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004710127414191524,
+      "loss": 0.5058,
+      "step": 78090
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047097034068985987,
+      "loss": 0.4797,
+      "step": 78100
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004709279399605674,
+      "loss": 0.52,
+      "step": 78110
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047088553923127475,
+      "loss": 0.5706,
+      "step": 78120
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047084313850198227,
+      "loss": 0.4606,
+      "step": 78130
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047080073777268973,
+      "loss": 0.4405,
+      "step": 78140
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047075833704339715,
+      "loss": 0.4801,
+      "step": 78150
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004707159363141046,
+      "loss": 0.6313,
+      "step": 78160
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047067353558481213,
+      "loss": 0.4597,
+      "step": 78170
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004706311348555195,
+      "loss": 0.5006,
+      "step": 78180
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.000470588734126227,
+      "loss": 0.4894,
+      "step": 78190
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004705463333969345,
+      "loss": 0.5283,
+      "step": 78200
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004705039326676419,
+      "loss": 0.5121,
+      "step": 78210
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047046153193834936,
+      "loss": 0.6172,
+      "step": 78220
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004704191312090568,
+      "loss": 0.5118,
+      "step": 78230
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047037673047976424,
+      "loss": 0.493,
+      "step": 78240
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047033432975047176,
+      "loss": 0.5701,
+      "step": 78250
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047029192902117923,
+      "loss": 0.5941,
+      "step": 78260
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047024952829188664,
+      "loss": 0.5032,
+      "step": 78270
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004702071275625941,
+      "loss": 0.4835,
+      "step": 78280
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004701647268333016,
+      "loss": 0.5355,
+      "step": 78290
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.000470122326104009,
+      "loss": 0.4992,
+      "step": 78300
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047007992537471645,
+      "loss": 0.427,
+      "step": 78310
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.000470037524645424,
+      "loss": 0.5313,
+      "step": 78320
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004699951239161314,
+      "loss": 0.5094,
+      "step": 78330
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00046995272318683885,
+      "loss": 0.5913,
+      "step": 78340
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00046991032245754627,
+      "loss": 0.5217,
+      "step": 78350
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00046986792172825373,
+      "loss": 0.5733,
+      "step": 78360
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004698255209989612,
+      "loss": 0.4892,
+      "step": 78370
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004697831202696686,
+      "loss": 0.5264,
+      "step": 78380
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00046974071954037613,
+      "loss": 0.5522,
+      "step": 78390
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004696983188110836,
+      "loss": 0.5134,
+      "step": 78400
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000469655918081791,
+      "loss": 0.5031,
+      "step": 78410
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004696135173524985,
+      "loss": 0.4955,
+      "step": 78420
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00046957111662320595,
+      "loss": 0.6239,
+      "step": 78430
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00046952871589391336,
+      "loss": 0.4901,
+      "step": 78440
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004694863151646208,
+      "loss": 0.5437,
+      "step": 78450
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00046944391443532835,
+      "loss": 0.5054,
+      "step": 78460
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00046940151370603576,
+      "loss": 0.5168,
+      "step": 78470
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004693591129767432,
+      "loss": 0.5396,
+      "step": 78480
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004693167122474507,
+      "loss": 0.5613,
+      "step": 78490
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004692743115181581,
+      "loss": 0.5335,
+      "step": 78500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00046923191078886557,
+      "loss": 0.4796,
+      "step": 78510
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004691895100595731,
+      "loss": 0.503,
+      "step": 78520
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00046914710933028045,
+      "loss": 0.4648,
+      "step": 78530
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00046910470860098797,
+      "loss": 0.5999,
+      "step": 78540
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00046906230787169544,
+      "loss": 0.5072,
+      "step": 78550
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00046901990714240285,
+      "loss": 0.5321,
+      "step": 78560
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004689775064131103,
+      "loss": 0.5419,
+      "step": 78570
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00046893510568381784,
+      "loss": 0.4863,
+      "step": 78580
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004688927049545252,
+      "loss": 0.6213,
+      "step": 78590
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004688503042252327,
+      "loss": 0.5466,
+      "step": 78600
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004688079034959402,
+      "loss": 0.5072,
+      "step": 78610
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004687655027666476,
+      "loss": 0.5147,
+      "step": 78620
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046872310203735506,
+      "loss": 0.5214,
+      "step": 78630
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004686807013080626,
+      "loss": 0.5302,
+      "step": 78640
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046863830057876994,
+      "loss": 0.5548,
+      "step": 78650
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046859589984947747,
+      "loss": 0.5373,
+      "step": 78660
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004685534991201848,
+      "loss": 0.512,
+      "step": 78670
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046851109839089234,
+      "loss": 0.6199,
+      "step": 78680
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004684686976615998,
+      "loss": 0.5322,
+      "step": 78690
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004684262969323072,
+      "loss": 0.4938,
+      "step": 78700
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004683838962030147,
+      "loss": 0.5166,
+      "step": 78710
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004683414954737222,
+      "loss": 0.5319,
+      "step": 78720
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046829909474442957,
+      "loss": 0.6023,
+      "step": 78730
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004682566940151371,
+      "loss": 0.5089,
+      "step": 78740
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046821429328584456,
+      "loss": 0.5156,
+      "step": 78750
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046817189255655197,
+      "loss": 0.519,
+      "step": 78760
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046812949182725944,
+      "loss": 0.4357,
+      "step": 78770
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046808709109796696,
+      "loss": 0.5004,
+      "step": 78780
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004680446903686743,
+      "loss": 0.5748,
+      "step": 78790
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046800228963938184,
+      "loss": 0.575,
+      "step": 78800
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004679598889100893,
+      "loss": 0.5357,
+      "step": 78810
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004679174881807967,
+      "loss": 0.5592,
+      "step": 78820
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004678750874515042,
+      "loss": 0.5436,
+      "step": 78830
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046783268672221165,
+      "loss": 0.5701,
+      "step": 78840
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046779028599291906,
+      "loss": 0.4055,
+      "step": 78850
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004677478852636266,
+      "loss": 0.5461,
+      "step": 78860
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046770548453433405,
+      "loss": 0.5097,
+      "step": 78870
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046766308380504146,
+      "loss": 0.5031,
+      "step": 78880
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046762068307574893,
+      "loss": 0.4804,
+      "step": 78890
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0004675782823464564,
+      "loss": 0.4972,
+      "step": 78900
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0004675358816171638,
+      "loss": 0.5617,
+      "step": 78910
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046749348088787133,
+      "loss": 0.5363,
+      "step": 78920
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0004674510801585788,
+      "loss": 0.5407,
+      "step": 78930
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0004674086794292862,
+      "loss": 0.5001,
+      "step": 78940
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0004673662786999937,
+      "loss": 0.4758,
+      "step": 78950
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046732387797070114,
+      "loss": 0.5059,
+      "step": 78960
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046728147724140856,
+      "loss": 0.524,
+      "step": 78970
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.000467239076512116,
+      "loss": 0.5423,
+      "step": 78980
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046719667578282344,
+      "loss": 0.6003,
+      "step": 78990
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046715427505353096,
+      "loss": 0.5398,
+      "step": 79000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0004671118743242384,
+      "loss": 0.5538,
+      "step": 79010
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046706947359494584,
+      "loss": 0.5319,
+      "step": 79020
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0004670270728656533,
+      "loss": 0.4928,
+      "step": 79030
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046698467213636077,
+      "loss": 0.499,
+      "step": 79040
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0004669422714070682,
+      "loss": 0.4768,
+      "step": 79050
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046689987067777565,
+      "loss": 0.4737,
+      "step": 79060
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046685746994848317,
+      "loss": 0.4453,
+      "step": 79070
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0004668150692191906,
+      "loss": 0.4771,
+      "step": 79080
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046677266848989805,
+      "loss": 0.5497,
+      "step": 79090
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004667302677606055,
+      "loss": 0.4014,
+      "step": 79100
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046668786703131293,
+      "loss": 0.5063,
+      "step": 79110
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004666454663020204,
+      "loss": 0.4955,
+      "step": 79120
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004666030655727279,
+      "loss": 0.5,
+      "step": 79130
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046656066484343533,
+      "loss": 0.5109,
+      "step": 79140
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004665182641141428,
+      "loss": 0.5023,
+      "step": 79150
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046647586338485026,
+      "loss": 0.5738,
+      "step": 79160
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004664334626555577,
+      "loss": 0.5026,
+      "step": 79170
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046639106192626514,
+      "loss": 0.5959,
+      "step": 79180
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046634866119697266,
+      "loss": 0.5153,
+      "step": 79190
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046630626046768,
+      "loss": 0.5151,
+      "step": 79200
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046626385973838754,
+      "loss": 0.4512,
+      "step": 79210
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.000466221459009095,
+      "loss": 0.6424,
+      "step": 79220
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004661790582798024,
+      "loss": 0.4702,
+      "step": 79230
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004661366575505099,
+      "loss": 0.5134,
+      "step": 79240
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004660942568212174,
+      "loss": 0.451,
+      "step": 79250
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046605185609192477,
+      "loss": 0.5722,
+      "step": 79260
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004660094553626323,
+      "loss": 0.5236,
+      "step": 79270
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046596705463333965,
+      "loss": 0.4958,
+      "step": 79280
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046592465390404717,
+      "loss": 0.6337,
+      "step": 79290
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046588225317475463,
+      "loss": 0.5613,
+      "step": 79300
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046583985244546205,
+      "loss": 0.4438,
+      "step": 79310
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004657974517161695,
+      "loss": 0.5383,
+      "step": 79320
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046575505098687704,
+      "loss": 0.5035,
+      "step": 79330
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004657126502575844,
+      "loss": 0.6884,
+      "step": 79340
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004656702495282919,
+      "loss": 0.5175,
+      "step": 79350
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004656278487989994,
+      "loss": 0.5082,
+      "step": 79360
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004655854480697068,
+      "loss": 0.5844,
+      "step": 79370
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046554304734041426,
+      "loss": 0.4739,
+      "step": 79380
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004655006466111218,
+      "loss": 0.4622,
+      "step": 79390
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046545824588182914,
+      "loss": 0.6094,
+      "step": 79400
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046541584515253666,
+      "loss": 0.4682,
+      "step": 79410
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046537344442324413,
+      "loss": 0.5531,
+      "step": 79420
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046533104369395154,
+      "loss": 0.4334,
+      "step": 79430
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.000465288642964659,
+      "loss": 0.5724,
+      "step": 79440
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046524624223536653,
+      "loss": 0.527,
+      "step": 79450
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004652038415060739,
+      "loss": 0.5527,
+      "step": 79460
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004651614407767814,
+      "loss": 0.5686,
+      "step": 79470
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004651190400474889,
+      "loss": 0.5562,
+      "step": 79480
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004650766393181963,
+      "loss": 0.523,
+      "step": 79490
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046503423858890375,
+      "loss": 0.5057,
+      "step": 79500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004649918378596112,
+      "loss": 0.4205,
+      "step": 79510
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046494943713031863,
+      "loss": 0.5258,
+      "step": 79520
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046490703640102615,
+      "loss": 0.4094,
+      "step": 79530
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004648646356717336,
+      "loss": 0.5119,
+      "step": 79540
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046482223494244103,
+      "loss": 0.5036,
+      "step": 79550
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004647798342131485,
+      "loss": 0.5161,
+      "step": 79560
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046473743348385597,
+      "loss": 0.5624,
+      "step": 79570
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004646950327545634,
+      "loss": 0.5417,
+      "step": 79580
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046465263202527085,
+      "loss": 0.4978,
+      "step": 79590
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046461023129597826,
+      "loss": 0.6289,
+      "step": 79600
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004645678305666858,
+      "loss": 0.5259,
+      "step": 79610
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046452542983739325,
+      "loss": 0.4834,
+      "step": 79620
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046448302910810066,
+      "loss": 0.4916,
+      "step": 79630
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004644406283788081,
+      "loss": 0.5284,
+      "step": 79640
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004643982276495156,
+      "loss": 0.5404,
+      "step": 79650
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.000464355826920223,
+      "loss": 0.5326,
+      "step": 79660
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004643134261909305,
+      "loss": 0.5076,
+      "step": 79670
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.000464271025461638,
+      "loss": 0.4532,
+      "step": 79680
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004642286247323454,
+      "loss": 0.5021,
+      "step": 79690
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046418622400305287,
+      "loss": 0.5198,
+      "step": 79700
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046414382327376034,
+      "loss": 0.6371,
+      "step": 79710
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046410142254446775,
+      "loss": 0.546,
+      "step": 79720
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004640590218151752,
+      "loss": 0.5171,
+      "step": 79730
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046401662108588274,
+      "loss": 0.5401,
+      "step": 79740
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046397422035659015,
+      "loss": 0.4441,
+      "step": 79750
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004639318196272976,
+      "loss": 0.5267,
+      "step": 79760
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004638894188980051,
+      "loss": 0.5362,
+      "step": 79770
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004638470181687125,
+      "loss": 0.6054,
+      "step": 79780
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046380461743941997,
+      "loss": 0.5124,
+      "step": 79790
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004637622167101275,
+      "loss": 0.4606,
+      "step": 79800
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046371981598083484,
+      "loss": 0.4983,
+      "step": 79810
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046367741525154237,
+      "loss": 0.4681,
+      "step": 79820
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046363501452224983,
+      "loss": 0.4496,
+      "step": 79830
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046359261379295725,
+      "loss": 0.4359,
+      "step": 79840
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004635502130636647,
+      "loss": 0.5247,
+      "step": 79850
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046350781233437223,
+      "loss": 0.3983,
+      "step": 79860
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004634654116050796,
+      "loss": 0.5179,
+      "step": 79870
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004634230108757871,
+      "loss": 0.507,
+      "step": 79880
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004633806101464945,
+      "loss": 0.4863,
+      "step": 79890
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.000463338209417202,
+      "loss": 0.4881,
+      "step": 79900
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046329580868790946,
+      "loss": 0.566,
+      "step": 79910
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046325340795861687,
+      "loss": 0.5129,
+      "step": 79920
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046321100722932434,
+      "loss": 0.5945,
+      "step": 79930
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046316860650003186,
+      "loss": 0.4884,
+      "step": 79940
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004631262057707392,
+      "loss": 0.528,
+      "step": 79950
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046308380504144674,
+      "loss": 0.477,
+      "step": 79960
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004630414043121542,
+      "loss": 0.4978,
+      "step": 79970
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004629990035828616,
+      "loss": 0.475,
+      "step": 79980
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004629566028535691,
+      "loss": 0.5057,
+      "step": 79990
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004629142021242766,
+      "loss": 0.4942,
+      "step": 80000
+    },
+    {
+      "epoch": 3.38,
+      "eval_loss": 0.6122093796730042,
+      "eval_runtime": 338.4,
+      "eval_samples_per_second": 15.529,
+      "eval_steps_per_second": 3.883,
+      "step": 80000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046287180139498396,
+      "loss": 0.4082,
+      "step": 80010
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004628294006656915,
+      "loss": 0.5666,
+      "step": 80020
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046278699993639895,
+      "loss": 0.4753,
+      "step": 80030
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046274459920710636,
+      "loss": 0.5011,
+      "step": 80040
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046270219847781383,
+      "loss": 0.4901,
+      "step": 80050
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046265979774852135,
+      "loss": 0.5085,
+      "step": 80060
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004626173970192287,
+      "loss": 0.4822,
+      "step": 80070
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046257499628993623,
+      "loss": 0.5356,
+      "step": 80080
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004625325955606437,
+      "loss": 0.6019,
+      "step": 80090
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004624901948313511,
+      "loss": 0.5721,
+      "step": 80100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004624477941020586,
+      "loss": 0.4752,
+      "step": 80110
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004624053933727661,
+      "loss": 0.5088,
+      "step": 80120
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046236299264347346,
+      "loss": 0.5359,
+      "step": 80130
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.000462320591914181,
+      "loss": 0.5309,
+      "step": 80140
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046227819118488844,
+      "loss": 0.5341,
+      "step": 80150
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046223579045559586,
+      "loss": 0.5814,
+      "step": 80160
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004621933897263033,
+      "loss": 0.6042,
+      "step": 80170
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004621509889970108,
+      "loss": 0.5697,
+      "step": 80180
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004621085882677182,
+      "loss": 0.5131,
+      "step": 80190
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004620661875384257,
+      "loss": 0.5079,
+      "step": 80200
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004620237868091331,
+      "loss": 0.5079,
+      "step": 80210
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004619813860798406,
+      "loss": 0.5032,
+      "step": 80220
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046193898535054807,
+      "loss": 0.4044,
+      "step": 80230
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004618965846212555,
+      "loss": 0.5015,
+      "step": 80240
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046185418389196295,
+      "loss": 0.5037,
+      "step": 80250
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004618117831626704,
+      "loss": 0.5532,
+      "step": 80260
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046176938243337783,
+      "loss": 0.5879,
+      "step": 80270
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046172698170408535,
+      "loss": 0.5359,
+      "step": 80280
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004616845809747928,
+      "loss": 0.5016,
+      "step": 80290
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046164218024550023,
+      "loss": 0.6164,
+      "step": 80300
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004615997795162077,
+      "loss": 0.5227,
+      "step": 80310
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046155737878691516,
+      "loss": 0.5624,
+      "step": 80320
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004615149780576226,
+      "loss": 0.4982,
+      "step": 80330
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046147257732833004,
+      "loss": 0.522,
+      "step": 80340
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046143017659903756,
+      "loss": 0.5027,
+      "step": 80350
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.000461387775869745,
+      "loss": 0.5325,
+      "step": 80360
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046134537514045244,
+      "loss": 0.4683,
+      "step": 80370
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004613029744111599,
+      "loss": 0.5384,
+      "step": 80380
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004612605736818673,
+      "loss": 0.4719,
+      "step": 80390
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004612181729525748,
+      "loss": 0.512,
+      "step": 80400
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004611757722232823,
+      "loss": 0.5083,
+      "step": 80410
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004611333714939897,
+      "loss": 0.4943,
+      "step": 80420
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004610909707646972,
+      "loss": 0.5388,
+      "step": 80430
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046104857003540466,
+      "loss": 0.4786,
+      "step": 80440
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046100616930611207,
+      "loss": 0.5195,
+      "step": 80450
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046096376857681954,
+      "loss": 0.4849,
+      "step": 80460
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046092136784752706,
+      "loss": 0.4619,
+      "step": 80470
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004608789671182344,
+      "loss": 0.4356,
+      "step": 80480
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046083656638894194,
+      "loss": 0.4706,
+      "step": 80490
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046079416565964935,
+      "loss": 0.4908,
+      "step": 80500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004607517649303568,
+      "loss": 0.4665,
+      "step": 80510
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004607093642010643,
+      "loss": 0.5556,
+      "step": 80520
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004606669634717717,
+      "loss": 0.5441,
+      "step": 80530
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046062456274247916,
+      "loss": 0.4858,
+      "step": 80540
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004605821620131867,
+      "loss": 0.5106,
+      "step": 80550
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046053976128389404,
+      "loss": 0.5486,
+      "step": 80560
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046049736055460156,
+      "loss": 0.5059,
+      "step": 80570
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046045495982530903,
+      "loss": 0.3893,
+      "step": 80580
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046041255909601644,
+      "loss": 0.5148,
+      "step": 80590
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004603701583667239,
+      "loss": 0.5648,
+      "step": 80600
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046032775763743143,
+      "loss": 0.436,
+      "step": 80610
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004602853569081388,
+      "loss": 0.6451,
+      "step": 80620
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004602429561788463,
+      "loss": 0.5732,
+      "step": 80630
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004602005554495538,
+      "loss": 0.4868,
+      "step": 80640
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004601581547202612,
+      "loss": 0.5091,
+      "step": 80650
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046011575399096865,
+      "loss": 0.4533,
+      "step": 80660
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004600733532616762,
+      "loss": 0.4441,
+      "step": 80670
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046003095253238353,
+      "loss": 0.5597,
+      "step": 80680
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00045998855180309105,
+      "loss": 0.5628,
+      "step": 80690
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004599461510737985,
+      "loss": 0.5432,
+      "step": 80700
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00045990375034450593,
+      "loss": 0.5332,
+      "step": 80710
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004598613496152134,
+      "loss": 0.516,
+      "step": 80720
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004598189488859209,
+      "loss": 0.5998,
+      "step": 80730
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004597765481566283,
+      "loss": 0.5582,
+      "step": 80740
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004597341474273358,
+      "loss": 0.5704,
+      "step": 80750
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00045969174669804327,
+      "loss": 0.6038,
+      "step": 80760
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004596493459687507,
+      "loss": 0.4745,
+      "step": 80770
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00045960694523945815,
+      "loss": 0.4955,
+      "step": 80780
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004595645445101656,
+      "loss": 0.5732,
+      "step": 80790
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.000459522143780873,
+      "loss": 0.5424,
+      "step": 80800
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00045947974305158055,
+      "loss": 0.4858,
+      "step": 80810
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004594373423222879,
+      "loss": 0.5128,
+      "step": 80820
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00045939494159299543,
+      "loss": 0.5234,
+      "step": 80830
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004593525408637029,
+      "loss": 0.4326,
+      "step": 80840
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004593101401344103,
+      "loss": 0.496,
+      "step": 80850
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004592677394051178,
+      "loss": 0.5151,
+      "step": 80860
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004592253386758253,
+      "loss": 0.5356,
+      "step": 80870
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00045918293794653265,
+      "loss": 0.5105,
+      "step": 80880
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004591405372172402,
+      "loss": 0.4664,
+      "step": 80890
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00045909813648794764,
+      "loss": 0.4944,
+      "step": 80900
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00045905573575865505,
+      "loss": 0.4627,
+      "step": 80910
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004590133350293625,
+      "loss": 0.4559,
+      "step": 80920
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00045897093430007,
+      "loss": 0.5585,
+      "step": 80930
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004589285335707774,
+      "loss": 0.4767,
+      "step": 80940
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004588861328414849,
+      "loss": 0.5086,
+      "step": 80950
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004588437321121924,
+      "loss": 0.4738,
+      "step": 80960
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004588013313828998,
+      "loss": 0.553,
+      "step": 80970
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00045875893065360727,
+      "loss": 0.5742,
+      "step": 80980
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00045871652992431473,
+      "loss": 0.5178,
+      "step": 80990
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045867412919502215,
+      "loss": 0.4982,
+      "step": 81000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004586317284657296,
+      "loss": 0.4358,
+      "step": 81010
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045858932773643713,
+      "loss": 0.5442,
+      "step": 81020
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045854692700714455,
+      "loss": 0.5227,
+      "step": 81030
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.000458504526277852,
+      "loss": 0.4837,
+      "step": 81040
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004584621255485595,
+      "loss": 0.5415,
+      "step": 81050
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004584197248192669,
+      "loss": 0.5626,
+      "step": 81060
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045837732408997436,
+      "loss": 0.6067,
+      "step": 81070
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004583349233606819,
+      "loss": 0.4479,
+      "step": 81080
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045829252263138924,
+      "loss": 0.5558,
+      "step": 81090
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045825012190209676,
+      "loss": 0.6024,
+      "step": 81100
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004582077211728042,
+      "loss": 0.4793,
+      "step": 81110
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045816532044351164,
+      "loss": 0.4667,
+      "step": 81120
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004581229197142191,
+      "loss": 0.522,
+      "step": 81130
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004580805189849265,
+      "loss": 0.5487,
+      "step": 81140
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.000458038118255634,
+      "loss": 0.5486,
+      "step": 81150
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004579957175263415,
+      "loss": 0.5599,
+      "step": 81160
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004579533167970489,
+      "loss": 0.5028,
+      "step": 81170
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004579109160677564,
+      "loss": 0.4728,
+      "step": 81180
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045786851533846385,
+      "loss": 0.4658,
+      "step": 81190
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045782611460917126,
+      "loss": 0.5624,
+      "step": 81200
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045778371387987873,
+      "loss": 0.5132,
+      "step": 81210
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045774131315058625,
+      "loss": 0.5158,
+      "step": 81220
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004576989124212936,
+      "loss": 0.4703,
+      "step": 81230
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045765651169200113,
+      "loss": 0.5479,
+      "step": 81240
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004576141109627086,
+      "loss": 0.5464,
+      "step": 81250
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.000457571710233416,
+      "loss": 0.457,
+      "step": 81260
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004575293095041235,
+      "loss": 0.5089,
+      "step": 81270
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.000457486908774831,
+      "loss": 0.5214,
+      "step": 81280
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045744450804553836,
+      "loss": 0.5,
+      "step": 81290
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004574021073162459,
+      "loss": 0.4528,
+      "step": 81300
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045735970658695335,
+      "loss": 0.474,
+      "step": 81310
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045731730585766076,
+      "loss": 0.4784,
+      "step": 81320
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004572749051283682,
+      "loss": 0.5163,
+      "step": 81330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045723250439907575,
+      "loss": 0.4624,
+      "step": 81340
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004571901036697831,
+      "loss": 0.549,
+      "step": 81350
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004571477029404906,
+      "loss": 0.5126,
+      "step": 81360
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004571053022111981,
+      "loss": 0.5391,
+      "step": 81370
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004570629014819055,
+      "loss": 0.5185,
+      "step": 81380
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045702050075261297,
+      "loss": 0.5017,
+      "step": 81390
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004569781000233205,
+      "loss": 0.5432,
+      "step": 81400
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045693569929402785,
+      "loss": 0.5678,
+      "step": 81410
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045689329856473537,
+      "loss": 0.5031,
+      "step": 81420
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045685089783544273,
+      "loss": 0.4915,
+      "step": 81430
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045680849710615025,
+      "loss": 0.5898,
+      "step": 81440
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004567660963768577,
+      "loss": 0.5806,
+      "step": 81450
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045672369564756513,
+      "loss": 0.5183,
+      "step": 81460
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004566812949182726,
+      "loss": 0.5584,
+      "step": 81470
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004566388941889801,
+      "loss": 0.5242,
+      "step": 81480
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004565964934596875,
+      "loss": 0.5042,
+      "step": 81490
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.000456554092730395,
+      "loss": 0.5844,
+      "step": 81500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045651169200110246,
+      "loss": 0.5123,
+      "step": 81510
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004564692912718099,
+      "loss": 0.5625,
+      "step": 81520
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045642689054251734,
+      "loss": 0.4909,
+      "step": 81530
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004563844898132248,
+      "loss": 0.5326,
+      "step": 81540
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004563420890839322,
+      "loss": 0.5314,
+      "step": 81550
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045629968835463974,
+      "loss": 0.5879,
+      "step": 81560
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004562572876253472,
+      "loss": 0.5022,
+      "step": 81570
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004562148868960546,
+      "loss": 0.5684,
+      "step": 81580
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004561724861667621,
+      "loss": 0.4716,
+      "step": 81590
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045613008543746956,
+      "loss": 0.4866,
+      "step": 81600
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045608768470817697,
+      "loss": 0.528,
+      "step": 81610
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045604528397888444,
+      "loss": 0.5541,
+      "step": 81620
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045600288324959196,
+      "loss": 0.5,
+      "step": 81630
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045596048252029937,
+      "loss": 0.5643,
+      "step": 81640
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045591808179100684,
+      "loss": 0.5087,
+      "step": 81650
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004558756810617143,
+      "loss": 0.5993,
+      "step": 81660
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004558332803324217,
+      "loss": 0.52,
+      "step": 81670
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004557908796031292,
+      "loss": 0.4685,
+      "step": 81680
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004557484788738367,
+      "loss": 0.4724,
+      "step": 81690
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004557060781445441,
+      "loss": 0.5762,
+      "step": 81700
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004556636774152516,
+      "loss": 0.495,
+      "step": 81710
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045562127668595905,
+      "loss": 0.4559,
+      "step": 81720
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045557887595666646,
+      "loss": 0.5523,
+      "step": 81730
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045553647522737393,
+      "loss": 0.5143,
+      "step": 81740
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045549407449808134,
+      "loss": 0.504,
+      "step": 81750
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004554516737687888,
+      "loss": 0.5626,
+      "step": 81760
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045540927303949633,
+      "loss": 0.4716,
+      "step": 81770
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045536687231020374,
+      "loss": 0.63,
+      "step": 81780
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004553244715809112,
+      "loss": 0.539,
+      "step": 81790
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004552820708516187,
+      "loss": 0.4813,
+      "step": 81800
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004552396701223261,
+      "loss": 0.5638,
+      "step": 81810
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045519726939303356,
+      "loss": 0.5177,
+      "step": 81820
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004551548686637411,
+      "loss": 0.5292,
+      "step": 81830
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045511246793444843,
+      "loss": 0.6428,
+      "step": 81840
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045507006720515596,
+      "loss": 0.5405,
+      "step": 81850
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004550276664758634,
+      "loss": 0.5132,
+      "step": 81860
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045498526574657083,
+      "loss": 0.5864,
+      "step": 81870
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004549428650172783,
+      "loss": 0.4627,
+      "step": 81880
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004549004642879858,
+      "loss": 0.6501,
+      "step": 81890
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004548580635586932,
+      "loss": 0.4923,
+      "step": 81900
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004548156628294007,
+      "loss": 0.5172,
+      "step": 81910
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045477326210010817,
+      "loss": 0.5591,
+      "step": 81920
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004547308613708156,
+      "loss": 0.577,
+      "step": 81930
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045468846064152305,
+      "loss": 0.6231,
+      "step": 81940
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045464605991223057,
+      "loss": 0.5523,
+      "step": 81950
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045460365918293793,
+      "loss": 0.5714,
+      "step": 81960
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045456125845364545,
+      "loss": 0.4711,
+      "step": 81970
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004545188577243529,
+      "loss": 0.5696,
+      "step": 81980
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045447645699506033,
+      "loss": 0.476,
+      "step": 81990
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004544340562657678,
+      "loss": 0.5441,
+      "step": 82000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004543916555364753,
+      "loss": 0.5386,
+      "step": 82010
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004543492548071827,
+      "loss": 0.4634,
+      "step": 82020
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004543068540778902,
+      "loss": 0.5125,
+      "step": 82030
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045426445334859755,
+      "loss": 0.4412,
+      "step": 82040
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004542220526193051,
+      "loss": 0.5387,
+      "step": 82050
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045417965189001254,
+      "loss": 0.5798,
+      "step": 82060
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045413725116071995,
+      "loss": 0.4738,
+      "step": 82070
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004540948504314274,
+      "loss": 0.4823,
+      "step": 82080
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045405244970213494,
+      "loss": 0.5516,
+      "step": 82090
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004540100489728423,
+      "loss": 0.5518,
+      "step": 82100
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004539676482435498,
+      "loss": 0.4681,
+      "step": 82110
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004539252475142573,
+      "loss": 0.5436,
+      "step": 82120
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004538828467849647,
+      "loss": 0.5578,
+      "step": 82130
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045384044605567217,
+      "loss": 0.4621,
+      "step": 82140
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004537980453263797,
+      "loss": 0.5388,
+      "step": 82150
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045375564459708705,
+      "loss": 0.5444,
+      "step": 82160
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045371324386779457,
+      "loss": 0.4769,
+      "step": 82170
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045367084313850203,
+      "loss": 0.5739,
+      "step": 82180
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045362844240920945,
+      "loss": 0.5447,
+      "step": 82190
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004535860416799169,
+      "loss": 0.5437,
+      "step": 82200
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004535436409506244,
+      "loss": 0.555,
+      "step": 82210
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004535012402213318,
+      "loss": 0.4813,
+      "step": 82220
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004534588394920393,
+      "loss": 0.532,
+      "step": 82230
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004534164387627468,
+      "loss": 0.5235,
+      "step": 82240
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004533740380334542,
+      "loss": 0.5056,
+      "step": 82250
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045333163730416166,
+      "loss": 0.5727,
+      "step": 82260
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004532892365748691,
+      "loss": 0.5287,
+      "step": 82270
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045324683584557654,
+      "loss": 0.4592,
+      "step": 82280
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.000453204435116284,
+      "loss": 0.6155,
+      "step": 82290
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045316203438699153,
+      "loss": 0.5061,
+      "step": 82300
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045311963365769894,
+      "loss": 0.5835,
+      "step": 82310
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004530772329284064,
+      "loss": 0.5108,
+      "step": 82320
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004530348321991139,
+      "loss": 0.5279,
+      "step": 82330
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004529924314698213,
+      "loss": 0.4711,
+      "step": 82340
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045295003074052875,
+      "loss": 0.4996,
+      "step": 82350
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045290763001123617,
+      "loss": 0.5757,
+      "step": 82360
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045286522928194363,
+      "loss": 0.4761,
+      "step": 82370
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045282282855265115,
+      "loss": 0.5496,
+      "step": 82380
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045278042782335857,
+      "loss": 0.4844,
+      "step": 82390
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045273802709406603,
+      "loss": 0.5401,
+      "step": 82400
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004526956263647735,
+      "loss": 0.5302,
+      "step": 82410
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004526532256354809,
+      "loss": 0.5183,
+      "step": 82420
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004526108249061884,
+      "loss": 0.5641,
+      "step": 82430
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004525684241768959,
+      "loss": 0.4539,
+      "step": 82440
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004525260234476033,
+      "loss": 0.5785,
+      "step": 82450
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004524836227183108,
+      "loss": 0.5916,
+      "step": 82460
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045244122198901825,
+      "loss": 0.5579,
+      "step": 82470
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045239882125972566,
+      "loss": 0.5276,
+      "step": 82480
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004523564205304331,
+      "loss": 0.4613,
+      "step": 82490
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045231401980114065,
+      "loss": 0.5174,
+      "step": 82500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000452271619071848,
+      "loss": 0.5421,
+      "step": 82510
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004522292183425555,
+      "loss": 0.5778,
+      "step": 82520
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000452186817613263,
+      "loss": 0.4922,
+      "step": 82530
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004521444168839704,
+      "loss": 0.5214,
+      "step": 82540
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045210201615467787,
+      "loss": 0.5612,
+      "step": 82550
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004520596154253854,
+      "loss": 0.5043,
+      "step": 82560
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045201721469609275,
+      "loss": 0.6314,
+      "step": 82570
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045197481396680027,
+      "loss": 0.6151,
+      "step": 82580
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045193241323750774,
+      "loss": 0.5662,
+      "step": 82590
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045189001250821515,
+      "loss": 0.5599,
+      "step": 82600
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004518476117789226,
+      "loss": 0.4748,
+      "step": 82610
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045180521104963014,
+      "loss": 0.5501,
+      "step": 82620
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004517628103203375,
+      "loss": 0.58,
+      "step": 82630
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000451720409591045,
+      "loss": 0.514,
+      "step": 82640
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004516780088617524,
+      "loss": 0.5529,
+      "step": 82650
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004516356081324599,
+      "loss": 0.5298,
+      "step": 82660
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045159320740316736,
+      "loss": 0.5562,
+      "step": 82670
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004515508066738748,
+      "loss": 0.5103,
+      "step": 82680
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045150840594458224,
+      "loss": 0.5277,
+      "step": 82690
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045146600521528977,
+      "loss": 0.5476,
+      "step": 82700
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004514236044859971,
+      "loss": 0.4783,
+      "step": 82710
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045138120375670464,
+      "loss": 0.5444,
+      "step": 82720
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004513388030274121,
+      "loss": 0.5361,
+      "step": 82730
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004512964022981195,
+      "loss": 0.6381,
+      "step": 82740
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.000451254001568827,
+      "loss": 0.5046,
+      "step": 82750
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004512116008395345,
+      "loss": 0.4788,
+      "step": 82760
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045116920011024187,
+      "loss": 0.4343,
+      "step": 82770
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004511267993809494,
+      "loss": 0.5258,
+      "step": 82780
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045108439865165686,
+      "loss": 0.5609,
+      "step": 82790
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045104199792236427,
+      "loss": 0.4592,
+      "step": 82800
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045099959719307174,
+      "loss": 0.5579,
+      "step": 82810
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004509571964637792,
+      "loss": 0.5347,
+      "step": 82820
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004509147957344866,
+      "loss": 0.5186,
+      "step": 82830
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045087239500519414,
+      "loss": 0.5301,
+      "step": 82840
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004508299942759016,
+      "loss": 0.5299,
+      "step": 82850
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.000450787593546609,
+      "loss": 0.5089,
+      "step": 82860
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004507451928173165,
+      "loss": 0.4958,
+      "step": 82870
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045070279208802395,
+      "loss": 0.5328,
+      "step": 82880
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045066039135873136,
+      "loss": 0.4385,
+      "step": 82890
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004506179906294389,
+      "loss": 0.5842,
+      "step": 82900
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045057558990014635,
+      "loss": 0.4933,
+      "step": 82910
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045053318917085376,
+      "loss": 0.5542,
+      "step": 82920
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045049078844156123,
+      "loss": 0.6322,
+      "step": 82930
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004504483877122687,
+      "loss": 0.6365,
+      "step": 82940
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004504059869829761,
+      "loss": 0.5494,
+      "step": 82950
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004503635862536836,
+      "loss": 0.501,
+      "step": 82960
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.000450321185524391,
+      "loss": 0.6392,
+      "step": 82970
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004502787847950985,
+      "loss": 0.5053,
+      "step": 82980
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.000450236384065806,
+      "loss": 0.5539,
+      "step": 82990
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004501939833365134,
+      "loss": 0.5205,
+      "step": 83000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045015158260722086,
+      "loss": 0.6119,
+      "step": 83010
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004501091818779283,
+      "loss": 0.5717,
+      "step": 83020
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045006678114863574,
+      "loss": 0.5324,
+      "step": 83030
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004500243804193432,
+      "loss": 0.4644,
+      "step": 83040
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004499819796900507,
+      "loss": 0.5668,
+      "step": 83050
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00044993957896075814,
+      "loss": 0.5465,
+      "step": 83060
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004498971782314656,
+      "loss": 0.5275,
+      "step": 83070
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00044985477750217307,
+      "loss": 0.4499,
+      "step": 83080
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004498123767728805,
+      "loss": 0.5026,
+      "step": 83090
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00044976997604358795,
+      "loss": 0.5261,
+      "step": 83100
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00044972757531429547,
+      "loss": 0.4873,
+      "step": 83110
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044968517458500283,
+      "loss": 0.5636,
+      "step": 83120
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044964277385571035,
+      "loss": 0.4283,
+      "step": 83130
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004496003731264178,
+      "loss": 0.4693,
+      "step": 83140
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044955797239712523,
+      "loss": 0.5161,
+      "step": 83150
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004495155716678327,
+      "loss": 0.6137,
+      "step": 83160
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004494731709385402,
+      "loss": 0.5253,
+      "step": 83170
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004494307702092476,
+      "loss": 0.5435,
+      "step": 83180
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004493883694799551,
+      "loss": 0.5428,
+      "step": 83190
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044934596875066256,
+      "loss": 0.5358,
+      "step": 83200
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044930356802137,
+      "loss": 0.5268,
+      "step": 83210
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044926116729207744,
+      "loss": 0.5296,
+      "step": 83220
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044921876656278496,
+      "loss": 0.5871,
+      "step": 83230
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004491763658334923,
+      "loss": 0.4401,
+      "step": 83240
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044913396510419984,
+      "loss": 0.5203,
+      "step": 83250
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004490915643749073,
+      "loss": 0.56,
+      "step": 83260
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004490491636456147,
+      "loss": 0.4519,
+      "step": 83270
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004490067629163222,
+      "loss": 0.5871,
+      "step": 83280
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004489643621870296,
+      "loss": 0.4771,
+      "step": 83290
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044892196145773707,
+      "loss": 0.4497,
+      "step": 83300
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004488795607284446,
+      "loss": 0.5287,
+      "step": 83310
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044883715999915195,
+      "loss": 0.5091,
+      "step": 83320
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044879475926985947,
+      "loss": 0.5507,
+      "step": 83330
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044875235854056694,
+      "loss": 0.5586,
+      "step": 83340
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044870995781127435,
+      "loss": 0.6183,
+      "step": 83350
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004486675570819818,
+      "loss": 0.4532,
+      "step": 83360
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00044862515635268934,
+      "loss": 0.6164,
+      "step": 83370
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004485827556233967,
+      "loss": 0.5126,
+      "step": 83380
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004485403548941042,
+      "loss": 0.5413,
+      "step": 83390
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004484979541648117,
+      "loss": 0.4547,
+      "step": 83400
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004484555534355191,
+      "loss": 0.4671,
+      "step": 83410
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00044841315270622656,
+      "loss": 0.505,
+      "step": 83420
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004483707519769341,
+      "loss": 0.5613,
+      "step": 83430
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00044832835124764144,
+      "loss": 0.5112,
+      "step": 83440
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00044828595051834896,
+      "loss": 0.5131,
+      "step": 83450
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00044824354978905643,
+      "loss": 0.5919,
+      "step": 83460
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00044820114905976384,
+      "loss": 0.6508,
+      "step": 83470
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004481587483304713,
+      "loss": 0.5672,
+      "step": 83480
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004481163476011788,
+      "loss": 0.5671,
+      "step": 83490
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004480739468718862,
+      "loss": 0.5119,
+      "step": 83500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004480315461425937,
+      "loss": 0.53,
+      "step": 83510
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004479891454133012,
+      "loss": 0.4594,
+      "step": 83520
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004479467446840086,
+      "loss": 0.5048,
+      "step": 83530
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00044790434395471605,
+      "loss": 0.5263,
+      "step": 83540
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004478619432254235,
+      "loss": 0.5455,
+      "step": 83550
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00044781954249613093,
+      "loss": 0.5206,
+      "step": 83560
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004477771417668384,
+      "loss": 0.4713,
+      "step": 83570
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004477347410375458,
+      "loss": 0.4864,
+      "step": 83580
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00044769234030825333,
+      "loss": 0.6233,
+      "step": 83590
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004476499395789608,
+      "loss": 0.6024,
+      "step": 83600
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004476075388496682,
+      "loss": 0.6078,
+      "step": 83610
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004475651381203757,
+      "loss": 0.5134,
+      "step": 83620
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00044752273739108315,
+      "loss": 0.5147,
+      "step": 83630
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00044748033666179056,
+      "loss": 0.5741,
+      "step": 83640
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004474379359324981,
+      "loss": 0.4602,
+      "step": 83650
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00044739553520320555,
+      "loss": 0.5221,
+      "step": 83660
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00044735313447391296,
+      "loss": 0.4531,
+      "step": 83670
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004473107337446204,
+      "loss": 0.4976,
+      "step": 83680
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004472683330153279,
+      "loss": 0.5993,
+      "step": 83690
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004472259322860353,
+      "loss": 0.4825,
+      "step": 83700
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00044718353155674277,
+      "loss": 0.6312,
+      "step": 83710
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004471411308274503,
+      "loss": 0.5747,
+      "step": 83720
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004470987300981577,
+      "loss": 0.5228,
+      "step": 83730
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004470563293688652,
+      "loss": 0.5486,
+      "step": 83740
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00044701392863957264,
+      "loss": 0.5338,
+      "step": 83750
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00044697152791028005,
+      "loss": 0.4703,
+      "step": 83760
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004469291271809875,
+      "loss": 0.5432,
+      "step": 83770
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00044688672645169504,
+      "loss": 0.542,
+      "step": 83780
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004468443257224024,
+      "loss": 0.5322,
+      "step": 83790
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004468019249931099,
+      "loss": 0.4817,
+      "step": 83800
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004467595242638174,
+      "loss": 0.4988,
+      "step": 83810
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004467171235345248,
+      "loss": 0.4825,
+      "step": 83820
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044667472280523227,
+      "loss": 0.4934,
+      "step": 83830
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004466323220759398,
+      "loss": 0.5376,
+      "step": 83840
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044658992134664714,
+      "loss": 0.5794,
+      "step": 83850
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044654752061735467,
+      "loss": 0.5744,
+      "step": 83860
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044650511988806213,
+      "loss": 0.5188,
+      "step": 83870
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044646271915876955,
+      "loss": 0.5872,
+      "step": 83880
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.000446420318429477,
+      "loss": 0.5117,
+      "step": 83890
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004463779177001844,
+      "loss": 0.5524,
+      "step": 83900
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004463355169708919,
+      "loss": 0.5982,
+      "step": 83910
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004462931162415994,
+      "loss": 0.5329,
+      "step": 83920
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044625071551230677,
+      "loss": 0.443,
+      "step": 83930
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004462083147830143,
+      "loss": 0.5514,
+      "step": 83940
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044616591405372176,
+      "loss": 0.5228,
+      "step": 83950
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044612351332442917,
+      "loss": 0.5166,
+      "step": 83960
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044608111259513664,
+      "loss": 0.505,
+      "step": 83970
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044603871186584416,
+      "loss": 0.4763,
+      "step": 83980
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004459963111365515,
+      "loss": 0.4607,
+      "step": 83990
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044595391040725904,
+      "loss": 0.5326,
+      "step": 84000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004459115096779665,
+      "loss": 0.5692,
+      "step": 84010
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004458691089486739,
+      "loss": 0.4286,
+      "step": 84020
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004458267082193814,
+      "loss": 0.4385,
+      "step": 84030
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004457843074900889,
+      "loss": 0.4507,
+      "step": 84040
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044574190676079626,
+      "loss": 0.4707,
+      "step": 84050
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004456995060315038,
+      "loss": 0.3809,
+      "step": 84060
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044565710530221125,
+      "loss": 0.5548,
+      "step": 84070
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044561470457291866,
+      "loss": 0.5125,
+      "step": 84080
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044557230384362613,
+      "loss": 0.5263,
+      "step": 84090
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004455299031143336,
+      "loss": 0.5338,
+      "step": 84100
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.000445487502385041,
+      "loss": 0.5481,
+      "step": 84110
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044544510165574853,
+      "loss": 0.5074,
+      "step": 84120
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.000445402700926456,
+      "loss": 0.5417,
+      "step": 84130
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004453603001971634,
+      "loss": 0.5721,
+      "step": 84140
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004453178994678709,
+      "loss": 0.5455,
+      "step": 84150
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044527549873857834,
+      "loss": 0.5335,
+      "step": 84160
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044523309800928576,
+      "loss": 0.5326,
+      "step": 84170
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004451906972799933,
+      "loss": 0.5758,
+      "step": 84180
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044514829655070064,
+      "loss": 0.4561,
+      "step": 84190
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044510589582140816,
+      "loss": 0.575,
+      "step": 84200
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004450634950921156,
+      "loss": 0.4696,
+      "step": 84210
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044502109436282304,
+      "loss": 0.5284,
+      "step": 84220
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004449786936335305,
+      "loss": 0.4735,
+      "step": 84230
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044493629290423797,
+      "loss": 0.5021,
+      "step": 84240
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004448938921749454,
+      "loss": 0.542,
+      "step": 84250
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004448514914456529,
+      "loss": 0.5449,
+      "step": 84260
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044480909071636037,
+      "loss": 0.6105,
+      "step": 84270
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004447666899870678,
+      "loss": 0.4518,
+      "step": 84280
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044472428925777525,
+      "loss": 0.5018,
+      "step": 84290
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004446818885284827,
+      "loss": 0.4776,
+      "step": 84300
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044463948779919013,
+      "loss": 0.522,
+      "step": 84310
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004445970870698976,
+      "loss": 0.5755,
+      "step": 84320
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004445546863406051,
+      "loss": 0.493,
+      "step": 84330
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044451228561131253,
+      "loss": 0.5082,
+      "step": 84340
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044446988488202,
+      "loss": 0.4678,
+      "step": 84350
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044442748415272746,
+      "loss": 0.5638,
+      "step": 84360
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004443850834234349,
+      "loss": 0.5563,
+      "step": 84370
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044434268269414234,
+      "loss": 0.4749,
+      "step": 84380
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044430028196484986,
+      "loss": 0.558,
+      "step": 84390
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004442578812355573,
+      "loss": 0.5229,
+      "step": 84400
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044421548050626474,
+      "loss": 0.679,
+      "step": 84410
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004441730797769722,
+      "loss": 0.6506,
+      "step": 84420
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004441306790476796,
+      "loss": 0.5528,
+      "step": 84430
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004440882783183871,
+      "loss": 0.5434,
+      "step": 84440
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004440458775890946,
+      "loss": 0.4858,
+      "step": 84450
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044400347685980197,
+      "loss": 0.4966,
+      "step": 84460
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004439610761305095,
+      "loss": 0.5439,
+      "step": 84470
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044391867540121696,
+      "loss": 0.441,
+      "step": 84480
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044387627467192437,
+      "loss": 0.5547,
+      "step": 84490
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044383387394263184,
+      "loss": 0.5049,
+      "step": 84500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044379147321333925,
+      "loss": 0.5213,
+      "step": 84510
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004437490724840467,
+      "loss": 0.5028,
+      "step": 84520
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044370667175475424,
+      "loss": 0.5814,
+      "step": 84530
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0004436642710254616,
+      "loss": 0.5443,
+      "step": 84540
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0004436218702961691,
+      "loss": 0.4903,
+      "step": 84550
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0004435794695668766,
+      "loss": 0.5407,
+      "step": 84560
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000443537068837584,
+      "loss": 0.4883,
+      "step": 84570
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044349466810829146,
+      "loss": 0.5595,
+      "step": 84580
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000443452267378999,
+      "loss": 0.4469,
+      "step": 84590
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044340986664970634,
+      "loss": 0.4961,
+      "step": 84600
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044336746592041386,
+      "loss": 0.5048,
+      "step": 84610
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044332506519112133,
+      "loss": 0.5406,
+      "step": 84620
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044328266446182874,
+      "loss": 0.4807,
+      "step": 84630
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0004432402637325362,
+      "loss": 0.4999,
+      "step": 84640
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044319786300324373,
+      "loss": 0.5355,
+      "step": 84650
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0004431554622739511,
+      "loss": 0.5461,
+      "step": 84660
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0004431130615446586,
+      "loss": 0.5225,
+      "step": 84670
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0004430706608153661,
+      "loss": 0.6021,
+      "step": 84680
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0004430282600860735,
+      "loss": 0.5528,
+      "step": 84690
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044298585935678095,
+      "loss": 0.6018,
+      "step": 84700
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0004429434586274885,
+      "loss": 0.4394,
+      "step": 84710
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044290105789819583,
+      "loss": 0.5551,
+      "step": 84720
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044285865716890336,
+      "loss": 0.4662,
+      "step": 84730
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0004428162564396108,
+      "loss": 0.5123,
+      "step": 84740
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044277385571031823,
+      "loss": 0.4977,
+      "step": 84750
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0004427314549810257,
+      "loss": 0.6125,
+      "step": 84760
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044268905425173317,
+      "loss": 0.5126,
+      "step": 84770
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004426466535224406,
+      "loss": 0.5547,
+      "step": 84780
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004426042527931481,
+      "loss": 0.577,
+      "step": 84790
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044256185206385546,
+      "loss": 0.6022,
+      "step": 84800
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.000442519451334563,
+      "loss": 0.4909,
+      "step": 84810
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044247705060527045,
+      "loss": 0.5884,
+      "step": 84820
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044243464987597786,
+      "loss": 0.5631,
+      "step": 84830
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004423922491466853,
+      "loss": 0.4358,
+      "step": 84840
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004423498484173928,
+      "loss": 0.4794,
+      "step": 84850
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004423074476881002,
+      "loss": 0.6476,
+      "step": 84860
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044226504695880773,
+      "loss": 0.414,
+      "step": 84870
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004422226462295152,
+      "loss": 0.6074,
+      "step": 84880
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004421802455002226,
+      "loss": 0.542,
+      "step": 84890
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004421378447709301,
+      "loss": 0.5114,
+      "step": 84900
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044209544404163754,
+      "loss": 0.5425,
+      "step": 84910
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044205304331234495,
+      "loss": 0.4501,
+      "step": 84920
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004420106425830525,
+      "loss": 0.474,
+      "step": 84930
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044196824185375994,
+      "loss": 0.4763,
+      "step": 84940
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044192584112446735,
+      "loss": 0.5761,
+      "step": 84950
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004418834403951748,
+      "loss": 0.6129,
+      "step": 84960
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004418410396658823,
+      "loss": 0.581,
+      "step": 84970
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004417986389365897,
+      "loss": 0.5368,
+      "step": 84980
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044175623820729717,
+      "loss": 0.5618,
+      "step": 84990
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004417138374780047,
+      "loss": 0.6195,
+      "step": 85000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004416714367487121,
+      "loss": 0.5983,
+      "step": 85010
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044162903601941957,
+      "loss": 0.516,
+      "step": 85020
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044158663529012703,
+      "loss": 0.4342,
+      "step": 85030
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044154423456083445,
+      "loss": 0.5415,
+      "step": 85040
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004415018338315419,
+      "loss": 0.5459,
+      "step": 85050
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044145943310224943,
+      "loss": 0.616,
+      "step": 85060
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004414170323729568,
+      "loss": 0.4723,
+      "step": 85070
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004413746316436643,
+      "loss": 0.4865,
+      "step": 85080
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004413322309143718,
+      "loss": 0.4965,
+      "step": 85090
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004412898301850792,
+      "loss": 0.497,
+      "step": 85100
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044124742945578666,
+      "loss": 0.5738,
+      "step": 85110
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044120502872649407,
+      "loss": 0.4881,
+      "step": 85120
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044116262799720154,
+      "loss": 0.4178,
+      "step": 85130
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044112022726790906,
+      "loss": 0.5749,
+      "step": 85140
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044107782653861647,
+      "loss": 0.5208,
+      "step": 85150
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044103542580932394,
+      "loss": 0.5684,
+      "step": 85160
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004409930250800314,
+      "loss": 0.4905,
+      "step": 85170
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004409506243507388,
+      "loss": 0.5708,
+      "step": 85180
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004409082236214463,
+      "loss": 0.5222,
+      "step": 85190
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004408658228921538,
+      "loss": 0.5718,
+      "step": 85200
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044082342216286116,
+      "loss": 0.5746,
+      "step": 85210
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004407810214335687,
+      "loss": 0.5455,
+      "step": 85220
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044073862070427615,
+      "loss": 0.5344,
+      "step": 85230
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044069621997498356,
+      "loss": 0.5264,
+      "step": 85240
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044065381924569103,
+      "loss": 0.4885,
+      "step": 85250
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044061141851639855,
+      "loss": 0.4787,
+      "step": 85260
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004405690177871059,
+      "loss": 0.4727,
+      "step": 85270
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044052661705781343,
+      "loss": 0.5517,
+      "step": 85280
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004404842163285209,
+      "loss": 0.4539,
+      "step": 85290
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004404418155992283,
+      "loss": 0.5866,
+      "step": 85300
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004403994148699358,
+      "loss": 0.5412,
+      "step": 85310
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004403570141406433,
+      "loss": 0.5663,
+      "step": 85320
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044031461341135066,
+      "loss": 0.5476,
+      "step": 85330
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004402722126820582,
+      "loss": 0.4708,
+      "step": 85340
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044022981195276565,
+      "loss": 0.5706,
+      "step": 85350
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044018741122347306,
+      "loss": 0.4997,
+      "step": 85360
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004401450104941805,
+      "loss": 0.5027,
+      "step": 85370
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.000440102609764888,
+      "loss": 0.5442,
+      "step": 85380
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004400602090355954,
+      "loss": 0.604,
+      "step": 85390
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004400178083063029,
+      "loss": 0.5525,
+      "step": 85400
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004399754075770104,
+      "loss": 0.6223,
+      "step": 85410
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004399330068477178,
+      "loss": 0.4008,
+      "step": 85420
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00043989060611842527,
+      "loss": 0.4679,
+      "step": 85430
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004398482053891327,
+      "loss": 0.5575,
+      "step": 85440
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00043980580465984015,
+      "loss": 0.4792,
+      "step": 85450
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00043976340393054767,
+      "loss": 0.4965,
+      "step": 85460
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00043972100320125503,
+      "loss": 0.5742,
+      "step": 85470
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00043967860247196255,
+      "loss": 0.504,
+      "step": 85480
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00043963620174267,
+      "loss": 0.5266,
+      "step": 85490
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00043959380101337743,
+      "loss": 0.5285,
+      "step": 85500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004395514002840849,
+      "loss": 0.495,
+      "step": 85510
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00043950899955479236,
+      "loss": 0.5104,
+      "step": 85520
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004394665988254998,
+      "loss": 0.5056,
+      "step": 85530
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004394241980962073,
+      "loss": 0.5206,
+      "step": 85540
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00043938179736691476,
+      "loss": 0.4619,
+      "step": 85550
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004393393966376222,
+      "loss": 0.5217,
+      "step": 85560
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00043929699590832964,
+      "loss": 0.535,
+      "step": 85570
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004392545951790371,
+      "loss": 0.5163,
+      "step": 85580
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004392121944497445,
+      "loss": 0.5574,
+      "step": 85590
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.000439169793720452,
+      "loss": 0.5208,
+      "step": 85600
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004391273929911595,
+      "loss": 0.5403,
+      "step": 85610
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004390849922618669,
+      "loss": 0.5184,
+      "step": 85620
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004390425915325744,
+      "loss": 0.5457,
+      "step": 85630
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00043900019080328186,
+      "loss": 0.5083,
+      "step": 85640
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00043895779007398927,
+      "loss": 0.6399,
+      "step": 85650
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00043891538934469674,
+      "loss": 0.5405,
+      "step": 85660
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00043887298861540426,
+      "loss": 0.5363,
+      "step": 85670
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00043883058788611167,
+      "loss": 0.5443,
+      "step": 85680
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00043878818715681914,
+      "loss": 0.5858,
+      "step": 85690
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004387457864275266,
+      "loss": 0.4625,
+      "step": 85700
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.000438703385698234,
+      "loss": 0.5875,
+      "step": 85710
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004386609849689415,
+      "loss": 0.5467,
+      "step": 85720
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004386185842396489,
+      "loss": 0.5551,
+      "step": 85730
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043857618351035636,
+      "loss": 0.4893,
+      "step": 85740
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004385337827810639,
+      "loss": 0.5175,
+      "step": 85750
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004384913820517713,
+      "loss": 0.5652,
+      "step": 85760
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043844898132247876,
+      "loss": 0.4485,
+      "step": 85770
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043840658059318623,
+      "loss": 0.5182,
+      "step": 85780
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043836417986389364,
+      "loss": 0.6477,
+      "step": 85790
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004383217791346011,
+      "loss": 0.5421,
+      "step": 85800
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043827937840530863,
+      "loss": 0.5363,
+      "step": 85810
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.000438236977676016,
+      "loss": 0.5127,
+      "step": 85820
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004381945769467235,
+      "loss": 0.4745,
+      "step": 85830
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.000438152176217431,
+      "loss": 0.5126,
+      "step": 85840
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004381097754881384,
+      "loss": 0.5022,
+      "step": 85850
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043806737475884586,
+      "loss": 0.5648,
+      "step": 85860
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004380249740295534,
+      "loss": 0.567,
+      "step": 85870
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043798257330026073,
+      "loss": 0.5832,
+      "step": 85880
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043794017257096826,
+      "loss": 0.5353,
+      "step": 85890
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004378977718416757,
+      "loss": 0.512,
+      "step": 85900
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043785537111238314,
+      "loss": 0.5716,
+      "step": 85910
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004378129703830906,
+      "loss": 0.5152,
+      "step": 85920
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004377705696537981,
+      "loss": 0.5015,
+      "step": 85930
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004377281689245055,
+      "loss": 0.4709,
+      "step": 85940
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.000437685768195213,
+      "loss": 0.4795,
+      "step": 85950
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043764336746592047,
+      "loss": 0.5041,
+      "step": 85960
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004376009667366279,
+      "loss": 0.4605,
+      "step": 85970
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043755856600733535,
+      "loss": 0.5105,
+      "step": 85980
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043751616527804287,
+      "loss": 0.5354,
+      "step": 85990
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043747376454875023,
+      "loss": 0.5571,
+      "step": 86000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043743136381945775,
+      "loss": 0.5551,
+      "step": 86010
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004373889630901652,
+      "loss": 0.4443,
+      "step": 86020
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043734656236087263,
+      "loss": 0.5295,
+      "step": 86030
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004373041616315801,
+      "loss": 0.5739,
+      "step": 86040
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004372617609022875,
+      "loss": 0.5418,
+      "step": 86050
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.000437219360172995,
+      "loss": 0.4685,
+      "step": 86060
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004371769594437025,
+      "loss": 0.4347,
+      "step": 86070
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043713455871440985,
+      "loss": 0.528,
+      "step": 86080
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004370921579851174,
+      "loss": 0.5047,
+      "step": 86090
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043704975725582484,
+      "loss": 0.5114,
+      "step": 86100
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043700735652653225,
+      "loss": 0.5152,
+      "step": 86110
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004369649557972397,
+      "loss": 0.6092,
+      "step": 86120
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004369225550679472,
+      "loss": 0.5384,
+      "step": 86130
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004368801543386546,
+      "loss": 0.5526,
+      "step": 86140
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004368377536093621,
+      "loss": 0.5762,
+      "step": 86150
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004367953528800696,
+      "loss": 0.5755,
+      "step": 86160
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.000436752952150777,
+      "loss": 0.4948,
+      "step": 86170
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043671055142148447,
+      "loss": 0.6771,
+      "step": 86180
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043666815069219193,
+      "loss": 0.5445,
+      "step": 86190
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043662574996289935,
+      "loss": 0.5199,
+      "step": 86200
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043658334923360687,
+      "loss": 0.5837,
+      "step": 86210
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043654094850431433,
+      "loss": 0.5741,
+      "step": 86220
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043649854777502175,
+      "loss": 0.5084,
+      "step": 86230
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004364561470457292,
+      "loss": 0.568,
+      "step": 86240
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004364137463164367,
+      "loss": 0.5,
+      "step": 86250
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004363713455871441,
+      "loss": 0.5629,
+      "step": 86260
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043632894485785156,
+      "loss": 0.534,
+      "step": 86270
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004362865441285591,
+      "loss": 0.4145,
+      "step": 86280
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004362441433992665,
+      "loss": 0.4546,
+      "step": 86290
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043620174266997396,
+      "loss": 0.5297,
+      "step": 86300
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043615934194068143,
+      "loss": 0.5635,
+      "step": 86310
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043611694121138884,
+      "loss": 0.6177,
+      "step": 86320
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004360745404820963,
+      "loss": 0.536,
+      "step": 86330
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004360321397528037,
+      "loss": 0.5687,
+      "step": 86340
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004359897390235112,
+      "loss": 0.4659,
+      "step": 86350
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004359473382942187,
+      "loss": 0.4922,
+      "step": 86360
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004359049375649261,
+      "loss": 0.5146,
+      "step": 86370
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004358625368356336,
+      "loss": 0.5641,
+      "step": 86380
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043582013610634105,
+      "loss": 0.6432,
+      "step": 86390
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043577773537704847,
+      "loss": 0.5069,
+      "step": 86400
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043573533464775593,
+      "loss": 0.4869,
+      "step": 86410
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043569293391846345,
+      "loss": 0.62,
+      "step": 86420
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043565053318917087,
+      "loss": 0.5567,
+      "step": 86430
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043560813245987833,
+      "loss": 0.5288,
+      "step": 86440
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004355657317305858,
+      "loss": 0.5456,
+      "step": 86450
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004355233310012932,
+      "loss": 0.553,
+      "step": 86460
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004354809302720007,
+      "loss": 0.4192,
+      "step": 86470
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004354385295427082,
+      "loss": 0.5029,
+      "step": 86480
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043539612881341556,
+      "loss": 0.6469,
+      "step": 86490
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004353537280841231,
+      "loss": 0.5495,
+      "step": 86500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043531132735483055,
+      "loss": 0.5194,
+      "step": 86510
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043526892662553796,
+      "loss": 0.527,
+      "step": 86520
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004352265258962454,
+      "loss": 0.4841,
+      "step": 86530
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043518412516695295,
+      "loss": 0.6061,
+      "step": 86540
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004351417244376603,
+      "loss": 0.5062,
+      "step": 86550
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004350993237083678,
+      "loss": 0.5021,
+      "step": 86560
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004350569229790753,
+      "loss": 0.5191,
+      "step": 86570
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004350145222497827,
+      "loss": 0.6089,
+      "step": 86580
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043497212152049017,
+      "loss": 0.5568,
+      "step": 86590
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004349297207911977,
+      "loss": 0.5194,
+      "step": 86600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043488732006190505,
+      "loss": 0.5036,
+      "step": 86610
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043484491933261257,
+      "loss": 0.4623,
+      "step": 86620
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043480251860332004,
+      "loss": 0.5337,
+      "step": 86630
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043476011787402745,
+      "loss": 0.4612,
+      "step": 86640
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004347177171447349,
+      "loss": 0.5264,
+      "step": 86650
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043467531641544233,
+      "loss": 0.4986,
+      "step": 86660
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004346329156861498,
+      "loss": 0.5437,
+      "step": 86670
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004345905149568573,
+      "loss": 0.5328,
+      "step": 86680
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004345481142275647,
+      "loss": 0.4962,
+      "step": 86690
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004345057134982722,
+      "loss": 0.51,
+      "step": 86700
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043446331276897967,
+      "loss": 0.5283,
+      "step": 86710
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004344209120396871,
+      "loss": 0.5473,
+      "step": 86720
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043437851131039454,
+      "loss": 0.6265,
+      "step": 86730
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043433611058110207,
+      "loss": 0.5634,
+      "step": 86740
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004342937098518094,
+      "loss": 0.5212,
+      "step": 86750
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043425130912251694,
+      "loss": 0.4861,
+      "step": 86760
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004342089083932244,
+      "loss": 0.5387,
+      "step": 86770
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004341665076639318,
+      "loss": 0.493,
+      "step": 86780
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004341241069346393,
+      "loss": 0.5704,
+      "step": 86790
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043408170620534676,
+      "loss": 0.4432,
+      "step": 86800
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043403930547605417,
+      "loss": 0.5499,
+      "step": 86810
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004339969047467617,
+      "loss": 0.5931,
+      "step": 86820
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043395450401746916,
+      "loss": 0.589,
+      "step": 86830
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043391210328817657,
+      "loss": 0.5723,
+      "step": 86840
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043386970255888404,
+      "loss": 0.5657,
+      "step": 86850
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004338273018295915,
+      "loss": 0.5261,
+      "step": 86860
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004337849011002989,
+      "loss": 0.4944,
+      "step": 86870
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004337425003710064,
+      "loss": 0.4734,
+      "step": 86880
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004337000996417139,
+      "loss": 0.5,
+      "step": 86890
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004336576989124213,
+      "loss": 0.5582,
+      "step": 86900
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004336152981831288,
+      "loss": 0.463,
+      "step": 86910
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043357289745383625,
+      "loss": 0.5304,
+      "step": 86920
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043353049672454366,
+      "loss": 0.4997,
+      "step": 86930
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043348809599525113,
+      "loss": 0.4673,
+      "step": 86940
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043344569526595854,
+      "loss": 0.6519,
+      "step": 86950
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043340329453666606,
+      "loss": 0.5253,
+      "step": 86960
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043336089380737353,
+      "loss": 0.636,
+      "step": 86970
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043331849307808094,
+      "loss": 0.4972,
+      "step": 86980
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004332760923487884,
+      "loss": 0.6198,
+      "step": 86990
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004332336916194959,
+      "loss": 0.469,
+      "step": 87000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004331912908902033,
+      "loss": 0.5797,
+      "step": 87010
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043314889016091076,
+      "loss": 0.6235,
+      "step": 87020
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004331064894316183,
+      "loss": 0.5785,
+      "step": 87030
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004330640887023257,
+      "loss": 0.4992,
+      "step": 87040
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043302168797303316,
+      "loss": 0.5376,
+      "step": 87050
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004329792872437406,
+      "loss": 0.4156,
+      "step": 87060
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043293688651444804,
+      "loss": 0.4991,
+      "step": 87070
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004328944857851555,
+      "loss": 0.5344,
+      "step": 87080
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.000432852085055863,
+      "loss": 0.5204,
+      "step": 87090
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004328096843265704,
+      "loss": 0.6039,
+      "step": 87100
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004327672835972779,
+      "loss": 0.5539,
+      "step": 87110
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043272488286798537,
+      "loss": 0.5106,
+      "step": 87120
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004326824821386928,
+      "loss": 0.5626,
+      "step": 87130
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043264008140940025,
+      "loss": 0.5268,
+      "step": 87140
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043259768068010777,
+      "loss": 0.5176,
+      "step": 87150
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043255527995081513,
+      "loss": 0.4934,
+      "step": 87160
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043251287922152265,
+      "loss": 0.5697,
+      "step": 87170
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004324704784922301,
+      "loss": 0.5499,
+      "step": 87180
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043242807776293753,
+      "loss": 0.5673,
+      "step": 87190
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.000432385677033645,
+      "loss": 0.532,
+      "step": 87200
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004323432763043525,
+      "loss": 0.4873,
+      "step": 87210
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004323008755750599,
+      "loss": 0.5136,
+      "step": 87220
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004322584748457674,
+      "loss": 0.4783,
+      "step": 87230
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043221607411647486,
+      "loss": 0.5454,
+      "step": 87240
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004321736733871823,
+      "loss": 0.4978,
+      "step": 87250
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043213127265788974,
+      "loss": 0.4931,
+      "step": 87260
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043208887192859715,
+      "loss": 0.5452,
+      "step": 87270
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004320464711993046,
+      "loss": 0.4454,
+      "step": 87280
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043200407047001214,
+      "loss": 0.4802,
+      "step": 87290
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004319616697407195,
+      "loss": 0.5758,
+      "step": 87300
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.000431919269011427,
+      "loss": 0.443,
+      "step": 87310
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004318768682821345,
+      "loss": 0.5306,
+      "step": 87320
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004318344675528419,
+      "loss": 0.4744,
+      "step": 87330
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043179206682354937,
+      "loss": 0.5922,
+      "step": 87340
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004317496660942569,
+      "loss": 0.5659,
+      "step": 87350
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043170726536496425,
+      "loss": 0.5585,
+      "step": 87360
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043166486463567177,
+      "loss": 0.7003,
+      "step": 87370
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043162246390637924,
+      "loss": 0.5822,
+      "step": 87380
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043158006317708665,
+      "loss": 0.5574,
+      "step": 87390
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004315376624477941,
+      "loss": 0.5228,
+      "step": 87400
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043149526171850164,
+      "loss": 0.4817,
+      "step": 87410
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.000431452860989209,
+      "loss": 0.5107,
+      "step": 87420
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004314104602599165,
+      "loss": 0.4782,
+      "step": 87430
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.000431368059530624,
+      "loss": 0.4693,
+      "step": 87440
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004313256588013314,
+      "loss": 0.6263,
+      "step": 87450
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043128325807203886,
+      "loss": 0.4928,
+      "step": 87460
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043124085734274633,
+      "loss": 0.5142,
+      "step": 87470
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043119845661345374,
+      "loss": 0.5517,
+      "step": 87480
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043115605588416126,
+      "loss": 0.5393,
+      "step": 87490
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043111365515486873,
+      "loss": 0.5983,
+      "step": 87500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043107125442557614,
+      "loss": 0.5266,
+      "step": 87510
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004310288536962836,
+      "loss": 0.5685,
+      "step": 87520
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004309864529669911,
+      "loss": 0.5181,
+      "step": 87530
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004309440522376985,
+      "loss": 0.4949,
+      "step": 87540
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043090165150840595,
+      "loss": 0.4883,
+      "step": 87550
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043085925077911337,
+      "loss": 0.4847,
+      "step": 87560
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004308168500498209,
+      "loss": 0.5835,
+      "step": 87570
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043077444932052835,
+      "loss": 0.5219,
+      "step": 87580
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043073204859123577,
+      "loss": 0.5542,
+      "step": 87590
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043068964786194323,
+      "loss": 0.6945,
+      "step": 87600
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004306472471326507,
+      "loss": 0.5206,
+      "step": 87610
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004306048464033581,
+      "loss": 0.5347,
+      "step": 87620
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004305624456740656,
+      "loss": 0.543,
+      "step": 87630
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004305200449447731,
+      "loss": 0.576,
+      "step": 87640
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004304776442154805,
+      "loss": 0.5513,
+      "step": 87650
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.000430435243486188,
+      "loss": 0.5384,
+      "step": 87660
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00043039284275689545,
+      "loss": 0.5489,
+      "step": 87670
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00043035044202760286,
+      "loss": 0.4677,
+      "step": 87680
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004303080412983103,
+      "loss": 0.5045,
+      "step": 87690
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00043026564056901785,
+      "loss": 0.5755,
+      "step": 87700
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00043022323983972526,
+      "loss": 0.4964,
+      "step": 87710
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004301808391104327,
+      "loss": 0.5189,
+      "step": 87720
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004301384383811402,
+      "loss": 0.4733,
+      "step": 87730
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004300960376518476,
+      "loss": 0.5589,
+      "step": 87740
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00043005363692255507,
+      "loss": 0.6515,
+      "step": 87750
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004300112361932626,
+      "loss": 0.587,
+      "step": 87760
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00042996883546396995,
+      "loss": 0.5701,
+      "step": 87770
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004299264347346775,
+      "loss": 0.6282,
+      "step": 87780
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00042988403400538494,
+      "loss": 0.5699,
+      "step": 87790
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00042984163327609235,
+      "loss": 0.5541,
+      "step": 87800
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004297992325467998,
+      "loss": 0.5345,
+      "step": 87810
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00042975683181750734,
+      "loss": 0.4805,
+      "step": 87820
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004297144310882147,
+      "loss": 0.5942,
+      "step": 87830
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004296720303589222,
+      "loss": 0.56,
+      "step": 87840
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004296296296296297,
+      "loss": 0.6051,
+      "step": 87850
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004295872289003371,
+      "loss": 0.5569,
+      "step": 87860
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00042954482817104457,
+      "loss": 0.4933,
+      "step": 87870
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.000429502427441752,
+      "loss": 0.5274,
+      "step": 87880
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00042946002671245945,
+      "loss": 0.5127,
+      "step": 87890
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00042941762598316697,
+      "loss": 0.5668,
+      "step": 87900
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004293752252538743,
+      "loss": 0.532,
+      "step": 87910
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00042933282452458185,
+      "loss": 0.525,
+      "step": 87920
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004292904237952893,
+      "loss": 0.5724,
+      "step": 87930
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004292480230659967,
+      "loss": 0.6056,
+      "step": 87940
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004292056223367042,
+      "loss": 0.4738,
+      "step": 87950
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004291632216074117,
+      "loss": 0.5088,
+      "step": 87960
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00042912082087811907,
+      "loss": 0.5548,
+      "step": 87970
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004290784201488266,
+      "loss": 0.5236,
+      "step": 87980
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00042903601941953406,
+      "loss": 0.5458,
+      "step": 87990
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00042899361869024147,
+      "loss": 0.6178,
+      "step": 88000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00042895121796094894,
+      "loss": 0.4888,
+      "step": 88010
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00042890881723165646,
+      "loss": 0.4738,
+      "step": 88020
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004288664165023638,
+      "loss": 0.4946,
+      "step": 88030
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00042882401577307134,
+      "loss": 0.5707,
+      "step": 88040
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004287816150437788,
+      "loss": 0.5212,
+      "step": 88050
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004287392143144862,
+      "loss": 0.5221,
+      "step": 88060
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004286968135851937,
+      "loss": 0.4305,
+      "step": 88070
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00042865441285590115,
+      "loss": 0.5432,
+      "step": 88080
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042861201212660856,
+      "loss": 0.4819,
+      "step": 88090
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004285696113973161,
+      "loss": 0.5526,
+      "step": 88100
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042852721066802355,
+      "loss": 0.5476,
+      "step": 88110
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042848480993873096,
+      "loss": 0.4591,
+      "step": 88120
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042844240920943843,
+      "loss": 0.5158,
+      "step": 88130
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004284000084801459,
+      "loss": 0.5396,
+      "step": 88140
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004283576077508533,
+      "loss": 0.5455,
+      "step": 88150
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042831520702156083,
+      "loss": 0.4556,
+      "step": 88160
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004282728062922683,
+      "loss": 0.5008,
+      "step": 88170
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004282304055629757,
+      "loss": 0.6215,
+      "step": 88180
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004281880048336832,
+      "loss": 0.6145,
+      "step": 88190
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004281456041043906,
+      "loss": 0.5144,
+      "step": 88200
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042810320337509806,
+      "loss": 0.512,
+      "step": 88210
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004280608026458055,
+      "loss": 0.4794,
+      "step": 88220
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042801840191651294,
+      "loss": 0.4936,
+      "step": 88230
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042797600118722046,
+      "loss": 0.5949,
+      "step": 88240
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004279336004579279,
+      "loss": 0.6048,
+      "step": 88250
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042789119972863534,
+      "loss": 0.4564,
+      "step": 88260
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004278487989993428,
+      "loss": 0.5551,
+      "step": 88270
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042780639827005027,
+      "loss": 0.5587,
+      "step": 88280
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004277639975407577,
+      "loss": 0.4728,
+      "step": 88290
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042772159681146515,
+      "loss": 0.4974,
+      "step": 88300
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042767919608217267,
+      "loss": 0.4846,
+      "step": 88310
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004276367953528801,
+      "loss": 0.4857,
+      "step": 88320
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042759439462358755,
+      "loss": 0.5705,
+      "step": 88330
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.000427551993894295,
+      "loss": 0.6223,
+      "step": 88340
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042750959316500243,
+      "loss": 0.6948,
+      "step": 88350
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004274671924357099,
+      "loss": 0.5185,
+      "step": 88360
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004274247917064174,
+      "loss": 0.5263,
+      "step": 88370
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004273823909771248,
+      "loss": 0.6081,
+      "step": 88380
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004273399902478323,
+      "loss": 0.6076,
+      "step": 88390
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042729758951853976,
+      "loss": 0.5677,
+      "step": 88400
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004272551887892472,
+      "loss": 0.5123,
+      "step": 88410
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042721278805995464,
+      "loss": 0.5541,
+      "step": 88420
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042717038733066216,
+      "loss": 0.4296,
+      "step": 88430
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004271279866013695,
+      "loss": 0.51,
+      "step": 88440
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042708558587207704,
+      "loss": 0.5002,
+      "step": 88450
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004270431851427845,
+      "loss": 0.5832,
+      "step": 88460
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004270007844134919,
+      "loss": 0.5247,
+      "step": 88470
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004269583836841994,
+      "loss": 0.4818,
+      "step": 88480
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004269159829549068,
+      "loss": 0.5523,
+      "step": 88490
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042687358222561427,
+      "loss": 0.5616,
+      "step": 88500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004268311814963218,
+      "loss": 0.4886,
+      "step": 88510
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042678878076702915,
+      "loss": 0.5465,
+      "step": 88520
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042674638003773667,
+      "loss": 0.506,
+      "step": 88530
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042670397930844414,
+      "loss": 0.5303,
+      "step": 88540
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042666157857915155,
+      "loss": 0.522,
+      "step": 88550
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.000426619177849859,
+      "loss": 0.5048,
+      "step": 88560
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042657677712056654,
+      "loss": 0.4647,
+      "step": 88570
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004265343763912739,
+      "loss": 0.4578,
+      "step": 88580
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004264919756619814,
+      "loss": 0.4945,
+      "step": 88590
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004264495749326889,
+      "loss": 0.5246,
+      "step": 88600
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004264071742033963,
+      "loss": 0.4359,
+      "step": 88610
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042636477347410376,
+      "loss": 0.478,
+      "step": 88620
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004263223727448113,
+      "loss": 0.4938,
+      "step": 88630
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042627997201551864,
+      "loss": 0.5392,
+      "step": 88640
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042623757128622616,
+      "loss": 0.519,
+      "step": 88650
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042619517055693363,
+      "loss": 0.4912,
+      "step": 88660
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042615276982764104,
+      "loss": 0.4366,
+      "step": 88670
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004261103690983485,
+      "loss": 0.48,
+      "step": 88680
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042606796836905603,
+      "loss": 0.4271,
+      "step": 88690
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004260255676397634,
+      "loss": 0.4868,
+      "step": 88700
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004259831669104709,
+      "loss": 0.5431,
+      "step": 88710
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004259407661811784,
+      "loss": 0.5644,
+      "step": 88720
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004258983654518858,
+      "loss": 0.5712,
+      "step": 88730
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042585596472259325,
+      "loss": 0.5712,
+      "step": 88740
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004258135639933007,
+      "loss": 0.4472,
+      "step": 88750
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042577116326400813,
+      "loss": 0.5373,
+      "step": 88760
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042572876253471566,
+      "loss": 0.4917,
+      "step": 88770
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004256863618054231,
+      "loss": 0.5566,
+      "step": 88780
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042564396107613053,
+      "loss": 0.5335,
+      "step": 88790
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.000425601560346838,
+      "loss": 0.468,
+      "step": 88800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004255591596175454,
+      "loss": 0.5526,
+      "step": 88810
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004255167588882529,
+      "loss": 0.5053,
+      "step": 88820
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042547435815896035,
+      "loss": 0.4864,
+      "step": 88830
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042543195742966776,
+      "loss": 0.6192,
+      "step": 88840
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004253895567003753,
+      "loss": 0.5106,
+      "step": 88850
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042534715597108275,
+      "loss": 0.566,
+      "step": 88860
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042530475524179016,
+      "loss": 0.5262,
+      "step": 88870
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042526235451249763,
+      "loss": 0.5293,
+      "step": 88880
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004252199537832051,
+      "loss": 0.5709,
+      "step": 88890
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004251775530539125,
+      "loss": 0.5345,
+      "step": 88900
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042513515232462003,
+      "loss": 0.5218,
+      "step": 88910
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004250927515953275,
+      "loss": 0.5793,
+      "step": 88920
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004250503508660349,
+      "loss": 0.4565,
+      "step": 88930
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004250079501367424,
+      "loss": 0.4838,
+      "step": 88940
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042496554940744984,
+      "loss": 0.5819,
+      "step": 88950
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042492314867815725,
+      "loss": 0.5683,
+      "step": 88960
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004248807479488647,
+      "loss": 0.5328,
+      "step": 88970
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042483834721957224,
+      "loss": 0.5135,
+      "step": 88980
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042479594649027965,
+      "loss": 0.5597,
+      "step": 88990
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004247535457609871,
+      "loss": 0.4936,
+      "step": 89000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004247111450316946,
+      "loss": 0.5465,
+      "step": 89010
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.000424668744302402,
+      "loss": 0.5741,
+      "step": 89020
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042462634357310947,
+      "loss": 0.5071,
+      "step": 89030
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.000424583942843817,
+      "loss": 0.5346,
+      "step": 89040
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042454154211452435,
+      "loss": 0.5355,
+      "step": 89050
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042449914138523187,
+      "loss": 0.5484,
+      "step": 89060
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042445674065593933,
+      "loss": 0.6231,
+      "step": 89070
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042441433992664675,
+      "loss": 0.5035,
+      "step": 89080
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004243719391973542,
+      "loss": 0.5163,
+      "step": 89090
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004243295384680616,
+      "loss": 0.5299,
+      "step": 89100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004242871377387691,
+      "loss": 0.4932,
+      "step": 89110
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004242447370094766,
+      "loss": 0.5707,
+      "step": 89120
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042420233628018397,
+      "loss": 0.5435,
+      "step": 89130
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004241599355508915,
+      "loss": 0.5846,
+      "step": 89140
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042411753482159896,
+      "loss": 0.4945,
+      "step": 89150
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042407513409230637,
+      "loss": 0.5759,
+      "step": 89160
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042403273336301384,
+      "loss": 0.5365,
+      "step": 89170
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042399033263372136,
+      "loss": 0.5395,
+      "step": 89180
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004239479319044287,
+      "loss": 0.535,
+      "step": 89190
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042390553117513624,
+      "loss": 0.5079,
+      "step": 89200
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004238631304458437,
+      "loss": 0.5702,
+      "step": 89210
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004238207297165511,
+      "loss": 0.4987,
+      "step": 89220
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004237783289872586,
+      "loss": 0.5177,
+      "step": 89230
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004237359282579661,
+      "loss": 0.4967,
+      "step": 89240
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042369352752867346,
+      "loss": 0.5135,
+      "step": 89250
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.000423651126799381,
+      "loss": 0.5129,
+      "step": 89260
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042360872607008845,
+      "loss": 0.5459,
+      "step": 89270
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042356632534079587,
+      "loss": 0.4796,
+      "step": 89280
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042352392461150333,
+      "loss": 0.5611,
+      "step": 89290
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042348152388221085,
+      "loss": 0.5478,
+      "step": 89300
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004234391231529182,
+      "loss": 0.579,
+      "step": 89310
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042339672242362573,
+      "loss": 0.5116,
+      "step": 89320
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004233543216943332,
+      "loss": 0.5585,
+      "step": 89330
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004233119209650406,
+      "loss": 0.4798,
+      "step": 89340
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004232695202357481,
+      "loss": 0.5564,
+      "step": 89350
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042322711950645555,
+      "loss": 0.5102,
+      "step": 89360
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042318471877716296,
+      "loss": 0.4969,
+      "step": 89370
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004231423180478705,
+      "loss": 0.495,
+      "step": 89380
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042309991731857795,
+      "loss": 0.5318,
+      "step": 89390
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042305751658928536,
+      "loss": 0.5625,
+      "step": 89400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004230151158599928,
+      "loss": 0.5348,
+      "step": 89410
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042297271513070024,
+      "loss": 0.5264,
+      "step": 89420
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004229303144014077,
+      "loss": 0.5032,
+      "step": 89430
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004228879136721152,
+      "loss": 0.5668,
+      "step": 89440
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004228455129428226,
+      "loss": 0.5475,
+      "step": 89450
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004228031122135301,
+      "loss": 0.531,
+      "step": 89460
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042276071148423757,
+      "loss": 0.4393,
+      "step": 89470
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.000422718310754945,
+      "loss": 0.4252,
+      "step": 89480
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042267591002565245,
+      "loss": 0.5174,
+      "step": 89490
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004226335092963599,
+      "loss": 0.5479,
+      "step": 89500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042259110856706733,
+      "loss": 0.5303,
+      "step": 89510
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042254870783777485,
+      "loss": 0.5193,
+      "step": 89520
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004225063071084823,
+      "loss": 0.5488,
+      "step": 89530
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042246390637918973,
+      "loss": 0.5263,
+      "step": 89540
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004224215056498972,
+      "loss": 0.4405,
+      "step": 89550
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042237910492060466,
+      "loss": 0.4911,
+      "step": 89560
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004223367041913121,
+      "loss": 0.5193,
+      "step": 89570
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042229430346201954,
+      "loss": 0.5234,
+      "step": 89580
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042225190273272706,
+      "loss": 0.4931,
+      "step": 89590
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004222095020034345,
+      "loss": 0.4988,
+      "step": 89600
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042216710127414194,
+      "loss": 0.566,
+      "step": 89610
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004221247005448494,
+      "loss": 0.5661,
+      "step": 89620
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004220822998155568,
+      "loss": 0.5533,
+      "step": 89630
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004220398990862643,
+      "loss": 0.5427,
+      "step": 89640
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004219974983569718,
+      "loss": 0.5255,
+      "step": 89650
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004219550976276792,
+      "loss": 0.5646,
+      "step": 89660
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004219126968983867,
+      "loss": 0.4642,
+      "step": 89670
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042187029616909416,
+      "loss": 0.5141,
+      "step": 89680
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042182789543980157,
+      "loss": 0.5128,
+      "step": 89690
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042178549471050904,
+      "loss": 0.5598,
+      "step": 89700
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042174309398121645,
+      "loss": 0.5743,
+      "step": 89710
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004217006932519239,
+      "loss": 0.5676,
+      "step": 89720
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042165829252263144,
+      "loss": 0.5346,
+      "step": 89730
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042161589179333885,
+      "loss": 0.606,
+      "step": 89740
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004215734910640463,
+      "loss": 0.5293,
+      "step": 89750
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004215310903347538,
+      "loss": 0.4518,
+      "step": 89760
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004214886896054612,
+      "loss": 0.6161,
+      "step": 89770
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042144628887616866,
+      "loss": 0.5517,
+      "step": 89780
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004214038881468762,
+      "loss": 0.4909,
+      "step": 89790
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042136148741758354,
+      "loss": 0.5326,
+      "step": 89800
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042131908668829106,
+      "loss": 0.5167,
+      "step": 89810
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042127668595899853,
+      "loss": 0.4947,
+      "step": 89820
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042123428522970594,
+      "loss": 0.545,
+      "step": 89830
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004211918845004134,
+      "loss": 0.5648,
+      "step": 89840
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042114948377112093,
+      "loss": 0.5963,
+      "step": 89850
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004211070830418283,
+      "loss": 0.5431,
+      "step": 89860
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004210646823125358,
+      "loss": 0.543,
+      "step": 89870
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004210222815832433,
+      "loss": 0.4565,
+      "step": 89880
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004209798808539507,
+      "loss": 0.5885,
+      "step": 89890
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042093748012465816,
+      "loss": 0.5322,
+      "step": 89900
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004208950793953657,
+      "loss": 0.6068,
+      "step": 89910
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042085267866607303,
+      "loss": 0.5339,
+      "step": 89920
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042081027793678056,
+      "loss": 0.467,
+      "step": 89930
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.000420767877207488,
+      "loss": 0.5088,
+      "step": 89940
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042072547647819544,
+      "loss": 0.5071,
+      "step": 89950
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004206830757489029,
+      "loss": 0.5425,
+      "step": 89960
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004206406750196104,
+      "loss": 0.5028,
+      "step": 89970
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004205982742903178,
+      "loss": 0.5451,
+      "step": 89980
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004205558735610253,
+      "loss": 0.5287,
+      "step": 89990
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042051347283173277,
+      "loss": 0.5184,
+      "step": 90000
+    },
+    {
+      "epoch": 3.81,
+      "eval_loss": 0.6043498516082764,
+      "eval_runtime": 337.771,
+      "eval_samples_per_second": 15.558,
+      "eval_steps_per_second": 3.89,
+      "step": 90000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004204710721024402,
+      "loss": 0.4972,
+      "step": 90010
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042042867137314765,
+      "loss": 0.4611,
+      "step": 90020
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042038627064385506,
+      "loss": 0.4679,
+      "step": 90030
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042034386991456253,
+      "loss": 0.5363,
+      "step": 90040
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042030146918527005,
+      "loss": 0.5741,
+      "step": 90050
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004202590684559774,
+      "loss": 0.4266,
+      "step": 90060
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042021666772668493,
+      "loss": 0.5124,
+      "step": 90070
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004201742669973924,
+      "loss": 0.535,
+      "step": 90080
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004201318662680998,
+      "loss": 0.5862,
+      "step": 90090
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004200894655388073,
+      "loss": 0.5156,
+      "step": 90100
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042004706480951474,
+      "loss": 0.4629,
+      "step": 90110
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042000466408022215,
+      "loss": 0.484,
+      "step": 90120
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004199622633509297,
+      "loss": 0.5924,
+      "step": 90130
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00041991986262163714,
+      "loss": 0.5571,
+      "step": 90140
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00041987746189234455,
+      "loss": 0.6235,
+      "step": 90150
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.000419835061163052,
+      "loss": 0.4898,
+      "step": 90160
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004197926604337595,
+      "loss": 0.4908,
+      "step": 90170
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004197502597044669,
+      "loss": 0.5333,
+      "step": 90180
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004197078589751744,
+      "loss": 0.5706,
+      "step": 90190
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004196654582458819,
+      "loss": 0.503,
+      "step": 90200
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004196230575165893,
+      "loss": 0.4491,
+      "step": 90210
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041958065678729677,
+      "loss": 0.5667,
+      "step": 90220
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041953825605800423,
+      "loss": 0.5221,
+      "step": 90230
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041949585532871165,
+      "loss": 0.4915,
+      "step": 90240
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0004194534545994191,
+      "loss": 0.4662,
+      "step": 90250
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041941105387012663,
+      "loss": 0.6016,
+      "step": 90260
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041936865314083405,
+      "loss": 0.4769,
+      "step": 90270
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0004193262524115415,
+      "loss": 0.5131,
+      "step": 90280
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.000419283851682249,
+      "loss": 0.539,
+      "step": 90290
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0004192414509529564,
+      "loss": 0.5891,
+      "step": 90300
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041919905022366386,
+      "loss": 0.5043,
+      "step": 90310
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0004191566494943714,
+      "loss": 0.5528,
+      "step": 90320
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041911424876507874,
+      "loss": 0.5008,
+      "step": 90330
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041907184803578626,
+      "loss": 0.5357,
+      "step": 90340
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0004190294473064937,
+      "loss": 0.5288,
+      "step": 90350
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041898704657720114,
+      "loss": 0.4776,
+      "step": 90360
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0004189446458479086,
+      "loss": 0.5549,
+      "step": 90370
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.000418902245118616,
+      "loss": 0.5446,
+      "step": 90380
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0004188598443893235,
+      "loss": 0.5027,
+      "step": 90390
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.000418817443660031,
+      "loss": 0.4826,
+      "step": 90400
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0004187750429307384,
+      "loss": 0.3647,
+      "step": 90410
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0004187326422014459,
+      "loss": 0.4977,
+      "step": 90420
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041869024147215335,
+      "loss": 0.5878,
+      "step": 90430
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041864784074286077,
+      "loss": 0.4731,
+      "step": 90440
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041860544001356823,
+      "loss": 0.478,
+      "step": 90450
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041856303928427575,
+      "loss": 0.5701,
+      "step": 90460
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004185206385549831,
+      "loss": 0.5017,
+      "step": 90470
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041847823782569063,
+      "loss": 0.4694,
+      "step": 90480
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004184358370963981,
+      "loss": 0.5958,
+      "step": 90490
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004183934363671055,
+      "loss": 0.6148,
+      "step": 90500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.000418351035637813,
+      "loss": 0.4949,
+      "step": 90510
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004183086349085205,
+      "loss": 0.5254,
+      "step": 90520
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041826623417922786,
+      "loss": 0.5135,
+      "step": 90530
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004182238334499354,
+      "loss": 0.6176,
+      "step": 90540
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041818143272064285,
+      "loss": 0.5088,
+      "step": 90550
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041813903199135026,
+      "loss": 0.4523,
+      "step": 90560
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004180966312620577,
+      "loss": 0.5226,
+      "step": 90570
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041805423053276525,
+      "loss": 0.4966,
+      "step": 90580
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004180118298034726,
+      "loss": 0.5714,
+      "step": 90590
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004179694290741801,
+      "loss": 0.4944,
+      "step": 90600
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004179270283448876,
+      "loss": 0.5635,
+      "step": 90610
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.000417884627615595,
+      "loss": 0.5306,
+      "step": 90620
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041784222688630247,
+      "loss": 0.6078,
+      "step": 90630
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004177998261570099,
+      "loss": 0.502,
+      "step": 90640
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041775742542771735,
+      "loss": 0.6534,
+      "step": 90650
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041771502469842487,
+      "loss": 0.5158,
+      "step": 90660
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041767262396913223,
+      "loss": 0.685,
+      "step": 90670
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041763022323983975,
+      "loss": 0.5393,
+      "step": 90680
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004175878225105472,
+      "loss": 0.5298,
+      "step": 90690
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041754542178125463,
+      "loss": 0.5022,
+      "step": 90700
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004175030210519621,
+      "loss": 0.49,
+      "step": 90710
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004174606203226696,
+      "loss": 0.5174,
+      "step": 90720
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.000417418219593377,
+      "loss": 0.4317,
+      "step": 90730
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004173758188640845,
+      "loss": 0.5103,
+      "step": 90740
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041733341813479197,
+      "loss": 0.5383,
+      "step": 90750
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004172910174054994,
+      "loss": 0.4348,
+      "step": 90760
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041724861667620684,
+      "loss": 0.5221,
+      "step": 90770
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004172062159469143,
+      "loss": 0.5611,
+      "step": 90780
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004171638152176217,
+      "loss": 0.5257,
+      "step": 90790
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041712141448832924,
+      "loss": 0.5956,
+      "step": 90800
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004170790137590367,
+      "loss": 0.5082,
+      "step": 90810
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004170366130297441,
+      "loss": 0.5808,
+      "step": 90820
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004169942123004516,
+      "loss": 0.5975,
+      "step": 90830
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041695181157115906,
+      "loss": 0.5242,
+      "step": 90840
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041690941084186647,
+      "loss": 0.554,
+      "step": 90850
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041686701011257394,
+      "loss": 0.5624,
+      "step": 90860
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041682460938328146,
+      "loss": 0.4407,
+      "step": 90870
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041678220865398887,
+      "loss": 0.5498,
+      "step": 90880
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041673980792469634,
+      "loss": 0.5634,
+      "step": 90890
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004166974071954038,
+      "loss": 0.5022,
+      "step": 90900
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004166550064661112,
+      "loss": 0.5721,
+      "step": 90910
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004166126057368187,
+      "loss": 0.4812,
+      "step": 90920
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004165702050075262,
+      "loss": 0.5363,
+      "step": 90930
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004165278042782336,
+      "loss": 0.4668,
+      "step": 90940
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004164854035489411,
+      "loss": 0.59,
+      "step": 90950
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004164430028196485,
+      "loss": 0.5673,
+      "step": 90960
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041640060209035596,
+      "loss": 0.4765,
+      "step": 90970
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041635820136106343,
+      "loss": 0.5011,
+      "step": 90980
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041631580063177084,
+      "loss": 0.5694,
+      "step": 90990
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004162733999024783,
+      "loss": 0.5061,
+      "step": 91000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041623099917318583,
+      "loss": 0.5409,
+      "step": 91010
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041618859844389324,
+      "loss": 0.4512,
+      "step": 91020
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004161461977146007,
+      "loss": 0.4962,
+      "step": 91030
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004161037969853082,
+      "loss": 0.4985,
+      "step": 91040
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004160613962560156,
+      "loss": 0.4893,
+      "step": 91050
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041601899552672306,
+      "loss": 0.5919,
+      "step": 91060
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004159765947974306,
+      "loss": 0.629,
+      "step": 91070
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041593419406813794,
+      "loss": 0.4808,
+      "step": 91080
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041589179333884546,
+      "loss": 0.5395,
+      "step": 91090
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004158493926095529,
+      "loss": 0.4962,
+      "step": 91100
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041580699188026034,
+      "loss": 0.4942,
+      "step": 91110
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004157645911509678,
+      "loss": 0.5828,
+      "step": 91120
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004157221904216753,
+      "loss": 0.4904,
+      "step": 91130
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004156797896923827,
+      "loss": 0.5556,
+      "step": 91140
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004156373889630902,
+      "loss": 0.5048,
+      "step": 91150
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041559498823379767,
+      "loss": 0.4993,
+      "step": 91160
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004155525875045051,
+      "loss": 0.5936,
+      "step": 91170
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041551018677521255,
+      "loss": 0.5172,
+      "step": 91180
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041546778604592007,
+      "loss": 0.5227,
+      "step": 91190
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041542538531662743,
+      "loss": 0.4739,
+      "step": 91200
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041538298458733495,
+      "loss": 0.5624,
+      "step": 91210
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004153405838580424,
+      "loss": 0.5553,
+      "step": 91220
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041529818312874983,
+      "loss": 0.4594,
+      "step": 91230
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004152557823994573,
+      "loss": 0.4687,
+      "step": 91240
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004152133816701647,
+      "loss": 0.4591,
+      "step": 91250
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004151709809408722,
+      "loss": 0.5428,
+      "step": 91260
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004151285802115797,
+      "loss": 0.5919,
+      "step": 91270
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041508617948228705,
+      "loss": 0.4907,
+      "step": 91280
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004150437787529946,
+      "loss": 0.5052,
+      "step": 91290
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041500137802370204,
+      "loss": 0.4809,
+      "step": 91300
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041495897729440945,
+      "loss": 0.5812,
+      "step": 91310
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004149165765651169,
+      "loss": 0.571,
+      "step": 91320
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041487417583582444,
+      "loss": 0.5233,
+      "step": 91330
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004148317751065318,
+      "loss": 0.4688,
+      "step": 91340
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004147893743772393,
+      "loss": 0.4674,
+      "step": 91350
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004147469736479468,
+      "loss": 0.5323,
+      "step": 91360
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004147045729186542,
+      "loss": 0.4165,
+      "step": 91370
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041466217218936167,
+      "loss": 0.52,
+      "step": 91380
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041461977146006913,
+      "loss": 0.4655,
+      "step": 91390
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041457737073077655,
+      "loss": 0.5733,
+      "step": 91400
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041453497000148407,
+      "loss": 0.4595,
+      "step": 91410
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041449256927219154,
+      "loss": 0.5277,
+      "step": 91420
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041445016854289895,
+      "loss": 0.5252,
+      "step": 91430
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004144077678136064,
+      "loss": 0.4969,
+      "step": 91440
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004143653670843139,
+      "loss": 0.5613,
+      "step": 91450
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004143229663550213,
+      "loss": 0.5067,
+      "step": 91460
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004142805656257288,
+      "loss": 0.6234,
+      "step": 91470
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004142381648964363,
+      "loss": 0.5241,
+      "step": 91480
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004141957641671437,
+      "loss": 0.569,
+      "step": 91490
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041415336343785116,
+      "loss": 0.5263,
+      "step": 91500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041411096270855863,
+      "loss": 0.4858,
+      "step": 91510
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041406856197926604,
+      "loss": 0.5499,
+      "step": 91520
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004140261612499735,
+      "loss": 0.5708,
+      "step": 91530
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041398376052068103,
+      "loss": 0.5226,
+      "step": 91540
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041394135979138844,
+      "loss": 0.5039,
+      "step": 91550
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004138989590620959,
+      "loss": 0.4153,
+      "step": 91560
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004138565583328033,
+      "loss": 0.5966,
+      "step": 91570
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004138141576035108,
+      "loss": 0.5321,
+      "step": 91580
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041377175687421825,
+      "loss": 0.4228,
+      "step": 91590
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041372935614492567,
+      "loss": 0.568,
+      "step": 91600
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041368695541563313,
+      "loss": 0.5032,
+      "step": 91610
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041364455468634065,
+      "loss": 0.5123,
+      "step": 91620
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041360215395704807,
+      "loss": 0.5763,
+      "step": 91630
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041355975322775553,
+      "loss": 0.4659,
+      "step": 91640
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.000413517352498463,
+      "loss": 0.5479,
+      "step": 91650
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004134749517691704,
+      "loss": 0.5027,
+      "step": 91660
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004134325510398779,
+      "loss": 0.5568,
+      "step": 91670
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004133901503105854,
+      "loss": 0.5918,
+      "step": 91680
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004133477495812928,
+      "loss": 0.5031,
+      "step": 91690
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004133053488520003,
+      "loss": 0.5864,
+      "step": 91700
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041326294812270775,
+      "loss": 0.4803,
+      "step": 91710
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041322054739341516,
+      "loss": 0.5605,
+      "step": 91720
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004131781466641226,
+      "loss": 0.5164,
+      "step": 91730
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041313574593483015,
+      "loss": 0.4507,
+      "step": 91740
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004130933452055375,
+      "loss": 0.5121,
+      "step": 91750
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.000413050944476245,
+      "loss": 0.6755,
+      "step": 91760
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004130085437469525,
+      "loss": 0.6269,
+      "step": 91770
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004129661430176599,
+      "loss": 0.5626,
+      "step": 91780
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041292374228836737,
+      "loss": 0.5334,
+      "step": 91790
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004128813415590749,
+      "loss": 0.5431,
+      "step": 91800
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041283894082978225,
+      "loss": 0.5615,
+      "step": 91810
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004127965401004898,
+      "loss": 0.6489,
+      "step": 91820
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041275413937119724,
+      "loss": 0.5208,
+      "step": 91830
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041271173864190465,
+      "loss": 0.5385,
+      "step": 91840
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004126693379126121,
+      "loss": 0.5467,
+      "step": 91850
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041262693718331953,
+      "loss": 0.5053,
+      "step": 91860
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.000412584536454027,
+      "loss": 0.4759,
+      "step": 91870
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004125421357247345,
+      "loss": 0.5273,
+      "step": 91880
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004124997349954419,
+      "loss": 0.5672,
+      "step": 91890
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004124573342661494,
+      "loss": 0.4577,
+      "step": 91900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041241493353685687,
+      "loss": 0.448,
+      "step": 91910
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004123725328075643,
+      "loss": 0.5076,
+      "step": 91920
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041233013207827175,
+      "loss": 0.5238,
+      "step": 91930
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041228773134897927,
+      "loss": 0.5415,
+      "step": 91940
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004122453306196866,
+      "loss": 0.496,
+      "step": 91950
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041220292989039415,
+      "loss": 0.4662,
+      "step": 91960
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004121605291611016,
+      "loss": 0.4986,
+      "step": 91970
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.000412118128431809,
+      "loss": 0.4392,
+      "step": 91980
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004120757277025165,
+      "loss": 0.5562,
+      "step": 91990
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.000412033326973224,
+      "loss": 0.5829,
+      "step": 92000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041199092624393137,
+      "loss": 0.533,
+      "step": 92010
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004119485255146389,
+      "loss": 0.5527,
+      "step": 92020
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041190612478534636,
+      "loss": 0.523,
+      "step": 92030
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041186372405605377,
+      "loss": 0.5578,
+      "step": 92040
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041182132332676124,
+      "loss": 0.4742,
+      "step": 92050
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004117789225974687,
+      "loss": 0.5103,
+      "step": 92060
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004117365218681761,
+      "loss": 0.5207,
+      "step": 92070
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041169412113888364,
+      "loss": 0.5057,
+      "step": 92080
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004116517204095911,
+      "loss": 0.4989,
+      "step": 92090
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004116093196802985,
+      "loss": 0.4893,
+      "step": 92100
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.000411566918951006,
+      "loss": 0.4941,
+      "step": 92110
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041152451822171345,
+      "loss": 0.5881,
+      "step": 92120
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041148211749242086,
+      "loss": 0.5462,
+      "step": 92130
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041143971676312833,
+      "loss": 0.5012,
+      "step": 92140
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041139731603383585,
+      "loss": 0.4595,
+      "step": 92150
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041135491530454326,
+      "loss": 0.5015,
+      "step": 92160
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041131251457525073,
+      "loss": 0.5667,
+      "step": 92170
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041127011384595814,
+      "loss": 0.5645,
+      "step": 92180
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004112277131166656,
+      "loss": 0.4931,
+      "step": 92190
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004111853123873731,
+      "loss": 0.5266,
+      "step": 92200
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004111429116580805,
+      "loss": 0.5136,
+      "step": 92210
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.000411100510928788,
+      "loss": 0.4746,
+      "step": 92220
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004110581101994955,
+      "loss": 0.4926,
+      "step": 92230
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004110157094702029,
+      "loss": 0.5606,
+      "step": 92240
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041097330874091036,
+      "loss": 0.583,
+      "step": 92250
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004109309080116178,
+      "loss": 0.5829,
+      "step": 92260
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041088850728232524,
+      "loss": 0.5161,
+      "step": 92270
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004108461065530327,
+      "loss": 0.517,
+      "step": 92280
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004108037058237402,
+      "loss": 0.5864,
+      "step": 92290
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041076130509444764,
+      "loss": 0.5353,
+      "step": 92300
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004107189043651551,
+      "loss": 0.521,
+      "step": 92310
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041067650363586257,
+      "loss": 0.4913,
+      "step": 92320
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041063410290657,
+      "loss": 0.5151,
+      "step": 92330
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041059170217727745,
+      "loss": 0.6511,
+      "step": 92340
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00041054930144798497,
+      "loss": 0.5362,
+      "step": 92350
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00041050690071869233,
+      "loss": 0.6393,
+      "step": 92360
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00041046449998939985,
+      "loss": 0.6401,
+      "step": 92370
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004104220992601073,
+      "loss": 0.4962,
+      "step": 92380
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00041037969853081473,
+      "loss": 0.5931,
+      "step": 92390
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004103372978015222,
+      "loss": 0.5502,
+      "step": 92400
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004102948970722297,
+      "loss": 0.6143,
+      "step": 92410
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004102524963429371,
+      "loss": 0.4851,
+      "step": 92420
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004102100956136446,
+      "loss": 0.509,
+      "step": 92430
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00041016769488435206,
+      "loss": 0.5259,
+      "step": 92440
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004101252941550595,
+      "loss": 0.6084,
+      "step": 92450
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00041008289342576694,
+      "loss": 0.5895,
+      "step": 92460
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00041004049269647446,
+      "loss": 0.5308,
+      "step": 92470
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004099980919671818,
+      "loss": 0.5035,
+      "step": 92480
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00040995569123788934,
+      "loss": 0.5169,
+      "step": 92490
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004099132905085967,
+      "loss": 0.5239,
+      "step": 92500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004098708897793042,
+      "loss": 0.5971,
+      "step": 92510
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004098284890500117,
+      "loss": 0.515,
+      "step": 92520
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004097860883207191,
+      "loss": 0.5071,
+      "step": 92530
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00040974368759142657,
+      "loss": 0.5389,
+      "step": 92540
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004097012868621341,
+      "loss": 0.5107,
+      "step": 92550
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00040965888613284145,
+      "loss": 0.571,
+      "step": 92560
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00040961648540354897,
+      "loss": 0.5012,
+      "step": 92570
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040957408467425644,
+      "loss": 0.5825,
+      "step": 92580
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040953168394496385,
+      "loss": 0.5112,
+      "step": 92590
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004094892832156713,
+      "loss": 0.4767,
+      "step": 92600
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040944688248637884,
+      "loss": 0.5808,
+      "step": 92610
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004094044817570862,
+      "loss": 0.5246,
+      "step": 92620
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004093620810277937,
+      "loss": 0.5582,
+      "step": 92630
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004093196802985012,
+      "loss": 0.4842,
+      "step": 92640
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004092772795692086,
+      "loss": 0.67,
+      "step": 92650
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040923487883991606,
+      "loss": 0.479,
+      "step": 92660
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004091924781106236,
+      "loss": 0.5505,
+      "step": 92670
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040915007738133094,
+      "loss": 0.4715,
+      "step": 92680
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040910767665203846,
+      "loss": 0.4846,
+      "step": 92690
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040906527592274593,
+      "loss": 0.5285,
+      "step": 92700
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040902287519345334,
+      "loss": 0.4925,
+      "step": 92710
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004089804744641608,
+      "loss": 0.5246,
+      "step": 92720
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004089380737348683,
+      "loss": 0.5018,
+      "step": 92730
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004088956730055757,
+      "loss": 0.4529,
+      "step": 92740
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004088532722762832,
+      "loss": 0.6586,
+      "step": 92750
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004088108715469907,
+      "loss": 0.5618,
+      "step": 92760
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004087684708176981,
+      "loss": 0.5165,
+      "step": 92770
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040872607008840555,
+      "loss": 0.5486,
+      "step": 92780
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040868366935911297,
+      "loss": 0.4767,
+      "step": 92790
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040864126862982043,
+      "loss": 0.4931,
+      "step": 92800
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004085988679005279,
+      "loss": 0.5714,
+      "step": 92810
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004085564671712353,
+      "loss": 0.5342,
+      "step": 92820
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040851406644194283,
+      "loss": 0.5561,
+      "step": 92830
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004084716657126503,
+      "loss": 0.5219,
+      "step": 92840
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004084292649833577,
+      "loss": 0.4911,
+      "step": 92850
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004083868642540652,
+      "loss": 0.5457,
+      "step": 92860
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040834446352477265,
+      "loss": 0.5283,
+      "step": 92870
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040830206279548006,
+      "loss": 0.5557,
+      "step": 92880
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004082596620661875,
+      "loss": 0.558,
+      "step": 92890
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040821726133689505,
+      "loss": 0.5103,
+      "step": 92900
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040817486060760246,
+      "loss": 0.5264,
+      "step": 92910
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040813245987830993,
+      "loss": 0.5268,
+      "step": 92920
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004080900591490174,
+      "loss": 0.5532,
+      "step": 92930
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004080476584197248,
+      "loss": 0.4587,
+      "step": 92940
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004080052576904323,
+      "loss": 0.5493,
+      "step": 92950
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004079628569611398,
+      "loss": 0.5359,
+      "step": 92960
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004079204562318472,
+      "loss": 0.6037,
+      "step": 92970
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004078780555025547,
+      "loss": 0.5949,
+      "step": 92980
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040783565477326214,
+      "loss": 0.5163,
+      "step": 92990
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040779325404396955,
+      "loss": 0.5138,
+      "step": 93000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.000407750853314677,
+      "loss": 0.6217,
+      "step": 93010
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040770845258538454,
+      "loss": 0.5473,
+      "step": 93020
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004076660518560919,
+      "loss": 0.5168,
+      "step": 93030
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004076236511267994,
+      "loss": 0.4693,
+      "step": 93040
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004075812503975069,
+      "loss": 0.5514,
+      "step": 93050
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004075388496682143,
+      "loss": 0.5046,
+      "step": 93060
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040749644893892177,
+      "loss": 0.5455,
+      "step": 93070
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004074540482096293,
+      "loss": 0.5842,
+      "step": 93080
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040741164748033665,
+      "loss": 0.5222,
+      "step": 93090
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040736924675104417,
+      "loss": 0.5646,
+      "step": 93100
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004073268460217515,
+      "loss": 0.5892,
+      "step": 93110
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040728444529245905,
+      "loss": 0.5407,
+      "step": 93120
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004072420445631665,
+      "loss": 0.6352,
+      "step": 93130
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004071996438338739,
+      "loss": 0.5534,
+      "step": 93140
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004071572431045814,
+      "loss": 0.577,
+      "step": 93150
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004071148423752889,
+      "loss": 0.4931,
+      "step": 93160
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040707244164599627,
+      "loss": 0.4565,
+      "step": 93170
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004070300409167038,
+      "loss": 0.4646,
+      "step": 93180
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040698764018741126,
+      "loss": 0.5615,
+      "step": 93190
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040694523945811867,
+      "loss": 0.595,
+      "step": 93200
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040690283872882614,
+      "loss": 0.5281,
+      "step": 93210
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040686043799953366,
+      "loss": 0.6543,
+      "step": 93220
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.000406818037270241,
+      "loss": 0.4985,
+      "step": 93230
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040677563654094854,
+      "loss": 0.4817,
+      "step": 93240
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.000406733235811656,
+      "loss": 0.4952,
+      "step": 93250
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004066908350823634,
+      "loss": 0.5003,
+      "step": 93260
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004066484343530709,
+      "loss": 0.6419,
+      "step": 93270
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004066060336237784,
+      "loss": 0.5162,
+      "step": 93280
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040656363289448576,
+      "loss": 0.489,
+      "step": 93290
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004065212321651933,
+      "loss": 0.5209,
+      "step": 93300
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040647883143590075,
+      "loss": 0.5228,
+      "step": 93310
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040643643070660817,
+      "loss": 0.5394,
+      "step": 93320
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040639402997731563,
+      "loss": 0.4801,
+      "step": 93330
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004063516292480231,
+      "loss": 0.4803,
+      "step": 93340
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004063092285187305,
+      "loss": 0.521,
+      "step": 93350
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040626682778943803,
+      "loss": 0.5539,
+      "step": 93360
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004062244270601455,
+      "loss": 0.5832,
+      "step": 93370
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004061820263308529,
+      "loss": 0.5612,
+      "step": 93380
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004061396256015604,
+      "loss": 0.512,
+      "step": 93390
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004060972248722678,
+      "loss": 0.536,
+      "step": 93400
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040605482414297526,
+      "loss": 0.5223,
+      "step": 93410
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004060124234136828,
+      "loss": 0.5754,
+      "step": 93420
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040597002268439014,
+      "loss": 0.5612,
+      "step": 93430
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040592762195509766,
+      "loss": 0.4364,
+      "step": 93440
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004058852212258051,
+      "loss": 0.5766,
+      "step": 93450
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040584282049651254,
+      "loss": 0.5721,
+      "step": 93460
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040580041976722,
+      "loss": 0.5364,
+      "step": 93470
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040575801903792747,
+      "loss": 0.6067,
+      "step": 93480
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004057156183086349,
+      "loss": 0.631,
+      "step": 93490
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004056732175793424,
+      "loss": 0.5805,
+      "step": 93500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040563081685004987,
+      "loss": 0.5113,
+      "step": 93510
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004055884161207573,
+      "loss": 0.5727,
+      "step": 93520
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040554601539146475,
+      "loss": 0.472,
+      "step": 93530
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004055036146621722,
+      "loss": 0.4793,
+      "step": 93540
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040546121393287963,
+      "loss": 0.5495,
+      "step": 93550
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004054188132035871,
+      "loss": 0.4998,
+      "step": 93560
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004053764124742946,
+      "loss": 0.5746,
+      "step": 93570
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040533401174500203,
+      "loss": 0.55,
+      "step": 93580
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004052916110157095,
+      "loss": 0.6305,
+      "step": 93590
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040524921028641696,
+      "loss": 0.5137,
+      "step": 93600
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004052068095571244,
+      "loss": 0.5035,
+      "step": 93610
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040516440882783184,
+      "loss": 0.5134,
+      "step": 93620
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040512200809853936,
+      "loss": 0.5068,
+      "step": 93630
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004050796073692467,
+      "loss": 0.5489,
+      "step": 93640
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040503720663995424,
+      "loss": 0.5548,
+      "step": 93650
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004049948059106617,
+      "loss": 0.5947,
+      "step": 93660
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004049524051813691,
+      "loss": 0.501,
+      "step": 93670
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004049100044520766,
+      "loss": 0.5387,
+      "step": 93680
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004048676037227841,
+      "loss": 0.6137,
+      "step": 93690
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040482520299349147,
+      "loss": 0.5107,
+      "step": 93700
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.000404782802264199,
+      "loss": 0.5551,
+      "step": 93710
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004047404015349064,
+      "loss": 0.5107,
+      "step": 93720
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040469800080561387,
+      "loss": 0.4905,
+      "step": 93730
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040465560007632134,
+      "loss": 0.5181,
+      "step": 93740
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040461319934702875,
+      "loss": 0.5367,
+      "step": 93750
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004045707986177362,
+      "loss": 0.4941,
+      "step": 93760
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040452839788844374,
+      "loss": 0.5366,
+      "step": 93770
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004044859971591511,
+      "loss": 0.4319,
+      "step": 93780
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004044435964298586,
+      "loss": 0.6072,
+      "step": 93790
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004044011957005661,
+      "loss": 0.5317,
+      "step": 93800
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004043587949712735,
+      "loss": 0.7465,
+      "step": 93810
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040431639424198096,
+      "loss": 0.5089,
+      "step": 93820
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004042739935126885,
+      "loss": 0.5102,
+      "step": 93830
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040423159278339584,
+      "loss": 0.4869,
+      "step": 93840
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040418919205410336,
+      "loss": 0.4816,
+      "step": 93850
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040414679132481083,
+      "loss": 0.5329,
+      "step": 93860
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040410439059551824,
+      "loss": 0.7165,
+      "step": 93870
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004040619898662257,
+      "loss": 0.5129,
+      "step": 93880
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040401958913693323,
+      "loss": 0.5452,
+      "step": 93890
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004039771884076406,
+      "loss": 0.5669,
+      "step": 93900
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004039347876783481,
+      "loss": 0.462,
+      "step": 93910
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004038923869490556,
+      "loss": 0.5452,
+      "step": 93920
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.000403849986219763,
+      "loss": 0.5644,
+      "step": 93930
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040380758549047046,
+      "loss": 0.5424,
+      "step": 93940
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.000403765184761178,
+      "loss": 0.524,
+      "step": 93950
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040372278403188533,
+      "loss": 0.5129,
+      "step": 93960
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040368038330259286,
+      "loss": 0.4136,
+      "step": 93970
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004036379825733003,
+      "loss": 0.556,
+      "step": 93980
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040359558184400774,
+      "loss": 0.4477,
+      "step": 93990
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004035531811147152,
+      "loss": 0.4738,
+      "step": 94000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004035107803854226,
+      "loss": 0.5402,
+      "step": 94010
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004034683796561301,
+      "loss": 0.5599,
+      "step": 94020
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004034259789268376,
+      "loss": 0.5062,
+      "step": 94030
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040338357819754496,
+      "loss": 0.4681,
+      "step": 94040
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004033411774682525,
+      "loss": 0.5212,
+      "step": 94050
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040329877673895995,
+      "loss": 0.569,
+      "step": 94060
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040325637600966736,
+      "loss": 0.4875,
+      "step": 94070
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040321397528037483,
+      "loss": 0.4897,
+      "step": 94080
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004031715745510823,
+      "loss": 0.5268,
+      "step": 94090
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004031291738217897,
+      "loss": 0.4776,
+      "step": 94100
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040308677309249723,
+      "loss": 0.539,
+      "step": 94110
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004030443723632047,
+      "loss": 0.6051,
+      "step": 94120
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004030019716339121,
+      "loss": 0.5553,
+      "step": 94130
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004029595709046196,
+      "loss": 0.4881,
+      "step": 94140
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040291717017532704,
+      "loss": 0.5244,
+      "step": 94150
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040287476944603445,
+      "loss": 0.4655,
+      "step": 94160
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.000402832368716742,
+      "loss": 0.4705,
+      "step": 94170
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040278996798744944,
+      "loss": 0.4914,
+      "step": 94180
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040274756725815685,
+      "loss": 0.5742,
+      "step": 94190
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004027051665288643,
+      "loss": 0.5076,
+      "step": 94200
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004026627657995718,
+      "loss": 0.5307,
+      "step": 94210
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004026203650702792,
+      "loss": 0.5031,
+      "step": 94220
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040257796434098667,
+      "loss": 0.5419,
+      "step": 94230
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004025355636116942,
+      "loss": 0.5102,
+      "step": 94240
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004024931628824016,
+      "loss": 0.5271,
+      "step": 94250
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040245076215310907,
+      "loss": 0.5472,
+      "step": 94260
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040240836142381653,
+      "loss": 0.5157,
+      "step": 94270
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040236596069452395,
+      "loss": 0.5035,
+      "step": 94280
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004023235599652314,
+      "loss": 0.4929,
+      "step": 94290
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040228115923593893,
+      "loss": 0.4663,
+      "step": 94300
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004022387585066463,
+      "loss": 0.5273,
+      "step": 94310
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004021963577773538,
+      "loss": 0.499,
+      "step": 94320
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004021539570480612,
+      "loss": 0.562,
+      "step": 94330
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004021115563187687,
+      "loss": 0.5678,
+      "step": 94340
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040206915558947616,
+      "loss": 0.5679,
+      "step": 94350
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040202675486018357,
+      "loss": 0.4466,
+      "step": 94360
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040198435413089104,
+      "loss": 0.4638,
+      "step": 94370
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040194195340159856,
+      "loss": 0.4542,
+      "step": 94380
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004018995526723059,
+      "loss": 0.5485,
+      "step": 94390
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040185715194301344,
+      "loss": 0.6233,
+      "step": 94400
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004018147512137209,
+      "loss": 0.4927,
+      "step": 94410
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004017723504844283,
+      "loss": 0.48,
+      "step": 94420
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004017299497551358,
+      "loss": 0.6609,
+      "step": 94430
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004016875490258433,
+      "loss": 0.5263,
+      "step": 94440
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040164514829655067,
+      "loss": 0.5226,
+      "step": 94450
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004016027475672582,
+      "loss": 0.6089,
+      "step": 94460
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040156034683796565,
+      "loss": 0.5618,
+      "step": 94470
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040151794610867307,
+      "loss": 0.5268,
+      "step": 94480
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040147554537938053,
+      "loss": 0.5104,
+      "step": 94490
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040143314465008805,
+      "loss": 0.5089,
+      "step": 94500
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004013907439207954,
+      "loss": 0.4649,
+      "step": 94510
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040134834319150293,
+      "loss": 0.5386,
+      "step": 94520
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004013059424622104,
+      "loss": 0.6264,
+      "step": 94530
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004012635417329178,
+      "loss": 0.5977,
+      "step": 94540
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004012211410036253,
+      "loss": 0.4853,
+      "step": 94550
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004011787402743328,
+      "loss": 0.4641,
+      "step": 94560
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040113633954504016,
+      "loss": 0.5532,
+      "step": 94570
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004010939388157477,
+      "loss": 0.5826,
+      "step": 94580
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040105153808645515,
+      "loss": 0.573,
+      "step": 94590
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040100913735716256,
+      "loss": 0.5345,
+      "step": 94600
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040096673662787,
+      "loss": 0.4824,
+      "step": 94610
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004009243358985775,
+      "loss": 0.5414,
+      "step": 94620
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004008819351692849,
+      "loss": 0.4844,
+      "step": 94630
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004008395344399924,
+      "loss": 0.4921,
+      "step": 94640
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004007971337106998,
+      "loss": 0.4602,
+      "step": 94650
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004007547329814073,
+      "loss": 0.4741,
+      "step": 94660
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040071233225211477,
+      "loss": 0.4545,
+      "step": 94670
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004006699315228222,
+      "loss": 0.4483,
+      "step": 94680
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040062753079352965,
+      "loss": 0.4483,
+      "step": 94690
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040058513006423717,
+      "loss": 0.4667,
+      "step": 94700
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040054272933494453,
+      "loss": 0.4741,
+      "step": 94710
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040050032860565205,
+      "loss": 0.4716,
+      "step": 94720
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004004579278763595,
+      "loss": 0.4809,
+      "step": 94730
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040041552714706693,
+      "loss": 0.3972,
+      "step": 94740
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004003731264177744,
+      "loss": 0.4351,
+      "step": 94750
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040033072568848186,
+      "loss": 0.3956,
+      "step": 94760
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004002883249591893,
+      "loss": 0.5638,
+      "step": 94770
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004002459242298968,
+      "loss": 0.4518,
+      "step": 94780
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040020352350060427,
+      "loss": 0.4385,
+      "step": 94790
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004001611227713117,
+      "loss": 0.5426,
+      "step": 94800
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040011872204201914,
+      "loss": 0.4678,
+      "step": 94810
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004000763213127266,
+      "loss": 0.496,
+      "step": 94820
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.000400033920583434,
+      "loss": 0.4843,
+      "step": 94830
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0003999915198541415,
+      "loss": 0.4491,
+      "step": 94840
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00039994911912484896,
+      "loss": 0.4477,
+      "step": 94850
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0003999067183955564,
+      "loss": 0.4479,
+      "step": 94860
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0003998643176662639,
+      "loss": 0.4569,
+      "step": 94870
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0003998219169369713,
+      "loss": 0.5829,
+      "step": 94880
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00039977951620767877,
+      "loss": 0.444,
+      "step": 94890
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00039973711547838624,
+      "loss": 0.4692,
+      "step": 94900
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0003996947147490937,
+      "loss": 0.4274,
+      "step": 94910
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00039965231401980117,
+      "loss": 0.4928,
+      "step": 94920
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00039960991329050864,
+      "loss": 0.596,
+      "step": 94930
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00039956751256121605,
+      "loss": 0.4284,
+      "step": 94940
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003995251118319235,
+      "loss": 0.4383,
+      "step": 94950
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.000399482711102631,
+      "loss": 0.4631,
+      "step": 94960
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00039944031037333845,
+      "loss": 0.4764,
+      "step": 94970
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00039939790964404586,
+      "loss": 0.5072,
+      "step": 94980
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003993555089147534,
+      "loss": 0.4626,
+      "step": 94990
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003993131081854608,
+      "loss": 0.4766,
+      "step": 95000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00039927070745616826,
+      "loss": 0.4916,
+      "step": 95010
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003992283067268757,
+      "loss": 0.4339,
+      "step": 95020
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003991859059975832,
+      "loss": 0.5471,
+      "step": 95030
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003991435052682906,
+      "loss": 0.4572,
+      "step": 95040
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003991011045389981,
+      "loss": 0.485,
+      "step": 95050
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00039905870380970554,
+      "loss": 0.4821,
+      "step": 95060
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.000399016303080413,
+      "loss": 0.5379,
+      "step": 95070
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003989739023511204,
+      "loss": 0.4164,
+      "step": 95080
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00039893150162182794,
+      "loss": 0.3894,
+      "step": 95090
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00039888910089253536,
+      "loss": 0.4528,
+      "step": 95100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003988467001632428,
+      "loss": 0.5317,
+      "step": 95110
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003988042994339503,
+      "loss": 0.4585,
+      "step": 95120
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00039876189870465776,
+      "loss": 0.528,
+      "step": 95130
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00039871949797536517,
+      "loss": 0.4344,
+      "step": 95140
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003986770972460727,
+      "loss": 0.584,
+      "step": 95150
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003986346965167801,
+      "loss": 0.4631,
+      "step": 95160
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00039859229578748757,
+      "loss": 0.4432,
+      "step": 95170
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.000398549895058195,
+      "loss": 0.5725,
+      "step": 95180
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003985074943289025,
+      "loss": 0.4526,
+      "step": 95190
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003984650935996099,
+      "loss": 0.5108,
+      "step": 95200
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003984226928703174,
+      "loss": 0.4305,
+      "step": 95210
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00039838029214102485,
+      "loss": 0.4318,
+      "step": 95220
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003983378914117323,
+      "loss": 0.5558,
+      "step": 95230
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00039829549068243973,
+      "loss": 0.4631,
+      "step": 95240
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00039825308995314725,
+      "loss": 0.5002,
+      "step": 95250
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00039821068922385466,
+      "loss": 0.4408,
+      "step": 95260
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00039816828849456213,
+      "loss": 0.4102,
+      "step": 95270
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003981258877652696,
+      "loss": 0.5139,
+      "step": 95280
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00039808348703597706,
+      "loss": 0.4366,
+      "step": 95290
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003980410863066845,
+      "loss": 0.4692,
+      "step": 95300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.000397998685577392,
+      "loss": 0.4697,
+      "step": 95310
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003979562848480994,
+      "loss": 0.3841,
+      "step": 95320
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003979138841188069,
+      "loss": 0.4393,
+      "step": 95330
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003978714833895143,
+      "loss": 0.4713,
+      "step": 95340
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003978290826602218,
+      "loss": 0.4351,
+      "step": 95350
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003977866819309292,
+      "loss": 0.4879,
+      "step": 95360
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003977442812016367,
+      "loss": 0.5067,
+      "step": 95370
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00039770188047234416,
+      "loss": 0.4959,
+      "step": 95380
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003976594797430516,
+      "loss": 0.3938,
+      "step": 95390
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00039761707901375903,
+      "loss": 0.452,
+      "step": 95400
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003975746782844665,
+      "loss": 0.4825,
+      "step": 95410
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039753227755517397,
+      "loss": 0.3996,
+      "step": 95420
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039748987682588144,
+      "loss": 0.5005,
+      "step": 95430
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003974474760965889,
+      "loss": 0.4829,
+      "step": 95440
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039740507536729637,
+      "loss": 0.5097,
+      "step": 95450
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003973626746380038,
+      "loss": 0.4815,
+      "step": 95460
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039732027390871125,
+      "loss": 0.442,
+      "step": 95470
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003972778731794187,
+      "loss": 0.4451,
+      "step": 95480
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003972354724501262,
+      "loss": 0.4121,
+      "step": 95490
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003971930717208336,
+      "loss": 0.4242,
+      "step": 95500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039715067099154106,
+      "loss": 0.5157,
+      "step": 95510
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039710827026224853,
+      "loss": 0.4918,
+      "step": 95520
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.000397065869532956,
+      "loss": 0.5394,
+      "step": 95530
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039702346880366346,
+      "loss": 0.4983,
+      "step": 95540
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003969810680743709,
+      "loss": 0.5035,
+      "step": 95550
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039693866734507834,
+      "loss": 0.4643,
+      "step": 95560
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003968962666157858,
+      "loss": 0.422,
+      "step": 95570
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003968538658864933,
+      "loss": 0.421,
+      "step": 95580
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003968114651572007,
+      "loss": 0.4223,
+      "step": 95590
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003967690644279082,
+      "loss": 0.5198,
+      "step": 95600
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003967266636986156,
+      "loss": 0.5642,
+      "step": 95610
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003966842629693231,
+      "loss": 0.4865,
+      "step": 95620
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003966418622400305,
+      "loss": 0.5249,
+      "step": 95630
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.000396599461510738,
+      "loss": 0.4655,
+      "step": 95640
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039655706078144543,
+      "loss": 0.4124,
+      "step": 95650
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003965146600521529,
+      "loss": 0.4791,
+      "step": 95660
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039647225932286037,
+      "loss": 0.3911,
+      "step": 95670
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039642985859356783,
+      "loss": 0.5073,
+      "step": 95680
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039638745786427525,
+      "loss": 0.4179,
+      "step": 95690
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039634505713498277,
+      "loss": 0.5227,
+      "step": 95700
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003963026564056902,
+      "loss": 0.4904,
+      "step": 95710
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039626025567639765,
+      "loss": 0.47,
+      "step": 95720
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003962178549471051,
+      "loss": 0.5209,
+      "step": 95730
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003961754542178126,
+      "loss": 0.4696,
+      "step": 95740
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039613305348852,
+      "loss": 0.5272,
+      "step": 95750
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003960906527592275,
+      "loss": 0.4536,
+      "step": 95760
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003960482520299349,
+      "loss": 0.415,
+      "step": 95770
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003960058513006424,
+      "loss": 0.424,
+      "step": 95780
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003959634505713498,
+      "loss": 0.464,
+      "step": 95790
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003959210498420573,
+      "loss": 0.4969,
+      "step": 95800
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039587864911276474,
+      "loss": 0.5042,
+      "step": 95810
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003958362483834722,
+      "loss": 0.4704,
+      "step": 95820
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003957938476541797,
+      "loss": 0.4999,
+      "step": 95830
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039575144692488714,
+      "loss": 0.4309,
+      "step": 95840
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039570904619559455,
+      "loss": 0.4319,
+      "step": 95850
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003956666454663021,
+      "loss": 0.482,
+      "step": 95860
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003956242447370095,
+      "loss": 0.4202,
+      "step": 95870
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039558184400771695,
+      "loss": 0.4843,
+      "step": 95880
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003955394432784244,
+      "loss": 0.5194,
+      "step": 95890
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003954970425491319,
+      "loss": 0.5083,
+      "step": 95900
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003954546418198393,
+      "loss": 0.3892,
+      "step": 95910
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003954122410905468,
+      "loss": 0.449,
+      "step": 95920
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039536984036125423,
+      "loss": 0.4672,
+      "step": 95930
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003953274396319617,
+      "loss": 0.5553,
+      "step": 95940
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003952850389026691,
+      "loss": 0.4667,
+      "step": 95950
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039524263817337663,
+      "loss": 0.4841,
+      "step": 95960
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039520023744408405,
+      "loss": 0.5358,
+      "step": 95970
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003951578367147915,
+      "loss": 0.4394,
+      "step": 95980
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.000395115435985499,
+      "loss": 0.5158,
+      "step": 95990
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039507303525620645,
+      "loss": 0.4772,
+      "step": 96000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039503063452691386,
+      "loss": 0.4481,
+      "step": 96010
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003949882337976214,
+      "loss": 0.479,
+      "step": 96020
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003949458330683288,
+      "loss": 0.3874,
+      "step": 96030
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039490343233903626,
+      "loss": 0.4316,
+      "step": 96040
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003948610316097437,
+      "loss": 0.4906,
+      "step": 96050
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003948186308804512,
+      "loss": 0.597,
+      "step": 96060
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003947762301511586,
+      "loss": 0.5307,
+      "step": 96070
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039473382942186607,
+      "loss": 0.3736,
+      "step": 96080
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039469142869257354,
+      "loss": 0.4193,
+      "step": 96090
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.000394649027963281,
+      "loss": 0.3996,
+      "step": 96100
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003946066272339884,
+      "loss": 0.5047,
+      "step": 96110
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003945642265046959,
+      "loss": 0.5089,
+      "step": 96120
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039452182577540335,
+      "loss": 0.4591,
+      "step": 96130
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003944794250461108,
+      "loss": 0.4939,
+      "step": 96140
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003944370243168183,
+      "loss": 0.4817,
+      "step": 96150
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003943946235875257,
+      "loss": 0.5551,
+      "step": 96160
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039435222285823316,
+      "loss": 0.502,
+      "step": 96170
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039430982212894063,
+      "loss": 0.5783,
+      "step": 96180
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003942674213996481,
+      "loss": 0.5376,
+      "step": 96190
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039422502067035556,
+      "loss": 0.5237,
+      "step": 96200
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039418261994106303,
+      "loss": 0.4875,
+      "step": 96210
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039414021921177044,
+      "loss": 0.4719,
+      "step": 96220
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003940978184824779,
+      "loss": 0.4378,
+      "step": 96230
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003940554177531854,
+      "loss": 0.4231,
+      "step": 96240
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039401301702389284,
+      "loss": 0.4636,
+      "step": 96250
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039397061629460026,
+      "loss": 0.474,
+      "step": 96260
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003939282155653077,
+      "loss": 0.467,
+      "step": 96270
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003938858148360152,
+      "loss": 0.4729,
+      "step": 96280
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039384341410672266,
+      "loss": 0.4401,
+      "step": 96290
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039380101337743007,
+      "loss": 0.5589,
+      "step": 96300
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003937586126481376,
+      "loss": 0.4501,
+      "step": 96310
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.000393716211918845,
+      "loss": 0.4337,
+      "step": 96320
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039367381118955247,
+      "loss": 0.4523,
+      "step": 96330
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039363141046025994,
+      "loss": 0.5298,
+      "step": 96340
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003935890097309674,
+      "loss": 0.5092,
+      "step": 96350
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003935466090016748,
+      "loss": 0.4508,
+      "step": 96360
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039350420827238234,
+      "loss": 0.4336,
+      "step": 96370
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039346180754308975,
+      "loss": 0.4628,
+      "step": 96380
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003934194068137972,
+      "loss": 0.4938,
+      "step": 96390
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039337700608450463,
+      "loss": 0.547,
+      "step": 96400
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039333460535521215,
+      "loss": 0.4567,
+      "step": 96410
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039329220462591956,
+      "loss": 0.4711,
+      "step": 96420
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039324980389662703,
+      "loss": 0.4571,
+      "step": 96430
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003932074031673345,
+      "loss": 0.5278,
+      "step": 96440
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039316500243804196,
+      "loss": 0.4761,
+      "step": 96450
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003931226017087494,
+      "loss": 0.4586,
+      "step": 96460
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003930802009794569,
+      "loss": 0.5102,
+      "step": 96470
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003930378002501643,
+      "loss": 0.4727,
+      "step": 96480
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003929953995208718,
+      "loss": 0.5176,
+      "step": 96490
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039295299879157924,
+      "loss": 0.4586,
+      "step": 96500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003929105980622867,
+      "loss": 0.5411,
+      "step": 96510
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003928681973329941,
+      "loss": 0.5188,
+      "step": 96520
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039282579660370164,
+      "loss": 0.4543,
+      "step": 96530
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039278339587440906,
+      "loss": 0.4383,
+      "step": 96540
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003927409951451165,
+      "loss": 0.5262,
+      "step": 96550
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039269859441582394,
+      "loss": 0.4256,
+      "step": 96560
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039265619368653146,
+      "loss": 0.4418,
+      "step": 96570
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039261379295723887,
+      "loss": 0.4116,
+      "step": 96580
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039257139222794634,
+      "loss": 0.489,
+      "step": 96590
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003925289914986538,
+      "loss": 0.4934,
+      "step": 96600
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039248659076936127,
+      "loss": 0.4959,
+      "step": 96610
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003924441900400687,
+      "loss": 0.518,
+      "step": 96620
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003924017893107762,
+      "loss": 0.417,
+      "step": 96630
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003923593885814836,
+      "loss": 0.4185,
+      "step": 96640
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003923169878521911,
+      "loss": 0.4338,
+      "step": 96650
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039227458712289855,
+      "loss": 0.528,
+      "step": 96660
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.000392232186393606,
+      "loss": 0.5685,
+      "step": 96670
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039218978566431343,
+      "loss": 0.4494,
+      "step": 96680
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039214738493502095,
+      "loss": 0.4542,
+      "step": 96690
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039210498420572836,
+      "loss": 0.4849,
+      "step": 96700
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039206258347643583,
+      "loss": 0.5136,
+      "step": 96710
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039202018274714324,
+      "loss": 0.5586,
+      "step": 96720
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039197778201785076,
+      "loss": 0.4868,
+      "step": 96730
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003919353812885582,
+      "loss": 0.5016,
+      "step": 96740
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039189298055926564,
+      "loss": 0.4546,
+      "step": 96750
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003918505798299731,
+      "loss": 0.4951,
+      "step": 96760
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003918081791006806,
+      "loss": 0.4818,
+      "step": 96770
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.000391765778371388,
+      "loss": 0.4039,
+      "step": 96780
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039172337764209545,
+      "loss": 0.3697,
+      "step": 96790
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003916809769128029,
+      "loss": 0.4654,
+      "step": 96800
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003916385761835104,
+      "loss": 0.5017,
+      "step": 96810
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039159617545421786,
+      "loss": 0.4317,
+      "step": 96820
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039155377472492527,
+      "loss": 0.4901,
+      "step": 96830
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039151137399563273,
+      "loss": 0.4857,
+      "step": 96840
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003914689732663402,
+      "loss": 0.5235,
+      "step": 96850
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039142657253704767,
+      "loss": 0.4437,
+      "step": 96860
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003913841718077551,
+      "loss": 0.4663,
+      "step": 96870
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039134177107846255,
+      "loss": 0.4361,
+      "step": 96880
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039129937034917,
+      "loss": 0.4142,
+      "step": 96890
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003912569696198775,
+      "loss": 0.4221,
+      "step": 96900
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003912145688905849,
+      "loss": 0.5427,
+      "step": 96910
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003911721681612924,
+      "loss": 0.5554,
+      "step": 96920
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039112976743199983,
+      "loss": 0.5151,
+      "step": 96930
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003910873667027073,
+      "loss": 0.4496,
+      "step": 96940
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039104496597341476,
+      "loss": 0.4385,
+      "step": 96950
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039100256524412223,
+      "loss": 0.4683,
+      "step": 96960
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039096016451482964,
+      "loss": 0.4876,
+      "step": 96970
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039091776378553716,
+      "loss": 0.4694,
+      "step": 96980
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003908753630562446,
+      "loss": 0.4382,
+      "step": 96990
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039083296232695204,
+      "loss": 0.5246,
+      "step": 97000
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039079056159765945,
+      "loss": 0.4855,
+      "step": 97010
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.000390748160868367,
+      "loss": 0.5445,
+      "step": 97020
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003907057601390744,
+      "loss": 0.4906,
+      "step": 97030
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039066335940978185,
+      "loss": 0.5156,
+      "step": 97040
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003906209586804893,
+      "loss": 0.5039,
+      "step": 97050
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003905785579511968,
+      "loss": 0.3811,
+      "step": 97060
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003905361572219042,
+      "loss": 0.4462,
+      "step": 97070
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003904937564926117,
+      "loss": 0.5163,
+      "step": 97080
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039045135576331913,
+      "loss": 0.4384,
+      "step": 97090
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003904089550340266,
+      "loss": 0.4726,
+      "step": 97100
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039036655430473407,
+      "loss": 0.4689,
+      "step": 97110
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039032415357544153,
+      "loss": 0.4517,
+      "step": 97120
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039028175284614895,
+      "loss": 0.4629,
+      "step": 97130
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039023935211685647,
+      "loss": 0.5007,
+      "step": 97140
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003901969513875639,
+      "loss": 0.4392,
+      "step": 97150
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039015455065827135,
+      "loss": 0.543,
+      "step": 97160
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039011214992897876,
+      "loss": 0.5684,
+      "step": 97170
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003900697491996863,
+      "loss": 0.5694,
+      "step": 97180
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003900273484703937,
+      "loss": 0.521,
+      "step": 97190
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00038998494774110116,
+      "loss": 0.4695,
+      "step": 97200
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003899425470118086,
+      "loss": 0.403,
+      "step": 97210
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003899001462825161,
+      "loss": 0.4519,
+      "step": 97220
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003898577455532235,
+      "loss": 0.4644,
+      "step": 97230
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.000389815344823931,
+      "loss": 0.4333,
+      "step": 97240
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00038977294409463844,
+      "loss": 0.5098,
+      "step": 97250
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003897305433653459,
+      "loss": 0.4364,
+      "step": 97260
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00038968814263605337,
+      "loss": 0.5358,
+      "step": 97270
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00038964574190676084,
+      "loss": 0.4475,
+      "step": 97280
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00038960334117746825,
+      "loss": 0.4379,
+      "step": 97290
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003895609404481758,
+      "loss": 0.5201,
+      "step": 97300
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003895185397188832,
+      "loss": 0.3804,
+      "step": 97310
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038947613898959065,
+      "loss": 0.4451,
+      "step": 97320
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038943373826029806,
+      "loss": 0.5364,
+      "step": 97330
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003893913375310056,
+      "loss": 0.4478,
+      "step": 97340
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.000389348936801713,
+      "loss": 0.4783,
+      "step": 97350
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038930653607242047,
+      "loss": 0.4111,
+      "step": 97360
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038926413534312793,
+      "loss": 0.485,
+      "step": 97370
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003892217346138354,
+      "loss": 0.5128,
+      "step": 97380
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003891793338845428,
+      "loss": 0.4971,
+      "step": 97390
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003891369331552503,
+      "loss": 0.4828,
+      "step": 97400
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038909453242595775,
+      "loss": 0.5476,
+      "step": 97410
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003890521316966652,
+      "loss": 0.5443,
+      "step": 97420
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003890097309673727,
+      "loss": 0.4828,
+      "step": 97430
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038896733023808015,
+      "loss": 0.4526,
+      "step": 97440
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038892492950878756,
+      "loss": 0.5007,
+      "step": 97450
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.000388882528779495,
+      "loss": 0.4236,
+      "step": 97460
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003888401280502025,
+      "loss": 0.4574,
+      "step": 97470
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038879772732090996,
+      "loss": 0.5109,
+      "step": 97480
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038875532659161737,
+      "loss": 0.4673,
+      "step": 97490
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038871292586232484,
+      "loss": 0.4868,
+      "step": 97500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003886705251330323,
+      "loss": 0.3506,
+      "step": 97510
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038862812440373977,
+      "loss": 0.4401,
+      "step": 97520
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038858572367444724,
+      "loss": 0.4874,
+      "step": 97530
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038854332294515465,
+      "loss": 0.4959,
+      "step": 97540
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003885009222158621,
+      "loss": 0.4939,
+      "step": 97550
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003884585214865696,
+      "loss": 0.4449,
+      "step": 97560
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00038841612075727705,
+      "loss": 0.4768,
+      "step": 97570
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00038837372002798446,
+      "loss": 0.5877,
+      "step": 97580
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.000388331319298692,
+      "loss": 0.4701,
+      "step": 97590
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003882889185693994,
+      "loss": 0.4629,
+      "step": 97600
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00038824651784010686,
+      "loss": 0.4641,
+      "step": 97610
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003882041171108143,
+      "loss": 0.4669,
+      "step": 97620
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003881617163815218,
+      "loss": 0.4725,
+      "step": 97630
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003881193156522292,
+      "loss": 0.5087,
+      "step": 97640
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003880769149229367,
+      "loss": 0.418,
+      "step": 97650
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00038803451419364414,
+      "loss": 0.5191,
+      "step": 97660
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003879921134643516,
+      "loss": 0.4028,
+      "step": 97670
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.000387949712735059,
+      "loss": 0.4205,
+      "step": 97680
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00038790731200576654,
+      "loss": 0.4399,
+      "step": 97690
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00038786491127647396,
+      "loss": 0.4639,
+      "step": 97700
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003878225105471814,
+      "loss": 0.5034,
+      "step": 97710
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003877801098178889,
+      "loss": 0.4934,
+      "step": 97720
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00038773770908859636,
+      "loss": 0.492,
+      "step": 97730
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00038769530835930377,
+      "loss": 0.4656,
+      "step": 97740
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003876529076300113,
+      "loss": 0.4129,
+      "step": 97750
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003876105069007187,
+      "loss": 0.4729,
+      "step": 97760
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00038756810617142617,
+      "loss": 0.5426,
+      "step": 97770
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003875257054421336,
+      "loss": 0.4186,
+      "step": 97780
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003874833047128411,
+      "loss": 0.4532,
+      "step": 97790
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003874409039835485,
+      "loss": 0.4787,
+      "step": 97800
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.000387398503254256,
+      "loss": 0.4714,
+      "step": 97810
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038735610252496345,
+      "loss": 0.4903,
+      "step": 97820
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003873137017956709,
+      "loss": 0.4519,
+      "step": 97830
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038727130106637833,
+      "loss": 0.5078,
+      "step": 97840
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038722890033708585,
+      "loss": 0.4832,
+      "step": 97850
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038718649960779326,
+      "loss": 0.5123,
+      "step": 97860
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038714409887850073,
+      "loss": 0.4526,
+      "step": 97870
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003871016981492082,
+      "loss": 0.5277,
+      "step": 97880
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038705929741991566,
+      "loss": 0.4701,
+      "step": 97890
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003870168966906231,
+      "loss": 0.4577,
+      "step": 97900
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003869744959613306,
+      "loss": 0.4158,
+      "step": 97910
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.000386932095232038,
+      "loss": 0.4899,
+      "step": 97920
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003868896945027455,
+      "loss": 0.601,
+      "step": 97930
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003868472937734529,
+      "loss": 0.4681,
+      "step": 97940
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003868048930441604,
+      "loss": 0.4641,
+      "step": 97950
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003867624923148678,
+      "loss": 0.5471,
+      "step": 97960
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003867200915855753,
+      "loss": 0.5078,
+      "step": 97970
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038667769085628276,
+      "loss": 0.4574,
+      "step": 97980
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003866352901269902,
+      "loss": 0.4623,
+      "step": 97990
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038659288939769764,
+      "loss": 0.518,
+      "step": 98000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038655048866840516,
+      "loss": 0.5162,
+      "step": 98010
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038650808793911257,
+      "loss": 0.4825,
+      "step": 98020
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038646568720982004,
+      "loss": 0.5356,
+      "step": 98030
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003864232864805275,
+      "loss": 0.386,
+      "step": 98040
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038638088575123497,
+      "loss": 0.4012,
+      "step": 98050
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003863384850219424,
+      "loss": 0.4765,
+      "step": 98060
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038629608429264985,
+      "loss": 0.4969,
+      "step": 98070
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003862536835633573,
+      "loss": 0.5055,
+      "step": 98080
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003862112828340648,
+      "loss": 0.4125,
+      "step": 98090
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003861688821047722,
+      "loss": 0.5299,
+      "step": 98100
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038612648137547966,
+      "loss": 0.4013,
+      "step": 98110
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038608408064618713,
+      "loss": 0.5099,
+      "step": 98120
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003860416799168946,
+      "loss": 0.3924,
+      "step": 98130
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038599927918760206,
+      "loss": 0.4424,
+      "step": 98140
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003859568784583095,
+      "loss": 0.495,
+      "step": 98150
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038591447772901694,
+      "loss": 0.4864,
+      "step": 98160
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003858720769997244,
+      "loss": 0.458,
+      "step": 98170
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003858296762704319,
+      "loss": 0.4734,
+      "step": 98180
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038578727554113934,
+      "loss": 0.5201,
+      "step": 98190
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003857448748118468,
+      "loss": 0.5129,
+      "step": 98200
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003857024740825542,
+      "loss": 0.4544,
+      "step": 98210
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003856600733532617,
+      "loss": 0.4701,
+      "step": 98220
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038561767262396915,
+      "loss": 0.4781,
+      "step": 98230
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003855752718946766,
+      "loss": 0.4573,
+      "step": 98240
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038553287116538403,
+      "loss": 0.462,
+      "step": 98250
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003854904704360915,
+      "loss": 0.5033,
+      "step": 98260
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038544806970679897,
+      "loss": 0.4977,
+      "step": 98270
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038540566897750643,
+      "loss": 0.5035,
+      "step": 98280
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038536326824821385,
+      "loss": 0.4727,
+      "step": 98290
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038532086751892137,
+      "loss": 0.4455,
+      "step": 98300
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003852784667896288,
+      "loss": 0.5257,
+      "step": 98310
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038523606606033625,
+      "loss": 0.5465,
+      "step": 98320
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003851936653310437,
+      "loss": 0.4572,
+      "step": 98330
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003851512646017512,
+      "loss": 0.4427,
+      "step": 98340
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003851088638724586,
+      "loss": 0.5449,
+      "step": 98350
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003850664631431661,
+      "loss": 0.4233,
+      "step": 98360
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003850240624138735,
+      "loss": 0.4447,
+      "step": 98370
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.000384981661684581,
+      "loss": 0.4358,
+      "step": 98380
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003849392609552884,
+      "loss": 0.4116,
+      "step": 98390
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038489686022599593,
+      "loss": 0.4757,
+      "step": 98400
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038485445949670334,
+      "loss": 0.4725,
+      "step": 98410
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003848120587674108,
+      "loss": 0.4837,
+      "step": 98420
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003847696580381183,
+      "loss": 0.4455,
+      "step": 98430
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038472725730882574,
+      "loss": 0.4123,
+      "step": 98440
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038468485657953315,
+      "loss": 0.4927,
+      "step": 98450
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003846424558502407,
+      "loss": 0.4349,
+      "step": 98460
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003846000551209481,
+      "loss": 0.5418,
+      "step": 98470
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038455765439165555,
+      "loss": 0.5195,
+      "step": 98480
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.000384515253662363,
+      "loss": 0.4384,
+      "step": 98490
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003844728529330705,
+      "loss": 0.3954,
+      "step": 98500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003844304522037779,
+      "loss": 0.4554,
+      "step": 98510
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003843880514744854,
+      "loss": 0.4049,
+      "step": 98520
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038434565074519283,
+      "loss": 0.4664,
+      "step": 98530
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003843032500159003,
+      "loss": 0.4463,
+      "step": 98540
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003842608492866077,
+      "loss": 0.4708,
+      "step": 98550
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038421844855731523,
+      "loss": 0.5715,
+      "step": 98560
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038417604782802265,
+      "loss": 0.5209,
+      "step": 98570
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003841336470987301,
+      "loss": 0.4115,
+      "step": 98580
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003840912463694376,
+      "loss": 0.4621,
+      "step": 98590
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038404884564014505,
+      "loss": 0.4495,
+      "step": 98600
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038400644491085246,
+      "loss": 0.4998,
+      "step": 98610
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038396404418156,
+      "loss": 0.4962,
+      "step": 98620
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003839216434522674,
+      "loss": 0.4991,
+      "step": 98630
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038387924272297486,
+      "loss": 0.4656,
+      "step": 98640
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003838368419936823,
+      "loss": 0.4955,
+      "step": 98650
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003837944412643898,
+      "loss": 0.5307,
+      "step": 98660
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003837520405350972,
+      "loss": 0.5815,
+      "step": 98670
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003837096398058047,
+      "loss": 0.4562,
+      "step": 98680
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038366723907651214,
+      "loss": 0.5252,
+      "step": 98690
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003836248383472196,
+      "loss": 0.5637,
+      "step": 98700
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.000383582437617927,
+      "loss": 0.5278,
+      "step": 98710
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038354003688863454,
+      "loss": 0.522,
+      "step": 98720
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038349763615934195,
+      "loss": 0.588,
+      "step": 98730
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003834552354300494,
+      "loss": 0.4258,
+      "step": 98740
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003834128347007569,
+      "loss": 0.4393,
+      "step": 98750
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038337043397146435,
+      "loss": 0.4584,
+      "step": 98760
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038332803324217176,
+      "loss": 0.4852,
+      "step": 98770
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038328563251287923,
+      "loss": 0.5235,
+      "step": 98780
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003832432317835867,
+      "loss": 0.4592,
+      "step": 98790
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038320083105429417,
+      "loss": 0.5108,
+      "step": 98800
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038315843032500163,
+      "loss": 0.4966,
+      "step": 98810
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038311602959570904,
+      "loss": 0.5211,
+      "step": 98820
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003830736288664165,
+      "loss": 0.5422,
+      "step": 98830
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.000383031228137124,
+      "loss": 0.5236,
+      "step": 98840
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038298882740783144,
+      "loss": 0.4361,
+      "step": 98850
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038294642667853886,
+      "loss": 0.4658,
+      "step": 98860
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003829040259492463,
+      "loss": 0.4,
+      "step": 98870
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003828616252199538,
+      "loss": 0.3967,
+      "step": 98880
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038281922449066126,
+      "loss": 0.5085,
+      "step": 98890
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038277682376136867,
+      "loss": 0.4977,
+      "step": 98900
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003827344230320762,
+      "loss": 0.5347,
+      "step": 98910
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003826920223027836,
+      "loss": 0.5013,
+      "step": 98920
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038264962157349107,
+      "loss": 0.4075,
+      "step": 98930
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038260722084419854,
+      "loss": 0.4998,
+      "step": 98940
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.000382564820114906,
+      "loss": 0.4893,
+      "step": 98950
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003825224193856134,
+      "loss": 0.4438,
+      "step": 98960
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038248001865632094,
+      "loss": 0.4782,
+      "step": 98970
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038243761792702835,
+      "loss": 0.4554,
+      "step": 98980
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003823952171977358,
+      "loss": 0.4415,
+      "step": 98990
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038235281646844323,
+      "loss": 0.5155,
+      "step": 99000
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038231041573915075,
+      "loss": 0.4368,
+      "step": 99010
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038226801500985816,
+      "loss": 0.5128,
+      "step": 99020
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038222561428056563,
+      "loss": 0.5842,
+      "step": 99030
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003821832135512731,
+      "loss": 0.5101,
+      "step": 99040
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038214081282198056,
+      "loss": 0.4631,
+      "step": 99050
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.000382098412092688,
+      "loss": 0.4853,
+      "step": 99060
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003820560113633955,
+      "loss": 0.5459,
+      "step": 99070
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003820136106341029,
+      "loss": 0.4844,
+      "step": 99080
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003819712099048104,
+      "loss": 0.4817,
+      "step": 99090
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038192880917551784,
+      "loss": 0.4181,
+      "step": 99100
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003818864084462253,
+      "loss": 0.4381,
+      "step": 99110
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003818440077169327,
+      "loss": 0.4025,
+      "step": 99120
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038180160698764024,
+      "loss": 0.4582,
+      "step": 99130
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038175920625834766,
+      "loss": 0.4998,
+      "step": 99140
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003817168055290551,
+      "loss": 0.417,
+      "step": 99150
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038167440479976254,
+      "loss": 0.4082,
+      "step": 99160
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038163200407047006,
+      "loss": 0.5306,
+      "step": 99170
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038158960334117747,
+      "loss": 0.4372,
+      "step": 99180
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038154720261188494,
+      "loss": 0.5193,
+      "step": 99190
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003815048018825924,
+      "loss": 0.5284,
+      "step": 99200
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038146240115329987,
+      "loss": 0.455,
+      "step": 99210
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003814200004240073,
+      "loss": 0.4665,
+      "step": 99220
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003813775996947148,
+      "loss": 0.5504,
+      "step": 99230
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003813351989654222,
+      "loss": 0.5824,
+      "step": 99240
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003812927982361297,
+      "loss": 0.4241,
+      "step": 99250
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038125039750683715,
+      "loss": 0.4434,
+      "step": 99260
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003812079967775446,
+      "loss": 0.4456,
+      "step": 99270
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038116559604825203,
+      "loss": 0.4553,
+      "step": 99280
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038112319531895955,
+      "loss": 0.494,
+      "step": 99290
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038108079458966696,
+      "loss": 0.571,
+      "step": 99300
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038103839386037443,
+      "loss": 0.4913,
+      "step": 99310
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038099599313108184,
+      "loss": 0.5229,
+      "step": 99320
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038095359240178936,
+      "loss": 0.5163,
+      "step": 99330
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003809111916724968,
+      "loss": 0.4645,
+      "step": 99340
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038086879094320424,
+      "loss": 0.5403,
+      "step": 99350
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003808263902139117,
+      "loss": 0.5003,
+      "step": 99360
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003807839894846192,
+      "loss": 0.4849,
+      "step": 99370
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003807415887553266,
+      "loss": 0.4452,
+      "step": 99380
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038069918802603406,
+      "loss": 0.4926,
+      "step": 99390
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003806567872967415,
+      "loss": 0.5515,
+      "step": 99400
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.000380614386567449,
+      "loss": 0.4629,
+      "step": 99410
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038057198583815646,
+      "loss": 0.5346,
+      "step": 99420
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003805295851088639,
+      "loss": 0.4315,
+      "step": 99430
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038048718437957133,
+      "loss": 0.4702,
+      "step": 99440
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003804447836502788,
+      "loss": 0.5576,
+      "step": 99450
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038040238292098627,
+      "loss": 0.4938,
+      "step": 99460
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038035998219169374,
+      "loss": 0.4466,
+      "step": 99470
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038031758146240115,
+      "loss": 0.4295,
+      "step": 99480
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003802751807331086,
+      "loss": 0.5209,
+      "step": 99490
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003802327800038161,
+      "loss": 0.4908,
+      "step": 99500
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038019037927452355,
+      "loss": 0.4614,
+      "step": 99510
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.000380147978545231,
+      "loss": 0.5161,
+      "step": 99520
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038010557781593843,
+      "loss": 0.6379,
+      "step": 99530
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003800631770866459,
+      "loss": 0.4807,
+      "step": 99540
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038002077635735336,
+      "loss": 0.5525,
+      "step": 99550
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00037997837562806083,
+      "loss": 0.4906,
+      "step": 99560
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00037993597489876824,
+      "loss": 0.4352,
+      "step": 99570
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00037989357416947576,
+      "loss": 0.4641,
+      "step": 99580
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003798511734401832,
+      "loss": 0.5135,
+      "step": 99590
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00037980877271089064,
+      "loss": 0.4699,
+      "step": 99600
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003797663719815981,
+      "loss": 0.5312,
+      "step": 99610
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003797239712523056,
+      "loss": 0.4993,
+      "step": 99620
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.000379681570523013,
+      "loss": 0.5444,
+      "step": 99630
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00037963916979372045,
+      "loss": 0.5327,
+      "step": 99640
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003795967690644279,
+      "loss": 0.4989,
+      "step": 99650
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003795543683351354,
+      "loss": 0.4361,
+      "step": 99660
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003795119676058428,
+      "loss": 0.4836,
+      "step": 99670
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003794695668765503,
+      "loss": 0.547,
+      "step": 99680
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00037942716614725773,
+      "loss": 0.4711,
+      "step": 99690
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003793847654179652,
+      "loss": 0.5111,
+      "step": 99700
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00037934236468867267,
+      "loss": 0.5302,
+      "step": 99710
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00037929996395938013,
+      "loss": 0.4733,
+      "step": 99720
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00037925756323008755,
+      "loss": 0.5167,
+      "step": 99730
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00037921516250079507,
+      "loss": 0.4723,
+      "step": 99740
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003791727617715025,
+      "loss": 0.4516,
+      "step": 99750
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00037913036104220995,
+      "loss": 0.4104,
+      "step": 99760
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00037908796031291736,
+      "loss": 0.4714,
+      "step": 99770
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003790455595836249,
+      "loss": 0.4702,
+      "step": 99780
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003790031588543323,
+      "loss": 0.5058,
+      "step": 99790
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00037896075812503976,
+      "loss": 0.4766,
+      "step": 99800
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003789183573957472,
+      "loss": 0.476,
+      "step": 99810
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003788759566664547,
+      "loss": 0.4197,
+      "step": 99820
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003788335559371621,
+      "loss": 0.4561,
+      "step": 99830
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003787911552078696,
+      "loss": 0.495,
+      "step": 99840
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00037874875447857704,
+      "loss": 0.4526,
+      "step": 99850
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003787063537492845,
+      "loss": 0.4722,
+      "step": 99860
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.000378663953019992,
+      "loss": 0.5302,
+      "step": 99870
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00037862155229069944,
+      "loss": 0.4386,
+      "step": 99880
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00037857915156140685,
+      "loss": 0.4804,
+      "step": 99890
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003785367508321144,
+      "loss": 0.4799,
+      "step": 99900
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003784943501028218,
+      "loss": 0.4876,
+      "step": 99910
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037845194937352925,
+      "loss": 0.5028,
+      "step": 99920
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037840954864423667,
+      "loss": 0.4858,
+      "step": 99930
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003783671479149442,
+      "loss": 0.5198,
+      "step": 99940
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003783247471856516,
+      "loss": 0.5352,
+      "step": 99950
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037828234645635907,
+      "loss": 0.4749,
+      "step": 99960
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037823994572706653,
+      "loss": 0.5112,
+      "step": 99970
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.000378197544997774,
+      "loss": 0.474,
+      "step": 99980
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003781551442684814,
+      "loss": 0.5256,
+      "step": 99990
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037811274353918893,
+      "loss": 0.5201,
+      "step": 100000
+    },
+    {
+      "epoch": 4.23,
+      "eval_loss": 0.6122934222221375,
+      "eval_runtime": 337.7306,
+      "eval_samples_per_second": 15.56,
+      "eval_steps_per_second": 3.891,
+      "step": 100000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037807034280989635,
+      "loss": 0.5115,
+      "step": 100010
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003780279420806038,
+      "loss": 0.479,
+      "step": 100020
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003779855413513113,
+      "loss": 0.4572,
+      "step": 100030
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037794314062201875,
+      "loss": 0.5408,
+      "step": 100040
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037790073989272616,
+      "loss": 0.458,
+      "step": 100050
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003778583391634336,
+      "loss": 0.5142,
+      "step": 100060
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003778159384341411,
+      "loss": 0.4192,
+      "step": 100070
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037777353770484856,
+      "loss": 0.4939,
+      "step": 100080
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037773113697555597,
+      "loss": 0.5021,
+      "step": 100090
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037768873624626344,
+      "loss": 0.5499,
+      "step": 100100
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003776463355169709,
+      "loss": 0.4566,
+      "step": 100110
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037760393478767837,
+      "loss": 0.4493,
+      "step": 100120
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037756153405838584,
+      "loss": 0.4751,
+      "step": 100130
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037751913332909325,
+      "loss": 0.5277,
+      "step": 100140
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003774767325998007,
+      "loss": 0.5718,
+      "step": 100150
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003774343318705082,
+      "loss": 0.4864,
+      "step": 100160
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037739193114121565,
+      "loss": 0.4527,
+      "step": 100170
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037734953041192306,
+      "loss": 0.4825,
+      "step": 100180
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003773071296826306,
+      "loss": 0.3646,
+      "step": 100190
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.000377264728953338,
+      "loss": 0.4667,
+      "step": 100200
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037722232822404546,
+      "loss": 0.4921,
+      "step": 100210
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037717992749475293,
+      "loss": 0.4794,
+      "step": 100220
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003771375267654604,
+      "loss": 0.5227,
+      "step": 100230
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003770951260361678,
+      "loss": 0.503,
+      "step": 100240
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003770527253068753,
+      "loss": 0.4732,
+      "step": 100250
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037701032457758274,
+      "loss": 0.4712,
+      "step": 100260
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003769679238482902,
+      "loss": 0.4727,
+      "step": 100270
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003769255231189976,
+      "loss": 0.3906,
+      "step": 100280
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037688312238970514,
+      "loss": 0.5432,
+      "step": 100290
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037684072166041256,
+      "loss": 0.464,
+      "step": 100300
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037679832093112,
+      "loss": 0.4892,
+      "step": 100310
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003767559202018275,
+      "loss": 0.4791,
+      "step": 100320
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037671351947253496,
+      "loss": 0.4881,
+      "step": 100330
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037667111874324237,
+      "loss": 0.4679,
+      "step": 100340
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003766287180139499,
+      "loss": 0.4985,
+      "step": 100350
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003765863172846573,
+      "loss": 0.4248,
+      "step": 100360
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037654391655536477,
+      "loss": 0.5163,
+      "step": 100370
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037650151582607224,
+      "loss": 0.4325,
+      "step": 100380
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003764591150967797,
+      "loss": 0.411,
+      "step": 100390
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003764167143674871,
+      "loss": 0.4943,
+      "step": 100400
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003763743136381946,
+      "loss": 0.4491,
+      "step": 100410
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037633191290890205,
+      "loss": 0.4697,
+      "step": 100420
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003762895121796095,
+      "loss": 0.5559,
+      "step": 100430
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037624711145031693,
+      "loss": 0.4303,
+      "step": 100440
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037620471072102445,
+      "loss": 0.6466,
+      "step": 100450
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037616230999173186,
+      "loss": 0.4771,
+      "step": 100460
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037611990926243933,
+      "loss": 0.4422,
+      "step": 100470
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003760775085331468,
+      "loss": 0.5213,
+      "step": 100480
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037603510780385426,
+      "loss": 0.4292,
+      "step": 100490
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003759927070745617,
+      "loss": 0.5017,
+      "step": 100500
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003759503063452692,
+      "loss": 0.4536,
+      "step": 100510
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003759079056159766,
+      "loss": 0.5045,
+      "step": 100520
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003758655048866841,
+      "loss": 0.5174,
+      "step": 100530
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003758231041573915,
+      "loss": 0.5312,
+      "step": 100540
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.000375780703428099,
+      "loss": 0.4634,
+      "step": 100550
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003757383026988064,
+      "loss": 0.5226,
+      "step": 100560
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003756959019695139,
+      "loss": 0.4747,
+      "step": 100570
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037565350124022136,
+      "loss": 0.469,
+      "step": 100580
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003756111005109288,
+      "loss": 0.4342,
+      "step": 100590
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037556869978163624,
+      "loss": 0.4665,
+      "step": 100600
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037552629905234376,
+      "loss": 0.4794,
+      "step": 100610
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037548389832305117,
+      "loss": 0.4671,
+      "step": 100620
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037544149759375864,
+      "loss": 0.4681,
+      "step": 100630
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003753990968644661,
+      "loss": 0.5686,
+      "step": 100640
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037535669613517357,
+      "loss": 0.4833,
+      "step": 100650
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.000375314295405881,
+      "loss": 0.4997,
+      "step": 100660
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037527189467658845,
+      "loss": 0.4928,
+      "step": 100670
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003752294939472959,
+      "loss": 0.4515,
+      "step": 100680
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003751870932180034,
+      "loss": 0.4354,
+      "step": 100690
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003751446924887108,
+      "loss": 0.4629,
+      "step": 100700
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003751022917594183,
+      "loss": 0.6308,
+      "step": 100710
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037505989103012573,
+      "loss": 0.5108,
+      "step": 100720
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003750174903008332,
+      "loss": 0.5576,
+      "step": 100730
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037497508957154066,
+      "loss": 0.457,
+      "step": 100740
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037493268884224813,
+      "loss": 0.4403,
+      "step": 100750
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037489028811295554,
+      "loss": 0.5121,
+      "step": 100760
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.000374847887383663,
+      "loss": 0.4065,
+      "step": 100770
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003748054866543705,
+      "loss": 0.4476,
+      "step": 100780
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037476308592507794,
+      "loss": 0.5822,
+      "step": 100790
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003747206851957854,
+      "loss": 0.544,
+      "step": 100800
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003746782844664928,
+      "loss": 0.54,
+      "step": 100810
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003746358837372003,
+      "loss": 0.4811,
+      "step": 100820
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037459348300790775,
+      "loss": 0.4448,
+      "step": 100830
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003745510822786152,
+      "loss": 0.5341,
+      "step": 100840
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037450868154932263,
+      "loss": 0.5665,
+      "step": 100850
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003744662808200301,
+      "loss": 0.4939,
+      "step": 100860
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037442388009073757,
+      "loss": 0.4745,
+      "step": 100870
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037438147936144503,
+      "loss": 0.4471,
+      "step": 100880
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037433907863215245,
+      "loss": 0.5303,
+      "step": 100890
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037429667790285997,
+      "loss": 0.4711,
+      "step": 100900
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003742542771735674,
+      "loss": 0.4201,
+      "step": 100910
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037421187644427485,
+      "loss": 0.4695,
+      "step": 100920
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003741694757149823,
+      "loss": 0.4731,
+      "step": 100930
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003741270749856898,
+      "loss": 0.4791,
+      "step": 100940
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003740846742563972,
+      "loss": 0.4711,
+      "step": 100950
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003740422735271047,
+      "loss": 0.4971,
+      "step": 100960
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037399987279781213,
+      "loss": 0.4488,
+      "step": 100970
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003739574720685196,
+      "loss": 0.449,
+      "step": 100980
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037391507133922706,
+      "loss": 0.5899,
+      "step": 100990
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037387267060993453,
+      "loss": 0.4882,
+      "step": 101000
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037383026988064194,
+      "loss": 0.5395,
+      "step": 101010
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003737878691513494,
+      "loss": 0.618,
+      "step": 101020
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003737454684220569,
+      "loss": 0.4357,
+      "step": 101030
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037370306769276434,
+      "loss": 0.4591,
+      "step": 101040
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037366066696347175,
+      "loss": 0.4628,
+      "step": 101050
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003736182662341793,
+      "loss": 0.4982,
+      "step": 101060
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003735758655048867,
+      "loss": 0.4522,
+      "step": 101070
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037353346477559415,
+      "loss": 0.456,
+      "step": 101080
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003734910640463016,
+      "loss": 0.4378,
+      "step": 101090
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003734486633170091,
+      "loss": 0.4571,
+      "step": 101100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003734062625877165,
+      "loss": 0.5577,
+      "step": 101110
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.000373363861858424,
+      "loss": 0.4907,
+      "step": 101120
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00037332146112913143,
+      "loss": 0.4421,
+      "step": 101130
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003732790603998389,
+      "loss": 0.4506,
+      "step": 101140
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003732366596705463,
+      "loss": 0.4444,
+      "step": 101150
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00037319425894125383,
+      "loss": 0.4696,
+      "step": 101160
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00037315185821196125,
+      "loss": 0.4072,
+      "step": 101170
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003731094574826687,
+      "loss": 0.3991,
+      "step": 101180
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003730670567533762,
+      "loss": 0.392,
+      "step": 101190
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00037302465602408365,
+      "loss": 0.4471,
+      "step": 101200
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00037298225529479106,
+      "loss": 0.5004,
+      "step": 101210
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003729398545654986,
+      "loss": 0.4755,
+      "step": 101220
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.000372897453836206,
+      "loss": 0.4983,
+      "step": 101230
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00037285505310691346,
+      "loss": 0.504,
+      "step": 101240
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003728126523776209,
+      "loss": 0.5152,
+      "step": 101250
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003727702516483284,
+      "loss": 0.5447,
+      "step": 101260
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003727278509190358,
+      "loss": 0.4825,
+      "step": 101270
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003726854501897433,
+      "loss": 0.5222,
+      "step": 101280
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00037264304946045074,
+      "loss": 0.4397,
+      "step": 101290
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003726006487311582,
+      "loss": 0.5157,
+      "step": 101300
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003725582480018656,
+      "loss": 0.4055,
+      "step": 101310
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00037251584727257314,
+      "loss": 0.4765,
+      "step": 101320
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037247344654328055,
+      "loss": 0.4089,
+      "step": 101330
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.000372431045813988,
+      "loss": 0.469,
+      "step": 101340
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003723886450846955,
+      "loss": 0.539,
+      "step": 101350
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037234624435540295,
+      "loss": 0.5122,
+      "step": 101360
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037230384362611037,
+      "loss": 0.5208,
+      "step": 101370
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037226144289681783,
+      "loss": 0.6185,
+      "step": 101380
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003722190421675253,
+      "loss": 0.4555,
+      "step": 101390
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037217664143823277,
+      "loss": 0.47,
+      "step": 101400
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037213424070894023,
+      "loss": 0.4936,
+      "step": 101410
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037209183997964764,
+      "loss": 0.5135,
+      "step": 101420
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003720494392503551,
+      "loss": 0.462,
+      "step": 101430
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003720070385210626,
+      "loss": 0.4825,
+      "step": 101440
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037196463779177005,
+      "loss": 0.4943,
+      "step": 101450
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003719222370624775,
+      "loss": 0.4534,
+      "step": 101460
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003718798363331849,
+      "loss": 0.4931,
+      "step": 101470
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003718374356038924,
+      "loss": 0.4604,
+      "step": 101480
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037179503487459986,
+      "loss": 0.4067,
+      "step": 101490
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003717526341453073,
+      "loss": 0.4985,
+      "step": 101500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003717102334160148,
+      "loss": 0.4664,
+      "step": 101510
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003716678326867222,
+      "loss": 0.4352,
+      "step": 101520
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037162543195742967,
+      "loss": 0.4618,
+      "step": 101530
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037158303122813714,
+      "loss": 0.5526,
+      "step": 101540
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003715406304988446,
+      "loss": 0.4533,
+      "step": 101550
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.000371498229769552,
+      "loss": 0.5558,
+      "step": 101560
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037145582904025954,
+      "loss": 0.4319,
+      "step": 101570
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037141342831096695,
+      "loss": 0.5158,
+      "step": 101580
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003713710275816744,
+      "loss": 0.6012,
+      "step": 101590
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003713286268523819,
+      "loss": 0.5404,
+      "step": 101600
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037128622612308935,
+      "loss": 0.4806,
+      "step": 101610
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037124382539379676,
+      "loss": 0.4804,
+      "step": 101620
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037120142466450423,
+      "loss": 0.4985,
+      "step": 101630
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003711590239352117,
+      "loss": 0.4649,
+      "step": 101640
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037111662320591916,
+      "loss": 0.4198,
+      "step": 101650
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003710742224766266,
+      "loss": 0.5085,
+      "step": 101660
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003710318217473341,
+      "loss": 0.4929,
+      "step": 101670
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003709894210180415,
+      "loss": 0.5219,
+      "step": 101680
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.000370947020288749,
+      "loss": 0.4805,
+      "step": 101690
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037090461955945644,
+      "loss": 0.3792,
+      "step": 101700
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003708622188301639,
+      "loss": 0.5078,
+      "step": 101710
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003708198181008713,
+      "loss": 0.5032,
+      "step": 101720
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037077741737157884,
+      "loss": 0.4606,
+      "step": 101730
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037073501664228626,
+      "loss": 0.5181,
+      "step": 101740
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003706926159129937,
+      "loss": 0.4921,
+      "step": 101750
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003706502151837012,
+      "loss": 0.5124,
+      "step": 101760
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037060781445440866,
+      "loss": 0.4922,
+      "step": 101770
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037056541372511607,
+      "loss": 0.4976,
+      "step": 101780
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037052301299582354,
+      "loss": 0.565,
+      "step": 101790
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.000370480612266531,
+      "loss": 0.4461,
+      "step": 101800
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037043821153723847,
+      "loss": 0.4763,
+      "step": 101810
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003703958108079459,
+      "loss": 0.4917,
+      "step": 101820
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003703534100786534,
+      "loss": 0.4658,
+      "step": 101830
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003703110093493608,
+      "loss": 0.5279,
+      "step": 101840
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003702686086200683,
+      "loss": 0.5185,
+      "step": 101850
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037022620789077575,
+      "loss": 0.4976,
+      "step": 101860
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003701838071614832,
+      "loss": 0.4343,
+      "step": 101870
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037014140643219063,
+      "loss": 0.5151,
+      "step": 101880
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037009900570289815,
+      "loss": 0.4488,
+      "step": 101890
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037005660497360556,
+      "loss": 0.5107,
+      "step": 101900
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037001420424431303,
+      "loss": 0.414,
+      "step": 101910
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00036997180351502044,
+      "loss": 0.4319,
+      "step": 101920
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00036992940278572796,
+      "loss": 0.5021,
+      "step": 101930
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003698870020564354,
+      "loss": 0.4721,
+      "step": 101940
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00036984460132714284,
+      "loss": 0.4593,
+      "step": 101950
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003698022005978503,
+      "loss": 0.6094,
+      "step": 101960
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003697597998685578,
+      "loss": 0.5516,
+      "step": 101970
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003697173991392652,
+      "loss": 0.5456,
+      "step": 101980
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003696749984099727,
+      "loss": 0.5488,
+      "step": 101990
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003696325976806801,
+      "loss": 0.5308,
+      "step": 102000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003695901969513876,
+      "loss": 0.4163,
+      "step": 102010
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00036954779622209506,
+      "loss": 0.5638,
+      "step": 102020
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003695053954928025,
+      "loss": 0.5331,
+      "step": 102030
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036946299476350994,
+      "loss": 0.4481,
+      "step": 102040
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003694205940342174,
+      "loss": 0.4986,
+      "step": 102050
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036937819330492487,
+      "loss": 0.4702,
+      "step": 102060
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036933579257563234,
+      "loss": 0.4666,
+      "step": 102070
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036929339184633975,
+      "loss": 0.5633,
+      "step": 102080
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003692509911170472,
+      "loss": 0.4454,
+      "step": 102090
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003692085903877547,
+      "loss": 0.5684,
+      "step": 102100
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036916618965846215,
+      "loss": 0.5848,
+      "step": 102110
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003691237889291696,
+      "loss": 0.5835,
+      "step": 102120
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036908138819987703,
+      "loss": 0.4701,
+      "step": 102130
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003690389874705845,
+      "loss": 0.4939,
+      "step": 102140
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036899658674129196,
+      "loss": 0.5113,
+      "step": 102150
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036895418601199943,
+      "loss": 0.489,
+      "step": 102160
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036891178528270684,
+      "loss": 0.3875,
+      "step": 102170
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036886938455341436,
+      "loss": 0.5317,
+      "step": 102180
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003688269838241218,
+      "loss": 0.558,
+      "step": 102190
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036878458309482924,
+      "loss": 0.4438,
+      "step": 102200
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003687421823655367,
+      "loss": 0.486,
+      "step": 102210
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003686997816362442,
+      "loss": 0.4561,
+      "step": 102220
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003686573809069516,
+      "loss": 0.4441,
+      "step": 102230
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036861498017765905,
+      "loss": 0.5718,
+      "step": 102240
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003685725794483665,
+      "loss": 0.4413,
+      "step": 102250
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.000368530178719074,
+      "loss": 0.5337,
+      "step": 102260
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003684877779897814,
+      "loss": 0.5053,
+      "step": 102270
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003684453772604889,
+      "loss": 0.523,
+      "step": 102280
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036840297653119633,
+      "loss": 0.5179,
+      "step": 102290
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003683605758019038,
+      "loss": 0.4743,
+      "step": 102300
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036831817507261127,
+      "loss": 0.4615,
+      "step": 102310
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036827577434331873,
+      "loss": 0.4905,
+      "step": 102320
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036823337361402615,
+      "loss": 0.4638,
+      "step": 102330
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036819097288473367,
+      "loss": 0.5252,
+      "step": 102340
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003681485721554411,
+      "loss": 0.4362,
+      "step": 102350
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036810617142614855,
+      "loss": 0.4802,
+      "step": 102360
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.000368063770696856,
+      "loss": 0.5025,
+      "step": 102370
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003680213699675635,
+      "loss": 0.4221,
+      "step": 102380
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003679789692382709,
+      "loss": 0.5063,
+      "step": 102390
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036793656850897836,
+      "loss": 0.6129,
+      "step": 102400
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003678941677796858,
+      "loss": 0.4768,
+      "step": 102410
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003678517670503933,
+      "loss": 0.482,
+      "step": 102420
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003678093663211007,
+      "loss": 0.5466,
+      "step": 102430
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036776696559180823,
+      "loss": 0.5793,
+      "step": 102440
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036772456486251564,
+      "loss": 0.5644,
+      "step": 102450
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003676821641332231,
+      "loss": 0.4425,
+      "step": 102460
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003676397634039306,
+      "loss": 0.4284,
+      "step": 102470
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036759736267463804,
+      "loss": 0.4455,
+      "step": 102480
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036755496194534545,
+      "loss": 0.4762,
+      "step": 102490
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.000367512561216053,
+      "loss": 0.5369,
+      "step": 102500
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003674701604867604,
+      "loss": 0.5148,
+      "step": 102510
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036742775975746785,
+      "loss": 0.4669,
+      "step": 102520
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003673853590281753,
+      "loss": 0.5429,
+      "step": 102530
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003673429582988828,
+      "loss": 0.5196,
+      "step": 102540
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003673005575695902,
+      "loss": 0.4884,
+      "step": 102550
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036725815684029767,
+      "loss": 0.4548,
+      "step": 102560
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036721575611100513,
+      "loss": 0.4906,
+      "step": 102570
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003671733553817126,
+      "loss": 0.4371,
+      "step": 102580
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036713095465242,
+      "loss": 0.4545,
+      "step": 102590
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036708855392312753,
+      "loss": 0.4472,
+      "step": 102600
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036704615319383495,
+      "loss": 0.4258,
+      "step": 102610
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003670037524645424,
+      "loss": 0.5266,
+      "step": 102620
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003669613517352499,
+      "loss": 0.4329,
+      "step": 102630
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036691895100595735,
+      "loss": 0.5059,
+      "step": 102640
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036687655027666476,
+      "loss": 0.4846,
+      "step": 102650
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003668341495473722,
+      "loss": 0.5789,
+      "step": 102660
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003667917488180797,
+      "loss": 0.4559,
+      "step": 102670
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036674934808878716,
+      "loss": 0.4843,
+      "step": 102680
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036670694735949457,
+      "loss": 0.4956,
+      "step": 102690
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003666645466302021,
+      "loss": 0.5291,
+      "step": 102700
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003666221459009095,
+      "loss": 0.5475,
+      "step": 102710
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036657974517161697,
+      "loss": 0.4643,
+      "step": 102720
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036653734444232444,
+      "loss": 0.5001,
+      "step": 102730
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003664949437130319,
+      "loss": 0.4551,
+      "step": 102740
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003664525429837393,
+      "loss": 0.5982,
+      "step": 102750
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003664101422544468,
+      "loss": 0.5116,
+      "step": 102760
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036636774152515425,
+      "loss": 0.4814,
+      "step": 102770
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003663253407958617,
+      "loss": 0.5746,
+      "step": 102780
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003662829400665692,
+      "loss": 0.3834,
+      "step": 102790
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003662405393372766,
+      "loss": 0.5123,
+      "step": 102800
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036619813860798406,
+      "loss": 0.4858,
+      "step": 102810
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036615573787869153,
+      "loss": 0.4419,
+      "step": 102820
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.000366113337149399,
+      "loss": 0.4902,
+      "step": 102830
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003660709364201064,
+      "loss": 0.4806,
+      "step": 102840
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003660285356908139,
+      "loss": 0.494,
+      "step": 102850
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036598613496152134,
+      "loss": 0.513,
+      "step": 102860
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003659437342322288,
+      "loss": 0.5162,
+      "step": 102870
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003659013335029362,
+      "loss": 0.4529,
+      "step": 102880
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036585893277364374,
+      "loss": 0.531,
+      "step": 102890
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036581653204435116,
+      "loss": 0.4299,
+      "step": 102900
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003657741313150586,
+      "loss": 0.551,
+      "step": 102910
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003657317305857661,
+      "loss": 0.4751,
+      "step": 102920
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036568932985647356,
+      "loss": 0.4236,
+      "step": 102930
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036564692912718097,
+      "loss": 0.5195,
+      "step": 102940
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003656045283978885,
+      "loss": 0.501,
+      "step": 102950
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003655621276685959,
+      "loss": 0.5366,
+      "step": 102960
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036551972693930337,
+      "loss": 0.4711,
+      "step": 102970
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036547732621001084,
+      "loss": 0.4757,
+      "step": 102980
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003654349254807183,
+      "loss": 0.5217,
+      "step": 102990
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003653925247514257,
+      "loss": 0.4649,
+      "step": 103000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003653501240221332,
+      "loss": 0.4923,
+      "step": 103010
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036530772329284065,
+      "loss": 0.5585,
+      "step": 103020
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003652653225635481,
+      "loss": 0.5184,
+      "step": 103030
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036522292183425553,
+      "loss": 0.5643,
+      "step": 103040
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036518052110496305,
+      "loss": 0.5209,
+      "step": 103050
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036513812037567046,
+      "loss": 0.4463,
+      "step": 103060
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036509571964637793,
+      "loss": 0.5097,
+      "step": 103070
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003650533189170854,
+      "loss": 0.4961,
+      "step": 103080
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036501091818779286,
+      "loss": 0.4846,
+      "step": 103090
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003649685174585003,
+      "loss": 0.5334,
+      "step": 103100
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003649261167292078,
+      "loss": 0.4753,
+      "step": 103110
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003648837159999152,
+      "loss": 0.5223,
+      "step": 103120
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003648413152706227,
+      "loss": 0.517,
+      "step": 103130
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036479891454133014,
+      "loss": 0.5345,
+      "step": 103140
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003647565138120376,
+      "loss": 0.5344,
+      "step": 103150
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.000364714113082745,
+      "loss": 0.5162,
+      "step": 103160
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003646717123534525,
+      "loss": 0.553,
+      "step": 103170
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036462931162415996,
+      "loss": 0.5309,
+      "step": 103180
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003645869108948674,
+      "loss": 0.5348,
+      "step": 103190
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036454451016557484,
+      "loss": 0.5191,
+      "step": 103200
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036450210943628236,
+      "loss": 0.4434,
+      "step": 103210
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036445970870698977,
+      "loss": 0.5649,
+      "step": 103220
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036441730797769724,
+      "loss": 0.4572,
+      "step": 103230
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003643749072484047,
+      "loss": 0.4595,
+      "step": 103240
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036433250651911217,
+      "loss": 0.4614,
+      "step": 103250
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003642901057898196,
+      "loss": 0.5647,
+      "step": 103260
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003642477050605271,
+      "loss": 0.465,
+      "step": 103270
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003642053043312345,
+      "loss": 0.4129,
+      "step": 103280
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.000364162903601942,
+      "loss": 0.4536,
+      "step": 103290
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003641205028726494,
+      "loss": 0.5146,
+      "step": 103300
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003640781021433569,
+      "loss": 0.5298,
+      "step": 103310
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036403570141406433,
+      "loss": 0.4482,
+      "step": 103320
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003639933006847718,
+      "loss": 0.5,
+      "step": 103330
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036395089995547926,
+      "loss": 0.4675,
+      "step": 103340
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036390849922618673,
+      "loss": 0.4835,
+      "step": 103350
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036386609849689414,
+      "loss": 0.5281,
+      "step": 103360
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003638236977676016,
+      "loss": 0.5208,
+      "step": 103370
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003637812970383091,
+      "loss": 0.5056,
+      "step": 103380
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036373889630901654,
+      "loss": 0.4169,
+      "step": 103390
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.000363696495579724,
+      "loss": 0.4949,
+      "step": 103400
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003636540948504314,
+      "loss": 0.4416,
+      "step": 103410
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003636116941211389,
+      "loss": 0.5237,
+      "step": 103420
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036356929339184636,
+      "loss": 0.5434,
+      "step": 103430
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003635268926625538,
+      "loss": 0.4383,
+      "step": 103440
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003634844919332613,
+      "loss": 0.574,
+      "step": 103450
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003634420912039687,
+      "loss": 0.5021,
+      "step": 103460
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036339969047467617,
+      "loss": 0.5078,
+      "step": 103470
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036335728974538363,
+      "loss": 0.4568,
+      "step": 103480
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003633148890160911,
+      "loss": 0.4943,
+      "step": 103490
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036327248828679857,
+      "loss": 0.4891,
+      "step": 103500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.000363230087557506,
+      "loss": 0.4637,
+      "step": 103510
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036318768682821345,
+      "loss": 0.4625,
+      "step": 103520
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003631452860989209,
+      "loss": 0.542,
+      "step": 103530
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003631028853696284,
+      "loss": 0.5357,
+      "step": 103540
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003630604846403358,
+      "loss": 0.521,
+      "step": 103550
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003630180839110433,
+      "loss": 0.4793,
+      "step": 103560
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036297568318175073,
+      "loss": 0.4963,
+      "step": 103570
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003629332824524582,
+      "loss": 0.4769,
+      "step": 103580
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036289088172316566,
+      "loss": 0.4633,
+      "step": 103590
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036284848099387313,
+      "loss": 0.5571,
+      "step": 103600
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036280608026458054,
+      "loss": 0.4937,
+      "step": 103610
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.000362763679535288,
+      "loss": 0.4919,
+      "step": 103620
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003627212788059955,
+      "loss": 0.4584,
+      "step": 103630
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036267887807670294,
+      "loss": 0.5242,
+      "step": 103640
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036263647734741035,
+      "loss": 0.485,
+      "step": 103650
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003625940766181179,
+      "loss": 0.3861,
+      "step": 103660
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003625516758888253,
+      "loss": 0.5732,
+      "step": 103670
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036250927515953275,
+      "loss": 0.4918,
+      "step": 103680
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003624668744302402,
+      "loss": 0.5439,
+      "step": 103690
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003624244737009477,
+      "loss": 0.5105,
+      "step": 103700
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003623820729716551,
+      "loss": 0.5255,
+      "step": 103710
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003623396722423626,
+      "loss": 0.4703,
+      "step": 103720
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00036229727151307003,
+      "loss": 0.539,
+      "step": 103730
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003622548707837775,
+      "loss": 0.5117,
+      "step": 103740
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00036221247005448497,
+      "loss": 0.4562,
+      "step": 103750
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00036217006932519243,
+      "loss": 0.5883,
+      "step": 103760
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00036212766859589985,
+      "loss": 0.4963,
+      "step": 103770
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003620852678666073,
+      "loss": 0.5133,
+      "step": 103780
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003620428671373148,
+      "loss": 0.4832,
+      "step": 103790
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00036200046640802225,
+      "loss": 0.4865,
+      "step": 103800
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00036195806567872966,
+      "loss": 0.4634,
+      "step": 103810
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003619156649494372,
+      "loss": 0.4517,
+      "step": 103820
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003618732642201446,
+      "loss": 0.5837,
+      "step": 103830
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00036183086349085206,
+      "loss": 0.5501,
+      "step": 103840
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003617884627615595,
+      "loss": 0.4845,
+      "step": 103850
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.000361746062032267,
+      "loss": 0.4675,
+      "step": 103860
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003617036613029744,
+      "loss": 0.4699,
+      "step": 103870
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003616612605736819,
+      "loss": 0.4945,
+      "step": 103880
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00036161885984438934,
+      "loss": 0.5303,
+      "step": 103890
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003615764591150968,
+      "loss": 0.4926,
+      "step": 103900
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003615340583858043,
+      "loss": 0.4721,
+      "step": 103910
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00036149165765651174,
+      "loss": 0.4442,
+      "step": 103920
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036144925692721915,
+      "loss": 0.5286,
+      "step": 103930
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003614068561979266,
+      "loss": 0.5439,
+      "step": 103940
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003613644554686341,
+      "loss": 0.5191,
+      "step": 103950
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036132205473934155,
+      "loss": 0.5007,
+      "step": 103960
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036127965401004897,
+      "loss": 0.5406,
+      "step": 103970
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003612372532807565,
+      "loss": 0.5764,
+      "step": 103980
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003611948525514639,
+      "loss": 0.5082,
+      "step": 103990
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036115245182217137,
+      "loss": 0.5358,
+      "step": 104000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036111005109287883,
+      "loss": 0.474,
+      "step": 104010
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003610676503635863,
+      "loss": 0.5554,
+      "step": 104020
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003610252496342937,
+      "loss": 0.5797,
+      "step": 104030
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003609828489050012,
+      "loss": 0.4829,
+      "step": 104040
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036094044817570865,
+      "loss": 0.3736,
+      "step": 104050
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003608980474464161,
+      "loss": 0.501,
+      "step": 104060
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003608556467171235,
+      "loss": 0.5699,
+      "step": 104070
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.000360813245987831,
+      "loss": 0.4828,
+      "step": 104080
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036077084525853846,
+      "loss": 0.5193,
+      "step": 104090
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003607284445292459,
+      "loss": 0.5074,
+      "step": 104100
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003606860437999534,
+      "loss": 0.4394,
+      "step": 104110
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003606436430706608,
+      "loss": 0.5012,
+      "step": 104120
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036060124234136827,
+      "loss": 0.3702,
+      "step": 104130
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036055884161207574,
+      "loss": 0.5003,
+      "step": 104140
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003605164408827832,
+      "loss": 0.4448,
+      "step": 104150
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003604740401534906,
+      "loss": 0.5133,
+      "step": 104160
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036043163942419814,
+      "loss": 0.4981,
+      "step": 104170
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036038923869490555,
+      "loss": 0.4623,
+      "step": 104180
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.000360346837965613,
+      "loss": 0.5479,
+      "step": 104190
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003603044372363205,
+      "loss": 0.6045,
+      "step": 104200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036026203650702795,
+      "loss": 0.423,
+      "step": 104210
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036021963577773536,
+      "loss": 0.5066,
+      "step": 104220
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036017723504844283,
+      "loss": 0.4476,
+      "step": 104230
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003601348343191503,
+      "loss": 0.4595,
+      "step": 104240
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036009243358985776,
+      "loss": 0.4719,
+      "step": 104250
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003600500328605652,
+      "loss": 0.4919,
+      "step": 104260
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003600076321312727,
+      "loss": 0.5467,
+      "step": 104270
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003599652314019801,
+      "loss": 0.5131,
+      "step": 104280
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003599228306726876,
+      "loss": 0.5379,
+      "step": 104290
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00035988042994339504,
+      "loss": 0.4731,
+      "step": 104300
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003598380292141025,
+      "loss": 0.3841,
+      "step": 104310
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003597956284848099,
+      "loss": 0.5442,
+      "step": 104320
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00035975322775551744,
+      "loss": 0.6518,
+      "step": 104330
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00035971082702622486,
+      "loss": 0.491,
+      "step": 104340
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003596684262969323,
+      "loss": 0.4487,
+      "step": 104350
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003596260255676398,
+      "loss": 0.598,
+      "step": 104360
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00035958362483834726,
+      "loss": 0.4188,
+      "step": 104370
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00035954122410905467,
+      "loss": 0.4783,
+      "step": 104380
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00035949882337976214,
+      "loss": 0.485,
+      "step": 104390
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003594564226504696,
+      "loss": 0.4888,
+      "step": 104400
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00035941402192117707,
+      "loss": 0.5,
+      "step": 104410
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003593716211918845,
+      "loss": 0.4957,
+      "step": 104420
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.000359329220462592,
+      "loss": 0.4225,
+      "step": 104430
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003592868197332994,
+      "loss": 0.5398,
+      "step": 104440
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003592444190040069,
+      "loss": 0.4807,
+      "step": 104450
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00035920201827471435,
+      "loss": 0.4688,
+      "step": 104460
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003591596175454218,
+      "loss": 0.4567,
+      "step": 104470
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00035911721681612923,
+      "loss": 0.5426,
+      "step": 104480
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00035907481608683675,
+      "loss": 0.4507,
+      "step": 104490
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00035903241535754416,
+      "loss": 0.477,
+      "step": 104500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00035899001462825163,
+      "loss": 0.4739,
+      "step": 104510
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003589476138989591,
+      "loss": 0.4973,
+      "step": 104520
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00035890521316966656,
+      "loss": 0.6004,
+      "step": 104530
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.000358862812440374,
+      "loss": 0.4649,
+      "step": 104540
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00035882041171108144,
+      "loss": 0.4739,
+      "step": 104550
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003587780109817889,
+      "loss": 0.5195,
+      "step": 104560
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003587356102524964,
+      "loss": 0.4426,
+      "step": 104570
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003586932095232038,
+      "loss": 0.563,
+      "step": 104580
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003586508087939113,
+      "loss": 0.573,
+      "step": 104590
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003586084080646187,
+      "loss": 0.5555,
+      "step": 104600
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003585660073353262,
+      "loss": 0.4835,
+      "step": 104610
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00035852360660603366,
+      "loss": 0.4872,
+      "step": 104620
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003584812058767411,
+      "loss": 0.4917,
+      "step": 104630
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00035843880514744854,
+      "loss": 0.5178,
+      "step": 104640
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.000358396404418156,
+      "loss": 0.4519,
+      "step": 104650
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00035835400368886347,
+      "loss": 0.5319,
+      "step": 104660
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00035831160295957094,
+      "loss": 0.4505,
+      "step": 104670
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003582692022302784,
+      "loss": 0.4873,
+      "step": 104680
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003582268015009858,
+      "loss": 0.4098,
+      "step": 104690
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003581844007716933,
+      "loss": 0.5111,
+      "step": 104700
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00035814200004240075,
+      "loss": 0.5149,
+      "step": 104710
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003580995993131082,
+      "loss": 0.5276,
+      "step": 104720
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003580571985838157,
+      "loss": 0.4428,
+      "step": 104730
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003580147978545231,
+      "loss": 0.6269,
+      "step": 104740
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00035797239712523056,
+      "loss": 0.5209,
+      "step": 104750
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00035792999639593803,
+      "loss": 0.4949,
+      "step": 104760
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003578875956666455,
+      "loss": 0.4982,
+      "step": 104770
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00035784519493735296,
+      "loss": 0.5091,
+      "step": 104780
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003578027942080604,
+      "loss": 0.5044,
+      "step": 104790
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00035776039347876784,
+      "loss": 0.4891,
+      "step": 104800
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003577179927494753,
+      "loss": 0.4099,
+      "step": 104810
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003576755920201828,
+      "loss": 0.4263,
+      "step": 104820
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003576331912908902,
+      "loss": 0.4671,
+      "step": 104830
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00035759079056159765,
+      "loss": 0.5133,
+      "step": 104840
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003575483898323051,
+      "loss": 0.5197,
+      "step": 104850
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003575059891030126,
+      "loss": 0.5085,
+      "step": 104860
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00035746358837372,
+      "loss": 0.4961,
+      "step": 104870
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003574211876444275,
+      "loss": 0.4951,
+      "step": 104880
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035737878691513493,
+      "loss": 0.5375,
+      "step": 104890
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003573363861858424,
+      "loss": 0.498,
+      "step": 104900
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035729398545654987,
+      "loss": 0.5091,
+      "step": 104910
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035725158472725733,
+      "loss": 0.5227,
+      "step": 104920
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035720918399796475,
+      "loss": 0.4665,
+      "step": 104930
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035716678326867227,
+      "loss": 0.4684,
+      "step": 104940
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003571243825393797,
+      "loss": 0.5529,
+      "step": 104950
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035708198181008715,
+      "loss": 0.5686,
+      "step": 104960
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003570395810807946,
+      "loss": 0.5361,
+      "step": 104970
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003569971803515021,
+      "loss": 0.4467,
+      "step": 104980
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003569547796222095,
+      "loss": 0.4745,
+      "step": 104990
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035691237889291696,
+      "loss": 0.44,
+      "step": 105000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035686997816362443,
+      "loss": 0.4929,
+      "step": 105010
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003568275774343319,
+      "loss": 0.5419,
+      "step": 105020
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003567851767050393,
+      "loss": 0.4348,
+      "step": 105030
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035674277597574683,
+      "loss": 0.618,
+      "step": 105040
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035670037524645424,
+      "loss": 0.5684,
+      "step": 105050
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003566579745171617,
+      "loss": 0.525,
+      "step": 105060
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003566155737878692,
+      "loss": 0.5373,
+      "step": 105070
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035657317305857664,
+      "loss": 0.5271,
+      "step": 105080
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035653077232928405,
+      "loss": 0.5136,
+      "step": 105090
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003564883715999916,
+      "loss": 0.444,
+      "step": 105100
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.000356445970870699,
+      "loss": 0.5035,
+      "step": 105110
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035640357014140645,
+      "loss": 0.4363,
+      "step": 105120
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003563611694121139,
+      "loss": 0.5455,
+      "step": 105130
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003563187686828214,
+      "loss": 0.576,
+      "step": 105140
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003562763679535288,
+      "loss": 0.4509,
+      "step": 105150
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035623396722423627,
+      "loss": 0.4989,
+      "step": 105160
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035619156649494373,
+      "loss": 0.4125,
+      "step": 105170
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003561491657656512,
+      "loss": 0.519,
+      "step": 105180
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003561067650363586,
+      "loss": 0.484,
+      "step": 105190
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035606436430706613,
+      "loss": 0.4996,
+      "step": 105200
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035602196357777355,
+      "loss": 0.4741,
+      "step": 105210
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.000355979562848481,
+      "loss": 0.4549,
+      "step": 105220
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003559371621191885,
+      "loss": 0.4845,
+      "step": 105230
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035589476138989595,
+      "loss": 0.4593,
+      "step": 105240
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035585236066060336,
+      "loss": 0.5409,
+      "step": 105250
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003558099599313109,
+      "loss": 0.4813,
+      "step": 105260
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003557675592020183,
+      "loss": 0.4658,
+      "step": 105270
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035572515847272576,
+      "loss": 0.4525,
+      "step": 105280
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003556827577434332,
+      "loss": 0.4674,
+      "step": 105290
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003556403570141407,
+      "loss": 0.4741,
+      "step": 105300
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003555979562848481,
+      "loss": 0.5298,
+      "step": 105310
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035555555555555557,
+      "loss": 0.4605,
+      "step": 105320
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035551315482626304,
+      "loss": 0.5392,
+      "step": 105330
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003554707540969705,
+      "loss": 0.51,
+      "step": 105340
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003554283533676779,
+      "loss": 0.4546,
+      "step": 105350
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003553859526383854,
+      "loss": 0.4861,
+      "step": 105360
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035534355190909285,
+      "loss": 0.608,
+      "step": 105370
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003553011511798003,
+      "loss": 0.5053,
+      "step": 105380
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003552587504505078,
+      "loss": 0.4714,
+      "step": 105390
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003552163497212152,
+      "loss": 0.504,
+      "step": 105400
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035517394899192267,
+      "loss": 0.5187,
+      "step": 105410
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035513154826263013,
+      "loss": 0.4688,
+      "step": 105420
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003550891475333376,
+      "loss": 0.5242,
+      "step": 105430
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.000355046746804045,
+      "loss": 0.3869,
+      "step": 105440
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003550043460747525,
+      "loss": 0.4179,
+      "step": 105450
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035496194534545995,
+      "loss": 0.4486,
+      "step": 105460
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003549195446161674,
+      "loss": 0.4635,
+      "step": 105470
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003548771438868749,
+      "loss": 0.5008,
+      "step": 105480
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035483474315758235,
+      "loss": 0.4662,
+      "step": 105490
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035479234242828976,
+      "loss": 0.5475,
+      "step": 105500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003547499416989972,
+      "loss": 0.5898,
+      "step": 105510
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003547075409697047,
+      "loss": 0.513,
+      "step": 105520
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035466514024041216,
+      "loss": 0.5331,
+      "step": 105530
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035462273951111957,
+      "loss": 0.4987,
+      "step": 105540
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003545803387818271,
+      "loss": 0.4708,
+      "step": 105550
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003545379380525345,
+      "loss": 0.5129,
+      "step": 105560
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035449553732324197,
+      "loss": 0.501,
+      "step": 105570
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035445313659394944,
+      "loss": 0.5049,
+      "step": 105580
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003544107358646569,
+      "loss": 0.5147,
+      "step": 105590
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003543683351353643,
+      "loss": 0.3935,
+      "step": 105600
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003543259344060718,
+      "loss": 0.5518,
+      "step": 105610
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035428353367677925,
+      "loss": 0.5809,
+      "step": 105620
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003542411329474867,
+      "loss": 0.4651,
+      "step": 105630
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035419873221819413,
+      "loss": 0.4918,
+      "step": 105640
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035415633148890165,
+      "loss": 0.5046,
+      "step": 105650
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035411393075960906,
+      "loss": 0.4106,
+      "step": 105660
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035407153003031653,
+      "loss": 0.5549,
+      "step": 105670
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.000354029129301024,
+      "loss": 0.4425,
+      "step": 105680
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035398672857173146,
+      "loss": 0.5426,
+      "step": 105690
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003539443278424389,
+      "loss": 0.5294,
+      "step": 105700
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003539019271131464,
+      "loss": 0.48,
+      "step": 105710
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003538595263838538,
+      "loss": 0.5395,
+      "step": 105720
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003538171256545613,
+      "loss": 0.4375,
+      "step": 105730
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035377472492526874,
+      "loss": 0.5187,
+      "step": 105740
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003537323241959762,
+      "loss": 0.4333,
+      "step": 105750
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003536899234666836,
+      "loss": 0.4588,
+      "step": 105760
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003536475227373911,
+      "loss": 0.4458,
+      "step": 105770
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035360512200809856,
+      "loss": 0.4292,
+      "step": 105780
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.000353562721278806,
+      "loss": 0.5041,
+      "step": 105790
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035352032054951344,
+      "loss": 0.5579,
+      "step": 105800
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035347791982022096,
+      "loss": 0.4804,
+      "step": 105810
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035343551909092837,
+      "loss": 0.468,
+      "step": 105820
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00035339311836163584,
+      "loss": 0.4845,
+      "step": 105830
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003533507176323433,
+      "loss": 0.5108,
+      "step": 105840
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00035330831690305077,
+      "loss": 0.5436,
+      "step": 105850
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003532659161737582,
+      "loss": 0.5103,
+      "step": 105860
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003532235154444657,
+      "loss": 0.528,
+      "step": 105870
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003531811147151731,
+      "loss": 0.535,
+      "step": 105880
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003531387139858806,
+      "loss": 0.5673,
+      "step": 105890
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00035309631325658805,
+      "loss": 0.4695,
+      "step": 105900
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003530539125272955,
+      "loss": 0.5182,
+      "step": 105910
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00035301151179800293,
+      "loss": 0.509,
+      "step": 105920
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003529691110687104,
+      "loss": 0.4662,
+      "step": 105930
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00035292671033941786,
+      "loss": 0.5813,
+      "step": 105940
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00035288430961012533,
+      "loss": 0.6149,
+      "step": 105950
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00035284190888083274,
+      "loss": 0.49,
+      "step": 105960
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00035279950815154026,
+      "loss": 0.4402,
+      "step": 105970
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003527571074222477,
+      "loss": 0.4106,
+      "step": 105980
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00035271470669295514,
+      "loss": 0.5169,
+      "step": 105990
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003526723059636626,
+      "loss": 0.5891,
+      "step": 106000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003526299052343701,
+      "loss": 0.571,
+      "step": 106010
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003525875045050775,
+      "loss": 0.5261,
+      "step": 106020
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00035254510377578496,
+      "loss": 0.513,
+      "step": 106030
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003525027030464924,
+      "loss": 0.47,
+      "step": 106040
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003524603023171999,
+      "loss": 0.4799,
+      "step": 106050
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035241790158790736,
+      "loss": 0.5154,
+      "step": 106060
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035237550085861477,
+      "loss": 0.4341,
+      "step": 106070
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035233310012932224,
+      "loss": 0.5716,
+      "step": 106080
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003522906994000297,
+      "loss": 0.5518,
+      "step": 106090
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035224829867073717,
+      "loss": 0.4163,
+      "step": 106100
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003522058979414446,
+      "loss": 0.4499,
+      "step": 106110
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035216349721215205,
+      "loss": 0.5145,
+      "step": 106120
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003521210964828595,
+      "loss": 0.5179,
+      "step": 106130
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.000352078695753567,
+      "loss": 0.474,
+      "step": 106140
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003520362950242744,
+      "loss": 0.4918,
+      "step": 106150
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003519938942949819,
+      "loss": 0.5022,
+      "step": 106160
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035195149356568933,
+      "loss": 0.5773,
+      "step": 106170
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003519090928363968,
+      "loss": 0.5664,
+      "step": 106180
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035186669210710426,
+      "loss": 0.4396,
+      "step": 106190
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035182429137781173,
+      "loss": 0.4456,
+      "step": 106200
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035178189064851914,
+      "loss": 0.4985,
+      "step": 106210
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003517394899192266,
+      "loss": 0.4263,
+      "step": 106220
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003516970891899341,
+      "loss": 0.413,
+      "step": 106230
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035165468846064154,
+      "loss": 0.4729,
+      "step": 106240
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035161228773134895,
+      "loss": 0.4792,
+      "step": 106250
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003515698870020565,
+      "loss": 0.5171,
+      "step": 106260
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003515274862727639,
+      "loss": 0.4817,
+      "step": 106270
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035148508554347135,
+      "loss": 0.4817,
+      "step": 106280
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003514426848141788,
+      "loss": 0.4149,
+      "step": 106290
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003514002840848863,
+      "loss": 0.4842,
+      "step": 106300
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003513578833555937,
+      "loss": 0.4949,
+      "step": 106310
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003513154826263012,
+      "loss": 0.567,
+      "step": 106320
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035127308189700863,
+      "loss": 0.4502,
+      "step": 106330
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003512306811677161,
+      "loss": 0.5403,
+      "step": 106340
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035118828043842357,
+      "loss": 0.472,
+      "step": 106350
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035114587970913103,
+      "loss": 0.4921,
+      "step": 106360
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035110347897983845,
+      "loss": 0.5264,
+      "step": 106370
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003510610782505459,
+      "loss": 0.4602,
+      "step": 106380
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003510186775212534,
+      "loss": 0.571,
+      "step": 106390
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035097627679196085,
+      "loss": 0.3894,
+      "step": 106400
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035093387606266826,
+      "loss": 0.4524,
+      "step": 106410
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003508914753333758,
+      "loss": 0.4882,
+      "step": 106420
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003508490746040832,
+      "loss": 0.4774,
+      "step": 106430
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035080667387479066,
+      "loss": 0.4718,
+      "step": 106440
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035076427314549813,
+      "loss": 0.3744,
+      "step": 106450
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003507218724162056,
+      "loss": 0.5668,
+      "step": 106460
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.000350679471686913,
+      "loss": 0.6328,
+      "step": 106470
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035063707095762053,
+      "loss": 0.4928,
+      "step": 106480
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035059467022832794,
+      "loss": 0.5547,
+      "step": 106490
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003505522694990354,
+      "loss": 0.4658,
+      "step": 106500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003505098687697429,
+      "loss": 0.5365,
+      "step": 106510
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035046746804045034,
+      "loss": 0.5803,
+      "step": 106520
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035042506731115775,
+      "loss": 0.4629,
+      "step": 106530
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003503826665818652,
+      "loss": 0.5381,
+      "step": 106540
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003503402658525727,
+      "loss": 0.5841,
+      "step": 106550
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035029786512328015,
+      "loss": 0.4277,
+      "step": 106560
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035025546439398757,
+      "loss": 0.4285,
+      "step": 106570
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003502130636646951,
+      "loss": 0.529,
+      "step": 106580
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003501706629354025,
+      "loss": 0.4909,
+      "step": 106590
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035012826220610997,
+      "loss": 0.4684,
+      "step": 106600
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035008586147681743,
+      "loss": 0.534,
+      "step": 106610
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003500434607475249,
+      "loss": 0.4918,
+      "step": 106620
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003500010600182323,
+      "loss": 0.4488,
+      "step": 106630
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003499586592889398,
+      "loss": 0.4534,
+      "step": 106640
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00034991625855964725,
+      "loss": 0.5353,
+      "step": 106650
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003498738578303547,
+      "loss": 0.5378,
+      "step": 106660
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003498314571010622,
+      "loss": 0.5609,
+      "step": 106670
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003497890563717696,
+      "loss": 0.4754,
+      "step": 106680
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00034974665564247706,
+      "loss": 0.5381,
+      "step": 106690
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003497042549131845,
+      "loss": 0.4833,
+      "step": 106700
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.000349661854183892,
+      "loss": 0.448,
+      "step": 106710
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00034961945345459946,
+      "loss": 0.4436,
+      "step": 106720
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00034957705272530687,
+      "loss": 0.4929,
+      "step": 106730
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00034953465199601434,
+      "loss": 0.6984,
+      "step": 106740
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003494922512667218,
+      "loss": 0.4845,
+      "step": 106750
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00034944985053742927,
+      "loss": 0.4963,
+      "step": 106760
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034940744980813674,
+      "loss": 0.4141,
+      "step": 106770
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034936504907884415,
+      "loss": 0.4954,
+      "step": 106780
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003493226483495516,
+      "loss": 0.5065,
+      "step": 106790
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003492802476202591,
+      "loss": 0.5094,
+      "step": 106800
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034923784689096655,
+      "loss": 0.4835,
+      "step": 106810
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034919544616167396,
+      "loss": 0.5174,
+      "step": 106820
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034915304543238143,
+      "loss": 0.4896,
+      "step": 106830
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003491106447030889,
+      "loss": 0.4268,
+      "step": 106840
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034906824397379637,
+      "loss": 0.4797,
+      "step": 106850
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003490258432445038,
+      "loss": 0.4695,
+      "step": 106860
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003489834425152113,
+      "loss": 0.5712,
+      "step": 106870
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003489410417859187,
+      "loss": 0.49,
+      "step": 106880
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003488986410566262,
+      "loss": 0.499,
+      "step": 106890
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034885624032733364,
+      "loss": 0.5207,
+      "step": 106900
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003488138395980411,
+      "loss": 0.503,
+      "step": 106910
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003487714388687485,
+      "loss": 0.5505,
+      "step": 106920
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034872903813945605,
+      "loss": 0.5236,
+      "step": 106930
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034868663741016346,
+      "loss": 0.4945,
+      "step": 106940
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003486442366808709,
+      "loss": 0.5301,
+      "step": 106950
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003486018359515784,
+      "loss": 0.579,
+      "step": 106960
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034855943522228586,
+      "loss": 0.6279,
+      "step": 106970
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034851703449299327,
+      "loss": 0.4421,
+      "step": 106980
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034847463376370074,
+      "loss": 0.5439,
+      "step": 106990
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003484322330344082,
+      "loss": 0.5025,
+      "step": 107000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00034838983230511567,
+      "loss": 0.5406,
+      "step": 107010
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003483474315758231,
+      "loss": 0.4764,
+      "step": 107020
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003483050308465306,
+      "loss": 0.4957,
+      "step": 107030
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.000348262630117238,
+      "loss": 0.3927,
+      "step": 107040
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003482202293879455,
+      "loss": 0.5063,
+      "step": 107050
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00034817782865865295,
+      "loss": 0.4234,
+      "step": 107060
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003481354279293604,
+      "loss": 0.5742,
+      "step": 107070
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00034809302720006783,
+      "loss": 0.4469,
+      "step": 107080
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00034805062647077535,
+      "loss": 0.5323,
+      "step": 107090
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00034800822574148276,
+      "loss": 0.4703,
+      "step": 107100
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00034796582501219023,
+      "loss": 0.5174,
+      "step": 107110
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003479234242828977,
+      "loss": 0.435,
+      "step": 107120
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00034788102355360516,
+      "loss": 0.5701,
+      "step": 107130
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003478386228243126,
+      "loss": 0.5459,
+      "step": 107140
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00034779622209502004,
+      "loss": 0.4502,
+      "step": 107150
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003477538213657275,
+      "loss": 0.4554,
+      "step": 107160
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.000347711420636435,
+      "loss": 0.4884,
+      "step": 107170
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003476690199071424,
+      "loss": 0.4633,
+      "step": 107180
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003476266191778499,
+      "loss": 0.4079,
+      "step": 107190
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003475842184485573,
+      "loss": 0.4479,
+      "step": 107200
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003475418177192648,
+      "loss": 0.4533,
+      "step": 107210
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00034749941698997226,
+      "loss": 0.4297,
+      "step": 107220
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003474570162606797,
+      "loss": 0.4894,
+      "step": 107230
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034741461553138714,
+      "loss": 0.5471,
+      "step": 107240
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034737221480209466,
+      "loss": 0.4712,
+      "step": 107250
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034732981407280207,
+      "loss": 0.4601,
+      "step": 107260
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034728741334350954,
+      "loss": 0.4511,
+      "step": 107270
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.000347245012614217,
+      "loss": 0.5375,
+      "step": 107280
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034720261188492447,
+      "loss": 0.5216,
+      "step": 107290
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003471602111556319,
+      "loss": 0.433,
+      "step": 107300
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034711781042633935,
+      "loss": 0.4578,
+      "step": 107310
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003470754096970468,
+      "loss": 0.5175,
+      "step": 107320
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003470330089677543,
+      "loss": 0.6046,
+      "step": 107330
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003469906082384617,
+      "loss": 0.5059,
+      "step": 107340
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034694820750916916,
+      "loss": 0.5152,
+      "step": 107350
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034690580677987663,
+      "loss": 0.4585,
+      "step": 107360
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003468634060505841,
+      "loss": 0.4774,
+      "step": 107370
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034682100532129156,
+      "loss": 0.4959,
+      "step": 107380
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.000346778604591999,
+      "loss": 0.5307,
+      "step": 107390
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034673620386270644,
+      "loss": 0.5356,
+      "step": 107400
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003466938031334139,
+      "loss": 0.5499,
+      "step": 107410
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003466514024041214,
+      "loss": 0.5504,
+      "step": 107420
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003466090016748288,
+      "loss": 0.4537,
+      "step": 107430
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003465666009455363,
+      "loss": 0.4826,
+      "step": 107440
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003465242002162437,
+      "loss": 0.5716,
+      "step": 107450
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003464817994869512,
+      "loss": 0.4838,
+      "step": 107460
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034643939875765866,
+      "loss": 0.4609,
+      "step": 107470
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003463969980283661,
+      "loss": 0.4426,
+      "step": 107480
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034635459729907353,
+      "loss": 0.5005,
+      "step": 107490
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.000346312196569781,
+      "loss": 0.5682,
+      "step": 107500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034626979584048847,
+      "loss": 0.5103,
+      "step": 107510
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034622739511119594,
+      "loss": 0.4696,
+      "step": 107520
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034618499438190335,
+      "loss": 0.493,
+      "step": 107530
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034614259365261087,
+      "loss": 0.4887,
+      "step": 107540
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003461001929233183,
+      "loss": 0.5209,
+      "step": 107550
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034605779219402575,
+      "loss": 0.4904,
+      "step": 107560
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003460153914647332,
+      "loss": 0.3924,
+      "step": 107570
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003459729907354407,
+      "loss": 0.5073,
+      "step": 107580
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003459305900061481,
+      "loss": 0.5398,
+      "step": 107590
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034588818927685556,
+      "loss": 0.555,
+      "step": 107600
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034584578854756303,
+      "loss": 0.5702,
+      "step": 107610
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003458033878182705,
+      "loss": 0.5026,
+      "step": 107620
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003457609870889779,
+      "loss": 0.4956,
+      "step": 107630
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034571858635968543,
+      "loss": 0.4388,
+      "step": 107640
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034567618563039284,
+      "loss": 0.4928,
+      "step": 107650
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003456337849011003,
+      "loss": 0.4737,
+      "step": 107660
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003455913841718078,
+      "loss": 0.4968,
+      "step": 107670
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034554898344251524,
+      "loss": 0.457,
+      "step": 107680
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034550658271322265,
+      "loss": 0.4246,
+      "step": 107690
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003454641819839302,
+      "loss": 0.4906,
+      "step": 107700
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003454217812546376,
+      "loss": 0.5061,
+      "step": 107710
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034537938052534505,
+      "loss": 0.4846,
+      "step": 107720
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003453369797960525,
+      "loss": 0.5337,
+      "step": 107730
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034529457906676,
+      "loss": 0.548,
+      "step": 107740
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003452521783374674,
+      "loss": 0.4874,
+      "step": 107750
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034520977760817487,
+      "loss": 0.5021,
+      "step": 107760
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034516737687888233,
+      "loss": 0.4684,
+      "step": 107770
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003451249761495898,
+      "loss": 0.3846,
+      "step": 107780
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003450825754202972,
+      "loss": 0.4669,
+      "step": 107790
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034504017469100473,
+      "loss": 0.5252,
+      "step": 107800
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034499777396171215,
+      "loss": 0.4335,
+      "step": 107810
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003449553732324196,
+      "loss": 0.4736,
+      "step": 107820
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003449129725031271,
+      "loss": 0.4428,
+      "step": 107830
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034487057177383455,
+      "loss": 0.556,
+      "step": 107840
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034482817104454196,
+      "loss": 0.4922,
+      "step": 107850
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003447857703152495,
+      "loss": 0.4925,
+      "step": 107860
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003447433695859569,
+      "loss": 0.5169,
+      "step": 107870
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034470096885666436,
+      "loss": 0.5352,
+      "step": 107880
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003446585681273718,
+      "loss": 0.5075,
+      "step": 107890
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003446161673980793,
+      "loss": 0.4733,
+      "step": 107900
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003445737666687867,
+      "loss": 0.5317,
+      "step": 107910
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003445313659394942,
+      "loss": 0.5063,
+      "step": 107920
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034448896521020164,
+      "loss": 0.5444,
+      "step": 107930
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003444465644809091,
+      "loss": 0.5286,
+      "step": 107940
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0003444041637516165,
+      "loss": 0.5924,
+      "step": 107950
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034436176302232404,
+      "loss": 0.4883,
+      "step": 107960
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034431936229303145,
+      "loss": 0.4851,
+      "step": 107970
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0003442769615637389,
+      "loss": 0.6215,
+      "step": 107980
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0003442345608344464,
+      "loss": 0.4841,
+      "step": 107990
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034419216010515385,
+      "loss": 0.5319,
+      "step": 108000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034414975937586127,
+      "loss": 0.4775,
+      "step": 108010
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034410735864656873,
+      "loss": 0.4822,
+      "step": 108020
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0003440649579172762,
+      "loss": 0.5886,
+      "step": 108030
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034402255718798367,
+      "loss": 0.5262,
+      "step": 108040
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034398015645869113,
+      "loss": 0.5194,
+      "step": 108050
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034393775572939855,
+      "loss": 0.534,
+      "step": 108060
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.000343895355000106,
+      "loss": 0.4929,
+      "step": 108070
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0003438529542708135,
+      "loss": 0.5975,
+      "step": 108080
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034381055354152095,
+      "loss": 0.5299,
+      "step": 108090
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034376815281222836,
+      "loss": 0.4774,
+      "step": 108100
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0003437257520829358,
+      "loss": 0.5331,
+      "step": 108110
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0003436833513536433,
+      "loss": 0.5488,
+      "step": 108120
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034364095062435076,
+      "loss": 0.4879,
+      "step": 108130
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034359854989505817,
+      "loss": 0.5142,
+      "step": 108140
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0003435561491657657,
+      "loss": 0.5254,
+      "step": 108150
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0003435137484364731,
+      "loss": 0.5427,
+      "step": 108160
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034347134770718057,
+      "loss": 0.5496,
+      "step": 108170
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034342894697788804,
+      "loss": 0.5269,
+      "step": 108180
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003433865462485955,
+      "loss": 0.4337,
+      "step": 108190
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003433441455193029,
+      "loss": 0.5217,
+      "step": 108200
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034330174479001044,
+      "loss": 0.4488,
+      "step": 108210
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034325934406071785,
+      "loss": 0.5362,
+      "step": 108220
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003432169433314253,
+      "loss": 0.4492,
+      "step": 108230
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034317454260213273,
+      "loss": 0.4603,
+      "step": 108240
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034313214187284025,
+      "loss": 0.4717,
+      "step": 108250
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034308974114354766,
+      "loss": 0.5156,
+      "step": 108260
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034304734041425513,
+      "loss": 0.4649,
+      "step": 108270
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003430049396849626,
+      "loss": 0.4972,
+      "step": 108280
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034296253895567006,
+      "loss": 0.5109,
+      "step": 108290
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003429201382263775,
+      "loss": 0.5172,
+      "step": 108300
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.000342877737497085,
+      "loss": 0.5499,
+      "step": 108310
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003428353367677924,
+      "loss": 0.4737,
+      "step": 108320
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003427929360384999,
+      "loss": 0.5354,
+      "step": 108330
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034275053530920734,
+      "loss": 0.5004,
+      "step": 108340
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003427081345799148,
+      "loss": 0.4944,
+      "step": 108350
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003426657338506222,
+      "loss": 0.6272,
+      "step": 108360
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003426233331213297,
+      "loss": 0.5181,
+      "step": 108370
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034258093239203716,
+      "loss": 0.5306,
+      "step": 108380
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003425385316627446,
+      "loss": 0.4948,
+      "step": 108390
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034249613093345204,
+      "loss": 0.5037,
+      "step": 108400
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034245373020415956,
+      "loss": 0.5172,
+      "step": 108410
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034241132947486697,
+      "loss": 0.4617,
+      "step": 108420
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034236892874557444,
+      "loss": 0.5727,
+      "step": 108430
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003423265280162819,
+      "loss": 0.4726,
+      "step": 108440
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034228412728698937,
+      "loss": 0.4441,
+      "step": 108450
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003422417265576968,
+      "loss": 0.5064,
+      "step": 108460
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003421993258284043,
+      "loss": 0.497,
+      "step": 108470
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003421569250991117,
+      "loss": 0.553,
+      "step": 108480
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003421145243698192,
+      "loss": 0.4618,
+      "step": 108490
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034207212364052665,
+      "loss": 0.4039,
+      "step": 108500
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003420297229112341,
+      "loss": 0.4445,
+      "step": 108510
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034198732218194153,
+      "loss": 0.5413,
+      "step": 108520
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.000341944921452649,
+      "loss": 0.5072,
+      "step": 108530
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034190252072335646,
+      "loss": 0.3748,
+      "step": 108540
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034186011999406393,
+      "loss": 0.499,
+      "step": 108550
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034181771926477134,
+      "loss": 0.5649,
+      "step": 108560
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034177531853547886,
+      "loss": 0.453,
+      "step": 108570
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003417329178061863,
+      "loss": 0.4569,
+      "step": 108580
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034169051707689374,
+      "loss": 0.5877,
+      "step": 108590
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003416481163476012,
+      "loss": 0.4822,
+      "step": 108600
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003416057156183087,
+      "loss": 0.5503,
+      "step": 108610
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003415633148890161,
+      "loss": 0.4607,
+      "step": 108620
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034152091415972356,
+      "loss": 0.5399,
+      "step": 108630
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.000341478513430431,
+      "loss": 0.4981,
+      "step": 108640
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003414361127011385,
+      "loss": 0.4212,
+      "step": 108650
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034139371197184596,
+      "loss": 0.4988,
+      "step": 108660
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034135131124255337,
+      "loss": 0.4604,
+      "step": 108670
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034130891051326084,
+      "loss": 0.5038,
+      "step": 108680
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0003412665097839683,
+      "loss": 0.4508,
+      "step": 108690
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034122410905467577,
+      "loss": 0.5555,
+      "step": 108700
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034118170832538324,
+      "loss": 0.5274,
+      "step": 108710
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034113930759609065,
+      "loss": 0.4886,
+      "step": 108720
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0003410969068667981,
+      "loss": 0.4547,
+      "step": 108730
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0003410545061375056,
+      "loss": 0.4666,
+      "step": 108740
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034101210540821305,
+      "loss": 0.4362,
+      "step": 108750
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0003409697046789205,
+      "loss": 0.4944,
+      "step": 108760
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034092730394962793,
+      "loss": 0.5456,
+      "step": 108770
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0003408849032203354,
+      "loss": 0.4279,
+      "step": 108780
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034084250249104286,
+      "loss": 0.4319,
+      "step": 108790
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034080010176175033,
+      "loss": 0.4287,
+      "step": 108800
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034075770103245774,
+      "loss": 0.375,
+      "step": 108810
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034071530030316526,
+      "loss": 0.4871,
+      "step": 108820
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0003406728995738727,
+      "loss": 0.5187,
+      "step": 108830
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034063049884458014,
+      "loss": 0.5983,
+      "step": 108840
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034058809811528755,
+      "loss": 0.4669,
+      "step": 108850
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0003405456973859951,
+      "loss": 0.4791,
+      "step": 108860
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0003405032966567025,
+      "loss": 0.4765,
+      "step": 108870
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034046089592740995,
+      "loss": 0.4531,
+      "step": 108880
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0003404184951981174,
+      "loss": 0.4705,
+      "step": 108890
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0003403760944688249,
+      "loss": 0.5924,
+      "step": 108900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0003403336937395323,
+      "loss": 0.5683,
+      "step": 108910
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0003402912930102398,
+      "loss": 0.4875,
+      "step": 108920
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034024889228094723,
+      "loss": 0.4569,
+      "step": 108930
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0003402064915516547,
+      "loss": 0.4816,
+      "step": 108940
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034016409082236217,
+      "loss": 0.4478,
+      "step": 108950
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034012169009306963,
+      "loss": 0.4939,
+      "step": 108960
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034007928936377705,
+      "loss": 0.4594,
+      "step": 108970
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0003400368886344845,
+      "loss": 0.4957,
+      "step": 108980
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.000339994487905192,
+      "loss": 0.5479,
+      "step": 108990
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00033995208717589945,
+      "loss": 0.4163,
+      "step": 109000
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00033990968644660686,
+      "loss": 0.6043,
+      "step": 109010
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0003398672857173144,
+      "loss": 0.4203,
+      "step": 109020
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0003398248849880218,
+      "loss": 0.4971,
+      "step": 109030
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00033978248425872926,
+      "loss": 0.5308,
+      "step": 109040
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00033974008352943673,
+      "loss": 0.4764,
+      "step": 109050
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0003396976828001442,
+      "loss": 0.4314,
+      "step": 109060
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0003396552820708516,
+      "loss": 0.5048,
+      "step": 109070
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00033961288134155913,
+      "loss": 0.5041,
+      "step": 109080
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00033957048061226654,
+      "loss": 0.4423,
+      "step": 109090
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.000339528079882974,
+      "loss": 0.515,
+      "step": 109100
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0003394856791536815,
+      "loss": 0.4034,
+      "step": 109110
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00033944327842438894,
+      "loss": 0.5175,
+      "step": 109120
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00033940087769509635,
+      "loss": 0.4981,
+      "step": 109130
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003393584769658038,
+      "loss": 0.4485,
+      "step": 109140
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003393160762365113,
+      "loss": 0.5905,
+      "step": 109150
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033927367550721875,
+      "loss": 0.4875,
+      "step": 109160
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033923127477792617,
+      "loss": 0.449,
+      "step": 109170
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003391888740486337,
+      "loss": 0.5813,
+      "step": 109180
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003391464733193411,
+      "loss": 0.4833,
+      "step": 109190
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033910407259004857,
+      "loss": 0.4621,
+      "step": 109200
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033906167186075603,
+      "loss": 0.5292,
+      "step": 109210
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003390192711314635,
+      "loss": 0.5061,
+      "step": 109220
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003389768704021709,
+      "loss": 0.4953,
+      "step": 109230
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033893446967287843,
+      "loss": 0.5209,
+      "step": 109240
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033889206894358585,
+      "loss": 0.4883,
+      "step": 109250
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003388496682142933,
+      "loss": 0.5746,
+      "step": 109260
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003388072674850008,
+      "loss": 0.4708,
+      "step": 109270
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033876486675570825,
+      "loss": 0.4688,
+      "step": 109280
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033872246602641566,
+      "loss": 0.4718,
+      "step": 109290
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003386800652971231,
+      "loss": 0.4368,
+      "step": 109300
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003386376645678306,
+      "loss": 0.5894,
+      "step": 109310
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033859526383853806,
+      "loss": 0.462,
+      "step": 109320
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033855286310924547,
+      "loss": 0.4448,
+      "step": 109330
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033851046237995294,
+      "loss": 0.4837,
+      "step": 109340
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003384680616506604,
+      "loss": 0.4997,
+      "step": 109350
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033842566092136787,
+      "loss": 0.4402,
+      "step": 109360
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033838326019207534,
+      "loss": 0.4664,
+      "step": 109370
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033834085946278275,
+      "loss": 0.5278,
+      "step": 109380
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003382984587334902,
+      "loss": 0.5284,
+      "step": 109390
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003382560580041977,
+      "loss": 0.4765,
+      "step": 109400
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033821365727490515,
+      "loss": 0.5916,
+      "step": 109410
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033817125654561257,
+      "loss": 0.618,
+      "step": 109420
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003381288558163201,
+      "loss": 0.4823,
+      "step": 109430
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003380864550870275,
+      "loss": 0.4825,
+      "step": 109440
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033804405435773497,
+      "loss": 0.5848,
+      "step": 109450
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033800165362844243,
+      "loss": 0.465,
+      "step": 109460
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003379592528991499,
+      "loss": 0.4794,
+      "step": 109470
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003379168521698573,
+      "loss": 0.5311,
+      "step": 109480
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003378744514405648,
+      "loss": 0.4644,
+      "step": 109490
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033783205071127225,
+      "loss": 0.4925,
+      "step": 109500
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003377896499819797,
+      "loss": 0.4459,
+      "step": 109510
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003377472492526871,
+      "loss": 0.5077,
+      "step": 109520
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033770484852339465,
+      "loss": 0.5126,
+      "step": 109530
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033766244779410206,
+      "loss": 0.4515,
+      "step": 109540
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003376200470648095,
+      "loss": 0.472,
+      "step": 109550
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.000337577646335517,
+      "loss": 0.4832,
+      "step": 109560
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033753524560622446,
+      "loss": 0.4661,
+      "step": 109570
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033749284487693187,
+      "loss": 0.4482,
+      "step": 109580
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003374504441476394,
+      "loss": 0.5055,
+      "step": 109590
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003374080434183468,
+      "loss": 0.4819,
+      "step": 109600
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033736564268905427,
+      "loss": 0.3648,
+      "step": 109610
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003373232419597617,
+      "loss": 0.4855,
+      "step": 109620
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003372808412304692,
+      "loss": 0.5572,
+      "step": 109630
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003372384405011766,
+      "loss": 0.482,
+      "step": 109640
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003371960397718841,
+      "loss": 0.5199,
+      "step": 109650
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033715363904259155,
+      "loss": 0.5626,
+      "step": 109660
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.000337111238313299,
+      "loss": 0.5461,
+      "step": 109670
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033706883758400643,
+      "loss": 0.5249,
+      "step": 109680
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033702643685471395,
+      "loss": 0.4616,
+      "step": 109690
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033698403612542136,
+      "loss": 0.5248,
+      "step": 109700
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033694163539612883,
+      "loss": 0.4879,
+      "step": 109710
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003368992346668363,
+      "loss": 0.4985,
+      "step": 109720
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033685683393754376,
+      "loss": 0.5683,
+      "step": 109730
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003368144332082512,
+      "loss": 0.5311,
+      "step": 109740
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033677203247895864,
+      "loss": 0.4919,
+      "step": 109750
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003367296317496661,
+      "loss": 0.5906,
+      "step": 109760
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003366872310203736,
+      "loss": 0.503,
+      "step": 109770
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.000336644830291081,
+      "loss": 0.5049,
+      "step": 109780
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003366024295617885,
+      "loss": 0.4431,
+      "step": 109790
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003365600288324959,
+      "loss": 0.3958,
+      "step": 109800
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003365176281032034,
+      "loss": 0.4799,
+      "step": 109810
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033647522737391086,
+      "loss": 0.3981,
+      "step": 109820
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003364328266446183,
+      "loss": 0.6033,
+      "step": 109830
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033639042591532574,
+      "loss": 0.6336,
+      "step": 109840
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033634802518603326,
+      "loss": 0.4697,
+      "step": 109850
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033630562445674067,
+      "loss": 0.4609,
+      "step": 109860
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033626322372744814,
+      "loss": 0.5227,
+      "step": 109870
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003362208229981556,
+      "loss": 0.4926,
+      "step": 109880
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033617842226886307,
+      "loss": 0.5061,
+      "step": 109890
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003361360215395705,
+      "loss": 0.5304,
+      "step": 109900
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033609362081027795,
+      "loss": 0.5917,
+      "step": 109910
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003360512200809854,
+      "loss": 0.491,
+      "step": 109920
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003360088193516929,
+      "loss": 0.4913,
+      "step": 109930
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003359664186224003,
+      "loss": 0.4361,
+      "step": 109940
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033592401789310776,
+      "loss": 0.518,
+      "step": 109950
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033588161716381523,
+      "loss": 0.4883,
+      "step": 109960
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003358392164345227,
+      "loss": 0.4736,
+      "step": 109970
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033579681570523016,
+      "loss": 0.42,
+      "step": 109980
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033575441497593763,
+      "loss": 0.498,
+      "step": 109990
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033571201424664504,
+      "loss": 0.4672,
+      "step": 110000
+    },
+    {
+      "epoch": 4.65,
+      "eval_loss": 0.6067126393318176,
+      "eval_runtime": 337.585,
+      "eval_samples_per_second": 15.566,
+      "eval_steps_per_second": 3.892,
+      "step": 110000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003356696135173525,
+      "loss": 0.4274,
+      "step": 110010
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033562721278806,
+      "loss": 0.4238,
+      "step": 110020
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033558481205876744,
+      "loss": 0.4934,
+      "step": 110030
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003355424113294749,
+      "loss": 0.4414,
+      "step": 110040
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003355000106001823,
+      "loss": 0.5067,
+      "step": 110050
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003354576098708898,
+      "loss": 0.5901,
+      "step": 110060
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033541520914159726,
+      "loss": 0.4406,
+      "step": 110070
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003353728084123047,
+      "loss": 0.5329,
+      "step": 110080
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033533040768301214,
+      "loss": 0.5422,
+      "step": 110090
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003352880069537196,
+      "loss": 0.5314,
+      "step": 110100
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033524560622442707,
+      "loss": 0.556,
+      "step": 110110
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033520320549513454,
+      "loss": 0.5898,
+      "step": 110120
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033516080476584195,
+      "loss": 0.4611,
+      "step": 110130
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033511840403654947,
+      "loss": 0.5194,
+      "step": 110140
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003350760033072569,
+      "loss": 0.4905,
+      "step": 110150
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033503360257796435,
+      "loss": 0.4019,
+      "step": 110160
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003349912018486718,
+      "loss": 0.4521,
+      "step": 110170
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003349488011193793,
+      "loss": 0.4644,
+      "step": 110180
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003349064003900867,
+      "loss": 0.4901,
+      "step": 110190
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003348639996607942,
+      "loss": 0.4841,
+      "step": 110200
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033482159893150163,
+      "loss": 0.4383,
+      "step": 110210
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003347791982022091,
+      "loss": 0.5671,
+      "step": 110220
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003347367974729165,
+      "loss": 0.4979,
+      "step": 110230
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033469439674362403,
+      "loss": 0.5806,
+      "step": 110240
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033465199601433144,
+      "loss": 0.4674,
+      "step": 110250
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003346095952850389,
+      "loss": 0.4922,
+      "step": 110260
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003345671945557464,
+      "loss": 0.5915,
+      "step": 110270
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033452479382645384,
+      "loss": 0.5798,
+      "step": 110280
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033448239309716125,
+      "loss": 0.5383,
+      "step": 110290
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003344399923678688,
+      "loss": 0.5043,
+      "step": 110300
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003343975916385762,
+      "loss": 0.4659,
+      "step": 110310
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033435519090928365,
+      "loss": 0.5022,
+      "step": 110320
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003343127901799911,
+      "loss": 0.4892,
+      "step": 110330
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003342703894506986,
+      "loss": 0.4185,
+      "step": 110340
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.000334227988721406,
+      "loss": 0.453,
+      "step": 110350
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003341855879921135,
+      "loss": 0.5077,
+      "step": 110360
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033414318726282093,
+      "loss": 0.5,
+      "step": 110370
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003341007865335284,
+      "loss": 0.5,
+      "step": 110380
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003340583858042358,
+      "loss": 0.5345,
+      "step": 110390
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033401598507494333,
+      "loss": 0.5014,
+      "step": 110400
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033397358434565075,
+      "loss": 0.4436,
+      "step": 110410
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003339311836163582,
+      "loss": 0.6237,
+      "step": 110420
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003338887828870657,
+      "loss": 0.5859,
+      "step": 110430
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033384638215777315,
+      "loss": 0.5128,
+      "step": 110440
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033380398142848056,
+      "loss": 0.4639,
+      "step": 110450
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003337615806991881,
+      "loss": 0.5145,
+      "step": 110460
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003337191799698955,
+      "loss": 0.5606,
+      "step": 110470
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033367677924060296,
+      "loss": 0.4392,
+      "step": 110480
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033363437851131043,
+      "loss": 0.3992,
+      "step": 110490
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003335919777820179,
+      "loss": 0.4788,
+      "step": 110500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003335495770527253,
+      "loss": 0.4611,
+      "step": 110510
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003335071763234328,
+      "loss": 0.5545,
+      "step": 110520
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033346477559414024,
+      "loss": 0.6042,
+      "step": 110530
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003334223748648477,
+      "loss": 0.3606,
+      "step": 110540
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003333799741355551,
+      "loss": 0.4672,
+      "step": 110550
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033333757340626264,
+      "loss": 0.4926,
+      "step": 110560
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033329517267697005,
+      "loss": 0.4707,
+      "step": 110570
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003332527719476775,
+      "loss": 0.4684,
+      "step": 110580
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.000333210371218385,
+      "loss": 0.4678,
+      "step": 110590
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033316797048909245,
+      "loss": 0.4138,
+      "step": 110600
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033312556975979987,
+      "loss": 0.4968,
+      "step": 110610
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033308316903050733,
+      "loss": 0.4683,
+      "step": 110620
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003330407683012148,
+      "loss": 0.4177,
+      "step": 110630
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033299836757192227,
+      "loss": 0.5396,
+      "step": 110640
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033295596684262973,
+      "loss": 0.4403,
+      "step": 110650
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033291356611333715,
+      "loss": 0.4717,
+      "step": 110660
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003328711653840446,
+      "loss": 0.4798,
+      "step": 110670
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003328287646547521,
+      "loss": 0.4852,
+      "step": 110680
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033278636392545955,
+      "loss": 0.4967,
+      "step": 110690
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033274396319616696,
+      "loss": 0.4354,
+      "step": 110700
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003327015624668744,
+      "loss": 0.3833,
+      "step": 110710
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003326591617375819,
+      "loss": 0.6045,
+      "step": 110720
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033261676100828936,
+      "loss": 0.551,
+      "step": 110730
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003325743602789968,
+      "loss": 0.511,
+      "step": 110740
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003325319595497043,
+      "loss": 0.6171,
+      "step": 110750
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003324895588204117,
+      "loss": 0.4,
+      "step": 110760
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033244715809111917,
+      "loss": 0.5761,
+      "step": 110770
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033240475736182664,
+      "loss": 0.4585,
+      "step": 110780
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003323623566325341,
+      "loss": 0.5371,
+      "step": 110790
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003323199559032415,
+      "loss": 0.4973,
+      "step": 110800
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033227755517394904,
+      "loss": 0.4346,
+      "step": 110810
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033223515444465645,
+      "loss": 0.4422,
+      "step": 110820
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003321927537153639,
+      "loss": 0.4622,
+      "step": 110830
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033215035298607133,
+      "loss": 0.4472,
+      "step": 110840
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033210795225677885,
+      "loss": 0.4683,
+      "step": 110850
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033206555152748626,
+      "loss": 0.4559,
+      "step": 110860
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033202315079819373,
+      "loss": 0.5344,
+      "step": 110870
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003319807500689012,
+      "loss": 0.4956,
+      "step": 110880
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033193834933960867,
+      "loss": 0.4332,
+      "step": 110890
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003318959486103161,
+      "loss": 0.5205,
+      "step": 110900
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003318535478810236,
+      "loss": 0.4883,
+      "step": 110910
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.000331811147151731,
+      "loss": 0.4676,
+      "step": 110920
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003317687464224385,
+      "loss": 0.448,
+      "step": 110930
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033172634569314594,
+      "loss": 0.4734,
+      "step": 110940
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003316839449638534,
+      "loss": 0.4144,
+      "step": 110950
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003316415442345608,
+      "loss": 0.423,
+      "step": 110960
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033159914350526835,
+      "loss": 0.4711,
+      "step": 110970
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033155674277597576,
+      "loss": 0.5471,
+      "step": 110980
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003315143420466832,
+      "loss": 0.5033,
+      "step": 110990
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033147194131739064,
+      "loss": 0.5116,
+      "step": 111000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033142954058809816,
+      "loss": 0.4365,
+      "step": 111010
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033138713985880557,
+      "loss": 0.4849,
+      "step": 111020
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033134473912951304,
+      "loss": 0.5259,
+      "step": 111030
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003313023384002205,
+      "loss": 0.5282,
+      "step": 111040
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033125993767092797,
+      "loss": 0.5133,
+      "step": 111050
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003312175369416354,
+      "loss": 0.4718,
+      "step": 111060
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003311751362123429,
+      "loss": 0.5402,
+      "step": 111070
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003311327354830503,
+      "loss": 0.4577,
+      "step": 111080
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003310903347537578,
+      "loss": 0.4029,
+      "step": 111090
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033104793402446525,
+      "loss": 0.4293,
+      "step": 111100
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003310055332951727,
+      "loss": 0.44,
+      "step": 111110
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033096313256588013,
+      "loss": 0.533,
+      "step": 111120
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003309207318365876,
+      "loss": 0.4127,
+      "step": 111130
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033087833110729506,
+      "loss": 0.5537,
+      "step": 111140
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033083593037800253,
+      "loss": 0.4938,
+      "step": 111150
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033079352964870994,
+      "loss": 0.4938,
+      "step": 111160
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033075112891941746,
+      "loss": 0.4769,
+      "step": 111170
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003307087281901249,
+      "loss": 0.5031,
+      "step": 111180
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033066632746083234,
+      "loss": 0.4395,
+      "step": 111190
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003306239267315398,
+      "loss": 0.4828,
+      "step": 111200
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003305815260022473,
+      "loss": 0.5023,
+      "step": 111210
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003305391252729547,
+      "loss": 0.5633,
+      "step": 111220
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003304967245436622,
+      "loss": 0.508,
+      "step": 111230
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003304543238143696,
+      "loss": 0.4762,
+      "step": 111240
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003304119230850771,
+      "loss": 0.4741,
+      "step": 111250
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033036952235578456,
+      "loss": 0.5344,
+      "step": 111260
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.000330327121626492,
+      "loss": 0.475,
+      "step": 111270
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033028472089719944,
+      "loss": 0.4422,
+      "step": 111280
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0003302423201679069,
+      "loss": 0.4617,
+      "step": 111290
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033019991943861437,
+      "loss": 0.5689,
+      "step": 111300
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033015751870932184,
+      "loss": 0.4594,
+      "step": 111310
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033011511798002925,
+      "loss": 0.5374,
+      "step": 111320
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0003300727172507367,
+      "loss": 0.4532,
+      "step": 111330
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0003300303165214442,
+      "loss": 0.4427,
+      "step": 111340
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00032998791579215165,
+      "loss": 0.4226,
+      "step": 111350
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0003299455150628591,
+      "loss": 0.5266,
+      "step": 111360
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00032990311433356653,
+      "loss": 0.541,
+      "step": 111370
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.000329860713604274,
+      "loss": 0.4821,
+      "step": 111380
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00032981831287498146,
+      "loss": 0.5041,
+      "step": 111390
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00032977591214568893,
+      "loss": 0.5455,
+      "step": 111400
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00032973351141639634,
+      "loss": 0.4832,
+      "step": 111410
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00032969111068710386,
+      "loss": 0.526,
+      "step": 111420
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0003296487099578113,
+      "loss": 0.4763,
+      "step": 111430
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00032960630922851874,
+      "loss": 0.521,
+      "step": 111440
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00032956390849922615,
+      "loss": 0.4367,
+      "step": 111450
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0003295215077699337,
+      "loss": 0.4031,
+      "step": 111460
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0003294791070406411,
+      "loss": 0.3974,
+      "step": 111470
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00032943670631134856,
+      "loss": 0.4428,
+      "step": 111480
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.000329394305582056,
+      "loss": 0.4851,
+      "step": 111490
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003293519048527635,
+      "loss": 0.4517,
+      "step": 111500
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003293095041234709,
+      "loss": 0.542,
+      "step": 111510
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003292671033941784,
+      "loss": 0.5398,
+      "step": 111520
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032922470266488583,
+      "loss": 0.5576,
+      "step": 111530
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003291823019355933,
+      "loss": 0.4784,
+      "step": 111540
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032913990120630077,
+      "loss": 0.5218,
+      "step": 111550
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032909750047700824,
+      "loss": 0.5257,
+      "step": 111560
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032905509974771565,
+      "loss": 0.4893,
+      "step": 111570
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032901269901842317,
+      "loss": 0.5859,
+      "step": 111580
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003289702982891306,
+      "loss": 0.4928,
+      "step": 111590
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032892789755983805,
+      "loss": 0.4575,
+      "step": 111600
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032888549683054546,
+      "loss": 0.3963,
+      "step": 111610
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.000328843096101253,
+      "loss": 0.4918,
+      "step": 111620
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003288006953719604,
+      "loss": 0.5402,
+      "step": 111630
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032875829464266786,
+      "loss": 0.5089,
+      "step": 111640
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032871589391337533,
+      "loss": 0.4593,
+      "step": 111650
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003286734931840828,
+      "loss": 0.4544,
+      "step": 111660
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003286310924547902,
+      "loss": 0.5428,
+      "step": 111670
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032858869172549773,
+      "loss": 0.4695,
+      "step": 111680
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032854629099620514,
+      "loss": 0.5507,
+      "step": 111690
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003285038902669126,
+      "loss": 0.6213,
+      "step": 111700
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003284614895376201,
+      "loss": 0.5477,
+      "step": 111710
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032841908880832754,
+      "loss": 0.5225,
+      "step": 111720
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032837668807903495,
+      "loss": 0.4647,
+      "step": 111730
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003283342873497425,
+      "loss": 0.4722,
+      "step": 111740
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003282918866204499,
+      "loss": 0.5223,
+      "step": 111750
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032824948589115735,
+      "loss": 0.4762,
+      "step": 111760
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032820708516186477,
+      "loss": 0.4761,
+      "step": 111770
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003281646844325723,
+      "loss": 0.4674,
+      "step": 111780
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003281222837032797,
+      "loss": 0.4865,
+      "step": 111790
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032807988297398717,
+      "loss": 0.5534,
+      "step": 111800
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032803748224469463,
+      "loss": 0.504,
+      "step": 111810
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003279950815154021,
+      "loss": 0.5923,
+      "step": 111820
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003279526807861095,
+      "loss": 0.5002,
+      "step": 111830
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032791028005681703,
+      "loss": 0.4922,
+      "step": 111840
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032786787932752445,
+      "loss": 0.4166,
+      "step": 111850
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003278254785982319,
+      "loss": 0.4206,
+      "step": 111860
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003277830778689394,
+      "loss": 0.4511,
+      "step": 111870
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032774067713964685,
+      "loss": 0.413,
+      "step": 111880
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032769827641035426,
+      "loss": 0.5162,
+      "step": 111890
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003276558756810617,
+      "loss": 0.4729,
+      "step": 111900
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003276134749517692,
+      "loss": 0.5134,
+      "step": 111910
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032757107422247666,
+      "loss": 0.4872,
+      "step": 111920
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032752867349318407,
+      "loss": 0.496,
+      "step": 111930
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032748627276389154,
+      "loss": 0.5904,
+      "step": 111940
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.000327443872034599,
+      "loss": 0.4329,
+      "step": 111950
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003274014713053065,
+      "loss": 0.4553,
+      "step": 111960
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00032735907057601394,
+      "loss": 0.4838,
+      "step": 111970
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003273166698467214,
+      "loss": 0.4876,
+      "step": 111980
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003272742691174288,
+      "loss": 0.4853,
+      "step": 111990
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003272318683881363,
+      "loss": 0.4446,
+      "step": 112000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00032718946765884375,
+      "loss": 0.5232,
+      "step": 112010
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003271470669295512,
+      "loss": 0.5934,
+      "step": 112020
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003271046662002587,
+      "loss": 0.5275,
+      "step": 112030
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003270622654709661,
+      "loss": 0.4315,
+      "step": 112040
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00032701986474167357,
+      "loss": 0.5369,
+      "step": 112050
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00032697746401238103,
+      "loss": 0.3887,
+      "step": 112060
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003269350632830885,
+      "loss": 0.4569,
+      "step": 112070
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003268926625537959,
+      "loss": 0.4865,
+      "step": 112080
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003268502618245034,
+      "loss": 0.5158,
+      "step": 112090
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00032680786109521085,
+      "loss": 0.4464,
+      "step": 112100
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003267654603659183,
+      "loss": 0.4816,
+      "step": 112110
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003267230596366257,
+      "loss": 0.544,
+      "step": 112120
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00032668065890733325,
+      "loss": 0.4313,
+      "step": 112130
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00032663825817804066,
+      "loss": 0.4999,
+      "step": 112140
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003265958574487481,
+      "loss": 0.4829,
+      "step": 112150
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003265534567194556,
+      "loss": 0.5368,
+      "step": 112160
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00032651105599016306,
+      "loss": 0.5355,
+      "step": 112170
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00032646865526087047,
+      "loss": 0.5052,
+      "step": 112180
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.000326426254531578,
+      "loss": 0.4438,
+      "step": 112190
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003263838538022854,
+      "loss": 0.5208,
+      "step": 112200
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032634145307299287,
+      "loss": 0.417,
+      "step": 112210
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003262990523437003,
+      "loss": 0.6108,
+      "step": 112220
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003262566516144078,
+      "loss": 0.4536,
+      "step": 112230
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003262142508851152,
+      "loss": 0.4932,
+      "step": 112240
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003261718501558227,
+      "loss": 0.4616,
+      "step": 112250
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032612944942653015,
+      "loss": 0.5858,
+      "step": 112260
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003260870486972376,
+      "loss": 0.4744,
+      "step": 112270
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032604464796794503,
+      "loss": 0.5593,
+      "step": 112280
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032600224723865255,
+      "loss": 0.4765,
+      "step": 112290
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032595984650935996,
+      "loss": 0.5703,
+      "step": 112300
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032591744578006743,
+      "loss": 0.4299,
+      "step": 112310
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003258750450507749,
+      "loss": 0.5629,
+      "step": 112320
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032583264432148236,
+      "loss": 0.4857,
+      "step": 112330
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003257902435921898,
+      "loss": 0.4966,
+      "step": 112340
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003257478428628973,
+      "loss": 0.5129,
+      "step": 112350
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003257054421336047,
+      "loss": 0.5088,
+      "step": 112360
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003256630414043122,
+      "loss": 0.4964,
+      "step": 112370
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003256206406750196,
+      "loss": 0.4798,
+      "step": 112380
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003255782399457271,
+      "loss": 0.437,
+      "step": 112390
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003255358392164345,
+      "loss": 0.4466,
+      "step": 112400
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.000325493438487142,
+      "loss": 0.552,
+      "step": 112410
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032545103775784946,
+      "loss": 0.4826,
+      "step": 112420
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003254086370285569,
+      "loss": 0.4499,
+      "step": 112430
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032536623629926434,
+      "loss": 0.4893,
+      "step": 112440
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032532383556997186,
+      "loss": 0.6257,
+      "step": 112450
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032528143484067927,
+      "loss": 0.4995,
+      "step": 112460
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032523903411138674,
+      "loss": 0.4074,
+      "step": 112470
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003251966333820942,
+      "loss": 0.5132,
+      "step": 112480
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032515423265280167,
+      "loss": 0.4563,
+      "step": 112490
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003251118319235091,
+      "loss": 0.427,
+      "step": 112500
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003250694311942166,
+      "loss": 0.441,
+      "step": 112510
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.000325027030464924,
+      "loss": 0.542,
+      "step": 112520
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003249846297356315,
+      "loss": 0.4457,
+      "step": 112530
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003249422290063389,
+      "loss": 0.5106,
+      "step": 112540
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003248998282770464,
+      "loss": 0.539,
+      "step": 112550
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032485742754775383,
+      "loss": 0.4972,
+      "step": 112560
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003248150268184613,
+      "loss": 0.4669,
+      "step": 112570
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032477262608916876,
+      "loss": 0.477,
+      "step": 112580
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032473022535987623,
+      "loss": 0.5037,
+      "step": 112590
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032468782463058364,
+      "loss": 0.4545,
+      "step": 112600
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003246454239012911,
+      "loss": 0.4606,
+      "step": 112610
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003246030231719986,
+      "loss": 0.5194,
+      "step": 112620
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032456062244270604,
+      "loss": 0.4526,
+      "step": 112630
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003245182217134135,
+      "loss": 0.4414,
+      "step": 112640
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003244758209841209,
+      "loss": 0.5027,
+      "step": 112650
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003244334202548284,
+      "loss": 0.5223,
+      "step": 112660
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032439101952553586,
+      "loss": 0.4685,
+      "step": 112670
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003243486187962433,
+      "loss": 0.4877,
+      "step": 112680
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032430621806695074,
+      "loss": 0.4397,
+      "step": 112690
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003242638173376582,
+      "loss": 0.4917,
+      "step": 112700
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032422141660836567,
+      "loss": 0.5463,
+      "step": 112710
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032417901587907314,
+      "loss": 0.4873,
+      "step": 112720
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003241366151497806,
+      "loss": 0.4686,
+      "step": 112730
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032409421442048807,
+      "loss": 0.5318,
+      "step": 112740
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003240518136911955,
+      "loss": 0.426,
+      "step": 112750
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032400941296190295,
+      "loss": 0.5139,
+      "step": 112760
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003239670122326104,
+      "loss": 0.4774,
+      "step": 112770
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003239246115033179,
+      "loss": 0.5127,
+      "step": 112780
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003238822107740253,
+      "loss": 0.5013,
+      "step": 112790
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003238398100447328,
+      "loss": 0.4923,
+      "step": 112800
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032379740931544023,
+      "loss": 0.5179,
+      "step": 112810
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003237550085861477,
+      "loss": 0.4463,
+      "step": 112820
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003237126078568551,
+      "loss": 0.4778,
+      "step": 112830
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032367020712756263,
+      "loss": 0.4798,
+      "step": 112840
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032362780639827004,
+      "loss": 0.4113,
+      "step": 112850
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003235854056689775,
+      "loss": 0.4777,
+      "step": 112860
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.000323543004939685,
+      "loss": 0.5544,
+      "step": 112870
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032350060421039244,
+      "loss": 0.3591,
+      "step": 112880
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032345820348109985,
+      "loss": 0.5048,
+      "step": 112890
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003234158027518074,
+      "loss": 0.4915,
+      "step": 112900
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003233734020225148,
+      "loss": 0.4465,
+      "step": 112910
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00032333100129322225,
+      "loss": 0.4395,
+      "step": 112920
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003232886005639297,
+      "loss": 0.4859,
+      "step": 112930
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003232461998346372,
+      "loss": 0.4694,
+      "step": 112940
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003232037991053446,
+      "loss": 0.4408,
+      "step": 112950
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003231613983760521,
+      "loss": 0.5286,
+      "step": 112960
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00032311899764675953,
+      "loss": 0.5261,
+      "step": 112970
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.000323076596917467,
+      "loss": 0.5615,
+      "step": 112980
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003230341961881744,
+      "loss": 0.5196,
+      "step": 112990
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00032299179545888193,
+      "loss": 0.4922,
+      "step": 113000
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00032294939472958935,
+      "loss": 0.5118,
+      "step": 113010
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003229069940002968,
+      "loss": 0.5086,
+      "step": 113020
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003228645932710043,
+      "loss": 0.555,
+      "step": 113030
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00032282219254171175,
+      "loss": 0.485,
+      "step": 113040
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00032277979181241916,
+      "loss": 0.5838,
+      "step": 113050
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003227373910831267,
+      "loss": 0.5133,
+      "step": 113060
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003226949903538341,
+      "loss": 0.5983,
+      "step": 113070
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00032265258962454156,
+      "loss": 0.4837,
+      "step": 113080
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00032261018889524903,
+      "loss": 0.4898,
+      "step": 113090
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003225677881659565,
+      "loss": 0.4557,
+      "step": 113100
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003225253874366639,
+      "loss": 0.4503,
+      "step": 113110
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00032248298670737143,
+      "loss": 0.5404,
+      "step": 113120
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00032244058597807884,
+      "loss": 0.4508,
+      "step": 113130
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003223981852487863,
+      "loss": 0.5207,
+      "step": 113140
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003223557845194937,
+      "loss": 0.5299,
+      "step": 113150
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00032231338379020124,
+      "loss": 0.529,
+      "step": 113160
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00032227098306090865,
+      "loss": 0.5357,
+      "step": 113170
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003222285823316161,
+      "loss": 0.499,
+      "step": 113180
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003221861816023236,
+      "loss": 0.4575,
+      "step": 113190
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00032214378087303105,
+      "loss": 0.5238,
+      "step": 113200
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00032210138014373847,
+      "loss": 0.5197,
+      "step": 113210
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.000322058979414446,
+      "loss": 0.5704,
+      "step": 113220
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003220165786851534,
+      "loss": 0.5029,
+      "step": 113230
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00032197417795586087,
+      "loss": 0.5265,
+      "step": 113240
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00032193177722656833,
+      "loss": 0.4628,
+      "step": 113250
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003218893764972758,
+      "loss": 0.4838,
+      "step": 113260
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003218469757679832,
+      "loss": 0.3985,
+      "step": 113270
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003218045750386907,
+      "loss": 0.5372,
+      "step": 113280
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00032176217430939815,
+      "loss": 0.4852,
+      "step": 113290
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003217197735801056,
+      "loss": 0.4497,
+      "step": 113300
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.000321677372850813,
+      "loss": 0.5343,
+      "step": 113310
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003216349721215205,
+      "loss": 0.345,
+      "step": 113320
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00032159257139222796,
+      "loss": 0.4662,
+      "step": 113330
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003215501706629354,
+      "loss": 0.5635,
+      "step": 113340
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003215077699336429,
+      "loss": 0.4866,
+      "step": 113350
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003214653692043503,
+      "loss": 0.5475,
+      "step": 113360
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00032142296847505777,
+      "loss": 0.5321,
+      "step": 113370
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00032138056774576524,
+      "loss": 0.4454,
+      "step": 113380
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003213381670164727,
+      "loss": 0.4072,
+      "step": 113390
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003212957662871801,
+      "loss": 0.5987,
+      "step": 113400
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032125336555788764,
+      "loss": 0.5297,
+      "step": 113410
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032121096482859505,
+      "loss": 0.4405,
+      "step": 113420
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003211685640993025,
+      "loss": 0.4695,
+      "step": 113430
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032112616337000993,
+      "loss": 0.5223,
+      "step": 113440
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032108376264071745,
+      "loss": 0.5386,
+      "step": 113450
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032104136191142487,
+      "loss": 0.5565,
+      "step": 113460
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032099896118213233,
+      "loss": 0.5017,
+      "step": 113470
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003209565604528398,
+      "loss": 0.5408,
+      "step": 113480
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032091415972354727,
+      "loss": 0.4998,
+      "step": 113490
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003208717589942547,
+      "loss": 0.4352,
+      "step": 113500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003208293582649622,
+      "loss": 0.5374,
+      "step": 113510
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003207869575356696,
+      "loss": 0.5198,
+      "step": 113520
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003207445568063771,
+      "loss": 0.5777,
+      "step": 113530
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032070215607708455,
+      "loss": 0.5467,
+      "step": 113540
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.000320659755347792,
+      "loss": 0.4996,
+      "step": 113550
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003206173546184994,
+      "loss": 0.4581,
+      "step": 113560
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032057495388920695,
+      "loss": 0.5455,
+      "step": 113570
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032053255315991436,
+      "loss": 0.4507,
+      "step": 113580
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003204901524306218,
+      "loss": 0.4605,
+      "step": 113590
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032044775170132924,
+      "loss": 0.5062,
+      "step": 113600
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032040535097203676,
+      "loss": 0.5319,
+      "step": 113610
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032036295024274417,
+      "loss": 0.4596,
+      "step": 113620
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00032032054951345164,
+      "loss": 0.4611,
+      "step": 113630
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003202781487841591,
+      "loss": 0.4532,
+      "step": 113640
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00032023574805486657,
+      "loss": 0.5782,
+      "step": 113650
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.000320193347325574,
+      "loss": 0.5156,
+      "step": 113660
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003201509465962815,
+      "loss": 0.4811,
+      "step": 113670
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003201085458669889,
+      "loss": 0.3971,
+      "step": 113680
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003200661451376964,
+      "loss": 0.5171,
+      "step": 113690
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00032002374440840385,
+      "loss": 0.5804,
+      "step": 113700
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003199813436791113,
+      "loss": 0.499,
+      "step": 113710
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00031993894294981873,
+      "loss": 0.42,
+      "step": 113720
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00031989654222052625,
+      "loss": 0.4609,
+      "step": 113730
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00031985414149123366,
+      "loss": 0.5767,
+      "step": 113740
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00031981174076194113,
+      "loss": 0.4071,
+      "step": 113750
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00031976934003264854,
+      "loss": 0.5568,
+      "step": 113760
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00031972693930335606,
+      "loss": 0.4933,
+      "step": 113770
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003196845385740635,
+      "loss": 0.4858,
+      "step": 113780
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00031964213784477094,
+      "loss": 0.4596,
+      "step": 113790
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003195997371154784,
+      "loss": 0.4825,
+      "step": 113800
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003195573363861859,
+      "loss": 0.4869,
+      "step": 113810
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003195149356568933,
+      "loss": 0.5184,
+      "step": 113820
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003194725349276008,
+      "loss": 0.5375,
+      "step": 113830
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003194301341983082,
+      "loss": 0.4961,
+      "step": 113840
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003193877334690157,
+      "loss": 0.5145,
+      "step": 113850
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00031934533273972316,
+      "loss": 0.4791,
+      "step": 113860
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003193029320104306,
+      "loss": 0.4572,
+      "step": 113870
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031926053128113804,
+      "loss": 0.4635,
+      "step": 113880
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003192181305518455,
+      "loss": 0.5268,
+      "step": 113890
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031917572982255297,
+      "loss": 0.4826,
+      "step": 113900
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031913332909326044,
+      "loss": 0.6019,
+      "step": 113910
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031909092836396785,
+      "loss": 0.5503,
+      "step": 113920
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003190485276346753,
+      "loss": 0.4099,
+      "step": 113930
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003190061269053828,
+      "loss": 0.3949,
+      "step": 113940
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031896372617609025,
+      "loss": 0.4721,
+      "step": 113950
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003189213254467977,
+      "loss": 0.4481,
+      "step": 113960
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031887892471750513,
+      "loss": 0.5375,
+      "step": 113970
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003188365239882126,
+      "loss": 0.4827,
+      "step": 113980
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031879412325892006,
+      "loss": 0.5709,
+      "step": 113990
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031875172252962753,
+      "loss": 0.4472,
+      "step": 114000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.000318709321800335,
+      "loss": 0.5047,
+      "step": 114010
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031866692107104246,
+      "loss": 0.5252,
+      "step": 114020
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003186245203417499,
+      "loss": 0.5519,
+      "step": 114030
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031858211961245734,
+      "loss": 0.5879,
+      "step": 114040
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003185397188831648,
+      "loss": 0.4675,
+      "step": 114050
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003184973181538723,
+      "loss": 0.5109,
+      "step": 114060
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003184549174245797,
+      "loss": 0.4334,
+      "step": 114070
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031841251669528716,
+      "loss": 0.5317,
+      "step": 114080
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003183701159659946,
+      "loss": 0.4365,
+      "step": 114090
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003183277152367021,
+      "loss": 0.4841,
+      "step": 114100
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003182853145074095,
+      "loss": 0.5572,
+      "step": 114110
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.000318242913778117,
+      "loss": 0.485,
+      "step": 114120
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031820051304882444,
+      "loss": 0.462,
+      "step": 114130
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003181581123195319,
+      "loss": 0.4468,
+      "step": 114140
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031811571159023937,
+      "loss": 0.4678,
+      "step": 114150
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031807331086094684,
+      "loss": 0.498,
+      "step": 114160
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031803091013165425,
+      "loss": 0.5165,
+      "step": 114170
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031798850940236177,
+      "loss": 0.5056,
+      "step": 114180
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003179461086730692,
+      "loss": 0.5435,
+      "step": 114190
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031790370794377665,
+      "loss": 0.4972,
+      "step": 114200
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031786130721448406,
+      "loss": 0.4603,
+      "step": 114210
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003178189064851916,
+      "loss": 0.5162,
+      "step": 114220
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.000317776505755899,
+      "loss": 0.4765,
+      "step": 114230
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031773410502660646,
+      "loss": 0.4669,
+      "step": 114240
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031769170429731393,
+      "loss": 0.5141,
+      "step": 114250
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003176493035680214,
+      "loss": 0.4732,
+      "step": 114260
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003176069028387288,
+      "loss": 0.468,
+      "step": 114270
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031756450210943633,
+      "loss": 0.594,
+      "step": 114280
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031752210138014374,
+      "loss": 0.5564,
+      "step": 114290
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003174797006508512,
+      "loss": 0.5708,
+      "step": 114300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003174372999215587,
+      "loss": 0.4275,
+      "step": 114310
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031739489919226614,
+      "loss": 0.5683,
+      "step": 114320
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031735249846297355,
+      "loss": 0.4596,
+      "step": 114330
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003173100977336811,
+      "loss": 0.5099,
+      "step": 114340
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003172676970043885,
+      "loss": 0.5392,
+      "step": 114350
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031722529627509595,
+      "loss": 0.6219,
+      "step": 114360
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031718289554580337,
+      "loss": 0.4468,
+      "step": 114370
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003171404948165109,
+      "loss": 0.4923,
+      "step": 114380
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003170980940872183,
+      "loss": 0.4412,
+      "step": 114390
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031705569335792577,
+      "loss": 0.5173,
+      "step": 114400
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031701329262863323,
+      "loss": 0.476,
+      "step": 114410
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003169708918993407,
+      "loss": 0.4938,
+      "step": 114420
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003169284911700481,
+      "loss": 0.4378,
+      "step": 114430
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031688609044075563,
+      "loss": 0.4715,
+      "step": 114440
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031684368971146305,
+      "loss": 0.5672,
+      "step": 114450
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003168012889821705,
+      "loss": 0.5265,
+      "step": 114460
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.000316758888252878,
+      "loss": 0.4606,
+      "step": 114470
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031671648752358545,
+      "loss": 0.6111,
+      "step": 114480
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031667408679429286,
+      "loss": 0.5009,
+      "step": 114490
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003166316860650004,
+      "loss": 0.5492,
+      "step": 114500
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003165892853357078,
+      "loss": 0.5697,
+      "step": 114510
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031654688460641526,
+      "loss": 0.5448,
+      "step": 114520
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003165044838771227,
+      "loss": 0.6029,
+      "step": 114530
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003164620831478302,
+      "loss": 0.4192,
+      "step": 114540
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003164196824185376,
+      "loss": 0.4536,
+      "step": 114550
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003163772816892451,
+      "loss": 0.4399,
+      "step": 114560
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031633488095995254,
+      "loss": 0.4574,
+      "step": 114570
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031629248023066,
+      "loss": 0.4795,
+      "step": 114580
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003162500795013674,
+      "loss": 0.3926,
+      "step": 114590
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003162076787720749,
+      "loss": 0.5373,
+      "step": 114600
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031616527804278235,
+      "loss": 0.4229,
+      "step": 114610
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003161228773134898,
+      "loss": 0.4483,
+      "step": 114620
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003160804765841973,
+      "loss": 0.4951,
+      "step": 114630
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003160380758549047,
+      "loss": 0.471,
+      "step": 114640
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031599567512561217,
+      "loss": 0.4534,
+      "step": 114650
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031595327439631963,
+      "loss": 0.5066,
+      "step": 114660
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003159108736670271,
+      "loss": 0.5055,
+      "step": 114670
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003158684729377345,
+      "loss": 0.5275,
+      "step": 114680
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.000315826072208442,
+      "loss": 0.5081,
+      "step": 114690
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031578367147914945,
+      "loss": 0.5429,
+      "step": 114700
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003157412707498569,
+      "loss": 0.4962,
+      "step": 114710
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003156988700205643,
+      "loss": 0.5458,
+      "step": 114720
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031565646929127185,
+      "loss": 0.509,
+      "step": 114730
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031561406856197926,
+      "loss": 0.518,
+      "step": 114740
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003155716678326867,
+      "loss": 0.5391,
+      "step": 114750
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003155292671033942,
+      "loss": 0.5587,
+      "step": 114760
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031548686637410166,
+      "loss": 0.5339,
+      "step": 114770
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031544446564480907,
+      "loss": 0.5164,
+      "step": 114780
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003154020649155166,
+      "loss": 0.4865,
+      "step": 114790
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.000315359664186224,
+      "loss": 0.4402,
+      "step": 114800
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031531726345693147,
+      "loss": 0.4831,
+      "step": 114810
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003152748627276389,
+      "loss": 0.5722,
+      "step": 114820
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003152324619983464,
+      "loss": 0.4701,
+      "step": 114830
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003151900612690538,
+      "loss": 0.5209,
+      "step": 114840
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003151476605397613,
+      "loss": 0.4506,
+      "step": 114850
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031510525981046875,
+      "loss": 0.487,
+      "step": 114860
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003150628590811762,
+      "loss": 0.4443,
+      "step": 114870
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031502045835188363,
+      "loss": 0.4866,
+      "step": 114880
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031497805762259115,
+      "loss": 0.5079,
+      "step": 114890
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031493565689329856,
+      "loss": 0.4953,
+      "step": 114900
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031489325616400603,
+      "loss": 0.563,
+      "step": 114910
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003148508554347135,
+      "loss": 0.5369,
+      "step": 114920
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031480845470542097,
+      "loss": 0.4302,
+      "step": 114930
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003147660539761284,
+      "loss": 0.4927,
+      "step": 114940
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003147236532468359,
+      "loss": 0.4455,
+      "step": 114950
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003146812525175433,
+      "loss": 0.5113,
+      "step": 114960
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003146388517882508,
+      "loss": 0.5523,
+      "step": 114970
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003145964510589582,
+      "loss": 0.4765,
+      "step": 114980
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003145540503296657,
+      "loss": 0.5701,
+      "step": 114990
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003145116496003731,
+      "loss": 0.4651,
+      "step": 115000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003144692488710806,
+      "loss": 0.5096,
+      "step": 115010
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031442684814178806,
+      "loss": 0.4172,
+      "step": 115020
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003143844474124955,
+      "loss": 0.5277,
+      "step": 115030
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031434204668320294,
+      "loss": 0.4566,
+      "step": 115040
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031429964595391046,
+      "loss": 0.4837,
+      "step": 115050
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031425724522461787,
+      "loss": 0.4456,
+      "step": 115060
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031421484449532534,
+      "loss": 0.4734,
+      "step": 115070
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003141724437660328,
+      "loss": 0.4867,
+      "step": 115080
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031413004303674027,
+      "loss": 0.4671,
+      "step": 115090
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003140876423074477,
+      "loss": 0.4839,
+      "step": 115100
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003140452415781552,
+      "loss": 0.546,
+      "step": 115110
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003140028408488626,
+      "loss": 0.4234,
+      "step": 115120
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003139604401195701,
+      "loss": 0.5237,
+      "step": 115130
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003139180393902775,
+      "loss": 0.4662,
+      "step": 115140
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.000313875638660985,
+      "loss": 0.5599,
+      "step": 115150
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031383323793169243,
+      "loss": 0.4377,
+      "step": 115160
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003137908372023999,
+      "loss": 0.427,
+      "step": 115170
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031374843647310736,
+      "loss": 0.5629,
+      "step": 115180
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031370603574381483,
+      "loss": 0.493,
+      "step": 115190
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031366363501452224,
+      "loss": 0.548,
+      "step": 115200
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003136212342852297,
+      "loss": 0.5536,
+      "step": 115210
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003135788335559372,
+      "loss": 0.4575,
+      "step": 115220
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031353643282664464,
+      "loss": 0.4625,
+      "step": 115230
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003134940320973521,
+      "loss": 0.4658,
+      "step": 115240
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003134516313680596,
+      "loss": 0.4834,
+      "step": 115250
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.000313409230638767,
+      "loss": 0.5417,
+      "step": 115260
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031336682990947446,
+      "loss": 0.4556,
+      "step": 115270
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003133244291801819,
+      "loss": 0.5109,
+      "step": 115280
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003132820284508894,
+      "loss": 0.5083,
+      "step": 115290
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003132396277215968,
+      "loss": 0.4481,
+      "step": 115300
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031319722699230427,
+      "loss": 0.447,
+      "step": 115310
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031315482626301174,
+      "loss": 0.4037,
+      "step": 115320
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003131124255337192,
+      "loss": 0.5308,
+      "step": 115330
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031307002480442667,
+      "loss": 0.4922,
+      "step": 115340
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003130276240751341,
+      "loss": 0.5148,
+      "step": 115350
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031298522334584155,
+      "loss": 0.5091,
+      "step": 115360
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.000312942822616549,
+      "loss": 0.4997,
+      "step": 115370
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003129004218872565,
+      "loss": 0.4475,
+      "step": 115380
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003128580211579639,
+      "loss": 0.5875,
+      "step": 115390
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003128156204286714,
+      "loss": 0.4616,
+      "step": 115400
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031277321969937883,
+      "loss": 0.4852,
+      "step": 115410
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003127308189700863,
+      "loss": 0.4862,
+      "step": 115420
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003126884182407937,
+      "loss": 0.4595,
+      "step": 115430
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031264601751150123,
+      "loss": 0.4773,
+      "step": 115440
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031260361678220864,
+      "loss": 0.4843,
+      "step": 115450
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003125612160529161,
+      "loss": 0.5217,
+      "step": 115460
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003125188153236236,
+      "loss": 0.4963,
+      "step": 115470
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031247641459433104,
+      "loss": 0.4834,
+      "step": 115480
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031243401386503845,
+      "loss": 0.4112,
+      "step": 115490
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.000312391613135746,
+      "loss": 0.4696,
+      "step": 115500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003123492124064534,
+      "loss": 0.5304,
+      "step": 115510
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031230681167716086,
+      "loss": 0.4048,
+      "step": 115520
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003122644109478683,
+      "loss": 0.5304,
+      "step": 115530
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003122220102185758,
+      "loss": 0.5368,
+      "step": 115540
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003121796094892832,
+      "loss": 0.5394,
+      "step": 115550
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003121372087599907,
+      "loss": 0.3897,
+      "step": 115560
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031209480803069814,
+      "loss": 0.585,
+      "step": 115570
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003120524073014056,
+      "loss": 0.4862,
+      "step": 115580
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.000312010006572113,
+      "loss": 0.5455,
+      "step": 115590
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031196760584282054,
+      "loss": 0.4814,
+      "step": 115600
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031192520511352795,
+      "loss": 0.5085,
+      "step": 115610
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003118828043842354,
+      "loss": 0.526,
+      "step": 115620
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003118404036549429,
+      "loss": 0.4809,
+      "step": 115630
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031179800292565035,
+      "loss": 0.4463,
+      "step": 115640
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031175560219635776,
+      "loss": 0.4926,
+      "step": 115650
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003117132014670653,
+      "loss": 0.4983,
+      "step": 115660
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003116708007377727,
+      "loss": 0.6062,
+      "step": 115670
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031162840000848016,
+      "loss": 0.4649,
+      "step": 115680
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031158599927918763,
+      "loss": 0.418,
+      "step": 115690
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003115435985498951,
+      "loss": 0.4652,
+      "step": 115700
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003115011978206025,
+      "loss": 0.4635,
+      "step": 115710
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031145879709131003,
+      "loss": 0.5309,
+      "step": 115720
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031141639636201744,
+      "loss": 0.5159,
+      "step": 115730
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003113739956327249,
+      "loss": 0.5028,
+      "step": 115740
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003113315949034323,
+      "loss": 0.4386,
+      "step": 115750
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031128919417413984,
+      "loss": 0.494,
+      "step": 115760
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031124679344484725,
+      "loss": 0.4982,
+      "step": 115770
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003112043927155547,
+      "loss": 0.4231,
+      "step": 115780
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003111619919862622,
+      "loss": 0.5679,
+      "step": 115790
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031111959125696965,
+      "loss": 0.5091,
+      "step": 115800
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031107719052767707,
+      "loss": 0.5177,
+      "step": 115810
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003110347897983846,
+      "loss": 0.4842,
+      "step": 115820
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.000310992389069092,
+      "loss": 0.4924,
+      "step": 115830
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031094998833979947,
+      "loss": 0.4781,
+      "step": 115840
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031090758761050693,
+      "loss": 0.5058,
+      "step": 115850
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003108651868812144,
+      "loss": 0.508,
+      "step": 115860
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003108227861519218,
+      "loss": 0.5045,
+      "step": 115870
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003107803854226293,
+      "loss": 0.4434,
+      "step": 115880
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031073798469333675,
+      "loss": 0.4744,
+      "step": 115890
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003106955839640442,
+      "loss": 0.5654,
+      "step": 115900
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003106531832347516,
+      "loss": 0.4187,
+      "step": 115910
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003106107825054591,
+      "loss": 0.4239,
+      "step": 115920
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031056838177616656,
+      "loss": 0.5591,
+      "step": 115930
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.000310525981046874,
+      "loss": 0.4386,
+      "step": 115940
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003104835803175815,
+      "loss": 0.4639,
+      "step": 115950
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003104411795882889,
+      "loss": 0.4677,
+      "step": 115960
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003103987788589964,
+      "loss": 0.5596,
+      "step": 115970
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031035637812970384,
+      "loss": 0.4698,
+      "step": 115980
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003103139774004113,
+      "loss": 0.5218,
+      "step": 115990
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003102715766711188,
+      "loss": 0.4619,
+      "step": 116000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00031022917594182624,
+      "loss": 0.545,
+      "step": 116010
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00031018677521253365,
+      "loss": 0.5864,
+      "step": 116020
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003101443744832411,
+      "loss": 0.5749,
+      "step": 116030
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003101019737539486,
+      "loss": 0.3888,
+      "step": 116040
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00031005957302465605,
+      "loss": 0.4275,
+      "step": 116050
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00031001717229536347,
+      "loss": 0.494,
+      "step": 116060
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00030997477156607093,
+      "loss": 0.475,
+      "step": 116070
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003099323708367784,
+      "loss": 0.4768,
+      "step": 116080
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00030988997010748587,
+      "loss": 0.4983,
+      "step": 116090
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003098475693781933,
+      "loss": 0.4784,
+      "step": 116100
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003098051686489008,
+      "loss": 0.5156,
+      "step": 116110
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003097627679196082,
+      "loss": 0.4454,
+      "step": 116120
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003097203671903157,
+      "loss": 0.4646,
+      "step": 116130
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00030967796646102315,
+      "loss": 0.5723,
+      "step": 116140
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003096355657317306,
+      "loss": 0.481,
+      "step": 116150
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.000309593165002438,
+      "loss": 0.516,
+      "step": 116160
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00030955076427314555,
+      "loss": 0.5028,
+      "step": 116170
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00030950836354385296,
+      "loss": 0.5142,
+      "step": 116180
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003094659628145604,
+      "loss": 0.5335,
+      "step": 116190
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00030942356208526784,
+      "loss": 0.5553,
+      "step": 116200
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00030938116135597536,
+      "loss": 0.5141,
+      "step": 116210
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00030933876062668277,
+      "loss": 0.4998,
+      "step": 116220
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030929635989739024,
+      "loss": 0.5709,
+      "step": 116230
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003092539591680977,
+      "loss": 0.4947,
+      "step": 116240
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030921155843880517,
+      "loss": 0.5475,
+      "step": 116250
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003091691577095126,
+      "loss": 0.5793,
+      "step": 116260
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003091267569802201,
+      "loss": 0.5252,
+      "step": 116270
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003090843562509275,
+      "loss": 0.5052,
+      "step": 116280
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.000309041955521635,
+      "loss": 0.4926,
+      "step": 116290
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030899955479234245,
+      "loss": 0.51,
+      "step": 116300
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003089571540630499,
+      "loss": 0.5582,
+      "step": 116310
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030891475333375733,
+      "loss": 0.5628,
+      "step": 116320
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030887235260446485,
+      "loss": 0.4366,
+      "step": 116330
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030882995187517226,
+      "loss": 0.505,
+      "step": 116340
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030878755114587973,
+      "loss": 0.4806,
+      "step": 116350
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030874515041658714,
+      "loss": 0.4482,
+      "step": 116360
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030870274968729467,
+      "loss": 0.5082,
+      "step": 116370
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003086603489580021,
+      "loss": 0.5105,
+      "step": 116380
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030861794822870954,
+      "loss": 0.4213,
+      "step": 116390
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.000308575547499417,
+      "loss": 0.5633,
+      "step": 116400
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003085331467701245,
+      "loss": 0.4861,
+      "step": 116410
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003084907460408319,
+      "loss": 0.5051,
+      "step": 116420
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003084483453115394,
+      "loss": 0.4623,
+      "step": 116430
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003084059445822468,
+      "loss": 0.5376,
+      "step": 116440
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003083635438529543,
+      "loss": 0.563,
+      "step": 116450
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030832114312366176,
+      "loss": 0.4177,
+      "step": 116460
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003082787423943692,
+      "loss": 0.4607,
+      "step": 116470
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030823634166507664,
+      "loss": 0.6,
+      "step": 116480
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030819394093578416,
+      "loss": 0.4971,
+      "step": 116490
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030815154020649157,
+      "loss": 0.4944,
+      "step": 116500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030810913947719904,
+      "loss": 0.3751,
+      "step": 116510
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030806673874790645,
+      "loss": 0.4569,
+      "step": 116520
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030802433801861397,
+      "loss": 0.4815,
+      "step": 116530
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003079819372893214,
+      "loss": 0.5019,
+      "step": 116540
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030793953656002885,
+      "loss": 0.4659,
+      "step": 116550
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003078971358307363,
+      "loss": 0.5544,
+      "step": 116560
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003078547351014438,
+      "loss": 0.4565,
+      "step": 116570
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003078123343721512,
+      "loss": 0.5113,
+      "step": 116580
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030776993364285866,
+      "loss": 0.4616,
+      "step": 116590
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030772753291356613,
+      "loss": 0.4732,
+      "step": 116600
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003076851321842736,
+      "loss": 0.5328,
+      "step": 116610
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030764273145498106,
+      "loss": 0.5333,
+      "step": 116620
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003076003307256885,
+      "loss": 0.5194,
+      "step": 116630
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030755792999639594,
+      "loss": 0.5152,
+      "step": 116640
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003075155292671034,
+      "loss": 0.4664,
+      "step": 116650
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003074731285378109,
+      "loss": 0.504,
+      "step": 116660
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003074307278085183,
+      "loss": 0.46,
+      "step": 116670
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030738832707922576,
+      "loss": 0.4192,
+      "step": 116680
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003073459263499332,
+      "loss": 0.4158,
+      "step": 116690
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003073035256206407,
+      "loss": 0.5494,
+      "step": 116700
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003072611248913481,
+      "loss": 0.5004,
+      "step": 116710
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003072187241620556,
+      "loss": 0.4949,
+      "step": 116720
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030717632343276304,
+      "loss": 0.4833,
+      "step": 116730
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003071339227034705,
+      "loss": 0.5164,
+      "step": 116740
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030709152197417797,
+      "loss": 0.5109,
+      "step": 116750
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030704912124488544,
+      "loss": 0.5301,
+      "step": 116760
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030700672051559285,
+      "loss": 0.4843,
+      "step": 116770
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030696431978630037,
+      "loss": 0.4544,
+      "step": 116780
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003069219190570078,
+      "loss": 0.5131,
+      "step": 116790
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030687951832771525,
+      "loss": 0.4595,
+      "step": 116800
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003068371175984227,
+      "loss": 0.4926,
+      "step": 116810
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003067947168691302,
+      "loss": 0.4475,
+      "step": 116820
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003067523161398376,
+      "loss": 0.4522,
+      "step": 116830
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030670991541054506,
+      "loss": 0.4752,
+      "step": 116840
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030666751468125253,
+      "loss": 0.4368,
+      "step": 116850
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030662511395196,
+      "loss": 0.4989,
+      "step": 116860
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003065827132226674,
+      "loss": 0.5151,
+      "step": 116870
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030654031249337493,
+      "loss": 0.5249,
+      "step": 116880
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030649791176408234,
+      "loss": 0.4855,
+      "step": 116890
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003064555110347898,
+      "loss": 0.5098,
+      "step": 116900
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003064131103054973,
+      "loss": 0.4798,
+      "step": 116910
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030637070957620474,
+      "loss": 0.4192,
+      "step": 116920
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030632830884691215,
+      "loss": 0.4261,
+      "step": 116930
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003062859081176197,
+      "loss": 0.5457,
+      "step": 116940
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003062435073883271,
+      "loss": 0.4903,
+      "step": 116950
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030620110665903456,
+      "loss": 0.5544,
+      "step": 116960
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030615870592974197,
+      "loss": 0.5157,
+      "step": 116970
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003061163052004495,
+      "loss": 0.4232,
+      "step": 116980
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003060739044711569,
+      "loss": 0.5069,
+      "step": 116990
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030603150374186437,
+      "loss": 0.5024,
+      "step": 117000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030598910301257183,
+      "loss": 0.5531,
+      "step": 117010
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003059467022832793,
+      "loss": 0.4584,
+      "step": 117020
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003059043015539867,
+      "loss": 0.4407,
+      "step": 117030
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030586190082469424,
+      "loss": 0.4469,
+      "step": 117040
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030581950009540165,
+      "loss": 0.5155,
+      "step": 117050
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003057770993661091,
+      "loss": 0.5061,
+      "step": 117060
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003057346986368166,
+      "loss": 0.5654,
+      "step": 117070
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030569229790752405,
+      "loss": 0.4628,
+      "step": 117080
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030564989717823146,
+      "loss": 0.5373,
+      "step": 117090
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.000305607496448939,
+      "loss": 0.5112,
+      "step": 117100
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003055650957196464,
+      "loss": 0.5358,
+      "step": 117110
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030552269499035386,
+      "loss": 0.5419,
+      "step": 117120
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003054802942610613,
+      "loss": 0.4735,
+      "step": 117130
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003054378935317688,
+      "loss": 0.4375,
+      "step": 117140
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003053954928024762,
+      "loss": 0.5052,
+      "step": 117150
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003053530920731837,
+      "loss": 0.5533,
+      "step": 117160
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030531069134389114,
+      "loss": 0.522,
+      "step": 117170
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003052682906145986,
+      "loss": 0.4975,
+      "step": 117180
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.000305225889885306,
+      "loss": 0.5206,
+      "step": 117190
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003051834891560135,
+      "loss": 0.56,
+      "step": 117200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030514108842672095,
+      "loss": 0.6001,
+      "step": 117210
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003050986876974284,
+      "loss": 0.5088,
+      "step": 117220
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003050562869681359,
+      "loss": 0.4581,
+      "step": 117230
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030501388623884335,
+      "loss": 0.5225,
+      "step": 117240
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030497148550955077,
+      "loss": 0.456,
+      "step": 117250
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030492908478025823,
+      "loss": 0.4829,
+      "step": 117260
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003048866840509657,
+      "loss": 0.4498,
+      "step": 117270
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030484428332167317,
+      "loss": 0.5406,
+      "step": 117280
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003048018825923806,
+      "loss": 0.448,
+      "step": 117290
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030475948186308805,
+      "loss": 0.4367,
+      "step": 117300
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003047170811337955,
+      "loss": 0.3891,
+      "step": 117310
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.000304674680404503,
+      "loss": 0.4346,
+      "step": 117320
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030463227967521045,
+      "loss": 0.455,
+      "step": 117330
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030458987894591786,
+      "loss": 0.5004,
+      "step": 117340
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003045474782166253,
+      "loss": 0.5391,
+      "step": 117350
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003045050774873328,
+      "loss": 0.4236,
+      "step": 117360
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030446267675804026,
+      "loss": 0.5025,
+      "step": 117370
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030442027602874767,
+      "loss": 0.5744,
+      "step": 117380
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003043778752994552,
+      "loss": 0.5166,
+      "step": 117390
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003043354745701626,
+      "loss": 0.5652,
+      "step": 117400
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030429307384087007,
+      "loss": 0.485,
+      "step": 117410
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030425067311157754,
+      "loss": 0.4593,
+      "step": 117420
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.000304208272382285,
+      "loss": 0.5312,
+      "step": 117430
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003041658716529924,
+      "loss": 0.4736,
+      "step": 117440
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003041234709236999,
+      "loss": 0.5684,
+      "step": 117450
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030408107019440735,
+      "loss": 0.492,
+      "step": 117460
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003040386694651148,
+      "loss": 0.5304,
+      "step": 117470
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030399626873582223,
+      "loss": 0.4837,
+      "step": 117480
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030395386800652975,
+      "loss": 0.4962,
+      "step": 117490
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030391146727723717,
+      "loss": 0.4648,
+      "step": 117500
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030386906654794463,
+      "loss": 0.5192,
+      "step": 117510
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003038266658186521,
+      "loss": 0.4705,
+      "step": 117520
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030378426508935957,
+      "loss": 0.5472,
+      "step": 117530
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.000303741864360067,
+      "loss": 0.499,
+      "step": 117540
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003036994636307745,
+      "loss": 0.5548,
+      "step": 117550
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003036570629014819,
+      "loss": 0.3976,
+      "step": 117560
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003036146621721894,
+      "loss": 0.4767,
+      "step": 117570
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003035722614428968,
+      "loss": 0.4652,
+      "step": 117580
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003035298607136043,
+      "loss": 0.5103,
+      "step": 117590
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003034874599843117,
+      "loss": 0.538,
+      "step": 117600
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003034450592550192,
+      "loss": 0.4392,
+      "step": 117610
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030340265852572666,
+      "loss": 0.4984,
+      "step": 117620
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003033602577964341,
+      "loss": 0.5534,
+      "step": 117630
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030331785706714154,
+      "loss": 0.5756,
+      "step": 117640
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030327545633784906,
+      "loss": 0.5292,
+      "step": 117650
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030323305560855647,
+      "loss": 0.482,
+      "step": 117660
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030319065487926394,
+      "loss": 0.4755,
+      "step": 117670
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003031482541499714,
+      "loss": 0.401,
+      "step": 117680
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030310585342067887,
+      "loss": 0.3924,
+      "step": 117690
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003030634526913863,
+      "loss": 0.6677,
+      "step": 117700
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003030210519620938,
+      "loss": 0.5163,
+      "step": 117710
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003029786512328012,
+      "loss": 0.5435,
+      "step": 117720
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003029362505035087,
+      "loss": 0.5297,
+      "step": 117730
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003028938497742161,
+      "loss": 0.4901,
+      "step": 117740
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003028514490449236,
+      "loss": 0.494,
+      "step": 117750
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030280904831563103,
+      "loss": 0.5158,
+      "step": 117760
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003027666475863385,
+      "loss": 0.4836,
+      "step": 117770
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030272424685704596,
+      "loss": 0.5246,
+      "step": 117780
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030268184612775343,
+      "loss": 0.5265,
+      "step": 117790
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030263944539846084,
+      "loss": 0.4729,
+      "step": 117800
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030259704466916836,
+      "loss": 0.4861,
+      "step": 117810
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003025546439398758,
+      "loss": 0.4759,
+      "step": 117820
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030251224321058324,
+      "loss": 0.5824,
+      "step": 117830
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003024698424812907,
+      "loss": 0.5232,
+      "step": 117840
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003024274417519982,
+      "loss": 0.4556,
+      "step": 117850
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003023850410227056,
+      "loss": 0.4461,
+      "step": 117860
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030234264029341306,
+      "loss": 0.506,
+      "step": 117870
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003023002395641205,
+      "loss": 0.433,
+      "step": 117880
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.000302257838834828,
+      "loss": 0.4924,
+      "step": 117890
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003022154381055354,
+      "loss": 0.4476,
+      "step": 117900
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030217303737624287,
+      "loss": 0.5044,
+      "step": 117910
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030213063664695034,
+      "loss": 0.5141,
+      "step": 117920
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003020882359176578,
+      "loss": 0.4987,
+      "step": 117930
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030204583518836527,
+      "loss": 0.5272,
+      "step": 117940
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003020034344590727,
+      "loss": 0.4535,
+      "step": 117950
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030196103372978015,
+      "loss": 0.5096,
+      "step": 117960
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003019186330004876,
+      "loss": 0.4555,
+      "step": 117970
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003018762322711951,
+      "loss": 0.4815,
+      "step": 117980
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030183383154190255,
+      "loss": 0.4404,
+      "step": 117990
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030179143081261,
+      "loss": 0.4458,
+      "step": 118000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030174903008331743,
+      "loss": 0.4571,
+      "step": 118010
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003017066293540249,
+      "loss": 0.4371,
+      "step": 118020
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030166422862473236,
+      "loss": 0.4887,
+      "step": 118030
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030162182789543983,
+      "loss": 0.5693,
+      "step": 118040
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030157942716614724,
+      "loss": 0.5188,
+      "step": 118050
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003015370264368547,
+      "loss": 0.4771,
+      "step": 118060
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003014946257075622,
+      "loss": 0.4985,
+      "step": 118070
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030145222497826964,
+      "loss": 0.5457,
+      "step": 118080
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030140982424897706,
+      "loss": 0.5065,
+      "step": 118090
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003013674235196846,
+      "loss": 0.4892,
+      "step": 118100
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.000301325022790392,
+      "loss": 0.424,
+      "step": 118110
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030128262206109946,
+      "loss": 0.4578,
+      "step": 118120
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003012402213318069,
+      "loss": 0.5222,
+      "step": 118130
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003011978206025144,
+      "loss": 0.5164,
+      "step": 118140
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003011554198732218,
+      "loss": 0.4121,
+      "step": 118150
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003011130191439293,
+      "loss": 0.5155,
+      "step": 118160
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030107061841463674,
+      "loss": 0.573,
+      "step": 118170
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003010282176853442,
+      "loss": 0.5218,
+      "step": 118180
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030098581695605167,
+      "loss": 0.5038,
+      "step": 118190
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030094341622675914,
+      "loss": 0.4662,
+      "step": 118200
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030090101549746655,
+      "loss": 0.5325,
+      "step": 118210
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.000300858614768174,
+      "loss": 0.5972,
+      "step": 118220
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003008162140388815,
+      "loss": 0.4889,
+      "step": 118230
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030077381330958895,
+      "loss": 0.4908,
+      "step": 118240
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030073141258029636,
+      "loss": 0.4795,
+      "step": 118250
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003006890118510039,
+      "loss": 0.3743,
+      "step": 118260
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003006466111217113,
+      "loss": 0.3919,
+      "step": 118270
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030060421039241876,
+      "loss": 0.4698,
+      "step": 118280
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030056180966312623,
+      "loss": 0.4007,
+      "step": 118290
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003005194089338337,
+      "loss": 0.4578,
+      "step": 118300
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003004770082045411,
+      "loss": 0.49,
+      "step": 118310
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030043460747524863,
+      "loss": 0.4356,
+      "step": 118320
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030039220674595604,
+      "loss": 0.4511,
+      "step": 118330
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003003498060166635,
+      "loss": 0.4216,
+      "step": 118340
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003003074052873709,
+      "loss": 0.4491,
+      "step": 118350
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030026500455807844,
+      "loss": 0.5323,
+      "step": 118360
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030022260382878585,
+      "loss": 0.4616,
+      "step": 118370
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0003001802030994933,
+      "loss": 0.3807,
+      "step": 118380
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0003001378023702008,
+      "loss": 0.4736,
+      "step": 118390
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030009540164090825,
+      "loss": 0.4643,
+      "step": 118400
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030005300091161567,
+      "loss": 0.4294,
+      "step": 118410
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0003000106001823232,
+      "loss": 0.4354,
+      "step": 118420
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002999681994530306,
+      "loss": 0.3918,
+      "step": 118430
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00029992579872373807,
+      "loss": 0.4468,
+      "step": 118440
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00029988339799444553,
+      "loss": 0.4394,
+      "step": 118450
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.000299840997265153,
+      "loss": 0.3717,
+      "step": 118460
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002997985965358604,
+      "loss": 0.4624,
+      "step": 118470
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002997561958065679,
+      "loss": 0.4499,
+      "step": 118480
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00029971379507727535,
+      "loss": 0.4401,
+      "step": 118490
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002996713943479828,
+      "loss": 0.433,
+      "step": 118500
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002996289936186902,
+      "loss": 0.4414,
+      "step": 118510
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00029958659288939775,
+      "loss": 0.4363,
+      "step": 118520
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00029954419216010516,
+      "loss": 0.3571,
+      "step": 118530
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00029950179143081263,
+      "loss": 0.4909,
+      "step": 118540
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002994593907015201,
+      "loss": 0.4166,
+      "step": 118550
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00029941698997222756,
+      "loss": 0.3843,
+      "step": 118560
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.000299374589242935,
+      "loss": 0.4603,
+      "step": 118570
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00029933218851364244,
+      "loss": 0.479,
+      "step": 118580
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002992897877843499,
+      "loss": 0.3815,
+      "step": 118590
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002992473870550574,
+      "loss": 0.4867,
+      "step": 118600
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029920498632576484,
+      "loss": 0.4444,
+      "step": 118610
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029916258559647225,
+      "loss": 0.4278,
+      "step": 118620
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002991201848671797,
+      "loss": 0.441,
+      "step": 118630
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002990777841378872,
+      "loss": 0.4316,
+      "step": 118640
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029903538340859465,
+      "loss": 0.4239,
+      "step": 118650
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029899298267930207,
+      "loss": 0.4164,
+      "step": 118660
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029895058195000953,
+      "loss": 0.5171,
+      "step": 118670
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.000298908181220717,
+      "loss": 0.3567,
+      "step": 118680
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029886578049142447,
+      "loss": 0.4179,
+      "step": 118690
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002988233797621319,
+      "loss": 0.407,
+      "step": 118700
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002987809790328394,
+      "loss": 0.4484,
+      "step": 118710
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002987385783035468,
+      "loss": 0.3892,
+      "step": 118720
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002986961775742543,
+      "loss": 0.3902,
+      "step": 118730
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029865377684496175,
+      "loss": 0.407,
+      "step": 118740
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002986113761156692,
+      "loss": 0.4316,
+      "step": 118750
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002985689753863766,
+      "loss": 0.4246,
+      "step": 118760
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029852657465708415,
+      "loss": 0.386,
+      "step": 118770
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029848417392779156,
+      "loss": 0.5008,
+      "step": 118780
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.000298441773198499,
+      "loss": 0.3723,
+      "step": 118790
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002983993724692065,
+      "loss": 0.4671,
+      "step": 118800
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029835697173991396,
+      "loss": 0.4579,
+      "step": 118810
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029831457101062137,
+      "loss": 0.508,
+      "step": 118820
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00029827217028132884,
+      "loss": 0.5177,
+      "step": 118830
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002982297695520363,
+      "loss": 0.4905,
+      "step": 118840
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00029818736882274377,
+      "loss": 0.509,
+      "step": 118850
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002981449680934512,
+      "loss": 0.4522,
+      "step": 118860
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002981025673641587,
+      "loss": 0.4388,
+      "step": 118870
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002980601666348661,
+      "loss": 0.4781,
+      "step": 118880
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002980177659055736,
+      "loss": 0.4532,
+      "step": 118890
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00029797536517628105,
+      "loss": 0.4153,
+      "step": 118900
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002979329644469885,
+      "loss": 0.4216,
+      "step": 118910
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00029789056371769593,
+      "loss": 0.4545,
+      "step": 118920
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00029784816298840345,
+      "loss": 0.3802,
+      "step": 118930
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00029780576225911087,
+      "loss": 0.4638,
+      "step": 118940
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00029776336152981833,
+      "loss": 0.4715,
+      "step": 118950
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00029772096080052574,
+      "loss": 0.4514,
+      "step": 118960
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00029767856007123327,
+      "loss": 0.4674,
+      "step": 118970
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002976361593419407,
+      "loss": 0.3853,
+      "step": 118980
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00029759375861264814,
+      "loss": 0.4092,
+      "step": 118990
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002975513578833556,
+      "loss": 0.3981,
+      "step": 119000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002975089571540631,
+      "loss": 0.4984,
+      "step": 119010
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002974665564247705,
+      "loss": 0.4744,
+      "step": 119020
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.000297424155695478,
+      "loss": 0.4588,
+      "step": 119030
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002973817549661854,
+      "loss": 0.4485,
+      "step": 119040
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002973393542368929,
+      "loss": 0.4269,
+      "step": 119050
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00029729695350760036,
+      "loss": 0.4061,
+      "step": 119060
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002972545527783078,
+      "loss": 0.3342,
+      "step": 119070
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029721215204901524,
+      "loss": 0.396,
+      "step": 119080
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029716975131972276,
+      "loss": 0.4375,
+      "step": 119090
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029712735059043017,
+      "loss": 0.4536,
+      "step": 119100
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029708494986113764,
+      "loss": 0.5258,
+      "step": 119110
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029704254913184505,
+      "loss": 0.5019,
+      "step": 119120
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029700014840255257,
+      "loss": 0.4427,
+      "step": 119130
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029695774767326,
+      "loss": 0.4917,
+      "step": 119140
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029691534694396745,
+      "loss": 0.476,
+      "step": 119150
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002968729462146749,
+      "loss": 0.4614,
+      "step": 119160
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002968305454853824,
+      "loss": 0.4864,
+      "step": 119170
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002967881447560898,
+      "loss": 0.4864,
+      "step": 119180
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029674574402679726,
+      "loss": 0.4458,
+      "step": 119190
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029670334329750473,
+      "loss": 0.4074,
+      "step": 119200
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002966609425682122,
+      "loss": 0.4787,
+      "step": 119210
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029661854183891966,
+      "loss": 0.4457,
+      "step": 119220
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002965761411096271,
+      "loss": 0.4461,
+      "step": 119230
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029653374038033454,
+      "loss": 0.4378,
+      "step": 119240
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.000296491339651042,
+      "loss": 0.4709,
+      "step": 119250
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002964489389217495,
+      "loss": 0.4948,
+      "step": 119260
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029640653819245694,
+      "loss": 0.4355,
+      "step": 119270
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029636413746316436,
+      "loss": 0.4555,
+      "step": 119280
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002963217367338718,
+      "loss": 0.4187,
+      "step": 119290
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002962793360045793,
+      "loss": 0.4291,
+      "step": 119300
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029623693527528676,
+      "loss": 0.4265,
+      "step": 119310
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002961945345459942,
+      "loss": 0.452,
+      "step": 119320
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029615213381670164,
+      "loss": 0.3971,
+      "step": 119330
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002961097330874091,
+      "loss": 0.4395,
+      "step": 119340
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029606733235811657,
+      "loss": 0.4639,
+      "step": 119350
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029602493162882404,
+      "loss": 0.4533,
+      "step": 119360
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029598253089953145,
+      "loss": 0.4588,
+      "step": 119370
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029594013017023897,
+      "loss": 0.4844,
+      "step": 119380
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002958977294409464,
+      "loss": 0.3619,
+      "step": 119390
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029585532871165385,
+      "loss": 0.4382,
+      "step": 119400
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002958129279823613,
+      "loss": 0.4357,
+      "step": 119410
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002957705272530688,
+      "loss": 0.4684,
+      "step": 119420
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002957281265237762,
+      "loss": 0.4458,
+      "step": 119430
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029568572579448366,
+      "loss": 0.3695,
+      "step": 119440
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029564332506519113,
+      "loss": 0.4284,
+      "step": 119450
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002956009243358986,
+      "loss": 0.472,
+      "step": 119460
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.000295558523606606,
+      "loss": 0.4735,
+      "step": 119470
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029551612287731353,
+      "loss": 0.481,
+      "step": 119480
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029547372214802094,
+      "loss": 0.4329,
+      "step": 119490
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002954313214187284,
+      "loss": 0.4135,
+      "step": 119500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002953889206894359,
+      "loss": 0.427,
+      "step": 119510
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029534651996014334,
+      "loss": 0.4082,
+      "step": 119520
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029530411923085076,
+      "loss": 0.4557,
+      "step": 119530
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002952617185015583,
+      "loss": 0.454,
+      "step": 119540
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002952193177722657,
+      "loss": 0.4995,
+      "step": 119550
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029517691704297316,
+      "loss": 0.409,
+      "step": 119560
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002951345163136806,
+      "loss": 0.4714,
+      "step": 119570
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002950921155843881,
+      "loss": 0.4417,
+      "step": 119580
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002950497148550955,
+      "loss": 0.4236,
+      "step": 119590
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029500731412580297,
+      "loss": 0.4311,
+      "step": 119600
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029496491339651044,
+      "loss": 0.3869,
+      "step": 119610
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002949225126672179,
+      "loss": 0.3998,
+      "step": 119620
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002948801119379253,
+      "loss": 0.4858,
+      "step": 119630
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029483771120863284,
+      "loss": 0.5396,
+      "step": 119640
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029479531047934025,
+      "loss": 0.4521,
+      "step": 119650
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002947529097500477,
+      "loss": 0.3859,
+      "step": 119660
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002947105090207552,
+      "loss": 0.406,
+      "step": 119670
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029466810829146265,
+      "loss": 0.4799,
+      "step": 119680
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029462570756217006,
+      "loss": 0.4349,
+      "step": 119690
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002945833068328776,
+      "loss": 0.4563,
+      "step": 119700
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.000294540906103585,
+      "loss": 0.3544,
+      "step": 119710
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029449850537429246,
+      "loss": 0.4368,
+      "step": 119720
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002944561046449999,
+      "loss": 0.4207,
+      "step": 119730
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002944137039157074,
+      "loss": 0.4692,
+      "step": 119740
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002943713031864148,
+      "loss": 0.4174,
+      "step": 119750
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002943289024571223,
+      "loss": 0.4493,
+      "step": 119760
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029428650172782974,
+      "loss": 0.4576,
+      "step": 119770
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002942441009985372,
+      "loss": 0.3887,
+      "step": 119780
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002942017002692446,
+      "loss": 0.4793,
+      "step": 119790
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029415929953995214,
+      "loss": 0.4482,
+      "step": 119800
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029411689881065955,
+      "loss": 0.4218,
+      "step": 119810
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.000294074498081367,
+      "loss": 0.4867,
+      "step": 119820
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002940320973520745,
+      "loss": 0.442,
+      "step": 119830
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029398969662278195,
+      "loss": 0.5015,
+      "step": 119840
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029394729589348937,
+      "loss": 0.476,
+      "step": 119850
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029390489516419683,
+      "loss": 0.4008,
+      "step": 119860
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002938624944349043,
+      "loss": 0.4272,
+      "step": 119870
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029382009370561177,
+      "loss": 0.4004,
+      "step": 119880
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002937776929763192,
+      "loss": 0.4223,
+      "step": 119890
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029373529224702665,
+      "loss": 0.5161,
+      "step": 119900
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002936928915177341,
+      "loss": 0.4125,
+      "step": 119910
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002936504907884416,
+      "loss": 0.3728,
+      "step": 119920
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029360809005914905,
+      "loss": 0.4935,
+      "step": 119930
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029356568932985646,
+      "loss": 0.3668,
+      "step": 119940
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002935232886005639,
+      "loss": 0.5353,
+      "step": 119950
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002934808878712714,
+      "loss": 0.406,
+      "step": 119960
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029343848714197886,
+      "loss": 0.3757,
+      "step": 119970
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029339608641268627,
+      "loss": 0.49,
+      "step": 119980
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002933536856833938,
+      "loss": 0.4489,
+      "step": 119990
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002933112849541012,
+      "loss": 0.4493,
+      "step": 120000
+    },
+    {
+      "epoch": 5.07,
+      "eval_loss": 0.6166871190071106,
+      "eval_runtime": 337.5835,
+      "eval_samples_per_second": 15.567,
+      "eval_steps_per_second": 3.892,
+      "step": 120000
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002932688842248087,
+      "loss": 0.4113,
+      "step": 120010
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029322648349551614,
+      "loss": 0.378,
+      "step": 120020
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002931840827662236,
+      "loss": 0.428,
+      "step": 120030
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.000293141682036931,
+      "loss": 0.4916,
+      "step": 120040
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002930992813076385,
+      "loss": 0.4533,
+      "step": 120050
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029305688057834595,
+      "loss": 0.4051,
+      "step": 120060
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002930144798490534,
+      "loss": 0.5,
+      "step": 120070
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029297207911976083,
+      "loss": 0.4556,
+      "step": 120080
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029292967839046835,
+      "loss": 0.5262,
+      "step": 120090
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029288727766117577,
+      "loss": 0.4996,
+      "step": 120100
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029284487693188323,
+      "loss": 0.3569,
+      "step": 120110
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002928024762025907,
+      "loss": 0.4744,
+      "step": 120120
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029276007547329817,
+      "loss": 0.4336,
+      "step": 120130
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002927176747440056,
+      "loss": 0.4766,
+      "step": 120140
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002926752740147131,
+      "loss": 0.5059,
+      "step": 120150
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002926328732854205,
+      "loss": 0.5101,
+      "step": 120160
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.000292590472556128,
+      "loss": 0.4061,
+      "step": 120170
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029254807182683545,
+      "loss": 0.4054,
+      "step": 120180
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002925056710975429,
+      "loss": 0.4351,
+      "step": 120190
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002924632703682503,
+      "loss": 0.4343,
+      "step": 120200
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002924208696389578,
+      "loss": 0.3402,
+      "step": 120210
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029237846890966526,
+      "loss": 0.3841,
+      "step": 120220
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002923360681803727,
+      "loss": 0.4945,
+      "step": 120230
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029229366745108014,
+      "loss": 0.439,
+      "step": 120240
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029225126672178766,
+      "loss": 0.4465,
+      "step": 120250
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029220886599249507,
+      "loss": 0.4644,
+      "step": 120260
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029216646526320254,
+      "loss": 0.4964,
+      "step": 120270
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029212406453391,
+      "loss": 0.3841,
+      "step": 120280
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029208166380461747,
+      "loss": 0.5011,
+      "step": 120290
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002920392630753249,
+      "loss": 0.4438,
+      "step": 120300
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002919968623460324,
+      "loss": 0.3761,
+      "step": 120310
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002919544616167398,
+      "loss": 0.404,
+      "step": 120320
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002919120608874473,
+      "loss": 0.4051,
+      "step": 120330
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029186966015815475,
+      "loss": 0.3904,
+      "step": 120340
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002918272594288622,
+      "loss": 0.453,
+      "step": 120350
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029178485869956963,
+      "loss": 0.4404,
+      "step": 120360
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002917424579702771,
+      "loss": 0.4397,
+      "step": 120370
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029170005724098456,
+      "loss": 0.3858,
+      "step": 120380
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029165765651169203,
+      "loss": 0.5937,
+      "step": 120390
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029161525578239944,
+      "loss": 0.5642,
+      "step": 120400
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029157285505310697,
+      "loss": 0.4428,
+      "step": 120410
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002915304543238144,
+      "loss": 0.4162,
+      "step": 120420
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029148805359452184,
+      "loss": 0.4062,
+      "step": 120430
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002914456528652293,
+      "loss": 0.3962,
+      "step": 120440
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002914032521359368,
+      "loss": 0.4585,
+      "step": 120450
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002913608514066442,
+      "loss": 0.3695,
+      "step": 120460
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029131845067735166,
+      "loss": 0.5408,
+      "step": 120470
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002912760499480591,
+      "loss": 0.4901,
+      "step": 120480
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002912336492187666,
+      "loss": 0.4666,
+      "step": 120490
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.000291191248489474,
+      "loss": 0.4244,
+      "step": 120500
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002911488477601815,
+      "loss": 0.4949,
+      "step": 120510
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029110644703088894,
+      "loss": 0.3635,
+      "step": 120520
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002910640463015964,
+      "loss": 0.4143,
+      "step": 120530
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029102164557230387,
+      "loss": 0.4768,
+      "step": 120540
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029097924484301134,
+      "loss": 0.4946,
+      "step": 120550
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029093684411371875,
+      "loss": 0.5449,
+      "step": 120560
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002908944433844262,
+      "loss": 0.4687,
+      "step": 120570
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002908520426551337,
+      "loss": 0.4103,
+      "step": 120580
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029080964192584115,
+      "loss": 0.4336,
+      "step": 120590
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002907672411965486,
+      "loss": 0.4058,
+      "step": 120600
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029072484046725603,
+      "loss": 0.4794,
+      "step": 120610
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002906824397379635,
+      "loss": 0.439,
+      "step": 120620
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029064003900867096,
+      "loss": 0.4371,
+      "step": 120630
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029059763827937843,
+      "loss": 0.5636,
+      "step": 120640
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029055523755008584,
+      "loss": 0.473,
+      "step": 120650
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002905128368207933,
+      "loss": 0.5103,
+      "step": 120660
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002904704360915008,
+      "loss": 0.4398,
+      "step": 120670
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029042803536220824,
+      "loss": 0.4569,
+      "step": 120680
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029038563463291566,
+      "loss": 0.4471,
+      "step": 120690
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002903432339036232,
+      "loss": 0.4572,
+      "step": 120700
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002903008331743306,
+      "loss": 0.44,
+      "step": 120710
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00029025843244503806,
+      "loss": 0.4542,
+      "step": 120720
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002902160317157455,
+      "loss": 0.4567,
+      "step": 120730
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.000290173630986453,
+      "loss": 0.5224,
+      "step": 120740
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002901312302571604,
+      "loss": 0.5153,
+      "step": 120750
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002900888295278679,
+      "loss": 0.4475,
+      "step": 120760
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00029004642879857534,
+      "loss": 0.4322,
+      "step": 120770
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002900040280692828,
+      "loss": 0.3882,
+      "step": 120780
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00028996162733999027,
+      "loss": 0.521,
+      "step": 120790
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00028991922661069774,
+      "loss": 0.5127,
+      "step": 120800
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00028987682588140515,
+      "loss": 0.4666,
+      "step": 120810
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002898344251521126,
+      "loss": 0.4165,
+      "step": 120820
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002897920244228201,
+      "loss": 0.5306,
+      "step": 120830
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00028974962369352755,
+      "loss": 0.437,
+      "step": 120840
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00028970722296423496,
+      "loss": 0.3806,
+      "step": 120850
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002896648222349425,
+      "loss": 0.3951,
+      "step": 120860
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002896224215056499,
+      "loss": 0.4742,
+      "step": 120870
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00028958002077635736,
+      "loss": 0.4839,
+      "step": 120880
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00028953762004706483,
+      "loss": 0.4528,
+      "step": 120890
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002894952193177723,
+      "loss": 0.5627,
+      "step": 120900
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002894528185884797,
+      "loss": 0.3479,
+      "step": 120910
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00028941041785918723,
+      "loss": 0.5019,
+      "step": 120920
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00028936801712989464,
+      "loss": 0.4358,
+      "step": 120930
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002893256164006021,
+      "loss": 0.3892,
+      "step": 120940
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002892832156713096,
+      "loss": 0.4206,
+      "step": 120950
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028924081494201704,
+      "loss": 0.4978,
+      "step": 120960
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028919841421272445,
+      "loss": 0.3915,
+      "step": 120970
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002891560134834319,
+      "loss": 0.4202,
+      "step": 120980
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002891136127541394,
+      "loss": 0.4517,
+      "step": 120990
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028907121202484686,
+      "loss": 0.4558,
+      "step": 121000
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028902881129555427,
+      "loss": 0.4595,
+      "step": 121010
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002889864105662618,
+      "loss": 0.4181,
+      "step": 121020
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002889440098369692,
+      "loss": 0.5052,
+      "step": 121030
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028890160910767667,
+      "loss": 0.5398,
+      "step": 121040
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028885920837838413,
+      "loss": 0.566,
+      "step": 121050
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002888168076490916,
+      "loss": 0.5004,
+      "step": 121060
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.000288774406919799,
+      "loss": 0.4986,
+      "step": 121070
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028873200619050654,
+      "loss": 0.4359,
+      "step": 121080
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028868960546121395,
+      "loss": 0.5071,
+      "step": 121090
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002886472047319214,
+      "loss": 0.4633,
+      "step": 121100
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028860480400262883,
+      "loss": 0.5563,
+      "step": 121110
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028856240327333635,
+      "loss": 0.4044,
+      "step": 121120
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028852000254404376,
+      "loss": 0.4218,
+      "step": 121130
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028847760181475123,
+      "loss": 0.4054,
+      "step": 121140
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002884352010854587,
+      "loss": 0.3974,
+      "step": 121150
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028839280035616616,
+      "loss": 0.4067,
+      "step": 121160
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002883503996268736,
+      "loss": 0.4957,
+      "step": 121170
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028830799889758104,
+      "loss": 0.4782,
+      "step": 121180
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002882655981682885,
+      "loss": 0.4268,
+      "step": 121190
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.000288223197438996,
+      "loss": 0.5038,
+      "step": 121200
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028818079670970344,
+      "loss": 0.4437,
+      "step": 121210
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028813839598041085,
+      "loss": 0.4872,
+      "step": 121220
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002880959952511183,
+      "loss": 0.4921,
+      "step": 121230
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002880535945218258,
+      "loss": 0.4166,
+      "step": 121240
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028801119379253325,
+      "loss": 0.4543,
+      "step": 121250
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002879687930632407,
+      "loss": 0.4418,
+      "step": 121260
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028792639233394813,
+      "loss": 0.4807,
+      "step": 121270
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002878839916046556,
+      "loss": 0.4386,
+      "step": 121280
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028784159087536307,
+      "loss": 0.4184,
+      "step": 121290
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028779919014607053,
+      "loss": 0.3907,
+      "step": 121300
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.000287756789416778,
+      "loss": 0.4035,
+      "step": 121310
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002877143886874854,
+      "loss": 0.4284,
+      "step": 121320
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002876719879581929,
+      "loss": 0.4882,
+      "step": 121330
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028762958722890035,
+      "loss": 0.4511,
+      "step": 121340
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002875871864996078,
+      "loss": 0.4439,
+      "step": 121350
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002875447857703152,
+      "loss": 0.4574,
+      "step": 121360
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028750238504102275,
+      "loss": 0.4437,
+      "step": 121370
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028745998431173016,
+      "loss": 0.448,
+      "step": 121380
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002874175835824376,
+      "loss": 0.5163,
+      "step": 121390
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002873751828531451,
+      "loss": 0.5089,
+      "step": 121400
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028733278212385256,
+      "loss": 0.4501,
+      "step": 121410
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028729038139455997,
+      "loss": 0.4568,
+      "step": 121420
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028724798066526744,
+      "loss": 0.497,
+      "step": 121430
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002872055799359749,
+      "loss": 0.4212,
+      "step": 121440
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028716317920668237,
+      "loss": 0.4151,
+      "step": 121450
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002871207784773898,
+      "loss": 0.4422,
+      "step": 121460
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002870783777480973,
+      "loss": 0.5071,
+      "step": 121470
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002870359770188047,
+      "loss": 0.4737,
+      "step": 121480
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002869935762895122,
+      "loss": 0.4624,
+      "step": 121490
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028695117556021965,
+      "loss": 0.4551,
+      "step": 121500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002869087748309271,
+      "loss": 0.4175,
+      "step": 121510
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028686637410163453,
+      "loss": 0.5211,
+      "step": 121520
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028682397337234205,
+      "loss": 0.4422,
+      "step": 121530
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028678157264304947,
+      "loss": 0.5069,
+      "step": 121540
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028673917191375693,
+      "loss": 0.421,
+      "step": 121550
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002866967711844644,
+      "loss": 0.4379,
+      "step": 121560
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028665437045517187,
+      "loss": 0.3807,
+      "step": 121570
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002866119697258793,
+      "loss": 0.4492,
+      "step": 121580
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028656956899658675,
+      "loss": 0.4565,
+      "step": 121590
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002865271682672942,
+      "loss": 0.438,
+      "step": 121600
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002864847675380017,
+      "loss": 0.4804,
+      "step": 121610
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002864423668087091,
+      "loss": 0.4862,
+      "step": 121620
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002863999660794166,
+      "loss": 0.392,
+      "step": 121630
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.000286357565350124,
+      "loss": 0.4048,
+      "step": 121640
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002863151646208315,
+      "loss": 0.4648,
+      "step": 121650
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028627276389153896,
+      "loss": 0.4442,
+      "step": 121660
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002862303631622464,
+      "loss": 0.4459,
+      "step": 121670
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028618796243295384,
+      "loss": 0.4546,
+      "step": 121680
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028614556170366136,
+      "loss": 0.4769,
+      "step": 121690
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028610316097436877,
+      "loss": 0.4888,
+      "step": 121700
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028606076024507624,
+      "loss": 0.4308,
+      "step": 121710
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002860183595157837,
+      "loss": 0.4324,
+      "step": 121720
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028597595878649117,
+      "loss": 0.4718,
+      "step": 121730
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002859335580571986,
+      "loss": 0.4581,
+      "step": 121740
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028589115732790605,
+      "loss": 0.4734,
+      "step": 121750
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002858487565986135,
+      "loss": 0.4611,
+      "step": 121760
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.000285806355869321,
+      "loss": 0.4264,
+      "step": 121770
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002857639551400284,
+      "loss": 0.3944,
+      "step": 121780
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002857215544107359,
+      "loss": 0.387,
+      "step": 121790
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028567915368144333,
+      "loss": 0.4801,
+      "step": 121800
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002856367529521508,
+      "loss": 0.4465,
+      "step": 121810
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028559435222285826,
+      "loss": 0.4145,
+      "step": 121820
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028555195149356573,
+      "loss": 0.4269,
+      "step": 121830
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028550955076427314,
+      "loss": 0.4888,
+      "step": 121840
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002854671500349806,
+      "loss": 0.4442,
+      "step": 121850
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002854247493056881,
+      "loss": 0.4209,
+      "step": 121860
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028538234857639554,
+      "loss": 0.3957,
+      "step": 121870
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028533994784710296,
+      "loss": 0.4653,
+      "step": 121880
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002852975471178104,
+      "loss": 0.4413,
+      "step": 121890
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002852551463885179,
+      "loss": 0.4254,
+      "step": 121900
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028521274565922536,
+      "loss": 0.4015,
+      "step": 121910
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002851703449299328,
+      "loss": 0.459,
+      "step": 121920
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028512794420064024,
+      "loss": 0.448,
+      "step": 121930
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002850855434713477,
+      "loss": 0.4354,
+      "step": 121940
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028504314274205517,
+      "loss": 0.4227,
+      "step": 121950
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028500074201276264,
+      "loss": 0.4349,
+      "step": 121960
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028495834128347005,
+      "loss": 0.3619,
+      "step": 121970
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028491594055417757,
+      "loss": 0.3729,
+      "step": 121980
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.000284873539824885,
+      "loss": 0.4761,
+      "step": 121990
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028483113909559245,
+      "loss": 0.3983,
+      "step": 122000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002847887383662999,
+      "loss": 0.4287,
+      "step": 122010
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002847463376370074,
+      "loss": 0.5014,
+      "step": 122020
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002847039369077148,
+      "loss": 0.4474,
+      "step": 122030
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028466153617842226,
+      "loss": 0.4679,
+      "step": 122040
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028461913544912973,
+      "loss": 0.4194,
+      "step": 122050
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002845767347198372,
+      "loss": 0.479,
+      "step": 122060
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002845343339905446,
+      "loss": 0.4271,
+      "step": 122070
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028449193326125213,
+      "loss": 0.4465,
+      "step": 122080
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028444953253195954,
+      "loss": 0.4853,
+      "step": 122090
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.000284407131802667,
+      "loss": 0.4676,
+      "step": 122100
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002843647310733745,
+      "loss": 0.4061,
+      "step": 122110
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028432233034408194,
+      "loss": 0.4212,
+      "step": 122120
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028427992961478936,
+      "loss": 0.4431,
+      "step": 122130
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002842375288854969,
+      "loss": 0.4238,
+      "step": 122140
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002841951281562043,
+      "loss": 0.3818,
+      "step": 122150
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028415272742691176,
+      "loss": 0.4852,
+      "step": 122160
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002841103266976192,
+      "loss": 0.4497,
+      "step": 122170
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002840679259683267,
+      "loss": 0.4292,
+      "step": 122180
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002840255252390341,
+      "loss": 0.4241,
+      "step": 122190
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028398312450974157,
+      "loss": 0.4711,
+      "step": 122200
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028394072378044904,
+      "loss": 0.4403,
+      "step": 122210
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002838983230511565,
+      "loss": 0.5022,
+      "step": 122220
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002838559223218639,
+      "loss": 0.4162,
+      "step": 122230
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028381352159257144,
+      "loss": 0.3993,
+      "step": 122240
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028377112086327885,
+      "loss": 0.4317,
+      "step": 122250
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002837287201339863,
+      "loss": 0.4381,
+      "step": 122260
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002836863194046938,
+      "loss": 0.4401,
+      "step": 122270
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028364391867540125,
+      "loss": 0.4768,
+      "step": 122280
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028360151794610866,
+      "loss": 0.4961,
+      "step": 122290
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002835591172168162,
+      "loss": 0.4615,
+      "step": 122300
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002835167164875236,
+      "loss": 0.4244,
+      "step": 122310
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028347431575823106,
+      "loss": 0.4404,
+      "step": 122320
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028343191502893853,
+      "loss": 0.405,
+      "step": 122330
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.000283389514299646,
+      "loss": 0.4865,
+      "step": 122340
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002833471135703534,
+      "loss": 0.4853,
+      "step": 122350
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002833047128410609,
+      "loss": 0.4307,
+      "step": 122360
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028326231211176834,
+      "loss": 0.4931,
+      "step": 122370
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002832199113824758,
+      "loss": 0.4183,
+      "step": 122380
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002831775106531832,
+      "loss": 0.4616,
+      "step": 122390
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028313510992389074,
+      "loss": 0.4826,
+      "step": 122400
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028309270919459815,
+      "loss": 0.4128,
+      "step": 122410
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002830503084653056,
+      "loss": 0.4757,
+      "step": 122420
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002830079077360131,
+      "loss": 0.4858,
+      "step": 122430
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028296550700672055,
+      "loss": 0.4586,
+      "step": 122440
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028292310627742797,
+      "loss": 0.5077,
+      "step": 122450
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028288070554813543,
+      "loss": 0.3923,
+      "step": 122460
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002828383048188429,
+      "loss": 0.4457,
+      "step": 122470
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028279590408955037,
+      "loss": 0.4135,
+      "step": 122480
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028275350336025783,
+      "loss": 0.4265,
+      "step": 122490
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002827111026309653,
+      "loss": 0.4152,
+      "step": 122500
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002826687019016727,
+      "loss": 0.5188,
+      "step": 122510
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002826263011723802,
+      "loss": 0.4373,
+      "step": 122520
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028258390044308765,
+      "loss": 0.4712,
+      "step": 122530
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002825414997137951,
+      "loss": 0.5437,
+      "step": 122540
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002824990989845025,
+      "loss": 0.4356,
+      "step": 122550
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028245669825521,
+      "loss": 0.357,
+      "step": 122560
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028241429752591746,
+      "loss": 0.3949,
+      "step": 122570
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028237189679662493,
+      "loss": 0.4009,
+      "step": 122580
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002823294960673324,
+      "loss": 0.4419,
+      "step": 122590
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002822870953380398,
+      "loss": 0.4019,
+      "step": 122600
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002822446946087473,
+      "loss": 0.4354,
+      "step": 122610
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028220229387945474,
+      "loss": 0.4039,
+      "step": 122620
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002821598931501622,
+      "loss": 0.3838,
+      "step": 122630
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002821174924208696,
+      "loss": 0.4611,
+      "step": 122640
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002820750916915771,
+      "loss": 0.4749,
+      "step": 122650
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028203269096228455,
+      "loss": 0.4462,
+      "step": 122660
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.000281990290232992,
+      "loss": 0.4242,
+      "step": 122670
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028194788950369943,
+      "loss": 0.3983,
+      "step": 122680
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028190548877440695,
+      "loss": 0.4835,
+      "step": 122690
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028186308804511437,
+      "loss": 0.4844,
+      "step": 122700
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028182068731582183,
+      "loss": 0.4815,
+      "step": 122710
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002817782865865293,
+      "loss": 0.4569,
+      "step": 122720
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028173588585723677,
+      "loss": 0.5103,
+      "step": 122730
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002816934851279442,
+      "loss": 0.3717,
+      "step": 122740
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002816510843986517,
+      "loss": 0.3608,
+      "step": 122750
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002816086836693591,
+      "loss": 0.4852,
+      "step": 122760
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002815662829400666,
+      "loss": 0.4328,
+      "step": 122770
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028152388221077405,
+      "loss": 0.4375,
+      "step": 122780
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002814814814814815,
+      "loss": 0.4582,
+      "step": 122790
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002814390807521889,
+      "loss": 0.4179,
+      "step": 122800
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002813966800228964,
+      "loss": 0.4329,
+      "step": 122810
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028135427929360386,
+      "loss": 0.5125,
+      "step": 122820
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002813118785643113,
+      "loss": 0.528,
+      "step": 122830
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028126947783501874,
+      "loss": 0.4979,
+      "step": 122840
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028122707710572626,
+      "loss": 0.5133,
+      "step": 122850
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028118467637643367,
+      "loss": 0.4794,
+      "step": 122860
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028114227564714114,
+      "loss": 0.5203,
+      "step": 122870
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002810998749178486,
+      "loss": 0.4147,
+      "step": 122880
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028105747418855607,
+      "loss": 0.448,
+      "step": 122890
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002810150734592635,
+      "loss": 0.4728,
+      "step": 122900
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.000280972672729971,
+      "loss": 0.4005,
+      "step": 122910
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002809302720006784,
+      "loss": 0.3939,
+      "step": 122920
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002808878712713859,
+      "loss": 0.4079,
+      "step": 122930
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028084547054209335,
+      "loss": 0.398,
+      "step": 122940
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002808030698128008,
+      "loss": 0.4103,
+      "step": 122950
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028076066908350823,
+      "loss": 0.4133,
+      "step": 122960
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002807182683542157,
+      "loss": 0.4136,
+      "step": 122970
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028067586762492317,
+      "loss": 0.4932,
+      "step": 122980
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028063346689563063,
+      "loss": 0.4305,
+      "step": 122990
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028059106616633804,
+      "loss": 0.4023,
+      "step": 123000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028054866543704557,
+      "loss": 0.4691,
+      "step": 123010
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.000280506264707753,
+      "loss": 0.5049,
+      "step": 123020
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028046386397846044,
+      "loss": 0.3833,
+      "step": 123030
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002804214632491679,
+      "loss": 0.3954,
+      "step": 123040
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002803790625198754,
+      "loss": 0.4383,
+      "step": 123050
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002803366617905828,
+      "loss": 0.4803,
+      "step": 123060
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002802942610612903,
+      "loss": 0.4313,
+      "step": 123070
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002802518603319977,
+      "loss": 0.3932,
+      "step": 123080
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002802094596027052,
+      "loss": 0.4287,
+      "step": 123090
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00028016705887341266,
+      "loss": 0.4515,
+      "step": 123100
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002801246581441201,
+      "loss": 0.4065,
+      "step": 123110
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00028008225741482754,
+      "loss": 0.4627,
+      "step": 123120
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.000280039856685535,
+      "loss": 0.3884,
+      "step": 123130
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00027999745595624247,
+      "loss": 0.4532,
+      "step": 123140
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00027995505522694994,
+      "loss": 0.4971,
+      "step": 123150
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00027991265449765735,
+      "loss": 0.3962,
+      "step": 123160
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002798702537683648,
+      "loss": 0.5079,
+      "step": 123170
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002798278530390723,
+      "loss": 0.4666,
+      "step": 123180
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00027978545230977975,
+      "loss": 0.429,
+      "step": 123190
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002797430515804872,
+      "loss": 0.4091,
+      "step": 123200
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00027970065085119463,
+      "loss": 0.4609,
+      "step": 123210
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002796582501219021,
+      "loss": 0.4191,
+      "step": 123220
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00027961584939260956,
+      "loss": 0.4297,
+      "step": 123230
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00027957344866331703,
+      "loss": 0.4071,
+      "step": 123240
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00027953104793402444,
+      "loss": 0.5149,
+      "step": 123250
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002794886472047319,
+      "loss": 0.4985,
+      "step": 123260
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002794462464754394,
+      "loss": 0.4279,
+      "step": 123270
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00027940384574614684,
+      "loss": 0.4312,
+      "step": 123280
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002793614450168543,
+      "loss": 0.5846,
+      "step": 123290
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002793190442875618,
+      "loss": 0.4303,
+      "step": 123300
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002792766435582692,
+      "loss": 0.3558,
+      "step": 123310
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027923424282897666,
+      "loss": 0.4156,
+      "step": 123320
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002791918420996841,
+      "loss": 0.5243,
+      "step": 123330
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002791494413703916,
+      "loss": 0.3582,
+      "step": 123340
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.000279107040641099,
+      "loss": 0.4567,
+      "step": 123350
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002790646399118065,
+      "loss": 0.5024,
+      "step": 123360
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027902223918251394,
+      "loss": 0.5132,
+      "step": 123370
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002789798384532214,
+      "loss": 0.4317,
+      "step": 123380
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027893743772392887,
+      "loss": 0.4379,
+      "step": 123390
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027889503699463634,
+      "loss": 0.4067,
+      "step": 123400
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027885263626534375,
+      "loss": 0.4338,
+      "step": 123410
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002788102355360512,
+      "loss": 0.3822,
+      "step": 123420
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002787678348067587,
+      "loss": 0.4605,
+      "step": 123430
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027872543407746615,
+      "loss": 0.415,
+      "step": 123440
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027868303334817356,
+      "loss": 0.4131,
+      "step": 123450
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002786406326188811,
+      "loss": 0.4123,
+      "step": 123460
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002785982318895885,
+      "loss": 0.4769,
+      "step": 123470
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027855583116029596,
+      "loss": 0.4737,
+      "step": 123480
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027851343043100343,
+      "loss": 0.3637,
+      "step": 123490
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002784710297017109,
+      "loss": 0.4735,
+      "step": 123500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002784286289724183,
+      "loss": 0.456,
+      "step": 123510
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027838622824312583,
+      "loss": 0.4262,
+      "step": 123520
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027834382751383324,
+      "loss": 0.513,
+      "step": 123530
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002783014267845407,
+      "loss": 0.4886,
+      "step": 123540
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002782590260552482,
+      "loss": 0.5833,
+      "step": 123550
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027821662532595564,
+      "loss": 0.4753,
+      "step": 123560
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027817422459666306,
+      "loss": 0.4114,
+      "step": 123570
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002781318238673705,
+      "loss": 0.4289,
+      "step": 123580
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.000278089423138078,
+      "loss": 0.4628,
+      "step": 123590
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027804702240878546,
+      "loss": 0.4134,
+      "step": 123600
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027800462167949287,
+      "loss": 0.4371,
+      "step": 123610
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002779622209502004,
+      "loss": 0.4074,
+      "step": 123620
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002779198202209078,
+      "loss": 0.44,
+      "step": 123630
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027787741949161527,
+      "loss": 0.4103,
+      "step": 123640
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027783501876232274,
+      "loss": 0.4108,
+      "step": 123650
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002777926180330302,
+      "loss": 0.4713,
+      "step": 123660
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002777502173037376,
+      "loss": 0.3928,
+      "step": 123670
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027770781657444514,
+      "loss": 0.4973,
+      "step": 123680
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027766541584515255,
+      "loss": 0.454,
+      "step": 123690
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027762301511586,
+      "loss": 0.488,
+      "step": 123700
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002775806143865675,
+      "loss": 0.4375,
+      "step": 123710
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027753821365727495,
+      "loss": 0.4224,
+      "step": 123720
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027749581292798236,
+      "loss": 0.3993,
+      "step": 123730
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027745341219868983,
+      "loss": 0.4775,
+      "step": 123740
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002774110114693973,
+      "loss": 0.4208,
+      "step": 123750
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027736861074010476,
+      "loss": 0.4917,
+      "step": 123760
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002773262100108122,
+      "loss": 0.4971,
+      "step": 123770
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002772838092815197,
+      "loss": 0.509,
+      "step": 123780
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002772414085522271,
+      "loss": 0.4591,
+      "step": 123790
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002771990078229346,
+      "loss": 0.449,
+      "step": 123800
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027715660709364204,
+      "loss": 0.4147,
+      "step": 123810
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002771142063643495,
+      "loss": 0.4262,
+      "step": 123820
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002770718056350569,
+      "loss": 0.4361,
+      "step": 123830
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002770294049057644,
+      "loss": 0.4402,
+      "step": 123840
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027698700417647185,
+      "loss": 0.3919,
+      "step": 123850
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002769446034471793,
+      "loss": 0.4073,
+      "step": 123860
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002769022027178868,
+      "loss": 0.4591,
+      "step": 123870
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002768598019885942,
+      "loss": 0.4773,
+      "step": 123880
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027681740125930167,
+      "loss": 0.5069,
+      "step": 123890
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027677500053000913,
+      "loss": 0.4302,
+      "step": 123900
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002767325998007166,
+      "loss": 0.5579,
+      "step": 123910
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.000276690199071424,
+      "loss": 0.4282,
+      "step": 123920
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002766477983421315,
+      "loss": 0.4541,
+      "step": 123930
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027660539761283895,
+      "loss": 0.472,
+      "step": 123940
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002765629968835464,
+      "loss": 0.4392,
+      "step": 123950
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002765205961542538,
+      "loss": 0.426,
+      "step": 123960
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027647819542496135,
+      "loss": 0.4761,
+      "step": 123970
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027643579469566876,
+      "loss": 0.4681,
+      "step": 123980
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002763933939663762,
+      "loss": 0.4696,
+      "step": 123990
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002763509932370837,
+      "loss": 0.4728,
+      "step": 124000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027630859250779116,
+      "loss": 0.4599,
+      "step": 124010
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027626619177849857,
+      "loss": 0.4041,
+      "step": 124020
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027622379104920604,
+      "loss": 0.3679,
+      "step": 124030
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002761813903199135,
+      "loss": 0.4134,
+      "step": 124040
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.000276138989590621,
+      "loss": 0.5854,
+      "step": 124050
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002760965888613284,
+      "loss": 0.5159,
+      "step": 124060
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002760541881320359,
+      "loss": 0.464,
+      "step": 124070
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002760117874027433,
+      "loss": 0.4533,
+      "step": 124080
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002759693866734508,
+      "loss": 0.4036,
+      "step": 124090
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027592698594415825,
+      "loss": 0.4771,
+      "step": 124100
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002758845852148657,
+      "loss": 0.4632,
+      "step": 124110
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027584218448557313,
+      "loss": 0.3907,
+      "step": 124120
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027579978375628065,
+      "loss": 0.326,
+      "step": 124130
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027575738302698807,
+      "loss": 0.4869,
+      "step": 124140
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027571498229769553,
+      "loss": 0.4004,
+      "step": 124150
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.000275672581568403,
+      "loss": 0.4434,
+      "step": 124160
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027563018083911047,
+      "loss": 0.4289,
+      "step": 124170
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002755877801098179,
+      "loss": 0.4901,
+      "step": 124180
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027554537938052535,
+      "loss": 0.4187,
+      "step": 124190
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002755029786512328,
+      "loss": 0.439,
+      "step": 124200
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002754605779219403,
+      "loss": 0.4481,
+      "step": 124210
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002754181771926477,
+      "loss": 0.4766,
+      "step": 124220
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002753757764633552,
+      "loss": 0.4589,
+      "step": 124230
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002753333757340626,
+      "loss": 0.4011,
+      "step": 124240
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002752909750047701,
+      "loss": 0.4234,
+      "step": 124250
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027524857427547756,
+      "loss": 0.473,
+      "step": 124260
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.000275206173546185,
+      "loss": 0.4088,
+      "step": 124270
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027516377281689244,
+      "loss": 0.4403,
+      "step": 124280
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027512137208759996,
+      "loss": 0.4461,
+      "step": 124290
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027507897135830737,
+      "loss": 0.4381,
+      "step": 124300
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027503657062901484,
+      "loss": 0.4959,
+      "step": 124310
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002749941698997223,
+      "loss": 0.4104,
+      "step": 124320
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027495176917042977,
+      "loss": 0.4855,
+      "step": 124330
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002749093684411372,
+      "loss": 0.3838,
+      "step": 124340
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027486696771184465,
+      "loss": 0.4665,
+      "step": 124350
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002748245669825521,
+      "loss": 0.4892,
+      "step": 124360
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002747821662532596,
+      "loss": 0.4536,
+      "step": 124370
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.000274739765523967,
+      "loss": 0.4188,
+      "step": 124380
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002746973647946745,
+      "loss": 0.4857,
+      "step": 124390
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027465496406538193,
+      "loss": 0.4115,
+      "step": 124400
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002746125633360894,
+      "loss": 0.4585,
+      "step": 124410
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027457016260679686,
+      "loss": 0.4539,
+      "step": 124420
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027452776187750433,
+      "loss": 0.4067,
+      "step": 124430
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027448536114821174,
+      "loss": 0.4479,
+      "step": 124440
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002744429604189192,
+      "loss": 0.5221,
+      "step": 124450
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002744005596896267,
+      "loss": 0.4624,
+      "step": 124460
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027435815896033414,
+      "loss": 0.4487,
+      "step": 124470
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002743157582310416,
+      "loss": 0.5287,
+      "step": 124480
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.000274273357501749,
+      "loss": 0.4204,
+      "step": 124490
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002742309567724565,
+      "loss": 0.3844,
+      "step": 124500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027418855604316396,
+      "loss": 0.484,
+      "step": 124510
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002741461553138714,
+      "loss": 0.4057,
+      "step": 124520
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002741037545845789,
+      "loss": 0.5035,
+      "step": 124530
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002740613538552863,
+      "loss": 0.4473,
+      "step": 124540
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027401895312599377,
+      "loss": 0.4374,
+      "step": 124550
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027397655239670124,
+      "loss": 0.4469,
+      "step": 124560
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002739341516674087,
+      "loss": 0.4383,
+      "step": 124570
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027389175093811617,
+      "loss": 0.4482,
+      "step": 124580
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002738493502088236,
+      "loss": 0.4425,
+      "step": 124590
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027380694947953105,
+      "loss": 0.4729,
+      "step": 124600
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002737645487502385,
+      "loss": 0.4852,
+      "step": 124610
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.000273722148020946,
+      "loss": 0.461,
+      "step": 124620
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002736797472916534,
+      "loss": 0.4916,
+      "step": 124630
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002736373465623609,
+      "loss": 0.4519,
+      "step": 124640
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027359494583306833,
+      "loss": 0.5399,
+      "step": 124650
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002735525451037758,
+      "loss": 0.4566,
+      "step": 124660
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002735101443744832,
+      "loss": 0.4663,
+      "step": 124670
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027346774364519073,
+      "loss": 0.513,
+      "step": 124680
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027342534291589814,
+      "loss": 0.4532,
+      "step": 124690
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002733829421866056,
+      "loss": 0.4733,
+      "step": 124700
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002733405414573131,
+      "loss": 0.4797,
+      "step": 124710
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027329814072802054,
+      "loss": 0.4835,
+      "step": 124720
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027325573999872796,
+      "loss": 0.3423,
+      "step": 124730
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002732133392694355,
+      "loss": 0.5036,
+      "step": 124740
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002731709385401429,
+      "loss": 0.5015,
+      "step": 124750
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027312853781085036,
+      "loss": 0.429,
+      "step": 124760
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002730861370815578,
+      "loss": 0.502,
+      "step": 124770
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002730437363522653,
+      "loss": 0.4075,
+      "step": 124780
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002730013356229727,
+      "loss": 0.4783,
+      "step": 124790
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027295893489368017,
+      "loss": 0.478,
+      "step": 124800
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027291653416438764,
+      "loss": 0.4389,
+      "step": 124810
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002728741334350951,
+      "loss": 0.5079,
+      "step": 124820
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002728317327058025,
+      "loss": 0.4433,
+      "step": 124830
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027278933197651004,
+      "loss": 0.4955,
+      "step": 124840
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027274693124721745,
+      "loss": 0.5201,
+      "step": 124850
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002727045305179249,
+      "loss": 0.4307,
+      "step": 124860
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002726621297886324,
+      "loss": 0.4115,
+      "step": 124870
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027261972905933985,
+      "loss": 0.3758,
+      "step": 124880
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027257732833004726,
+      "loss": 0.4072,
+      "step": 124890
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002725349276007548,
+      "loss": 0.4617,
+      "step": 124900
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002724925268714622,
+      "loss": 0.477,
+      "step": 124910
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027245012614216966,
+      "loss": 0.517,
+      "step": 124920
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027240772541287713,
+      "loss": 0.452,
+      "step": 124930
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002723653246835846,
+      "loss": 0.419,
+      "step": 124940
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.000272322923954292,
+      "loss": 0.5059,
+      "step": 124950
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002722805232249995,
+      "loss": 0.4757,
+      "step": 124960
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027223812249570694,
+      "loss": 0.449,
+      "step": 124970
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002721957217664144,
+      "loss": 0.5168,
+      "step": 124980
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002721533210371218,
+      "loss": 0.4956,
+      "step": 124990
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00027211092030782934,
+      "loss": 0.3737,
+      "step": 125000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00027206851957853675,
+      "loss": 0.397,
+      "step": 125010
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002720261188492442,
+      "loss": 0.3932,
+      "step": 125020
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002719837181199517,
+      "loss": 0.4864,
+      "step": 125030
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00027194131739065916,
+      "loss": 0.4049,
+      "step": 125040
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00027189891666136657,
+      "loss": 0.4584,
+      "step": 125050
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002718565159320741,
+      "loss": 0.5397,
+      "step": 125060
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002718141152027815,
+      "loss": 0.4496,
+      "step": 125070
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00027177171447348897,
+      "loss": 0.496,
+      "step": 125080
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00027172931374419644,
+      "loss": 0.3277,
+      "step": 125090
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002716869130149039,
+      "loss": 0.5019,
+      "step": 125100
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002716445122856113,
+      "loss": 0.4245,
+      "step": 125110
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002716021115563188,
+      "loss": 0.4305,
+      "step": 125120
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00027155971082702625,
+      "loss": 0.4861,
+      "step": 125130
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002715173100977337,
+      "loss": 0.501,
+      "step": 125140
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00027147490936844113,
+      "loss": 0.4242,
+      "step": 125150
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002714325086391486,
+      "loss": 0.4801,
+      "step": 125160
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00027139010790985606,
+      "loss": 0.4285,
+      "step": 125170
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00027134770718056353,
+      "loss": 0.4781,
+      "step": 125180
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.000271305306451271,
+      "loss": 0.5648,
+      "step": 125190
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002712629057219784,
+      "loss": 0.5845,
+      "step": 125200
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002712205049926859,
+      "loss": 0.4731,
+      "step": 125210
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027117810426339334,
+      "loss": 0.4691,
+      "step": 125220
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002711357035341008,
+      "loss": 0.4813,
+      "step": 125230
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002710933028048082,
+      "loss": 0.4072,
+      "step": 125240
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027105090207551574,
+      "loss": 0.4433,
+      "step": 125250
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027100850134622315,
+      "loss": 0.4147,
+      "step": 125260
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002709661006169306,
+      "loss": 0.5477,
+      "step": 125270
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002709236998876381,
+      "loss": 0.4633,
+      "step": 125280
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027088129915834555,
+      "loss": 0.5101,
+      "step": 125290
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027083889842905297,
+      "loss": 0.4536,
+      "step": 125300
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027079649769976043,
+      "loss": 0.4244,
+      "step": 125310
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002707540969704679,
+      "loss": 0.4287,
+      "step": 125320
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027071169624117537,
+      "loss": 0.5009,
+      "step": 125330
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002706692955118828,
+      "loss": 0.4533,
+      "step": 125340
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002706268947825903,
+      "loss": 0.4841,
+      "step": 125350
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002705844940532977,
+      "loss": 0.3431,
+      "step": 125360
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002705420933240052,
+      "loss": 0.4898,
+      "step": 125370
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027049969259471265,
+      "loss": 0.4859,
+      "step": 125380
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002704572918654201,
+      "loss": 0.5032,
+      "step": 125390
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002704148911361275,
+      "loss": 0.4704,
+      "step": 125400
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.000270372490406835,
+      "loss": 0.472,
+      "step": 125410
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027033008967754246,
+      "loss": 0.4939,
+      "step": 125420
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002702876889482499,
+      "loss": 0.3689,
+      "step": 125430
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027024528821895734,
+      "loss": 0.4865,
+      "step": 125440
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00027020288748966486,
+      "loss": 0.4662,
+      "step": 125450
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00027016048676037227,
+      "loss": 0.5151,
+      "step": 125460
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00027011808603107974,
+      "loss": 0.475,
+      "step": 125470
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002700756853017872,
+      "loss": 0.4379,
+      "step": 125480
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002700332845724947,
+      "loss": 0.5399,
+      "step": 125490
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002699908838432021,
+      "loss": 0.4878,
+      "step": 125500
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002699484831139096,
+      "loss": 0.4415,
+      "step": 125510
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.000269906082384617,
+      "loss": 0.475,
+      "step": 125520
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002698636816553245,
+      "loss": 0.5434,
+      "step": 125530
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00026982128092603195,
+      "loss": 0.4781,
+      "step": 125540
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002697788801967394,
+      "loss": 0.4766,
+      "step": 125550
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00026973647946744683,
+      "loss": 0.4626,
+      "step": 125560
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002696940787381543,
+      "loss": 0.4806,
+      "step": 125570
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00026965167800886177,
+      "loss": 0.4597,
+      "step": 125580
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00026960927727956923,
+      "loss": 0.4634,
+      "step": 125590
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00026956687655027664,
+      "loss": 0.4088,
+      "step": 125600
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00026952447582098417,
+      "loss": 0.4631,
+      "step": 125610
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002694820750916916,
+      "loss": 0.415,
+      "step": 125620
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00026943967436239905,
+      "loss": 0.4693,
+      "step": 125630
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002693972736331065,
+      "loss": 0.4571,
+      "step": 125640
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.000269354872903814,
+      "loss": 0.4095,
+      "step": 125650
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002693124721745214,
+      "loss": 0.502,
+      "step": 125660
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002692700714452289,
+      "loss": 0.4733,
+      "step": 125670
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002692276707159363,
+      "loss": 0.4514,
+      "step": 125680
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002691852699866438,
+      "loss": 0.4264,
+      "step": 125690
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026914286925735126,
+      "loss": 0.4115,
+      "step": 125700
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002691004685280587,
+      "loss": 0.4922,
+      "step": 125710
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026905806779876614,
+      "loss": 0.5101,
+      "step": 125720
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002690156670694736,
+      "loss": 0.4739,
+      "step": 125730
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026897326634018107,
+      "loss": 0.5404,
+      "step": 125740
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026893086561088854,
+      "loss": 0.4954,
+      "step": 125750
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026888846488159595,
+      "loss": 0.4753,
+      "step": 125760
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026884606415230347,
+      "loss": 0.4005,
+      "step": 125770
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002688036634230109,
+      "loss": 0.3893,
+      "step": 125780
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026876126269371835,
+      "loss": 0.4146,
+      "step": 125790
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002687188619644258,
+      "loss": 0.3445,
+      "step": 125800
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002686764612351333,
+      "loss": 0.4199,
+      "step": 125810
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002686340605058407,
+      "loss": 0.4469,
+      "step": 125820
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026859165977654816,
+      "loss": 0.4694,
+      "step": 125830
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026854925904725563,
+      "loss": 0.4978,
+      "step": 125840
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002685068583179631,
+      "loss": 0.4897,
+      "step": 125850
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026846445758867056,
+      "loss": 0.5421,
+      "step": 125860
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.000268422056859378,
+      "loss": 0.3836,
+      "step": 125870
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026837965613008544,
+      "loss": 0.4193,
+      "step": 125880
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002683372554007929,
+      "loss": 0.4615,
+      "step": 125890
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002682948546715004,
+      "loss": 0.3863,
+      "step": 125900
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002682524539422078,
+      "loss": 0.441,
+      "step": 125910
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026821005321291526,
+      "loss": 0.4514,
+      "step": 125920
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002681676524836227,
+      "loss": 0.5002,
+      "step": 125930
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002681252517543302,
+      "loss": 0.5125,
+      "step": 125940
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002680828510250376,
+      "loss": 0.4177,
+      "step": 125950
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002680404502957451,
+      "loss": 0.5146,
+      "step": 125960
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026799804956645254,
+      "loss": 0.5002,
+      "step": 125970
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026795564883716,
+      "loss": 0.3729,
+      "step": 125980
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026791324810786747,
+      "loss": 0.4739,
+      "step": 125990
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026787084737857494,
+      "loss": 0.4981,
+      "step": 126000
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026782844664928235,
+      "loss": 0.4225,
+      "step": 126010
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026778604591998987,
+      "loss": 0.4554,
+      "step": 126020
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002677436451906973,
+      "loss": 0.4431,
+      "step": 126030
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026770124446140475,
+      "loss": 0.4996,
+      "step": 126040
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026765884373211216,
+      "loss": 0.4845,
+      "step": 126050
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002676164430028197,
+      "loss": 0.4286,
+      "step": 126060
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002675740422735271,
+      "loss": 0.4594,
+      "step": 126070
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026753164154423456,
+      "loss": 0.5286,
+      "step": 126080
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026748924081494203,
+      "loss": 0.4677,
+      "step": 126090
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002674468400856495,
+      "loss": 0.3942,
+      "step": 126100
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002674044393563569,
+      "loss": 0.4611,
+      "step": 126110
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026736203862706443,
+      "loss": 0.4316,
+      "step": 126120
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026731963789777184,
+      "loss": 0.4033,
+      "step": 126130
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002672772371684793,
+      "loss": 0.5023,
+      "step": 126140
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002672348364391868,
+      "loss": 0.4793,
+      "step": 126150
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026719243570989424,
+      "loss": 0.3651,
+      "step": 126160
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026715003498060166,
+      "loss": 0.4234,
+      "step": 126170
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002671076342513091,
+      "loss": 0.4448,
+      "step": 126180
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002670652335220166,
+      "loss": 0.5347,
+      "step": 126190
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026702283279272406,
+      "loss": 0.4153,
+      "step": 126200
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026698043206343147,
+      "loss": 0.536,
+      "step": 126210
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.000266938031334139,
+      "loss": 0.4909,
+      "step": 126220
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002668956306048464,
+      "loss": 0.5215,
+      "step": 126230
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026685322987555387,
+      "loss": 0.3809,
+      "step": 126240
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026681082914626134,
+      "loss": 0.4875,
+      "step": 126250
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002667684284169688,
+      "loss": 0.4381,
+      "step": 126260
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002667260276876762,
+      "loss": 0.4149,
+      "step": 126270
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026668362695838374,
+      "loss": 0.5262,
+      "step": 126280
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026664122622909115,
+      "loss": 0.3535,
+      "step": 126290
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002665988254997986,
+      "loss": 0.4472,
+      "step": 126300
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002665564247705061,
+      "loss": 0.4604,
+      "step": 126310
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026651402404121355,
+      "loss": 0.4296,
+      "step": 126320
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026647162331192096,
+      "loss": 0.5606,
+      "step": 126330
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026642922258262843,
+      "loss": 0.4629,
+      "step": 126340
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002663868218533359,
+      "loss": 0.5076,
+      "step": 126350
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026634442112404336,
+      "loss": 0.4692,
+      "step": 126360
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002663020203947508,
+      "loss": 0.5232,
+      "step": 126370
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002662596196654583,
+      "loss": 0.4123,
+      "step": 126380
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002662172189361657,
+      "loss": 0.3984,
+      "step": 126390
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002661748182068732,
+      "loss": 0.4017,
+      "step": 126400
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026613241747758064,
+      "loss": 0.4516,
+      "step": 126410
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002660900167482881,
+      "loss": 0.4836,
+      "step": 126420
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002660476160189955,
+      "loss": 0.4305,
+      "step": 126430
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.000266005215289703,
+      "loss": 0.4709,
+      "step": 126440
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026596281456041045,
+      "loss": 0.4887,
+      "step": 126450
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002659204138311179,
+      "loss": 0.3958,
+      "step": 126460
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002658780131018254,
+      "loss": 0.3957,
+      "step": 126470
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002658356123725328,
+      "loss": 0.3739,
+      "step": 126480
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026579321164324027,
+      "loss": 0.4739,
+      "step": 126490
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026575081091394773,
+      "loss": 0.4495,
+      "step": 126500
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002657084101846552,
+      "loss": 0.4273,
+      "step": 126510
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026566600945536267,
+      "loss": 0.4736,
+      "step": 126520
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002656236087260701,
+      "loss": 0.4723,
+      "step": 126530
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026558120799677755,
+      "loss": 0.4537,
+      "step": 126540
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.000265538807267485,
+      "loss": 0.4917,
+      "step": 126550
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002654964065381925,
+      "loss": 0.5481,
+      "step": 126560
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026545400580889995,
+      "loss": 0.5047,
+      "step": 126570
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026541160507960736,
+      "loss": 0.419,
+      "step": 126580
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002653692043503148,
+      "loss": 0.4154,
+      "step": 126590
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002653268036210223,
+      "loss": 0.4481,
+      "step": 126600
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026528440289172976,
+      "loss": 0.4738,
+      "step": 126610
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002652420021624372,
+      "loss": 0.378,
+      "step": 126620
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002651996014331447,
+      "loss": 0.3775,
+      "step": 126630
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002651572007038521,
+      "loss": 0.4244,
+      "step": 126640
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002651147999745596,
+      "loss": 0.5271,
+      "step": 126650
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.000265072399245267,
+      "loss": 0.3789,
+      "step": 126660
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002650299985159745,
+      "loss": 0.4433,
+      "step": 126670
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002649875977866819,
+      "loss": 0.4257,
+      "step": 126680
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002649451970573894,
+      "loss": 0.4554,
+      "step": 126690
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026490279632809685,
+      "loss": 0.3907,
+      "step": 126700
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002648603955988043,
+      "loss": 0.4481,
+      "step": 126710
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026481799486951173,
+      "loss": 0.452,
+      "step": 126720
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026477559414021925,
+      "loss": 0.4116,
+      "step": 126730
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026473319341092667,
+      "loss": 0.4295,
+      "step": 126740
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026469079268163413,
+      "loss": 0.4852,
+      "step": 126750
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002646483919523416,
+      "loss": 0.4668,
+      "step": 126760
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026460599122304907,
+      "loss": 0.4951,
+      "step": 126770
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002645635904937565,
+      "loss": 0.4702,
+      "step": 126780
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.000264521189764464,
+      "loss": 0.3579,
+      "step": 126790
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002644787890351714,
+      "loss": 0.3863,
+      "step": 126800
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002644363883058789,
+      "loss": 0.6278,
+      "step": 126810
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002643939875765863,
+      "loss": 0.4689,
+      "step": 126820
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002643515868472938,
+      "loss": 0.4741,
+      "step": 126830
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002643091861180012,
+      "loss": 0.5173,
+      "step": 126840
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002642667853887087,
+      "loss": 0.4582,
+      "step": 126850
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026422438465941616,
+      "loss": 0.4374,
+      "step": 126860
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002641819839301236,
+      "loss": 0.5076,
+      "step": 126870
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026413958320083104,
+      "loss": 0.5059,
+      "step": 126880
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026409718247153856,
+      "loss": 0.4431,
+      "step": 126890
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026405478174224597,
+      "loss": 0.4762,
+      "step": 126900
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026401238101295344,
+      "loss": 0.4177,
+      "step": 126910
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002639699802836609,
+      "loss": 0.4356,
+      "step": 126920
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026392757955436837,
+      "loss": 0.5245,
+      "step": 126930
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002638851788250758,
+      "loss": 0.4502,
+      "step": 126940
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026384277809578325,
+      "loss": 0.464,
+      "step": 126950
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002638003773664907,
+      "loss": 0.4875,
+      "step": 126960
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002637579766371982,
+      "loss": 0.398,
+      "step": 126970
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002637155759079056,
+      "loss": 0.5378,
+      "step": 126980
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002636731751786131,
+      "loss": 0.4744,
+      "step": 126990
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026363077444932053,
+      "loss": 0.4561,
+      "step": 127000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.000263588373720028,
+      "loss": 0.5089,
+      "step": 127010
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026354597299073547,
+      "loss": 0.4472,
+      "step": 127020
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026350357226144293,
+      "loss": 0.394,
+      "step": 127030
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026346117153215034,
+      "loss": 0.4603,
+      "step": 127040
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026341877080285787,
+      "loss": 0.4426,
+      "step": 127050
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002633763700735653,
+      "loss": 0.4221,
+      "step": 127060
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026333396934427275,
+      "loss": 0.4806,
+      "step": 127070
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002632915686149802,
+      "loss": 0.5636,
+      "step": 127080
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002632491678856877,
+      "loss": 0.4188,
+      "step": 127090
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002632067671563951,
+      "loss": 0.5022,
+      "step": 127100
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026316436642710256,
+      "loss": 0.5236,
+      "step": 127110
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026312196569781,
+      "loss": 0.4529,
+      "step": 127120
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002630795649685175,
+      "loss": 0.3406,
+      "step": 127130
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002630371642392249,
+      "loss": 0.417,
+      "step": 127140
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026299476350993237,
+      "loss": 0.3906,
+      "step": 127150
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026295236278063984,
+      "loss": 0.4312,
+      "step": 127160
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002629099620513473,
+      "loss": 0.5238,
+      "step": 127170
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026286756132205477,
+      "loss": 0.4229,
+      "step": 127180
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002628251605927622,
+      "loss": 0.4509,
+      "step": 127190
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026278275986346965,
+      "loss": 0.5057,
+      "step": 127200
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002627403591341771,
+      "loss": 0.4427,
+      "step": 127210
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002626979584048846,
+      "loss": 0.4981,
+      "step": 127220
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.000262655557675592,
+      "loss": 0.3992,
+      "step": 127230
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002626131569462995,
+      "loss": 0.4517,
+      "step": 127240
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026257075621700693,
+      "loss": 0.5138,
+      "step": 127250
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002625283554877144,
+      "loss": 0.4821,
+      "step": 127260
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026248595475842186,
+      "loss": 0.4541,
+      "step": 127270
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026244355402912933,
+      "loss": 0.4519,
+      "step": 127280
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026240115329983674,
+      "loss": 0.3729,
+      "step": 127290
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002623587525705442,
+      "loss": 0.4479,
+      "step": 127300
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002623163518412517,
+      "loss": 0.4601,
+      "step": 127310
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026227395111195914,
+      "loss": 0.529,
+      "step": 127320
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026223155038266656,
+      "loss": 0.5133,
+      "step": 127330
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002621891496533741,
+      "loss": 0.4805,
+      "step": 127340
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002621467489240815,
+      "loss": 0.4923,
+      "step": 127350
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00026210434819478896,
+      "loss": 0.4865,
+      "step": 127360
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002620619474654964,
+      "loss": 0.4604,
+      "step": 127370
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002620195467362039,
+      "loss": 0.3934,
+      "step": 127380
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002619771460069113,
+      "loss": 0.4701,
+      "step": 127390
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002619347452776188,
+      "loss": 0.4038,
+      "step": 127400
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00026189234454832624,
+      "loss": 0.4181,
+      "step": 127410
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002618499438190337,
+      "loss": 0.4998,
+      "step": 127420
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002618075430897411,
+      "loss": 0.5698,
+      "step": 127430
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00026176514236044864,
+      "loss": 0.4473,
+      "step": 127440
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00026172274163115605,
+      "loss": 0.456,
+      "step": 127450
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002616803409018635,
+      "loss": 0.4943,
+      "step": 127460
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.000261637940172571,
+      "loss": 0.4144,
+      "step": 127470
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00026159553944327845,
+      "loss": 0.4639,
+      "step": 127480
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00026155313871398586,
+      "loss": 0.5062,
+      "step": 127490
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002615107379846934,
+      "loss": 0.393,
+      "step": 127500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002614683372554008,
+      "loss": 0.5046,
+      "step": 127510
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00026142593652610826,
+      "loss": 0.4413,
+      "step": 127520
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00026138353579681573,
+      "loss": 0.4411,
+      "step": 127530
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002613411350675232,
+      "loss": 0.3945,
+      "step": 127540
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002612987343382306,
+      "loss": 0.4287,
+      "step": 127550
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002612563336089381,
+      "loss": 0.5042,
+      "step": 127560
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00026121393287964554,
+      "loss": 0.446,
+      "step": 127570
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.000261171532150353,
+      "loss": 0.4095,
+      "step": 127580
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002611291314210604,
+      "loss": 0.3991,
+      "step": 127590
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026108673069176794,
+      "loss": 0.4187,
+      "step": 127600
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026104432996247536,
+      "loss": 0.4352,
+      "step": 127610
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002610019292331828,
+      "loss": 0.4709,
+      "step": 127620
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002609595285038903,
+      "loss": 0.3773,
+      "step": 127630
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026091712777459776,
+      "loss": 0.4728,
+      "step": 127640
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026087472704530517,
+      "loss": 0.3815,
+      "step": 127650
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002608323263160127,
+      "loss": 0.4875,
+      "step": 127660
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002607899255867201,
+      "loss": 0.4882,
+      "step": 127670
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026074752485742757,
+      "loss": 0.494,
+      "step": 127680
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026070512412813504,
+      "loss": 0.4077,
+      "step": 127690
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002606627233988425,
+      "loss": 0.4379,
+      "step": 127700
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002606203226695499,
+      "loss": 0.526,
+      "step": 127710
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002605779219402574,
+      "loss": 0.4622,
+      "step": 127720
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026053552121096485,
+      "loss": 0.4302,
+      "step": 127730
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002604931204816723,
+      "loss": 0.4636,
+      "step": 127740
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026045071975237973,
+      "loss": 0.4224,
+      "step": 127750
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002604083190230872,
+      "loss": 0.499,
+      "step": 127760
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026036591829379466,
+      "loss": 0.5035,
+      "step": 127770
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026032351756450213,
+      "loss": 0.4358,
+      "step": 127780
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002602811168352096,
+      "loss": 0.4894,
+      "step": 127790
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026023871610591706,
+      "loss": 0.589,
+      "step": 127800
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002601963153766245,
+      "loss": 0.478,
+      "step": 127810
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00026015391464733194,
+      "loss": 0.3481,
+      "step": 127820
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002601115139180394,
+      "loss": 0.5171,
+      "step": 127830
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002600691131887469,
+      "loss": 0.4682,
+      "step": 127840
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00026002671245945434,
+      "loss": 0.5093,
+      "step": 127850
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00025998431173016175,
+      "loss": 0.4245,
+      "step": 127860
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002599419110008692,
+      "loss": 0.4671,
+      "step": 127870
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002598995102715767,
+      "loss": 0.4454,
+      "step": 127880
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00025985710954228415,
+      "loss": 0.4112,
+      "step": 127890
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00025981470881299157,
+      "loss": 0.3858,
+      "step": 127900
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00025977230808369903,
+      "loss": 0.5184,
+      "step": 127910
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002597299073544065,
+      "loss": 0.4784,
+      "step": 127920
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00025968750662511397,
+      "loss": 0.3822,
+      "step": 127930
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002596451058958214,
+      "loss": 0.363,
+      "step": 127940
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002596027051665289,
+      "loss": 0.445,
+      "step": 127950
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002595603044372363,
+      "loss": 0.4612,
+      "step": 127960
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002595179037079438,
+      "loss": 0.4759,
+      "step": 127970
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00025947550297865125,
+      "loss": 0.4113,
+      "step": 127980
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002594331022493587,
+      "loss": 0.5499,
+      "step": 127990
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002593907015200661,
+      "loss": 0.4574,
+      "step": 128000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00025934830079077365,
+      "loss": 0.3969,
+      "step": 128010
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00025930590006148106,
+      "loss": 0.4673,
+      "step": 128020
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002592634993321885,
+      "loss": 0.433,
+      "step": 128030
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00025922109860289594,
+      "loss": 0.47,
+      "step": 128040
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025917869787360346,
+      "loss": 0.4405,
+      "step": 128050
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002591362971443109,
+      "loss": 0.4268,
+      "step": 128060
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025909389641501834,
+      "loss": 0.4932,
+      "step": 128070
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002590514956857258,
+      "loss": 0.4787,
+      "step": 128080
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002590090949564333,
+      "loss": 0.4181,
+      "step": 128090
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002589666942271407,
+      "loss": 0.3713,
+      "step": 128100
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002589242934978482,
+      "loss": 0.4096,
+      "step": 128110
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002588818927685556,
+      "loss": 0.4297,
+      "step": 128120
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002588394920392631,
+      "loss": 0.5462,
+      "step": 128130
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025879709130997055,
+      "loss": 0.5192,
+      "step": 128140
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.000258754690580678,
+      "loss": 0.4362,
+      "step": 128150
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025871228985138543,
+      "loss": 0.3699,
+      "step": 128160
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025866988912209295,
+      "loss": 0.43,
+      "step": 128170
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025862748839280037,
+      "loss": 0.393,
+      "step": 128180
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025858508766350783,
+      "loss": 0.4189,
+      "step": 128190
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025854268693421525,
+      "loss": 0.413,
+      "step": 128200
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025850028620492277,
+      "loss": 0.5135,
+      "step": 128210
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002584578854756302,
+      "loss": 0.4858,
+      "step": 128220
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025841548474633765,
+      "loss": 0.4561,
+      "step": 128230
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002583730840170451,
+      "loss": 0.4726,
+      "step": 128240
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002583306832877526,
+      "loss": 0.4825,
+      "step": 128250
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025828828255846,
+      "loss": 0.5216,
+      "step": 128260
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002582458818291675,
+      "loss": 0.5788,
+      "step": 128270
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002582034810998749,
+      "loss": 0.4245,
+      "step": 128280
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002581610803705824,
+      "loss": 0.4402,
+      "step": 128290
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025811867964128986,
+      "loss": 0.4615,
+      "step": 128300
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002580762789119973,
+      "loss": 0.4182,
+      "step": 128310
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025803387818270474,
+      "loss": 0.4437,
+      "step": 128320
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002579914774534122,
+      "loss": 0.4477,
+      "step": 128330
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025794907672411967,
+      "loss": 0.4113,
+      "step": 128340
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025790667599482714,
+      "loss": 0.4197,
+      "step": 128350
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025786427526553455,
+      "loss": 0.3795,
+      "step": 128360
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025782187453624207,
+      "loss": 0.4479,
+      "step": 128370
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002577794738069495,
+      "loss": 0.419,
+      "step": 128380
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025773707307765695,
+      "loss": 0.4112,
+      "step": 128390
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002576946723483644,
+      "loss": 0.4903,
+      "step": 128400
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002576522716190719,
+      "loss": 0.4794,
+      "step": 128410
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002576098708897793,
+      "loss": 0.4308,
+      "step": 128420
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025756747016048676,
+      "loss": 0.3622,
+      "step": 128430
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025752506943119423,
+      "loss": 0.4067,
+      "step": 128440
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002574826687019017,
+      "loss": 0.3668,
+      "step": 128450
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025744026797260917,
+      "loss": 0.4697,
+      "step": 128460
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002573978672433166,
+      "loss": 0.4828,
+      "step": 128470
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025735546651402404,
+      "loss": 0.4945,
+      "step": 128480
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002573130657847315,
+      "loss": 0.421,
+      "step": 128490
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.000257270665055439,
+      "loss": 0.505,
+      "step": 128500
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002572282643261464,
+      "loss": 0.4459,
+      "step": 128510
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025718586359685386,
+      "loss": 0.5361,
+      "step": 128520
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002571434628675613,
+      "loss": 0.4394,
+      "step": 128530
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002571010621382688,
+      "loss": 0.4506,
+      "step": 128540
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025705866140897626,
+      "loss": 0.4121,
+      "step": 128550
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002570162606796837,
+      "loss": 0.4098,
+      "step": 128560
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025697385995039114,
+      "loss": 0.4772,
+      "step": 128570
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002569314592210986,
+      "loss": 0.4739,
+      "step": 128580
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025688905849180607,
+      "loss": 0.4526,
+      "step": 128590
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025684665776251354,
+      "loss": 0.42,
+      "step": 128600
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025680425703322095,
+      "loss": 0.4367,
+      "step": 128610
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025676185630392847,
+      "loss": 0.5034,
+      "step": 128620
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002567194555746359,
+      "loss": 0.4062,
+      "step": 128630
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025667705484534335,
+      "loss": 0.4423,
+      "step": 128640
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025663465411605076,
+      "loss": 0.474,
+      "step": 128650
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002565922533867583,
+      "loss": 0.4532,
+      "step": 128660
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002565498526574657,
+      "loss": 0.4389,
+      "step": 128670
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025650745192817316,
+      "loss": 0.462,
+      "step": 128680
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025646505119888063,
+      "loss": 0.4802,
+      "step": 128690
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002564226504695881,
+      "loss": 0.4346,
+      "step": 128700
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002563802497402955,
+      "loss": 0.4848,
+      "step": 128710
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025633784901100303,
+      "loss": 0.4709,
+      "step": 128720
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025629544828171044,
+      "loss": 0.4429,
+      "step": 128730
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002562530475524179,
+      "loss": 0.4682,
+      "step": 128740
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002562106468231254,
+      "loss": 0.4954,
+      "step": 128750
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025616824609383284,
+      "loss": 0.5419,
+      "step": 128760
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025612584536454026,
+      "loss": 0.4956,
+      "step": 128770
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002560834446352478,
+      "loss": 0.3776,
+      "step": 128780
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002560410439059552,
+      "loss": 0.4733,
+      "step": 128790
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025599864317666266,
+      "loss": 0.5033,
+      "step": 128800
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025595624244737007,
+      "loss": 0.4125,
+      "step": 128810
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002559138417180776,
+      "loss": 0.4626,
+      "step": 128820
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.000255871440988785,
+      "loss": 0.4539,
+      "step": 128830
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025582904025949247,
+      "loss": 0.3871,
+      "step": 128840
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025578663953019994,
+      "loss": 0.4886,
+      "step": 128850
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002557442388009074,
+      "loss": 0.4688,
+      "step": 128860
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002557018380716148,
+      "loss": 0.3917,
+      "step": 128870
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025565943734232234,
+      "loss": 0.4683,
+      "step": 128880
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025561703661302975,
+      "loss": 0.5167,
+      "step": 128890
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002555746358837372,
+      "loss": 0.4346,
+      "step": 128900
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002555322351544447,
+      "loss": 0.4964,
+      "step": 128910
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025548983442515215,
+      "loss": 0.4955,
+      "step": 128920
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025544743369585956,
+      "loss": 0.4628,
+      "step": 128930
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002554050329665671,
+      "loss": 0.5195,
+      "step": 128940
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002553626322372745,
+      "loss": 0.4455,
+      "step": 128950
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025532023150798196,
+      "loss": 0.4941,
+      "step": 128960
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002552778307786894,
+      "loss": 0.3866,
+      "step": 128970
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002552354300493969,
+      "loss": 0.4085,
+      "step": 128980
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002551930293201043,
+      "loss": 0.426,
+      "step": 128990
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002551506285908118,
+      "loss": 0.4419,
+      "step": 129000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025510822786151924,
+      "loss": 0.4158,
+      "step": 129010
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002550658271322267,
+      "loss": 0.5339,
+      "step": 129020
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002550234264029341,
+      "loss": 0.5044,
+      "step": 129030
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025498102567364164,
+      "loss": 0.4911,
+      "step": 129040
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025493862494434906,
+      "loss": 0.417,
+      "step": 129050
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002548962242150565,
+      "loss": 0.4684,
+      "step": 129060
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.000254853823485764,
+      "loss": 0.3659,
+      "step": 129070
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025481142275647146,
+      "loss": 0.4303,
+      "step": 129080
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025476902202717887,
+      "loss": 0.4247,
+      "step": 129090
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025472662129788633,
+      "loss": 0.3772,
+      "step": 129100
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002546842205685938,
+      "loss": 0.4825,
+      "step": 129110
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025464181983930127,
+      "loss": 0.49,
+      "step": 129120
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002545994191100087,
+      "loss": 0.4754,
+      "step": 129130
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025455701838071615,
+      "loss": 0.3475,
+      "step": 129140
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002545146176514236,
+      "loss": 0.4699,
+      "step": 129150
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002544722169221311,
+      "loss": 0.4739,
+      "step": 129160
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025442981619283855,
+      "loss": 0.4621,
+      "step": 129170
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025438741546354596,
+      "loss": 0.5021,
+      "step": 129180
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025434501473425343,
+      "loss": 0.4638,
+      "step": 129190
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002543026140049609,
+      "loss": 0.4506,
+      "step": 129200
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025426021327566836,
+      "loss": 0.4732,
+      "step": 129210
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002542178125463758,
+      "loss": 0.4113,
+      "step": 129220
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002541754118170833,
+      "loss": 0.4781,
+      "step": 129230
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002541330110877907,
+      "loss": 0.4179,
+      "step": 129240
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002540906103584982,
+      "loss": 0.4411,
+      "step": 129250
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002540482096292056,
+      "loss": 0.5754,
+      "step": 129260
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002540058088999131,
+      "loss": 0.3932,
+      "step": 129270
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002539634081706205,
+      "loss": 0.3936,
+      "step": 129280
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.000253921007441328,
+      "loss": 0.429,
+      "step": 129290
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025387860671203545,
+      "loss": 0.4513,
+      "step": 129300
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002538362059827429,
+      "loss": 0.3776,
+      "step": 129310
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025379380525345033,
+      "loss": 0.3979,
+      "step": 129320
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025375140452415785,
+      "loss": 0.4457,
+      "step": 129330
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025370900379486527,
+      "loss": 0.426,
+      "step": 129340
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025366660306557273,
+      "loss": 0.4459,
+      "step": 129350
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002536242023362802,
+      "loss": 0.388,
+      "step": 129360
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025358180160698767,
+      "loss": 0.4172,
+      "step": 129370
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002535394008776951,
+      "loss": 0.487,
+      "step": 129380
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002534970001484026,
+      "loss": 0.4597,
+      "step": 129390
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025345459941911,
+      "loss": 0.3926,
+      "step": 129400
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002534121986898175,
+      "loss": 0.4983,
+      "step": 129410
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002533697979605249,
+      "loss": 0.5037,
+      "step": 129420
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002533273972312324,
+      "loss": 0.448,
+      "step": 129430
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002532849965019398,
+      "loss": 0.4672,
+      "step": 129440
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002532425957726473,
+      "loss": 0.5306,
+      "step": 129450
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025320019504335476,
+      "loss": 0.5274,
+      "step": 129460
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002531577943140622,
+      "loss": 0.4508,
+      "step": 129470
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025311539358476964,
+      "loss": 0.4033,
+      "step": 129480
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025307299285547716,
+      "loss": 0.4455,
+      "step": 129490
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025303059212618457,
+      "loss": 0.4352,
+      "step": 129500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025298819139689204,
+      "loss": 0.4052,
+      "step": 129510
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002529457906675995,
+      "loss": 0.466,
+      "step": 129520
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.000252903389938307,
+      "loss": 0.42,
+      "step": 129530
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002528609892090144,
+      "loss": 0.507,
+      "step": 129540
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002528185884797219,
+      "loss": 0.4606,
+      "step": 129550
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002527761877504293,
+      "loss": 0.4265,
+      "step": 129560
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002527337870211368,
+      "loss": 0.419,
+      "step": 129570
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002526913862918442,
+      "loss": 0.4458,
+      "step": 129580
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002526489855625517,
+      "loss": 0.4814,
+      "step": 129590
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025260658483325913,
+      "loss": 0.453,
+      "step": 129600
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002525641841039666,
+      "loss": 0.3841,
+      "step": 129610
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025252178337467407,
+      "loss": 0.43,
+      "step": 129620
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025247938264538153,
+      "loss": 0.4828,
+      "step": 129630
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025243698191608895,
+      "loss": 0.4429,
+      "step": 129640
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025239458118679647,
+      "loss": 0.4417,
+      "step": 129650
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002523521804575039,
+      "loss": 0.4788,
+      "step": 129660
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025230977972821135,
+      "loss": 0.4976,
+      "step": 129670
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002522673789989188,
+      "loss": 0.4409,
+      "step": 129680
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002522249782696263,
+      "loss": 0.3278,
+      "step": 129690
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002521825775403337,
+      "loss": 0.4725,
+      "step": 129700
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025214017681104116,
+      "loss": 0.5112,
+      "step": 129710
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002520977760817486,
+      "loss": 0.4201,
+      "step": 129720
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002520553753524561,
+      "loss": 0.5525,
+      "step": 129730
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002520129746231635,
+      "loss": 0.4309,
+      "step": 129740
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025197057389387097,
+      "loss": 0.4863,
+      "step": 129750
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025192817316457844,
+      "loss": 0.4536,
+      "step": 129760
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002518857724352859,
+      "loss": 0.4793,
+      "step": 129770
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025184337170599337,
+      "loss": 0.4943,
+      "step": 129780
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025180097097670084,
+      "loss": 0.5061,
+      "step": 129790
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025175857024740825,
+      "loss": 0.4825,
+      "step": 129800
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002517161695181157,
+      "loss": 0.469,
+      "step": 129810
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002516737687888232,
+      "loss": 0.5403,
+      "step": 129820
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025163136805953065,
+      "loss": 0.4106,
+      "step": 129830
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002515889673302381,
+      "loss": 0.4106,
+      "step": 129840
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025154656660094553,
+      "loss": 0.4373,
+      "step": 129850
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.000251504165871653,
+      "loss": 0.3759,
+      "step": 129860
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025146176514236046,
+      "loss": 0.4268,
+      "step": 129870
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025141936441306793,
+      "loss": 0.4225,
+      "step": 129880
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025137696368377534,
+      "loss": 0.4157,
+      "step": 129890
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002513345629544828,
+      "loss": 0.5216,
+      "step": 129900
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002512921622251903,
+      "loss": 0.4665,
+      "step": 129910
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025124976149589774,
+      "loss": 0.5125,
+      "step": 129920
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025120736076660516,
+      "loss": 0.5122,
+      "step": 129930
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002511649600373127,
+      "loss": 0.3717,
+      "step": 129940
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002511225593080201,
+      "loss": 0.4593,
+      "step": 129950
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025108015857872756,
+      "loss": 0.4757,
+      "step": 129960
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.000251037757849435,
+      "loss": 0.4289,
+      "step": 129970
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002509953571201425,
+      "loss": 0.4425,
+      "step": 129980
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002509529563908499,
+      "loss": 0.4299,
+      "step": 129990
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002509105556615574,
+      "loss": 0.4679,
+      "step": 130000
+    },
+    {
+      "epoch": 5.5,
+      "eval_loss": 0.6127136945724487,
+      "eval_runtime": 337.643,
+      "eval_samples_per_second": 15.564,
+      "eval_steps_per_second": 3.892,
+      "step": 130000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025086815493226484,
+      "loss": 0.4309,
+      "step": 130010
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002508257542029723,
+      "loss": 0.4469,
+      "step": 130020
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002507833534736797,
+      "loss": 0.3861,
+      "step": 130030
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025074095274438724,
+      "loss": 0.3999,
+      "step": 130040
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025069855201509465,
+      "loss": 0.4197,
+      "step": 130050
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002506561512858021,
+      "loss": 0.4518,
+      "step": 130060
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002506137505565096,
+      "loss": 0.4336,
+      "step": 130070
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025057134982721705,
+      "loss": 0.4566,
+      "step": 130080
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025052894909792446,
+      "loss": 0.4498,
+      "step": 130090
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.000250486548368632,
+      "loss": 0.4629,
+      "step": 130100
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002504441476393394,
+      "loss": 0.4943,
+      "step": 130110
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025040174691004686,
+      "loss": 0.4331,
+      "step": 130120
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025035934618075433,
+      "loss": 0.4425,
+      "step": 130130
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002503169454514618,
+      "loss": 0.4086,
+      "step": 130140
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002502745447221692,
+      "loss": 0.4575,
+      "step": 130150
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025023214399287673,
+      "loss": 0.5048,
+      "step": 130160
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025018974326358414,
+      "loss": 0.4861,
+      "step": 130170
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002501473425342916,
+      "loss": 0.4411,
+      "step": 130180
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.000250104941804999,
+      "loss": 0.508,
+      "step": 130190
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00025006254107570654,
+      "loss": 0.4431,
+      "step": 130200
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00025002014034641396,
+      "loss": 0.458,
+      "step": 130210
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002499777396171214,
+      "loss": 0.4728,
+      "step": 130220
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002499353388878289,
+      "loss": 0.4582,
+      "step": 130230
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024989293815853636,
+      "loss": 0.4168,
+      "step": 130240
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024985053742924377,
+      "loss": 0.4125,
+      "step": 130250
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002498081366999513,
+      "loss": 0.4813,
+      "step": 130260
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002497657359706587,
+      "loss": 0.5304,
+      "step": 130270
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024972333524136617,
+      "loss": 0.3963,
+      "step": 130280
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024968093451207364,
+      "loss": 0.4248,
+      "step": 130290
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002496385337827811,
+      "loss": 0.409,
+      "step": 130300
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002495961330534885,
+      "loss": 0.4499,
+      "step": 130310
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024955373232419604,
+      "loss": 0.479,
+      "step": 130320
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024951133159490345,
+      "loss": 0.4674,
+      "step": 130330
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002494689308656109,
+      "loss": 0.451,
+      "step": 130340
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024942653013631833,
+      "loss": 0.4361,
+      "step": 130350
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024938412940702585,
+      "loss": 0.4576,
+      "step": 130360
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024934172867773326,
+      "loss": 0.478,
+      "step": 130370
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024929932794844073,
+      "loss": 0.5265,
+      "step": 130380
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002492569272191482,
+      "loss": 0.4826,
+      "step": 130390
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024921452648985566,
+      "loss": 0.4756,
+      "step": 130400
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002491721257605631,
+      "loss": 0.3666,
+      "step": 130410
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024912972503127054,
+      "loss": 0.5519,
+      "step": 130420
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.000249087324301978,
+      "loss": 0.4781,
+      "step": 130430
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002490449235726855,
+      "loss": 0.3722,
+      "step": 130440
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024900252284339294,
+      "loss": 0.4238,
+      "step": 130450
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024896012211410035,
+      "loss": 0.4656,
+      "step": 130460
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002489177213848078,
+      "loss": 0.5112,
+      "step": 130470
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002488753206555153,
+      "loss": 0.439,
+      "step": 130480
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024883291992622275,
+      "loss": 0.4698,
+      "step": 130490
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024879051919693017,
+      "loss": 0.4497,
+      "step": 130500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024874811846763763,
+      "loss": 0.437,
+      "step": 130510
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002487057177383451,
+      "loss": 0.4478,
+      "step": 130520
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024866331700905257,
+      "loss": 0.4288,
+      "step": 130530
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024862091627976003,
+      "loss": 0.3795,
+      "step": 130540
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002485785155504675,
+      "loss": 0.4048,
+      "step": 130550
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002485361148211749,
+      "loss": 0.4414,
+      "step": 130560
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002484937140918824,
+      "loss": 0.5314,
+      "step": 130570
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024845131336258985,
+      "loss": 0.4457,
+      "step": 130580
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002484089126332973,
+      "loss": 0.5407,
+      "step": 130590
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002483665119040047,
+      "loss": 0.5084,
+      "step": 130600
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024832411117471225,
+      "loss": 0.524,
+      "step": 130610
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024828171044541966,
+      "loss": 0.4464,
+      "step": 130620
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024823930971612713,
+      "loss": 0.4625,
+      "step": 130630
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024819690898683454,
+      "loss": 0.4718,
+      "step": 130640
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024815450825754206,
+      "loss": 0.4774,
+      "step": 130650
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002481121075282495,
+      "loss": 0.4865,
+      "step": 130660
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024806970679895694,
+      "loss": 0.3866,
+      "step": 130670
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002480273060696644,
+      "loss": 0.4968,
+      "step": 130680
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002479849053403719,
+      "loss": 0.4642,
+      "step": 130690
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002479425046110793,
+      "loss": 0.5064,
+      "step": 130700
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002479001038817868,
+      "loss": 0.55,
+      "step": 130710
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002478577031524942,
+      "loss": 0.4496,
+      "step": 130720
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002478153024232017,
+      "loss": 0.4034,
+      "step": 130730
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024777290169390915,
+      "loss": 0.6204,
+      "step": 130740
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002477305009646166,
+      "loss": 0.4856,
+      "step": 130750
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024768810023532403,
+      "loss": 0.4269,
+      "step": 130760
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024764569950603155,
+      "loss": 0.4328,
+      "step": 130770
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024760329877673897,
+      "loss": 0.4681,
+      "step": 130780
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024756089804744643,
+      "loss": 0.4792,
+      "step": 130790
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024751849731815385,
+      "loss": 0.3939,
+      "step": 130800
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024747609658886137,
+      "loss": 0.4916,
+      "step": 130810
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002474336958595688,
+      "loss": 0.4871,
+      "step": 130820
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024739129513027625,
+      "loss": 0.3534,
+      "step": 130830
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002473488944009837,
+      "loss": 0.4695,
+      "step": 130840
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002473064936716912,
+      "loss": 0.4645,
+      "step": 130850
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002472640929423986,
+      "loss": 0.4462,
+      "step": 130860
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002472216922131061,
+      "loss": 0.4224,
+      "step": 130870
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002471792914838135,
+      "loss": 0.4561,
+      "step": 130880
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.000247136890754521,
+      "loss": 0.4656,
+      "step": 130890
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024709449002522846,
+      "loss": 0.4898,
+      "step": 130900
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002470520892959359,
+      "loss": 0.4614,
+      "step": 130910
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024700968856664334,
+      "loss": 0.4656,
+      "step": 130920
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024696728783735086,
+      "loss": 0.3796,
+      "step": 130930
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024692488710805827,
+      "loss": 0.4683,
+      "step": 130940
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024688248637876574,
+      "loss": 0.5279,
+      "step": 130950
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024684008564947315,
+      "loss": 0.4602,
+      "step": 130960
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024679768492018067,
+      "loss": 0.4597,
+      "step": 130970
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002467552841908881,
+      "loss": 0.462,
+      "step": 130980
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024671288346159555,
+      "loss": 0.4378,
+      "step": 130990
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.000246670482732303,
+      "loss": 0.4747,
+      "step": 131000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002466280820030105,
+      "loss": 0.4743,
+      "step": 131010
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002465856812737179,
+      "loss": 0.4269,
+      "step": 131020
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002465432805444254,
+      "loss": 0.4573,
+      "step": 131030
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024650087981513283,
+      "loss": 0.4598,
+      "step": 131040
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002464584790858403,
+      "loss": 0.4531,
+      "step": 131050
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024641607835654777,
+      "loss": 0.4261,
+      "step": 131060
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024637367762725523,
+      "loss": 0.4044,
+      "step": 131070
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024633127689796264,
+      "loss": 0.4225,
+      "step": 131080
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002462888761686701,
+      "loss": 0.4355,
+      "step": 131090
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002462464754393776,
+      "loss": 0.5009,
+      "step": 131100
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024620407471008505,
+      "loss": 0.4637,
+      "step": 131110
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024616167398079246,
+      "loss": 0.4586,
+      "step": 131120
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002461192732514999,
+      "loss": 0.5421,
+      "step": 131130
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002460768725222074,
+      "loss": 0.5207,
+      "step": 131140
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024603447179291486,
+      "loss": 0.4894,
+      "step": 131150
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002459920710636223,
+      "loss": 0.4054,
+      "step": 131160
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024594967033432974,
+      "loss": 0.4834,
+      "step": 131170
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002459072696050372,
+      "loss": 0.4408,
+      "step": 131180
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024586486887574467,
+      "loss": 0.4787,
+      "step": 131190
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024582246814645214,
+      "loss": 0.3763,
+      "step": 131200
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024578006741715955,
+      "loss": 0.4691,
+      "step": 131210
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024573766668786707,
+      "loss": 0.4857,
+      "step": 131220
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002456952659585745,
+      "loss": 0.4401,
+      "step": 131230
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024565286522928195,
+      "loss": 0.455,
+      "step": 131240
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024561046449998936,
+      "loss": 0.386,
+      "step": 131250
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002455680637706969,
+      "loss": 0.428,
+      "step": 131260
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002455256630414043,
+      "loss": 0.5376,
+      "step": 131270
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024548326231211176,
+      "loss": 0.47,
+      "step": 131280
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024544086158281923,
+      "loss": 0.4121,
+      "step": 131290
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002453984608535267,
+      "loss": 0.4817,
+      "step": 131300
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002453560601242341,
+      "loss": 0.4511,
+      "step": 131310
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024531365939494163,
+      "loss": 0.5175,
+      "step": 131320
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024527125866564904,
+      "loss": 0.4745,
+      "step": 131330
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002452288579363565,
+      "loss": 0.5107,
+      "step": 131340
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.000245186457207064,
+      "loss": 0.3634,
+      "step": 131350
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024514405647777144,
+      "loss": 0.4309,
+      "step": 131360
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024510165574847886,
+      "loss": 0.4568,
+      "step": 131370
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002450592550191864,
+      "loss": 0.4735,
+      "step": 131380
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002450168542898938,
+      "loss": 0.4354,
+      "step": 131390
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024497445356060126,
+      "loss": 0.5315,
+      "step": 131400
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024493205283130867,
+      "loss": 0.468,
+      "step": 131410
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002448896521020162,
+      "loss": 0.4249,
+      "step": 131420
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002448472513727236,
+      "loss": 0.485,
+      "step": 131430
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024480485064343107,
+      "loss": 0.5003,
+      "step": 131440
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024476244991413854,
+      "loss": 0.4565,
+      "step": 131450
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.000244720049184846,
+      "loss": 0.4301,
+      "step": 131460
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002446776484555534,
+      "loss": 0.4722,
+      "step": 131470
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024463524772626094,
+      "loss": 0.486,
+      "step": 131480
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024459284699696835,
+      "loss": 0.4708,
+      "step": 131490
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002445504462676758,
+      "loss": 0.4265,
+      "step": 131500
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002445080455383833,
+      "loss": 0.4149,
+      "step": 131510
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024446564480909075,
+      "loss": 0.5085,
+      "step": 131520
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024442324407979816,
+      "loss": 0.4489,
+      "step": 131530
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002443808433505057,
+      "loss": 0.3855,
+      "step": 131540
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002443384426212131,
+      "loss": 0.4765,
+      "step": 131550
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024429604189192056,
+      "loss": 0.4888,
+      "step": 131560
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.000244253641162628,
+      "loss": 0.4107,
+      "step": 131570
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002442112404333355,
+      "loss": 0.5054,
+      "step": 131580
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002441688397040429,
+      "loss": 0.4437,
+      "step": 131590
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024412643897475035,
+      "loss": 0.4619,
+      "step": 131600
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024408403824545784,
+      "loss": 0.3901,
+      "step": 131610
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024404163751616528,
+      "loss": 0.4833,
+      "step": 131620
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024399923678687272,
+      "loss": 0.4509,
+      "step": 131630
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024395683605758022,
+      "loss": 0.3997,
+      "step": 131640
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024391443532828766,
+      "loss": 0.4734,
+      "step": 131650
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002438720345989951,
+      "loss": 0.4813,
+      "step": 131660
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002438296338697026,
+      "loss": 0.4079,
+      "step": 131670
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024378723314041003,
+      "loss": 0.443,
+      "step": 131680
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024374483241111747,
+      "loss": 0.4718,
+      "step": 131690
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024370243168182496,
+      "loss": 0.3811,
+      "step": 131700
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002436600309525324,
+      "loss": 0.6331,
+      "step": 131710
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024361763022323984,
+      "loss": 0.4738,
+      "step": 131720
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024357522949394728,
+      "loss": 0.4939,
+      "step": 131730
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024353282876465478,
+      "loss": 0.3902,
+      "step": 131740
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024349042803536221,
+      "loss": 0.4817,
+      "step": 131750
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024344802730606965,
+      "loss": 0.4279,
+      "step": 131760
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024340562657677715,
+      "loss": 0.4698,
+      "step": 131770
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002433632258474846,
+      "loss": 0.4042,
+      "step": 131780
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024332082511819203,
+      "loss": 0.4371,
+      "step": 131790
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024327842438889952,
+      "loss": 0.4104,
+      "step": 131800
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024323602365960696,
+      "loss": 0.495,
+      "step": 131810
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002431936229303144,
+      "loss": 0.4523,
+      "step": 131820
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002431512222010219,
+      "loss": 0.4635,
+      "step": 131830
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024310882147172933,
+      "loss": 0.4205,
+      "step": 131840
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024306642074243677,
+      "loss": 0.4654,
+      "step": 131850
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024302402001314421,
+      "loss": 0.3801,
+      "step": 131860
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002429816192838517,
+      "loss": 0.428,
+      "step": 131870
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024293921855455915,
+      "loss": 0.5411,
+      "step": 131880
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002428968178252666,
+      "loss": 0.4418,
+      "step": 131890
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024285441709597408,
+      "loss": 0.4106,
+      "step": 131900
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024281201636668152,
+      "loss": 0.5252,
+      "step": 131910
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024276961563738896,
+      "loss": 0.3991,
+      "step": 131920
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024272721490809645,
+      "loss": 0.4314,
+      "step": 131930
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002426848141788039,
+      "loss": 0.516,
+      "step": 131940
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024264241344951133,
+      "loss": 0.4007,
+      "step": 131950
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024260001272021883,
+      "loss": 0.4274,
+      "step": 131960
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024255761199092627,
+      "loss": 0.4225,
+      "step": 131970
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002425152112616337,
+      "loss": 0.4819,
+      "step": 131980
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002424728105323412,
+      "loss": 0.5004,
+      "step": 131990
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024243040980304864,
+      "loss": 0.4286,
+      "step": 132000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024238800907375608,
+      "loss": 0.4781,
+      "step": 132010
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024234560834446352,
+      "loss": 0.5058,
+      "step": 132020
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024230320761517101,
+      "loss": 0.4995,
+      "step": 132030
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024226080688587845,
+      "loss": 0.4377,
+      "step": 132040
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002422184061565859,
+      "loss": 0.4288,
+      "step": 132050
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024217600542729336,
+      "loss": 0.4574,
+      "step": 132060
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024213360469800083,
+      "loss": 0.4114,
+      "step": 132070
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024209120396870827,
+      "loss": 0.4623,
+      "step": 132080
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024204880323941573,
+      "loss": 0.4163,
+      "step": 132090
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0002420064025101232,
+      "loss": 0.3597,
+      "step": 132100
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024196400178083064,
+      "loss": 0.5151,
+      "step": 132110
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0002419216010515381,
+      "loss": 0.4841,
+      "step": 132120
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024187920032224555,
+      "loss": 0.4358,
+      "step": 132130
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.000241836799592953,
+      "loss": 0.4854,
+      "step": 132140
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024179439886366048,
+      "loss": 0.503,
+      "step": 132150
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024175199813436792,
+      "loss": 0.4871,
+      "step": 132160
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024170959740507536,
+      "loss": 0.5192,
+      "step": 132170
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024166719667578283,
+      "loss": 0.5609,
+      "step": 132180
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0002416247959464903,
+      "loss": 0.4858,
+      "step": 132190
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024158239521719773,
+      "loss": 0.4977,
+      "step": 132200
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024153999448790517,
+      "loss": 0.4674,
+      "step": 132210
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024149759375861267,
+      "loss": 0.4462,
+      "step": 132220
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0002414551930293201,
+      "loss": 0.4921,
+      "step": 132230
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024141279230002755,
+      "loss": 0.4351,
+      "step": 132240
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024137039157073504,
+      "loss": 0.4831,
+      "step": 132250
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024132799084144248,
+      "loss": 0.4457,
+      "step": 132260
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024128559011214992,
+      "loss": 0.4741,
+      "step": 132270
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0002412431893828574,
+      "loss": 0.4757,
+      "step": 132280
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024120078865356485,
+      "loss": 0.4567,
+      "step": 132290
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0002411583879242723,
+      "loss": 0.4034,
+      "step": 132300
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024111598719497979,
+      "loss": 0.4515,
+      "step": 132310
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024107358646568723,
+      "loss": 0.4608,
+      "step": 132320
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024103118573639467,
+      "loss": 0.4141,
+      "step": 132330
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002409887850071021,
+      "loss": 0.5332,
+      "step": 132340
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002409463842778096,
+      "loss": 0.4427,
+      "step": 132350
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024090398354851704,
+      "loss": 0.4188,
+      "step": 132360
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024086158281922448,
+      "loss": 0.4277,
+      "step": 132370
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024081918208993197,
+      "loss": 0.4432,
+      "step": 132380
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002407767813606394,
+      "loss": 0.443,
+      "step": 132390
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024073438063134685,
+      "loss": 0.453,
+      "step": 132400
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024069197990205435,
+      "loss": 0.4597,
+      "step": 132410
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024064957917276179,
+      "loss": 0.4516,
+      "step": 132420
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024060717844346922,
+      "loss": 0.4831,
+      "step": 132430
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024056477771417672,
+      "loss": 0.4021,
+      "step": 132440
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024052237698488416,
+      "loss": 0.4796,
+      "step": 132450
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002404799762555916,
+      "loss": 0.4955,
+      "step": 132460
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002404375755262991,
+      "loss": 0.4238,
+      "step": 132470
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024039517479700653,
+      "loss": 0.5683,
+      "step": 132480
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024035277406771397,
+      "loss": 0.377,
+      "step": 132490
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002403103733384214,
+      "loss": 0.4218,
+      "step": 132500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002402679726091289,
+      "loss": 0.4739,
+      "step": 132510
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024022557187983634,
+      "loss": 0.4906,
+      "step": 132520
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024018317115054378,
+      "loss": 0.4706,
+      "step": 132530
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024014077042125128,
+      "loss": 0.4664,
+      "step": 132540
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024009836969195872,
+      "loss": 0.4498,
+      "step": 132550
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024005596896266616,
+      "loss": 0.426,
+      "step": 132560
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024001356823337365,
+      "loss": 0.5092,
+      "step": 132570
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0002399711675040811,
+      "loss": 0.4337,
+      "step": 132580
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023992876677478853,
+      "loss": 0.4767,
+      "step": 132590
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023988636604549602,
+      "loss": 0.4226,
+      "step": 132600
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023984396531620346,
+      "loss": 0.4798,
+      "step": 132610
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0002398015645869109,
+      "loss": 0.4762,
+      "step": 132620
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023975916385761834,
+      "loss": 0.5711,
+      "step": 132630
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023971676312832584,
+      "loss": 0.4756,
+      "step": 132640
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023967436239903328,
+      "loss": 0.4509,
+      "step": 132650
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023963196166974072,
+      "loss": 0.4357,
+      "step": 132660
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0002395895609404482,
+      "loss": 0.3958,
+      "step": 132670
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023954716021115565,
+      "loss": 0.4342,
+      "step": 132680
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0002395047594818631,
+      "loss": 0.4805,
+      "step": 132690
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023946235875257056,
+      "loss": 0.4059,
+      "step": 132700
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023941995802327802,
+      "loss": 0.4255,
+      "step": 132710
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023937755729398546,
+      "loss": 0.4999,
+      "step": 132720
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023933515656469293,
+      "loss": 0.4911,
+      "step": 132730
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0002392927558354004,
+      "loss": 0.4289,
+      "step": 132740
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023925035510610784,
+      "loss": 0.4462,
+      "step": 132750
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0002392079543768153,
+      "loss": 0.5012,
+      "step": 132760
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023916555364752274,
+      "loss": 0.424,
+      "step": 132770
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0002391231529182302,
+      "loss": 0.3998,
+      "step": 132780
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023908075218893765,
+      "loss": 0.4083,
+      "step": 132790
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023903835145964512,
+      "loss": 0.4099,
+      "step": 132800
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023899595073035256,
+      "loss": 0.4527,
+      "step": 132810
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023895355000106002,
+      "loss": 0.4954,
+      "step": 132820
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0002389111492717675,
+      "loss": 0.461,
+      "step": 132830
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023886874854247493,
+      "loss": 0.4651,
+      "step": 132840
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0002388263478131824,
+      "loss": 0.4762,
+      "step": 132850
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023878394708388986,
+      "loss": 0.457,
+      "step": 132860
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0002387415463545973,
+      "loss": 0.4222,
+      "step": 132870
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023869914562530474,
+      "loss": 0.4352,
+      "step": 132880
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023865674489601224,
+      "loss": 0.4315,
+      "step": 132890
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023861434416671968,
+      "loss": 0.4231,
+      "step": 132900
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023857194343742712,
+      "loss": 0.4835,
+      "step": 132910
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0002385295427081346,
+      "loss": 0.4479,
+      "step": 132920
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023848714197884205,
+      "loss": 0.3953,
+      "step": 132930
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0002384447412495495,
+      "loss": 0.4449,
+      "step": 132940
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023840234052025693,
+      "loss": 0.4107,
+      "step": 132950
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023835993979096442,
+      "loss": 0.4342,
+      "step": 132960
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023831753906167186,
+      "loss": 0.4559,
+      "step": 132970
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0002382751383323793,
+      "loss": 0.4886,
+      "step": 132980
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0002382327376030868,
+      "loss": 0.4393,
+      "step": 132990
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023819033687379424,
+      "loss": 0.4781,
+      "step": 133000
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023814793614450168,
+      "loss": 0.4899,
+      "step": 133010
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023810553541520917,
+      "loss": 0.432,
+      "step": 133020
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002380631346859166,
+      "loss": 0.4553,
+      "step": 133030
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023802073395662405,
+      "loss": 0.4564,
+      "step": 133040
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023797833322733154,
+      "loss": 0.5364,
+      "step": 133050
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023793593249803898,
+      "loss": 0.4775,
+      "step": 133060
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023789353176874642,
+      "loss": 0.5024,
+      "step": 133070
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023785113103945392,
+      "loss": 0.4937,
+      "step": 133080
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023780873031016136,
+      "loss": 0.4167,
+      "step": 133090
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002377663295808688,
+      "loss": 0.4475,
+      "step": 133100
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023772392885157623,
+      "loss": 0.4744,
+      "step": 133110
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023768152812228373,
+      "loss": 0.4594,
+      "step": 133120
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023763912739299117,
+      "loss": 0.5758,
+      "step": 133130
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002375967266636986,
+      "loss": 0.3897,
+      "step": 133140
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002375543259344061,
+      "loss": 0.491,
+      "step": 133150
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023751192520511354,
+      "loss": 0.4022,
+      "step": 133160
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023746952447582098,
+      "loss": 0.4621,
+      "step": 133170
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023742712374652848,
+      "loss": 0.3787,
+      "step": 133180
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023738472301723591,
+      "loss": 0.4708,
+      "step": 133190
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023734232228794335,
+      "loss": 0.4214,
+      "step": 133200
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023729992155865085,
+      "loss": 0.3633,
+      "step": 133210
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002372575208293583,
+      "loss": 0.4476,
+      "step": 133220
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023721512010006573,
+      "loss": 0.4499,
+      "step": 133230
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023717271937077317,
+      "loss": 0.4802,
+      "step": 133240
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023713031864148066,
+      "loss": 0.5186,
+      "step": 133250
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002370879179121881,
+      "loss": 0.4282,
+      "step": 133260
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023704551718289554,
+      "loss": 0.4147,
+      "step": 133270
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023700311645360303,
+      "loss": 0.4768,
+      "step": 133280
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023696071572431047,
+      "loss": 0.4444,
+      "step": 133290
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023691831499501791,
+      "loss": 0.4792,
+      "step": 133300
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002368759142657254,
+      "loss": 0.4684,
+      "step": 133310
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023683351353643285,
+      "loss": 0.5071,
+      "step": 133320
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002367911128071403,
+      "loss": 0.4814,
+      "step": 133330
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023674871207784775,
+      "loss": 0.4781,
+      "step": 133340
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023670631134855522,
+      "loss": 0.452,
+      "step": 133350
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023666391061926266,
+      "loss": 0.4028,
+      "step": 133360
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023662150988997013,
+      "loss": 0.4842,
+      "step": 133370
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002365791091606776,
+      "loss": 0.5428,
+      "step": 133380
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023653670843138503,
+      "loss": 0.5021,
+      "step": 133390
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023649430770209247,
+      "loss": 0.4718,
+      "step": 133400
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023645190697279994,
+      "loss": 0.4061,
+      "step": 133410
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002364095062435074,
+      "loss": 0.4146,
+      "step": 133420
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023636710551421485,
+      "loss": 0.5243,
+      "step": 133430
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002363247047849223,
+      "loss": 0.4355,
+      "step": 133440
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023628230405562975,
+      "loss": 0.3948,
+      "step": 133450
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023623990332633722,
+      "loss": 0.4378,
+      "step": 133460
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002361975025970447,
+      "loss": 0.4562,
+      "step": 133470
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023615510186775213,
+      "loss": 0.4271,
+      "step": 133480
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002361127011384596,
+      "loss": 0.4621,
+      "step": 133490
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023607030040916706,
+      "loss": 0.4219,
+      "step": 133500
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002360278996798745,
+      "loss": 0.4507,
+      "step": 133510
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023598549895058194,
+      "loss": 0.4482,
+      "step": 133520
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023594309822128943,
+      "loss": 0.4889,
+      "step": 133530
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023590069749199687,
+      "loss": 0.4564,
+      "step": 133540
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002358582967627043,
+      "loss": 0.4001,
+      "step": 133550
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023581589603341175,
+      "loss": 0.4209,
+      "step": 133560
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023577349530411925,
+      "loss": 0.4345,
+      "step": 133570
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023573109457482669,
+      "loss": 0.506,
+      "step": 133580
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023568869384553413,
+      "loss": 0.4244,
+      "step": 133590
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023564629311624162,
+      "loss": 0.3714,
+      "step": 133600
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023560389238694906,
+      "loss": 0.4901,
+      "step": 133610
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002355614916576565,
+      "loss": 0.3869,
+      "step": 133620
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.000235519090928364,
+      "loss": 0.4765,
+      "step": 133630
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023547669019907143,
+      "loss": 0.4975,
+      "step": 133640
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023543428946977887,
+      "loss": 0.4557,
+      "step": 133650
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023539188874048637,
+      "loss": 0.5623,
+      "step": 133660
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002353494880111938,
+      "loss": 0.5288,
+      "step": 133670
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023530708728190125,
+      "loss": 0.4574,
+      "step": 133680
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023526468655260874,
+      "loss": 0.389,
+      "step": 133690
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023522228582331618,
+      "loss": 0.4098,
+      "step": 133700
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023517988509402362,
+      "loss": 0.4338,
+      "step": 133710
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023513748436473106,
+      "loss": 0.4924,
+      "step": 133720
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023509508363543855,
+      "loss": 0.3825,
+      "step": 133730
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.000235052682906146,
+      "loss": 0.5269,
+      "step": 133740
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023501028217685343,
+      "loss": 0.5303,
+      "step": 133750
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023496788144756093,
+      "loss": 0.4723,
+      "step": 133760
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023492548071826837,
+      "loss": 0.4592,
+      "step": 133770
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0002348830799889758,
+      "loss": 0.5037,
+      "step": 133780
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0002348406792596833,
+      "loss": 0.4459,
+      "step": 133790
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023479827853039074,
+      "loss": 0.5153,
+      "step": 133800
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023475587780109818,
+      "loss": 0.4893,
+      "step": 133810
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023471347707180567,
+      "loss": 0.407,
+      "step": 133820
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0002346710763425131,
+      "loss": 0.4,
+      "step": 133830
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023462867561322055,
+      "loss": 0.47,
+      "step": 133840
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023458627488392805,
+      "loss": 0.42,
+      "step": 133850
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023454387415463548,
+      "loss": 0.4781,
+      "step": 133860
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023450147342534292,
+      "loss": 0.5447,
+      "step": 133870
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023445907269605036,
+      "loss": 0.4471,
+      "step": 133880
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023441667196675786,
+      "loss": 0.4492,
+      "step": 133890
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0002343742712374653,
+      "loss": 0.4352,
+      "step": 133900
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023433187050817274,
+      "loss": 0.3919,
+      "step": 133910
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023428946977888023,
+      "loss": 0.5184,
+      "step": 133920
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023424706904958767,
+      "loss": 0.479,
+      "step": 133930
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0002342046683202951,
+      "loss": 0.5686,
+      "step": 133940
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0002341622675910026,
+      "loss": 0.6078,
+      "step": 133950
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023411986686171004,
+      "loss": 0.4787,
+      "step": 133960
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023407746613241748,
+      "loss": 0.5169,
+      "step": 133970
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023403506540312498,
+      "loss": 0.4539,
+      "step": 133980
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023399266467383242,
+      "loss": 0.406,
+      "step": 133990
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023395026394453986,
+      "loss": 0.5042,
+      "step": 134000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002339078632152473,
+      "loss": 0.4391,
+      "step": 134010
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002338654624859548,
+      "loss": 0.4467,
+      "step": 134020
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023382306175666223,
+      "loss": 0.3878,
+      "step": 134030
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023378066102736967,
+      "loss": 0.4474,
+      "step": 134040
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023373826029807714,
+      "loss": 0.4788,
+      "step": 134050
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002336958595687846,
+      "loss": 0.4587,
+      "step": 134060
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023365345883949204,
+      "loss": 0.5032,
+      "step": 134070
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002336110581101995,
+      "loss": 0.5117,
+      "step": 134080
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023356865738090695,
+      "loss": 0.3946,
+      "step": 134090
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023352625665161442,
+      "loss": 0.4741,
+      "step": 134100
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023348385592232188,
+      "loss": 0.5057,
+      "step": 134110
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023344145519302932,
+      "loss": 0.5208,
+      "step": 134120
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002333990544637368,
+      "loss": 0.4992,
+      "step": 134130
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023335665373444426,
+      "loss": 0.4902,
+      "step": 134140
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002333142530051517,
+      "loss": 0.4812,
+      "step": 134150
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023327185227585914,
+      "loss": 0.43,
+      "step": 134160
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002332294515465666,
+      "loss": 0.3769,
+      "step": 134170
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023318705081727407,
+      "loss": 0.4265,
+      "step": 134180
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002331446500879815,
+      "loss": 0.4796,
+      "step": 134190
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023310224935868895,
+      "loss": 0.4562,
+      "step": 134200
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023305984862939644,
+      "loss": 0.4559,
+      "step": 134210
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023301744790010388,
+      "loss": 0.4339,
+      "step": 134220
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023297504717081132,
+      "loss": 0.4498,
+      "step": 134230
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023293264644151882,
+      "loss": 0.3999,
+      "step": 134240
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023289024571222626,
+      "loss": 0.4686,
+      "step": 134250
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002328478449829337,
+      "loss": 0.4313,
+      "step": 134260
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002328054442536412,
+      "loss": 0.4808,
+      "step": 134270
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023276304352434863,
+      "loss": 0.4263,
+      "step": 134280
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023272064279505607,
+      "loss": 0.4085,
+      "step": 134290
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023267824206576356,
+      "loss": 0.4119,
+      "step": 134300
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.000232635841336471,
+      "loss": 0.4545,
+      "step": 134310
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023259344060717844,
+      "loss": 0.3983,
+      "step": 134320
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023255103987788588,
+      "loss": 0.5322,
+      "step": 134330
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023250863914859338,
+      "loss": 0.4843,
+      "step": 134340
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023246623841930082,
+      "loss": 0.3851,
+      "step": 134350
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023242383769000826,
+      "loss": 0.4214,
+      "step": 134360
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023238143696071575,
+      "loss": 0.4989,
+      "step": 134370
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002323390362314232,
+      "loss": 0.4967,
+      "step": 134380
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023229663550213063,
+      "loss": 0.5017,
+      "step": 134390
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023225423477283812,
+      "loss": 0.5039,
+      "step": 134400
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023221183404354556,
+      "loss": 0.4413,
+      "step": 134410
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.000232169433314253,
+      "loss": 0.4313,
+      "step": 134420
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002321270325849605,
+      "loss": 0.4582,
+      "step": 134430
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023208463185566794,
+      "loss": 0.4248,
+      "step": 134440
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023204223112637537,
+      "loss": 0.5257,
+      "step": 134450
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023199983039708287,
+      "loss": 0.4554,
+      "step": 134460
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002319574296677903,
+      "loss": 0.4613,
+      "step": 134470
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023191502893849775,
+      "loss": 0.3785,
+      "step": 134480
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002318726282092052,
+      "loss": 0.3934,
+      "step": 134490
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023183022747991268,
+      "loss": 0.513,
+      "step": 134500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023178782675062012,
+      "loss": 0.5099,
+      "step": 134510
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023174542602132756,
+      "loss": 0.4316,
+      "step": 134520
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023170302529203505,
+      "loss": 0.437,
+      "step": 134530
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002316606245627425,
+      "loss": 0.4267,
+      "step": 134540
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023161822383344993,
+      "loss": 0.4991,
+      "step": 134550
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023157582310415743,
+      "loss": 0.4881,
+      "step": 134560
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023153342237486487,
+      "loss": 0.4727,
+      "step": 134570
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002314910216455723,
+      "loss": 0.5469,
+      "step": 134580
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002314486209162798,
+      "loss": 0.5538,
+      "step": 134590
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023140622018698724,
+      "loss": 0.4631,
+      "step": 134600
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023136381945769468,
+      "loss": 0.4615,
+      "step": 134610
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023132141872840217,
+      "loss": 0.3812,
+      "step": 134620
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023127901799910961,
+      "loss": 0.4773,
+      "step": 134630
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023123661726981705,
+      "loss": 0.4497,
+      "step": 134640
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002311942165405245,
+      "loss": 0.4854,
+      "step": 134650
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.000231151815811232,
+      "loss": 0.4168,
+      "step": 134660
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023110941508193943,
+      "loss": 0.4889,
+      "step": 134670
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023106701435264687,
+      "loss": 0.4084,
+      "step": 134680
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023102461362335433,
+      "loss": 0.4693,
+      "step": 134690
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002309822128940618,
+      "loss": 0.5205,
+      "step": 134700
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023093981216476924,
+      "loss": 0.4976,
+      "step": 134710
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002308974114354767,
+      "loss": 0.4541,
+      "step": 134720
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023085501070618417,
+      "loss": 0.4239,
+      "step": 134730
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023081260997689161,
+      "loss": 0.4331,
+      "step": 134740
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023077020924759908,
+      "loss": 0.4411,
+      "step": 134750
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023072780851830652,
+      "loss": 0.4922,
+      "step": 134760
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.000230685407789014,
+      "loss": 0.5046,
+      "step": 134770
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023064300705972143,
+      "loss": 0.4949,
+      "step": 134780
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002306006063304289,
+      "loss": 0.5441,
+      "step": 134790
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023055820560113633,
+      "loss": 0.4763,
+      "step": 134800
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002305158048718438,
+      "loss": 0.4624,
+      "step": 134810
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023047340414255127,
+      "loss": 0.5307,
+      "step": 134820
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002304310034132587,
+      "loss": 0.4778,
+      "step": 134830
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023038860268396615,
+      "loss": 0.6543,
+      "step": 134840
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023034620195467364,
+      "loss": 0.4539,
+      "step": 134850
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023030380122538108,
+      "loss": 0.4644,
+      "step": 134860
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023026140049608852,
+      "loss": 0.4047,
+      "step": 134870
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.000230218999766796,
+      "loss": 0.4068,
+      "step": 134880
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023017659903750345,
+      "loss": 0.4481,
+      "step": 134890
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002301341983082109,
+      "loss": 0.4719,
+      "step": 134900
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00023009179757891839,
+      "loss": 0.458,
+      "step": 134910
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00023004939684962583,
+      "loss": 0.4449,
+      "step": 134920
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00023000699612033327,
+      "loss": 0.4538,
+      "step": 134930
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0002299645953910407,
+      "loss": 0.4706,
+      "step": 134940
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0002299221946617482,
+      "loss": 0.4486,
+      "step": 134950
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022987979393245564,
+      "loss": 0.4703,
+      "step": 134960
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022983739320316308,
+      "loss": 0.5689,
+      "step": 134970
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022979499247387057,
+      "loss": 0.4425,
+      "step": 134980
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.000229752591744578,
+      "loss": 0.439,
+      "step": 134990
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022971019101528545,
+      "loss": 0.4384,
+      "step": 135000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022966779028599295,
+      "loss": 0.4576,
+      "step": 135010
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022962538955670039,
+      "loss": 0.5138,
+      "step": 135020
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022958298882740783,
+      "loss": 0.4155,
+      "step": 135030
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022954058809811532,
+      "loss": 0.5247,
+      "step": 135040
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022949818736882276,
+      "loss": 0.4185,
+      "step": 135050
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0002294557866395302,
+      "loss": 0.444,
+      "step": 135060
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0002294133859102377,
+      "loss": 0.435,
+      "step": 135070
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022937098518094513,
+      "loss": 0.436,
+      "step": 135080
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022932858445165257,
+      "loss": 0.4329,
+      "step": 135090
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022928618372236,
+      "loss": 0.474,
+      "step": 135100
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0002292437829930675,
+      "loss": 0.475,
+      "step": 135110
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022920138226377494,
+      "loss": 0.4254,
+      "step": 135120
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022915898153448238,
+      "loss": 0.4233,
+      "step": 135130
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022911658080518988,
+      "loss": 0.4445,
+      "step": 135140
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022907418007589732,
+      "loss": 0.4277,
+      "step": 135150
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022903177934660476,
+      "loss": 0.4113,
+      "step": 135160
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022898937861731225,
+      "loss": 0.4712,
+      "step": 135170
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002289469778880197,
+      "loss": 0.4392,
+      "step": 135180
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022890457715872713,
+      "loss": 0.589,
+      "step": 135190
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022886217642943463,
+      "loss": 0.4625,
+      "step": 135200
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022881977570014206,
+      "loss": 0.4077,
+      "step": 135210
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002287773749708495,
+      "loss": 0.4711,
+      "step": 135220
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.000228734974241557,
+      "loss": 0.5163,
+      "step": 135230
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022869257351226444,
+      "loss": 0.4653,
+      "step": 135240
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022865017278297188,
+      "loss": 0.4647,
+      "step": 135250
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022860777205367932,
+      "loss": 0.4066,
+      "step": 135260
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002285653713243868,
+      "loss": 0.4173,
+      "step": 135270
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022852297059509425,
+      "loss": 0.403,
+      "step": 135280
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002284805698658017,
+      "loss": 0.4775,
+      "step": 135290
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022843816913650918,
+      "loss": 0.4165,
+      "step": 135300
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022839576840721662,
+      "loss": 0.509,
+      "step": 135310
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022835336767792406,
+      "loss": 0.4771,
+      "step": 135320
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022831096694863153,
+      "loss": 0.5289,
+      "step": 135330
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.000228268566219339,
+      "loss": 0.4683,
+      "step": 135340
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022822616549004644,
+      "loss": 0.4737,
+      "step": 135350
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002281837647607539,
+      "loss": 0.4956,
+      "step": 135360
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022814136403146137,
+      "loss": 0.4924,
+      "step": 135370
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002280989633021688,
+      "loss": 0.4808,
+      "step": 135380
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022805656257287625,
+      "loss": 0.5183,
+      "step": 135390
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022801416184358372,
+      "loss": 0.5302,
+      "step": 135400
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022797176111429118,
+      "loss": 0.4821,
+      "step": 135410
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022792936038499862,
+      "loss": 0.4643,
+      "step": 135420
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002278869596557061,
+      "loss": 0.4666,
+      "step": 135430
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022784455892641353,
+      "loss": 0.4447,
+      "step": 135440
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.000227802158197121,
+      "loss": 0.4139,
+      "step": 135450
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022775975746782846,
+      "loss": 0.4212,
+      "step": 135460
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002277173567385359,
+      "loss": 0.4663,
+      "step": 135470
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022767495600924337,
+      "loss": 0.5104,
+      "step": 135480
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022763255527995084,
+      "loss": 0.4212,
+      "step": 135490
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022759015455065828,
+      "loss": 0.4767,
+      "step": 135500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022754775382136572,
+      "loss": 0.4249,
+      "step": 135510
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002275053530920732,
+      "loss": 0.4835,
+      "step": 135520
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022746295236278065,
+      "loss": 0.4983,
+      "step": 135530
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002274205516334881,
+      "loss": 0.4985,
+      "step": 135540
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022737815090419553,
+      "loss": 0.4076,
+      "step": 135550
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022733575017490302,
+      "loss": 0.5724,
+      "step": 135560
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022729334944561046,
+      "loss": 0.4971,
+      "step": 135570
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002272509487163179,
+      "loss": 0.3866,
+      "step": 135580
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002272085479870254,
+      "loss": 0.4149,
+      "step": 135590
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022716614725773284,
+      "loss": 0.4221,
+      "step": 135600
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022712374652844028,
+      "loss": 0.492,
+      "step": 135610
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022708134579914777,
+      "loss": 0.4722,
+      "step": 135620
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002270389450698552,
+      "loss": 0.5396,
+      "step": 135630
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022699654434056265,
+      "loss": 0.4661,
+      "step": 135640
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022695414361127014,
+      "loss": 0.505,
+      "step": 135650
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022691174288197758,
+      "loss": 0.4467,
+      "step": 135660
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022686934215268502,
+      "loss": 0.4268,
+      "step": 135670
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022682694142339252,
+      "loss": 0.4382,
+      "step": 135680
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022678454069409996,
+      "loss": 0.4664,
+      "step": 135690
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002267421399648074,
+      "loss": 0.5045,
+      "step": 135700
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022669973923551484,
+      "loss": 0.4181,
+      "step": 135710
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022665733850622233,
+      "loss": 0.4074,
+      "step": 135720
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022661493777692977,
+      "loss": 0.4894,
+      "step": 135730
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002265725370476372,
+      "loss": 0.397,
+      "step": 135740
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002265301363183447,
+      "loss": 0.4717,
+      "step": 135750
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022648773558905214,
+      "loss": 0.506,
+      "step": 135760
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022644533485975958,
+      "loss": 0.4887,
+      "step": 135770
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022640293413046708,
+      "loss": 0.4002,
+      "step": 135780
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022636053340117452,
+      "loss": 0.4637,
+      "step": 135790
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022631813267188195,
+      "loss": 0.4777,
+      "step": 135800
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022627573194258945,
+      "loss": 0.4355,
+      "step": 135810
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002262333312132969,
+      "loss": 0.4474,
+      "step": 135820
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022619093048400433,
+      "loss": 0.4377,
+      "step": 135830
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022614852975471182,
+      "loss": 0.4261,
+      "step": 135840
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022610612902541926,
+      "loss": 0.5369,
+      "step": 135850
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0002260637282961267,
+      "loss": 0.3658,
+      "step": 135860
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022602132756683414,
+      "loss": 0.4833,
+      "step": 135870
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022597892683754163,
+      "loss": 0.43,
+      "step": 135880
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022593652610824907,
+      "loss": 0.4585,
+      "step": 135890
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022589412537895651,
+      "loss": 0.4613,
+      "step": 135900
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.000225851724649664,
+      "loss": 0.4877,
+      "step": 135910
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022580932392037145,
+      "loss": 0.3974,
+      "step": 135920
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0002257669231910789,
+      "loss": 0.4574,
+      "step": 135930
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022572452246178638,
+      "loss": 0.463,
+      "step": 135940
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022568212173249382,
+      "loss": 0.4485,
+      "step": 135950
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022563972100320126,
+      "loss": 0.4282,
+      "step": 135960
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022559732027390873,
+      "loss": 0.4636,
+      "step": 135970
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0002255549195446162,
+      "loss": 0.413,
+      "step": 135980
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022551251881532363,
+      "loss": 0.4829,
+      "step": 135990
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0002254701180860311,
+      "loss": 0.4773,
+      "step": 136000
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022542771735673857,
+      "loss": 0.4684,
+      "step": 136010
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.000225385316627446,
+      "loss": 0.4196,
+      "step": 136020
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022534291589815345,
+      "loss": 0.4823,
+      "step": 136030
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022530051516886091,
+      "loss": 0.4753,
+      "step": 136040
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022525811443956838,
+      "loss": 0.4455,
+      "step": 136050
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022521571371027582,
+      "loss": 0.4214,
+      "step": 136060
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0002251733129809833,
+      "loss": 0.5117,
+      "step": 136070
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022513091225169073,
+      "loss": 0.4816,
+      "step": 136080
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0002250885115223982,
+      "loss": 0.5513,
+      "step": 136090
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022504611079310566,
+      "loss": 0.523,
+      "step": 136100
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0002250037100638131,
+      "loss": 0.3967,
+      "step": 136110
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022496130933452057,
+      "loss": 0.4359,
+      "step": 136120
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022491890860522803,
+      "loss": 0.4531,
+      "step": 136130
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022487650787593547,
+      "loss": 0.4137,
+      "step": 136140
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0002248341071466429,
+      "loss": 0.4645,
+      "step": 136150
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022479170641735038,
+      "loss": 0.4884,
+      "step": 136160
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022474930568805785,
+      "loss": 0.4337,
+      "step": 136170
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022470690495876529,
+      "loss": 0.3971,
+      "step": 136180
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022466450422947273,
+      "loss": 0.5368,
+      "step": 136190
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022462210350018022,
+      "loss": 0.5003,
+      "step": 136200
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022457970277088766,
+      "loss": 0.503,
+      "step": 136210
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0002245373020415951,
+      "loss": 0.4171,
+      "step": 136220
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0002244949013123026,
+      "loss": 0.4444,
+      "step": 136230
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022445250058301003,
+      "loss": 0.4972,
+      "step": 136240
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022441009985371747,
+      "loss": 0.4084,
+      "step": 136250
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022436769912442497,
+      "loss": 0.4414,
+      "step": 136260
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0002243252983951324,
+      "loss": 0.4563,
+      "step": 136270
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022428289766583985,
+      "loss": 0.4281,
+      "step": 136280
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022424049693654734,
+      "loss": 0.4561,
+      "step": 136290
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022419809620725478,
+      "loss": 0.4146,
+      "step": 136300
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022415569547796222,
+      "loss": 0.4908,
+      "step": 136310
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022411329474866966,
+      "loss": 0.4676,
+      "step": 136320
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022407089401937715,
+      "loss": 0.4885,
+      "step": 136330
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002240284932900846,
+      "loss": 0.5372,
+      "step": 136340
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022398609256079203,
+      "loss": 0.4775,
+      "step": 136350
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022394369183149953,
+      "loss": 0.3942,
+      "step": 136360
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022390129110220697,
+      "loss": 0.4442,
+      "step": 136370
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002238588903729144,
+      "loss": 0.4898,
+      "step": 136380
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002238164896436219,
+      "loss": 0.4483,
+      "step": 136390
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022377408891432934,
+      "loss": 0.4744,
+      "step": 136400
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022373168818503678,
+      "loss": 0.4319,
+      "step": 136410
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022368928745574427,
+      "loss": 0.4583,
+      "step": 136420
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002236468867264517,
+      "loss": 0.5123,
+      "step": 136430
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022360448599715915,
+      "loss": 0.4642,
+      "step": 136440
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022356208526786665,
+      "loss": 0.4512,
+      "step": 136450
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022351968453857409,
+      "loss": 0.4798,
+      "step": 136460
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022347728380928152,
+      "loss": 0.4894,
+      "step": 136470
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022343488307998896,
+      "loss": 0.3926,
+      "step": 136480
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022339248235069646,
+      "loss": 0.4291,
+      "step": 136490
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002233500816214039,
+      "loss": 0.4697,
+      "step": 136500
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022330768089211134,
+      "loss": 0.5094,
+      "step": 136510
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022326528016281883,
+      "loss": 0.3909,
+      "step": 136520
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022322287943352627,
+      "loss": 0.4375,
+      "step": 136530
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002231804787042337,
+      "loss": 0.4491,
+      "step": 136540
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002231380779749412,
+      "loss": 0.4028,
+      "step": 136550
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022309567724564864,
+      "loss": 0.378,
+      "step": 136560
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022305327651635608,
+      "loss": 0.4358,
+      "step": 136570
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022301087578706358,
+      "loss": 0.4733,
+      "step": 136580
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022296847505777102,
+      "loss": 0.4274,
+      "step": 136590
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022292607432847846,
+      "loss": 0.4752,
+      "step": 136600
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022288367359918595,
+      "loss": 0.3996,
+      "step": 136610
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0002228412728698934,
+      "loss": 0.4567,
+      "step": 136620
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022279887214060083,
+      "loss": 0.3825,
+      "step": 136630
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022275647141130827,
+      "loss": 0.4471,
+      "step": 136640
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022271407068201576,
+      "loss": 0.4492,
+      "step": 136650
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0002226716699527232,
+      "loss": 0.5035,
+      "step": 136660
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022262926922343064,
+      "loss": 0.4498,
+      "step": 136670
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0002225868684941381,
+      "loss": 0.5016,
+      "step": 136680
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022254446776484558,
+      "loss": 0.4204,
+      "step": 136690
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022250206703555302,
+      "loss": 0.4726,
+      "step": 136700
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022245966630626048,
+      "loss": 0.4334,
+      "step": 136710
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022241726557696792,
+      "loss": 0.5198,
+      "step": 136720
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0002223748648476754,
+      "loss": 0.4089,
+      "step": 136730
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022233246411838286,
+      "loss": 0.4129,
+      "step": 136740
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0002222900633890903,
+      "loss": 0.4392,
+      "step": 136750
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022224766265979776,
+      "loss": 0.4083,
+      "step": 136760
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022220526193050523,
+      "loss": 0.5089,
+      "step": 136770
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022216286120121267,
+      "loss": 0.4183,
+      "step": 136780
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0002221204604719201,
+      "loss": 0.4547,
+      "step": 136790
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022207805974262758,
+      "loss": 0.449,
+      "step": 136800
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022203565901333504,
+      "loss": 0.4407,
+      "step": 136810
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022199325828404248,
+      "loss": 0.5582,
+      "step": 136820
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022195085755474992,
+      "loss": 0.4349,
+      "step": 136830
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022190845682545742,
+      "loss": 0.4728,
+      "step": 136840
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022186605609616486,
+      "loss": 0.561,
+      "step": 136850
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002218236553668723,
+      "loss": 0.5399,
+      "step": 136860
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002217812546375798,
+      "loss": 0.3749,
+      "step": 136870
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022173885390828723,
+      "loss": 0.4991,
+      "step": 136880
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022169645317899467,
+      "loss": 0.4673,
+      "step": 136890
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022165405244970216,
+      "loss": 0.4365,
+      "step": 136900
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002216116517204096,
+      "loss": 0.4103,
+      "step": 136910
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022156925099111704,
+      "loss": 0.4266,
+      "step": 136920
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022152685026182448,
+      "loss": 0.4594,
+      "step": 136930
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022148444953253198,
+      "loss": 0.5715,
+      "step": 136940
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022144204880323942,
+      "loss": 0.4494,
+      "step": 136950
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022139964807394686,
+      "loss": 0.5257,
+      "step": 136960
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022135724734465435,
+      "loss": 0.3912,
+      "step": 136970
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002213148466153618,
+      "loss": 0.4761,
+      "step": 136980
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022127244588606923,
+      "loss": 0.4266,
+      "step": 136990
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022123004515677672,
+      "loss": 0.5021,
+      "step": 137000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022118764442748416,
+      "loss": 0.3853,
+      "step": 137010
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002211452436981916,
+      "loss": 0.4511,
+      "step": 137020
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002211028429688991,
+      "loss": 0.5454,
+      "step": 137030
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022106044223960654,
+      "loss": 0.4441,
+      "step": 137040
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022101804151031398,
+      "loss": 0.5351,
+      "step": 137050
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022097564078102147,
+      "loss": 0.5318,
+      "step": 137060
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002209332400517289,
+      "loss": 0.4721,
+      "step": 137070
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022089083932243635,
+      "loss": 0.5089,
+      "step": 137080
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002208484385931438,
+      "loss": 0.4651,
+      "step": 137090
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022080603786385128,
+      "loss": 0.5013,
+      "step": 137100
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022076363713455872,
+      "loss": 0.4365,
+      "step": 137110
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022072123640526616,
+      "loss": 0.4402,
+      "step": 137120
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022067883567597366,
+      "loss": 0.4227,
+      "step": 137130
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002206364349466811,
+      "loss": 0.4916,
+      "step": 137140
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022059403421738853,
+      "loss": 0.461,
+      "step": 137150
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022055163348809603,
+      "loss": 0.4272,
+      "step": 137160
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022050923275880347,
+      "loss": 0.4432,
+      "step": 137170
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002204668320295109,
+      "loss": 0.4281,
+      "step": 137180
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002204244313002184,
+      "loss": 0.4541,
+      "step": 137190
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022038203057092584,
+      "loss": 0.4509,
+      "step": 137200
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022033962984163328,
+      "loss": 0.429,
+      "step": 137210
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022029722911234078,
+      "loss": 0.444,
+      "step": 137220
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022025482838304821,
+      "loss": 0.4205,
+      "step": 137230
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022021242765375565,
+      "loss": 0.5531,
+      "step": 137240
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002201700269244631,
+      "loss": 0.3977,
+      "step": 137250
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002201276261951706,
+      "loss": 0.4263,
+      "step": 137260
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022008522546587803,
+      "loss": 0.4455,
+      "step": 137270
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00022004282473658547,
+      "loss": 0.4875,
+      "step": 137280
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00022000042400729296,
+      "loss": 0.4483,
+      "step": 137290
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002199580232780004,
+      "loss": 0.3466,
+      "step": 137300
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021991562254870784,
+      "loss": 0.4759,
+      "step": 137310
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002198732218194153,
+      "loss": 0.4656,
+      "step": 137320
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021983082109012277,
+      "loss": 0.5198,
+      "step": 137330
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021978842036083021,
+      "loss": 0.3796,
+      "step": 137340
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021974601963153768,
+      "loss": 0.4897,
+      "step": 137350
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021970361890224515,
+      "loss": 0.472,
+      "step": 137360
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002196612181729526,
+      "loss": 0.4385,
+      "step": 137370
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021961881744366005,
+      "loss": 0.4271,
+      "step": 137380
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002195764167143675,
+      "loss": 0.3951,
+      "step": 137390
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021953401598507496,
+      "loss": 0.4445,
+      "step": 137400
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002194916152557824,
+      "loss": 0.4876,
+      "step": 137410
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021944921452648987,
+      "loss": 0.4143,
+      "step": 137420
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002194068137971973,
+      "loss": 0.4934,
+      "step": 137430
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021936441306790477,
+      "loss": 0.368,
+      "step": 137440
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021932201233861224,
+      "loss": 0.4454,
+      "step": 137450
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021927961160931968,
+      "loss": 0.4123,
+      "step": 137460
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021923721088002712,
+      "loss": 0.5073,
+      "step": 137470
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002191948101507346,
+      "loss": 0.4567,
+      "step": 137480
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021915240942144205,
+      "loss": 0.4996,
+      "step": 137490
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002191100086921495,
+      "loss": 0.5326,
+      "step": 137500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.000219067607962857,
+      "loss": 0.48,
+      "step": 137510
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021902520723356443,
+      "loss": 0.4651,
+      "step": 137520
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021898280650427187,
+      "loss": 0.4009,
+      "step": 137530
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002189404057749793,
+      "loss": 0.4262,
+      "step": 137540
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002188980050456868,
+      "loss": 0.4265,
+      "step": 137550
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021885560431639424,
+      "loss": 0.4692,
+      "step": 137560
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021881320358710168,
+      "loss": 0.4746,
+      "step": 137570
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021877080285780917,
+      "loss": 0.4032,
+      "step": 137580
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002187284021285166,
+      "loss": 0.4417,
+      "step": 137590
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021868600139922405,
+      "loss": 0.4932,
+      "step": 137600
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021864360066993155,
+      "loss": 0.4553,
+      "step": 137610
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021860119994063899,
+      "loss": 0.5052,
+      "step": 137620
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021855879921134643,
+      "loss": 0.4811,
+      "step": 137630
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021851639848205392,
+      "loss": 0.4964,
+      "step": 137640
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021847399775276136,
+      "loss": 0.4497,
+      "step": 137650
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002184315970234688,
+      "loss": 0.4663,
+      "step": 137660
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002183891962941763,
+      "loss": 0.4391,
+      "step": 137670
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021834679556488373,
+      "loss": 0.4489,
+      "step": 137680
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021830439483559117,
+      "loss": 0.4504,
+      "step": 137690
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002182619941062986,
+      "loss": 0.4701,
+      "step": 137700
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002182195933770061,
+      "loss": 0.4326,
+      "step": 137710
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021817719264771355,
+      "loss": 0.4555,
+      "step": 137720
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021813479191842099,
+      "loss": 0.3884,
+      "step": 137730
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021809239118912848,
+      "loss": 0.5089,
+      "step": 137740
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021804999045983592,
+      "loss": 0.4026,
+      "step": 137750
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021800758973054336,
+      "loss": 0.48,
+      "step": 137760
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021796518900125085,
+      "loss": 0.485,
+      "step": 137770
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0002179227882719583,
+      "loss": 0.4663,
+      "step": 137780
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021788038754266573,
+      "loss": 0.4366,
+      "step": 137790
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021783798681337323,
+      "loss": 0.425,
+      "step": 137800
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021779558608408067,
+      "loss": 0.4206,
+      "step": 137810
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0002177531853547881,
+      "loss": 0.4518,
+      "step": 137820
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0002177107846254956,
+      "loss": 0.4394,
+      "step": 137830
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021766838389620304,
+      "loss": 0.5056,
+      "step": 137840
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021762598316691048,
+      "loss": 0.4918,
+      "step": 137850
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021758358243761792,
+      "loss": 0.4682,
+      "step": 137860
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0002175411817083254,
+      "loss": 0.3859,
+      "step": 137870
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021749878097903285,
+      "loss": 0.4939,
+      "step": 137880
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0002174563802497403,
+      "loss": 0.4555,
+      "step": 137890
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021741397952044778,
+      "loss": 0.4779,
+      "step": 137900
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021737157879115522,
+      "loss": 0.4317,
+      "step": 137910
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021732917806186266,
+      "loss": 0.5262,
+      "step": 137920
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021728677733257016,
+      "loss": 0.4191,
+      "step": 137930
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0002172443766032776,
+      "loss": 0.3928,
+      "step": 137940
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021720197587398504,
+      "loss": 0.5087,
+      "step": 137950
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0002171595751446925,
+      "loss": 0.4621,
+      "step": 137960
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021711717441539997,
+      "loss": 0.4279,
+      "step": 137970
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0002170747736861074,
+      "loss": 0.4447,
+      "step": 137980
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021703237295681488,
+      "loss": 0.4773,
+      "step": 137990
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021698997222752234,
+      "loss": 0.4155,
+      "step": 138000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021694757149822978,
+      "loss": 0.4336,
+      "step": 138010
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021690517076893722,
+      "loss": 0.4095,
+      "step": 138020
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002168627700396447,
+      "loss": 0.4773,
+      "step": 138030
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021682036931035216,
+      "loss": 0.5362,
+      "step": 138040
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002167779685810596,
+      "loss": 0.4438,
+      "step": 138050
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021673556785176706,
+      "loss": 0.3999,
+      "step": 138060
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002166931671224745,
+      "loss": 0.399,
+      "step": 138070
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021665076639318197,
+      "loss": 0.4866,
+      "step": 138080
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021660836566388944,
+      "loss": 0.3908,
+      "step": 138090
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021656596493459688,
+      "loss": 0.4284,
+      "step": 138100
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021652356420530434,
+      "loss": 0.4429,
+      "step": 138110
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002164811634760118,
+      "loss": 0.4407,
+      "step": 138120
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021643876274671925,
+      "loss": 0.449,
+      "step": 138130
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002163963620174267,
+      "loss": 0.4316,
+      "step": 138140
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021635396128813418,
+      "loss": 0.5016,
+      "step": 138150
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021631156055884162,
+      "loss": 0.4551,
+      "step": 138160
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021626915982954906,
+      "loss": 0.503,
+      "step": 138170
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002162267591002565,
+      "loss": 0.3877,
+      "step": 138180
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.000216184358370964,
+      "loss": 0.5389,
+      "step": 138190
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021614195764167144,
+      "loss": 0.4412,
+      "step": 138200
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021609955691237888,
+      "loss": 0.4659,
+      "step": 138210
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021605715618308637,
+      "loss": 0.4343,
+      "step": 138220
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002160147554537938,
+      "loss": 0.3774,
+      "step": 138230
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021597235472450125,
+      "loss": 0.418,
+      "step": 138240
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021592995399520874,
+      "loss": 0.4399,
+      "step": 138250
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021588755326591618,
+      "loss": 0.5241,
+      "step": 138260
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021584515253662362,
+      "loss": 0.471,
+      "step": 138270
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021580275180733112,
+      "loss": 0.4928,
+      "step": 138280
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021576035107803856,
+      "loss": 0.3881,
+      "step": 138290
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.000215717950348746,
+      "loss": 0.4516,
+      "step": 138300
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021567554961945344,
+      "loss": 0.4564,
+      "step": 138310
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021563314889016093,
+      "loss": 0.4703,
+      "step": 138320
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021559074816086837,
+      "loss": 0.4971,
+      "step": 138330
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002155483474315758,
+      "loss": 0.4078,
+      "step": 138340
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002155059467022833,
+      "loss": 0.4606,
+      "step": 138350
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021546354597299074,
+      "loss": 0.5039,
+      "step": 138360
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021542114524369818,
+      "loss": 0.4673,
+      "step": 138370
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021537874451440568,
+      "loss": 0.4584,
+      "step": 138380
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021533634378511312,
+      "loss": 0.3842,
+      "step": 138390
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021529394305582056,
+      "loss": 0.468,
+      "step": 138400
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021525154232652805,
+      "loss": 0.4422,
+      "step": 138410
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002152091415972355,
+      "loss": 0.4624,
+      "step": 138420
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021516674086794293,
+      "loss": 0.4506,
+      "step": 138430
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021512434013865042,
+      "loss": 0.3714,
+      "step": 138440
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021508193940935786,
+      "loss": 0.4725,
+      "step": 138450
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0002150395386800653,
+      "loss": 0.5363,
+      "step": 138460
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021499713795077274,
+      "loss": 0.4032,
+      "step": 138470
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021495473722148024,
+      "loss": 0.4893,
+      "step": 138480
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021491233649218767,
+      "loss": 0.4209,
+      "step": 138490
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021486993576289511,
+      "loss": 0.4426,
+      "step": 138500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0002148275350336026,
+      "loss": 0.4634,
+      "step": 138510
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021478513430431005,
+      "loss": 0.4702,
+      "step": 138520
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0002147427335750175,
+      "loss": 0.4486,
+      "step": 138530
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021470033284572498,
+      "loss": 0.4841,
+      "step": 138540
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021465793211643242,
+      "loss": 0.4767,
+      "step": 138550
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021461553138713986,
+      "loss": 0.5451,
+      "step": 138560
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021457313065784736,
+      "loss": 0.4826,
+      "step": 138570
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0002145307299285548,
+      "loss": 0.3677,
+      "step": 138580
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021448832919926223,
+      "loss": 0.4388,
+      "step": 138590
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0002144459284699697,
+      "loss": 0.4761,
+      "step": 138600
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021440352774067717,
+      "loss": 0.4595,
+      "step": 138610
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0002143611270113846,
+      "loss": 0.4504,
+      "step": 138620
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021431872628209205,
+      "loss": 0.4318,
+      "step": 138630
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021427632555279954,
+      "loss": 0.4357,
+      "step": 138640
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021423392482350698,
+      "loss": 0.4682,
+      "step": 138650
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021419152409421442,
+      "loss": 0.4502,
+      "step": 138660
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0002141491233649219,
+      "loss": 0.4606,
+      "step": 138670
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021410672263562935,
+      "loss": 0.5044,
+      "step": 138680
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0002140643219063368,
+      "loss": 0.4462,
+      "step": 138690
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021402192117704426,
+      "loss": 0.4724,
+      "step": 138700
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0002139795204477517,
+      "loss": 0.484,
+      "step": 138710
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021393711971845917,
+      "loss": 0.3903,
+      "step": 138720
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021389471898916663,
+      "loss": 0.4734,
+      "step": 138730
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021385231825987407,
+      "loss": 0.5154,
+      "step": 138740
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021380991753058154,
+      "loss": 0.506,
+      "step": 138750
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.000213767516801289,
+      "loss": 0.4108,
+      "step": 138760
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021372511607199645,
+      "loss": 0.4009,
+      "step": 138770
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0002136827153427039,
+      "loss": 0.4162,
+      "step": 138780
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021364031461341135,
+      "loss": 0.4591,
+      "step": 138790
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021359791388411882,
+      "loss": 0.443,
+      "step": 138800
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021355551315482626,
+      "loss": 0.4277,
+      "step": 138810
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0002135131124255337,
+      "loss": 0.6059,
+      "step": 138820
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0002134707116962412,
+      "loss": 0.436,
+      "step": 138830
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021342831096694863,
+      "loss": 0.5304,
+      "step": 138840
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021338591023765607,
+      "loss": 0.5081,
+      "step": 138850
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021334350950836357,
+      "loss": 0.4352,
+      "step": 138860
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.000213301108779071,
+      "loss": 0.4444,
+      "step": 138870
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021325870804977845,
+      "loss": 0.4739,
+      "step": 138880
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021321630732048594,
+      "loss": 0.5099,
+      "step": 138890
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021317390659119338,
+      "loss": 0.4564,
+      "step": 138900
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021313150586190082,
+      "loss": 0.3596,
+      "step": 138910
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0002130891051326083,
+      "loss": 0.4201,
+      "step": 138920
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021304670440331575,
+      "loss": 0.458,
+      "step": 138930
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002130043036740232,
+      "loss": 0.429,
+      "step": 138940
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021296190294473063,
+      "loss": 0.4224,
+      "step": 138950
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021291950221543813,
+      "loss": 0.4913,
+      "step": 138960
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021287710148614557,
+      "loss": 0.4379,
+      "step": 138970
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.000212834700756853,
+      "loss": 0.3837,
+      "step": 138980
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002127923000275605,
+      "loss": 0.4204,
+      "step": 138990
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021274989929826794,
+      "loss": 0.4901,
+      "step": 139000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021270749856897538,
+      "loss": 0.4159,
+      "step": 139010
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021266509783968287,
+      "loss": 0.4275,
+      "step": 139020
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002126226971103903,
+      "loss": 0.4891,
+      "step": 139030
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021258029638109775,
+      "loss": 0.4351,
+      "step": 139040
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021253789565180525,
+      "loss": 0.4155,
+      "step": 139050
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021249549492251269,
+      "loss": 0.4034,
+      "step": 139060
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021245309419322013,
+      "loss": 0.5147,
+      "step": 139070
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021241069346392757,
+      "loss": 0.4261,
+      "step": 139080
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021236829273463506,
+      "loss": 0.4618,
+      "step": 139090
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002123258920053425,
+      "loss": 0.4811,
+      "step": 139100
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021228349127604994,
+      "loss": 0.4579,
+      "step": 139110
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021224109054675743,
+      "loss": 0.4624,
+      "step": 139120
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021219868981746487,
+      "loss": 0.4563,
+      "step": 139130
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002121562890881723,
+      "loss": 0.4369,
+      "step": 139140
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002121138883588798,
+      "loss": 0.3875,
+      "step": 139150
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021207148762958725,
+      "loss": 0.4522,
+      "step": 139160
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021202908690029468,
+      "loss": 0.4307,
+      "step": 139170
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021198668617100218,
+      "loss": 0.5078,
+      "step": 139180
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021194428544170962,
+      "loss": 0.4428,
+      "step": 139190
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021190188471241706,
+      "loss": 0.4614,
+      "step": 139200
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021185948398312455,
+      "loss": 0.433,
+      "step": 139210
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.000211817083253832,
+      "loss": 0.4316,
+      "step": 139220
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021177468252453943,
+      "loss": 0.4697,
+      "step": 139230
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021173228179524687,
+      "loss": 0.5082,
+      "step": 139240
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021168988106595436,
+      "loss": 0.4202,
+      "step": 139250
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002116474803366618,
+      "loss": 0.3984,
+      "step": 139260
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021160507960736924,
+      "loss": 0.4396,
+      "step": 139270
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021156267887807674,
+      "loss": 0.386,
+      "step": 139280
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021152027814878418,
+      "loss": 0.4641,
+      "step": 139290
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021147787741949162,
+      "loss": 0.393,
+      "step": 139300
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021143547669019908,
+      "loss": 0.4663,
+      "step": 139310
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021139307596090655,
+      "loss": 0.4419,
+      "step": 139320
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.000211350675231614,
+      "loss": 0.4864,
+      "step": 139330
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021130827450232146,
+      "loss": 0.4792,
+      "step": 139340
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002112658737730289,
+      "loss": 0.4481,
+      "step": 139350
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021122347304373636,
+      "loss": 0.3896,
+      "step": 139360
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021118107231444383,
+      "loss": 0.539,
+      "step": 139370
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021113867158515127,
+      "loss": 0.408,
+      "step": 139380
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021109627085585874,
+      "loss": 0.4089,
+      "step": 139390
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021105387012656618,
+      "loss": 0.4153,
+      "step": 139400
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021101146939727364,
+      "loss": 0.4764,
+      "step": 139410
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021096906866798108,
+      "loss": 0.5316,
+      "step": 139420
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021092666793868855,
+      "loss": 0.4748,
+      "step": 139430
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021088426720939602,
+      "loss": 0.431,
+      "step": 139440
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021084186648010346,
+      "loss": 0.488,
+      "step": 139450
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002107994657508109,
+      "loss": 0.518,
+      "step": 139460
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002107570650215184,
+      "loss": 0.4565,
+      "step": 139470
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021071466429222583,
+      "loss": 0.411,
+      "step": 139480
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021067226356293327,
+      "loss": 0.439,
+      "step": 139490
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021062986283364076,
+      "loss": 0.4509,
+      "step": 139500
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002105874621043482,
+      "loss": 0.4913,
+      "step": 139510
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021054506137505564,
+      "loss": 0.4745,
+      "step": 139520
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021050266064576314,
+      "loss": 0.424,
+      "step": 139530
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021046025991647058,
+      "loss": 0.4331,
+      "step": 139540
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021041785918717802,
+      "loss": 0.4473,
+      "step": 139550
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021037545845788546,
+      "loss": 0.4625,
+      "step": 139560
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021033305772859295,
+      "loss": 0.5133,
+      "step": 139570
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002102906569993004,
+      "loss": 0.4238,
+      "step": 139580
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021024825627000783,
+      "loss": 0.4161,
+      "step": 139590
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021020585554071532,
+      "loss": 0.5078,
+      "step": 139600
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021016345481142276,
+      "loss": 0.496,
+      "step": 139610
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002101210540821302,
+      "loss": 0.5024,
+      "step": 139620
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002100786533528377,
+      "loss": 0.4132,
+      "step": 139630
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00021003625262354514,
+      "loss": 0.5155,
+      "step": 139640
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020999385189425258,
+      "loss": 0.482,
+      "step": 139650
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020995145116496007,
+      "loss": 0.4365,
+      "step": 139660
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0002099090504356675,
+      "loss": 0.5303,
+      "step": 139670
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020986664970637495,
+      "loss": 0.4323,
+      "step": 139680
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0002098242489770824,
+      "loss": 0.5299,
+      "step": 139690
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020978184824778988,
+      "loss": 0.4978,
+      "step": 139700
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020973944751849732,
+      "loss": 0.4797,
+      "step": 139710
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020969704678920476,
+      "loss": 0.4894,
+      "step": 139720
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020965464605991226,
+      "loss": 0.4908,
+      "step": 139730
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0002096122453306197,
+      "loss": 0.5069,
+      "step": 139740
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020956984460132714,
+      "loss": 0.445,
+      "step": 139750
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020952744387203463,
+      "loss": 0.4705,
+      "step": 139760
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020948504314274207,
+      "loss": 0.5959,
+      "step": 139770
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0002094426424134495,
+      "loss": 0.4569,
+      "step": 139780
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.000209400241684157,
+      "loss": 0.4059,
+      "step": 139790
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020935784095486444,
+      "loss": 0.4391,
+      "step": 139800
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020931544022557188,
+      "loss": 0.4742,
+      "step": 139810
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020927303949627938,
+      "loss": 0.4306,
+      "step": 139820
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020923063876698682,
+      "loss": 0.47,
+      "step": 139830
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020918823803769425,
+      "loss": 0.4165,
+      "step": 139840
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0002091458373084017,
+      "loss": 0.4242,
+      "step": 139850
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0002091034365791092,
+      "loss": 0.462,
+      "step": 139860
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020906103584981663,
+      "loss": 0.4482,
+      "step": 139870
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020901863512052407,
+      "loss": 0.4302,
+      "step": 139880
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020897623439123156,
+      "loss": 0.3772,
+      "step": 139890
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.000208933833661939,
+      "loss": 0.527,
+      "step": 139900
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020889143293264644,
+      "loss": 0.3955,
+      "step": 139910
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020884903220335394,
+      "loss": 0.4213,
+      "step": 139920
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020880663147406137,
+      "loss": 0.4523,
+      "step": 139930
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020876423074476881,
+      "loss": 0.3461,
+      "step": 139940
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020872183001547628,
+      "loss": 0.4427,
+      "step": 139950
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020867942928618375,
+      "loss": 0.4866,
+      "step": 139960
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0002086370285568912,
+      "loss": 0.5519,
+      "step": 139970
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020859462782759865,
+      "loss": 0.4811,
+      "step": 139980
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020855222709830612,
+      "loss": 0.4446,
+      "step": 139990
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020850982636901356,
+      "loss": 0.5399,
+      "step": 140000
+    },
+    {
+      "epoch": 5.92,
+      "eval_loss": 0.6056107878684998,
+      "eval_runtime": 337.6283,
+      "eval_samples_per_second": 15.564,
+      "eval_steps_per_second": 3.892,
+      "step": 140000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.000208467425639721,
+      "loss": 0.4435,
+      "step": 140010
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020842502491042847,
+      "loss": 0.3995,
+      "step": 140020
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020838262418113593,
+      "loss": 0.4419,
+      "step": 140030
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020834022345184337,
+      "loss": 0.3923,
+      "step": 140040
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020829782272255084,
+      "loss": 0.4462,
+      "step": 140050
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020825542199325828,
+      "loss": 0.5182,
+      "step": 140060
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020821302126396575,
+      "loss": 0.4705,
+      "step": 140070
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020817062053467321,
+      "loss": 0.4274,
+      "step": 140080
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020812821980538065,
+      "loss": 0.4781,
+      "step": 140090
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0002080858190760881,
+      "loss": 0.4013,
+      "step": 140100
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0002080434183467956,
+      "loss": 0.4162,
+      "step": 140110
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020800101761750303,
+      "loss": 0.4701,
+      "step": 140120
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020795861688821047,
+      "loss": 0.4789,
+      "step": 140130
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020791621615891796,
+      "loss": 0.4661,
+      "step": 140140
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0002078738154296254,
+      "loss": 0.4929,
+      "step": 140150
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020783141470033284,
+      "loss": 0.4461,
+      "step": 140160
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020778901397104028,
+      "loss": 0.3702,
+      "step": 140170
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020774661324174777,
+      "loss": 0.4704,
+      "step": 140180
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0002077042125124552,
+      "loss": 0.4845,
+      "step": 140190
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020766181178316265,
+      "loss": 0.5168,
+      "step": 140200
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020761941105387015,
+      "loss": 0.4443,
+      "step": 140210
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020757701032457759,
+      "loss": 0.4931,
+      "step": 140220
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020753460959528503,
+      "loss": 0.4857,
+      "step": 140230
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020749220886599252,
+      "loss": 0.455,
+      "step": 140240
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020744980813669996,
+      "loss": 0.4857,
+      "step": 140250
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0002074074074074074,
+      "loss": 0.4848,
+      "step": 140260
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0002073650066781149,
+      "loss": 0.4913,
+      "step": 140270
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020732260594882233,
+      "loss": 0.4615,
+      "step": 140280
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020728020521952977,
+      "loss": 0.3982,
+      "step": 140290
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020723780449023727,
+      "loss": 0.3955,
+      "step": 140300
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0002071954037609447,
+      "loss": 0.4264,
+      "step": 140310
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020715300303165215,
+      "loss": 0.4605,
+      "step": 140320
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020711060230235959,
+      "loss": 0.4678,
+      "step": 140330
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020706820157306708,
+      "loss": 0.4913,
+      "step": 140340
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020702580084377452,
+      "loss": 0.4678,
+      "step": 140350
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020698340011448196,
+      "loss": 0.4012,
+      "step": 140360
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020694099938518945,
+      "loss": 0.5381,
+      "step": 140370
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002068985986558969,
+      "loss": 0.4243,
+      "step": 140380
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020685619792660433,
+      "loss": 0.4447,
+      "step": 140390
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020681379719731183,
+      "loss": 0.4233,
+      "step": 140400
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020677139646801927,
+      "loss": 0.4134,
+      "step": 140410
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002067289957387267,
+      "loss": 0.4155,
+      "step": 140420
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002066865950094342,
+      "loss": 0.5145,
+      "step": 140430
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020664419428014164,
+      "loss": 0.4915,
+      "step": 140440
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020660179355084908,
+      "loss": 0.411,
+      "step": 140450
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020655939282155652,
+      "loss": 0.4694,
+      "step": 140460
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.000206516992092264,
+      "loss": 0.4513,
+      "step": 140470
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020647459136297145,
+      "loss": 0.5017,
+      "step": 140480
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002064321906336789,
+      "loss": 0.5045,
+      "step": 140490
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020638978990438639,
+      "loss": 0.4968,
+      "step": 140500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020634738917509383,
+      "loss": 0.4893,
+      "step": 140510
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020630498844580126,
+      "loss": 0.4275,
+      "step": 140520
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020626258771650876,
+      "loss": 0.4521,
+      "step": 140530
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002062201869872162,
+      "loss": 0.4474,
+      "step": 140540
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020617778625792364,
+      "loss": 0.4842,
+      "step": 140550
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020613538552863113,
+      "loss": 0.4281,
+      "step": 140560
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020609298479933857,
+      "loss": 0.4758,
+      "step": 140570
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.000206050584070046,
+      "loss": 0.4596,
+      "step": 140580
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020600818334075348,
+      "loss": 0.4603,
+      "step": 140590
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020596578261146094,
+      "loss": 0.4749,
+      "step": 140600
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020592338188216838,
+      "loss": 0.4512,
+      "step": 140610
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020588098115287582,
+      "loss": 0.5089,
+      "step": 140620
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020583858042358332,
+      "loss": 0.4436,
+      "step": 140630
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020579617969429076,
+      "loss": 0.4434,
+      "step": 140640
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002057537789649982,
+      "loss": 0.5214,
+      "step": 140650
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020571137823570566,
+      "loss": 0.4603,
+      "step": 140660
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020566897750641313,
+      "loss": 0.3975,
+      "step": 140670
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020562657677712057,
+      "loss": 0.4745,
+      "step": 140680
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020558417604782804,
+      "loss": 0.4703,
+      "step": 140690
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020554177531853548,
+      "loss": 0.5482,
+      "step": 140700
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020549937458924294,
+      "loss": 0.5312,
+      "step": 140710
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002054569738599504,
+      "loss": 0.5067,
+      "step": 140720
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020541457313065785,
+      "loss": 0.5307,
+      "step": 140730
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002053721724013653,
+      "loss": 0.499,
+      "step": 140740
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020532977167207278,
+      "loss": 0.4767,
+      "step": 140750
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020528737094278022,
+      "loss": 0.4559,
+      "step": 140760
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020524497021348766,
+      "loss": 0.5038,
+      "step": 140770
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020520256948419513,
+      "loss": 0.444,
+      "step": 140780
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002051601687549026,
+      "loss": 0.4647,
+      "step": 140790
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020511776802561004,
+      "loss": 0.4331,
+      "step": 140800
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020507536729631748,
+      "loss": 0.542,
+      "step": 140810
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020503296656702497,
+      "loss": 0.4823,
+      "step": 140820
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0002049905658377324,
+      "loss": 0.5285,
+      "step": 140830
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020494816510843985,
+      "loss": 0.471,
+      "step": 140840
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020490576437914734,
+      "loss": 0.4496,
+      "step": 140850
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020486336364985478,
+      "loss": 0.4394,
+      "step": 140860
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020482096292056222,
+      "loss": 0.5092,
+      "step": 140870
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020477856219126972,
+      "loss": 0.4786,
+      "step": 140880
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020473616146197716,
+      "loss": 0.4327,
+      "step": 140890
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0002046937607326846,
+      "loss": 0.4458,
+      "step": 140900
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0002046513600033921,
+      "loss": 0.4785,
+      "step": 140910
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020460895927409953,
+      "loss": 0.4165,
+      "step": 140920
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020456655854480697,
+      "loss": 0.4503,
+      "step": 140930
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0002045241578155144,
+      "loss": 0.4904,
+      "step": 140940
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0002044817570862219,
+      "loss": 0.4138,
+      "step": 140950
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020443935635692934,
+      "loss": 0.4491,
+      "step": 140960
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020439695562763678,
+      "loss": 0.401,
+      "step": 140970
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020435455489834428,
+      "loss": 0.435,
+      "step": 140980
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020431215416905172,
+      "loss": 0.4205,
+      "step": 140990
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020426975343975916,
+      "loss": 0.4694,
+      "step": 141000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020422735271046665,
+      "loss": 0.4325,
+      "step": 141010
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0002041849519811741,
+      "loss": 0.4604,
+      "step": 141020
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020414255125188153,
+      "loss": 0.4837,
+      "step": 141030
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020410015052258902,
+      "loss": 0.4719,
+      "step": 141040
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020405774979329646,
+      "loss": 0.4859,
+      "step": 141050
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0002040153490640039,
+      "loss": 0.4369,
+      "step": 141060
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0002039729483347114,
+      "loss": 0.4064,
+      "step": 141070
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020393054760541884,
+      "loss": 0.5766,
+      "step": 141080
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020388814687612628,
+      "loss": 0.431,
+      "step": 141090
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020384574614683372,
+      "loss": 0.5034,
+      "step": 141100
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0002038033454175412,
+      "loss": 0.4028,
+      "step": 141110
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020376094468824865,
+      "loss": 0.4808,
+      "step": 141120
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0002037185439589561,
+      "loss": 0.4421,
+      "step": 141130
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020367614322966358,
+      "loss": 0.4982,
+      "step": 141140
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020363374250037102,
+      "loss": 0.4477,
+      "step": 141150
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020359134177107846,
+      "loss": 0.4531,
+      "step": 141160
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020354894104178596,
+      "loss": 0.4731,
+      "step": 141170
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0002035065403124934,
+      "loss": 0.4431,
+      "step": 141180
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020346413958320083,
+      "loss": 0.4777,
+      "step": 141190
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020342173885390833,
+      "loss": 0.4413,
+      "step": 141200
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020337933812461577,
+      "loss": 0.4524,
+      "step": 141210
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0002033369373953232,
+      "loss": 0.4095,
+      "step": 141220
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020329453666603065,
+      "loss": 0.46,
+      "step": 141230
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020325213593673814,
+      "loss": 0.4905,
+      "step": 141240
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020320973520744558,
+      "loss": 0.4158,
+      "step": 141250
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020316733447815302,
+      "loss": 0.4762,
+      "step": 141260
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020312493374886051,
+      "loss": 0.4825,
+      "step": 141270
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020308253301956795,
+      "loss": 0.4171,
+      "step": 141280
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0002030401322902754,
+      "loss": 0.4303,
+      "step": 141290
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020299773156098286,
+      "loss": 0.4421,
+      "step": 141300
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020295533083169033,
+      "loss": 0.4495,
+      "step": 141310
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020291293010239777,
+      "loss": 0.4548,
+      "step": 141320
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020287052937310523,
+      "loss": 0.4955,
+      "step": 141330
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020282812864381267,
+      "loss": 0.4924,
+      "step": 141340
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020278572791452014,
+      "loss": 0.4803,
+      "step": 141350
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0002027433271852276,
+      "loss": 0.4846,
+      "step": 141360
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020270092645593505,
+      "loss": 0.539,
+      "step": 141370
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020265852572664251,
+      "loss": 0.4754,
+      "step": 141380
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020261612499734995,
+      "loss": 0.3854,
+      "step": 141390
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020257372426805742,
+      "loss": 0.4309,
+      "step": 141400
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020253132353876486,
+      "loss": 0.5001,
+      "step": 141410
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020248892280947233,
+      "loss": 0.4951,
+      "step": 141420
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0002024465220801798,
+      "loss": 0.4696,
+      "step": 141430
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020240412135088723,
+      "loss": 0.427,
+      "step": 141440
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020236172062159467,
+      "loss": 0.4574,
+      "step": 141450
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020231931989230217,
+      "loss": 0.382,
+      "step": 141460
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0002022769191630096,
+      "loss": 0.4455,
+      "step": 141470
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020223451843371705,
+      "loss": 0.5587,
+      "step": 141480
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020219211770442454,
+      "loss": 0.5636,
+      "step": 141490
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020214971697513198,
+      "loss": 0.4108,
+      "step": 141500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020210731624583942,
+      "loss": 0.4447,
+      "step": 141510
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020206491551654691,
+      "loss": 0.5121,
+      "step": 141520
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020202251478725435,
+      "loss": 0.3978,
+      "step": 141530
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002019801140579618,
+      "loss": 0.4973,
+      "step": 141540
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020193771332866923,
+      "loss": 0.4394,
+      "step": 141550
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020189531259937673,
+      "loss": 0.4478,
+      "step": 141560
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020185291187008417,
+      "loss": 0.4161,
+      "step": 141570
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002018105111407916,
+      "loss": 0.3954,
+      "step": 141580
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002017681104114991,
+      "loss": 0.4452,
+      "step": 141590
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020172570968220654,
+      "loss": 0.4731,
+      "step": 141600
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020168330895291398,
+      "loss": 0.4756,
+      "step": 141610
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020164090822362147,
+      "loss": 0.4408,
+      "step": 141620
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002015985074943289,
+      "loss": 0.443,
+      "step": 141630
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020155610676503635,
+      "loss": 0.4205,
+      "step": 141640
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020151370603574385,
+      "loss": 0.4268,
+      "step": 141650
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020147130530645129,
+      "loss": 0.3971,
+      "step": 141660
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020142890457715873,
+      "loss": 0.5806,
+      "step": 141670
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020138650384786622,
+      "loss": 0.5053,
+      "step": 141680
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020134410311857366,
+      "loss": 0.4947,
+      "step": 141690
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002013017023892811,
+      "loss": 0.4715,
+      "step": 141700
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020125930165998854,
+      "loss": 0.5126,
+      "step": 141710
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020121690093069603,
+      "loss": 0.4659,
+      "step": 141720
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020117450020140347,
+      "loss": 0.5113,
+      "step": 141730
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002011320994721109,
+      "loss": 0.4932,
+      "step": 141740
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002010896987428184,
+      "loss": 0.4056,
+      "step": 141750
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020104729801352585,
+      "loss": 0.5303,
+      "step": 141760
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020100489728423329,
+      "loss": 0.4841,
+      "step": 141770
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020096249655494078,
+      "loss": 0.5178,
+      "step": 141780
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020092009582564822,
+      "loss": 0.4291,
+      "step": 141790
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020087769509635566,
+      "loss": 0.483,
+      "step": 141800
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020083529436706315,
+      "loss": 0.5078,
+      "step": 141810
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0002007928936377706,
+      "loss": 0.5337,
+      "step": 141820
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020075049290847803,
+      "loss": 0.442,
+      "step": 141830
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020070809217918547,
+      "loss": 0.4391,
+      "step": 141840
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020066569144989297,
+      "loss": 0.5012,
+      "step": 141850
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0002006232907206004,
+      "loss": 0.5113,
+      "step": 141860
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020058088999130784,
+      "loss": 0.4494,
+      "step": 141870
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020053848926201534,
+      "loss": 0.4982,
+      "step": 141880
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020049608853272278,
+      "loss": 0.3851,
+      "step": 141890
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020045368780343022,
+      "loss": 0.3998,
+      "step": 141900
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0002004112870741377,
+      "loss": 0.4652,
+      "step": 141910
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020036888634484515,
+      "loss": 0.4644,
+      "step": 141920
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0002003264856155526,
+      "loss": 0.4457,
+      "step": 141930
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020028408488626006,
+      "loss": 0.4183,
+      "step": 141940
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020024168415696752,
+      "loss": 0.3659,
+      "step": 141950
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020019928342767496,
+      "loss": 0.408,
+      "step": 141960
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020015688269838243,
+      "loss": 0.3959,
+      "step": 141970
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020011448196908987,
+      "loss": 0.3814,
+      "step": 141980
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020007208123979734,
+      "loss": 0.3989,
+      "step": 141990
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020002968051050478,
+      "loss": 0.3814,
+      "step": 142000
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019998727978121224,
+      "loss": 0.417,
+      "step": 142010
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001999448790519197,
+      "loss": 0.3782,
+      "step": 142020
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019990247832262715,
+      "loss": 0.4113,
+      "step": 142030
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019986007759333462,
+      "loss": 0.4304,
+      "step": 142040
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019981767686404206,
+      "loss": 0.3995,
+      "step": 142050
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019977527613474952,
+      "loss": 0.3925,
+      "step": 142060
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019973287540545696,
+      "loss": 0.3916,
+      "step": 142070
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019969047467616443,
+      "loss": 0.3546,
+      "step": 142080
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001996480739468719,
+      "loss": 0.4423,
+      "step": 142090
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019960567321757934,
+      "loss": 0.4288,
+      "step": 142100
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001995632724882868,
+      "loss": 0.4947,
+      "step": 142110
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019952087175899424,
+      "loss": 0.3974,
+      "step": 142120
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001994784710297017,
+      "loss": 0.4346,
+      "step": 142130
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019943607030040918,
+      "loss": 0.4068,
+      "step": 142140
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019939366957111662,
+      "loss": 0.4233,
+      "step": 142150
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019935126884182408,
+      "loss": 0.4621,
+      "step": 142160
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019930886811253155,
+      "loss": 0.415,
+      "step": 142170
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.000199266467383239,
+      "loss": 0.4705,
+      "step": 142180
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019922406665394646,
+      "loss": 0.3809,
+      "step": 142190
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001991816659246539,
+      "loss": 0.3868,
+      "step": 142200
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019913926519536136,
+      "loss": 0.4102,
+      "step": 142210
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019909686446606883,
+      "loss": 0.4405,
+      "step": 142220
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019905446373677627,
+      "loss": 0.3791,
+      "step": 142230
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019901206300748374,
+      "loss": 0.3668,
+      "step": 142240
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001989696622781912,
+      "loss": 0.3944,
+      "step": 142250
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019892726154889864,
+      "loss": 0.44,
+      "step": 142260
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001988848608196061,
+      "loss": 0.3905,
+      "step": 142270
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019884246009031355,
+      "loss": 0.3644,
+      "step": 142280
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019880005936102102,
+      "loss": 0.4325,
+      "step": 142290
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019875765863172848,
+      "loss": 0.447,
+      "step": 142300
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019871525790243592,
+      "loss": 0.4266,
+      "step": 142310
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001986728571731434,
+      "loss": 0.3616,
+      "step": 142320
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019863045644385086,
+      "loss": 0.3751,
+      "step": 142330
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001985880557145583,
+      "loss": 0.424,
+      "step": 142340
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019854565498526576,
+      "loss": 0.4086,
+      "step": 142350
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001985032542559732,
+      "loss": 0.324,
+      "step": 142360
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019846085352668067,
+      "loss": 0.5179,
+      "step": 142370
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019841845279738814,
+      "loss": 0.376,
+      "step": 142380
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019837605206809558,
+      "loss": 0.4242,
+      "step": 142390
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019833365133880304,
+      "loss": 0.4544,
+      "step": 142400
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001982912506095105,
+      "loss": 0.313,
+      "step": 142410
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019824884988021795,
+      "loss": 0.4081,
+      "step": 142420
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019820644915092542,
+      "loss": 0.3852,
+      "step": 142430
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019816404842163286,
+      "loss": 0.3994,
+      "step": 142440
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019812164769234032,
+      "loss": 0.3987,
+      "step": 142450
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001980792469630478,
+      "loss": 0.3493,
+      "step": 142460
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019803684623375523,
+      "loss": 0.4797,
+      "step": 142470
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001979944455044627,
+      "loss": 0.4488,
+      "step": 142480
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019795204477517016,
+      "loss": 0.4633,
+      "step": 142490
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001979096440458776,
+      "loss": 0.3701,
+      "step": 142500
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019786724331658507,
+      "loss": 0.337,
+      "step": 142510
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001978248425872925,
+      "loss": 0.3821,
+      "step": 142520
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019778244185799998,
+      "loss": 0.4256,
+      "step": 142530
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019774004112870744,
+      "loss": 0.4003,
+      "step": 142540
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019769764039941488,
+      "loss": 0.4219,
+      "step": 142550
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019765523967012235,
+      "loss": 0.4388,
+      "step": 142560
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001976128389408298,
+      "loss": 0.4373,
+      "step": 142570
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019757043821153725,
+      "loss": 0.4111,
+      "step": 142580
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019752803748224472,
+      "loss": 0.3655,
+      "step": 142590
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019748563675295216,
+      "loss": 0.3782,
+      "step": 142600
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019744323602365963,
+      "loss": 0.4445,
+      "step": 142610
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019740083529436707,
+      "loss": 0.4083,
+      "step": 142620
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019735843456507453,
+      "loss": 0.4513,
+      "step": 142630
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.000197316033835782,
+      "loss": 0.4055,
+      "step": 142640
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019727363310648944,
+      "loss": 0.3697,
+      "step": 142650
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001972312323771969,
+      "loss": 0.35,
+      "step": 142660
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019718883164790435,
+      "loss": 0.3951,
+      "step": 142670
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019714643091861181,
+      "loss": 0.3938,
+      "step": 142680
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019710403018931925,
+      "loss": 0.3893,
+      "step": 142690
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019706162946002672,
+      "loss": 0.409,
+      "step": 142700
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019701922873073416,
+      "loss": 0.3647,
+      "step": 142710
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019697682800144163,
+      "loss": 0.3969,
+      "step": 142720
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019693442727214907,
+      "loss": 0.3918,
+      "step": 142730
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019689202654285653,
+      "loss": 0.453,
+      "step": 142740
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.000196849625813564,
+      "loss": 0.4115,
+      "step": 142750
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019680722508427144,
+      "loss": 0.4046,
+      "step": 142760
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001967648243549789,
+      "loss": 0.3763,
+      "step": 142770
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019672242362568637,
+      "loss": 0.3966,
+      "step": 142780
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001966800228963938,
+      "loss": 0.3863,
+      "step": 142790
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019663762216710128,
+      "loss": 0.3506,
+      "step": 142800
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019659522143780872,
+      "loss": 0.3985,
+      "step": 142810
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001965528207085162,
+      "loss": 0.3652,
+      "step": 142820
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019651041997922365,
+      "loss": 0.4266,
+      "step": 142830
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001964680192499311,
+      "loss": 0.4275,
+      "step": 142840
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019642561852063856,
+      "loss": 0.385,
+      "step": 142850
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019638321779134603,
+      "loss": 0.4651,
+      "step": 142860
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019634081706205347,
+      "loss": 0.4654,
+      "step": 142870
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019629841633276093,
+      "loss": 0.3457,
+      "step": 142880
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019625601560346837,
+      "loss": 0.3197,
+      "step": 142890
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019621361487417584,
+      "loss": 0.3503,
+      "step": 142900
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001961712141448833,
+      "loss": 0.3799,
+      "step": 142910
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019612881341559075,
+      "loss": 0.3325,
+      "step": 142920
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001960864126862982,
+      "loss": 0.4206,
+      "step": 142930
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019604401195700568,
+      "loss": 0.4224,
+      "step": 142940
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019600161122771312,
+      "loss": 0.4441,
+      "step": 142950
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019595921049842059,
+      "loss": 0.3402,
+      "step": 142960
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019591680976912803,
+      "loss": 0.383,
+      "step": 142970
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001958744090398355,
+      "loss": 0.3584,
+      "step": 142980
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019583200831054296,
+      "loss": 0.4178,
+      "step": 142990
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001957896075812504,
+      "loss": 0.3716,
+      "step": 143000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019574720685195787,
+      "loss": 0.4914,
+      "step": 143010
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019570480612266533,
+      "loss": 0.406,
+      "step": 143020
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019566240539337277,
+      "loss": 0.4118,
+      "step": 143030
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019562000466408024,
+      "loss": 0.4163,
+      "step": 143040
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019557760393478768,
+      "loss": 0.3787,
+      "step": 143050
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019553520320549515,
+      "loss": 0.49,
+      "step": 143060
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001954928024762026,
+      "loss": 0.4417,
+      "step": 143070
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019545040174691005,
+      "loss": 0.425,
+      "step": 143080
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019540800101761752,
+      "loss": 0.374,
+      "step": 143090
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019536560028832499,
+      "loss": 0.4207,
+      "step": 143100
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019532319955903243,
+      "loss": 0.4257,
+      "step": 143110
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001952807988297399,
+      "loss": 0.3861,
+      "step": 143120
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019523839810044733,
+      "loss": 0.3486,
+      "step": 143130
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001951959973711548,
+      "loss": 0.4175,
+      "step": 143140
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019515359664186227,
+      "loss": 0.4193,
+      "step": 143150
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001951111959125697,
+      "loss": 0.4292,
+      "step": 143160
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019506879518327717,
+      "loss": 0.3514,
+      "step": 143170
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019502639445398464,
+      "loss": 0.3567,
+      "step": 143180
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019498399372469208,
+      "loss": 0.4268,
+      "step": 143190
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019494159299539955,
+      "loss": 0.3809,
+      "step": 143200
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019489919226610698,
+      "loss": 0.3889,
+      "step": 143210
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019485679153681445,
+      "loss": 0.4764,
+      "step": 143220
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019481439080752192,
+      "loss": 0.3665,
+      "step": 143230
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019477199007822936,
+      "loss": 0.3566,
+      "step": 143240
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019472958934893682,
+      "loss": 0.3788,
+      "step": 143250
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019468718861964426,
+      "loss": 0.4452,
+      "step": 143260
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019464478789035173,
+      "loss": 0.3491,
+      "step": 143270
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0001946023871610592,
+      "loss": 0.4318,
+      "step": 143280
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019455998643176664,
+      "loss": 0.3821,
+      "step": 143290
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0001945175857024741,
+      "loss": 0.3312,
+      "step": 143300
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019447518497318154,
+      "loss": 0.3606,
+      "step": 143310
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.000194432784243889,
+      "loss": 0.3916,
+      "step": 143320
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019439038351459645,
+      "loss": 0.3984,
+      "step": 143330
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019434798278530392,
+      "loss": 0.4117,
+      "step": 143340
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019430558205601136,
+      "loss": 0.3753,
+      "step": 143350
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019426318132671882,
+      "loss": 0.465,
+      "step": 143360
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019422078059742626,
+      "loss": 0.4029,
+      "step": 143370
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019417837986813373,
+      "loss": 0.4144,
+      "step": 143380
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0001941359791388412,
+      "loss": 0.3753,
+      "step": 143390
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019409357840954864,
+      "loss": 0.355,
+      "step": 143400
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0001940511776802561,
+      "loss": 0.3914,
+      "step": 143410
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019400877695096354,
+      "loss": 0.4783,
+      "step": 143420
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.000193966376221671,
+      "loss": 0.3892,
+      "step": 143430
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019392397549237848,
+      "loss": 0.4223,
+      "step": 143440
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019388157476308592,
+      "loss": 0.4712,
+      "step": 143450
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019383917403379338,
+      "loss": 0.3834,
+      "step": 143460
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019379677330450085,
+      "loss": 0.4228,
+      "step": 143470
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001937543725752083,
+      "loss": 0.4347,
+      "step": 143480
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019371197184591576,
+      "loss": 0.395,
+      "step": 143490
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001936695711166232,
+      "loss": 0.4071,
+      "step": 143500
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019362717038733066,
+      "loss": 0.4367,
+      "step": 143510
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019358476965803813,
+      "loss": 0.3883,
+      "step": 143520
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019354236892874557,
+      "loss": 0.4616,
+      "step": 143530
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019349996819945304,
+      "loss": 0.393,
+      "step": 143540
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001934575674701605,
+      "loss": 0.3386,
+      "step": 143550
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019341516674086794,
+      "loss": 0.45,
+      "step": 143560
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001933727660115754,
+      "loss": 0.478,
+      "step": 143570
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019333036528228285,
+      "loss": 0.3925,
+      "step": 143580
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019328796455299032,
+      "loss": 0.3823,
+      "step": 143590
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019324556382369778,
+      "loss": 0.3828,
+      "step": 143600
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019320316309440522,
+      "loss": 0.4873,
+      "step": 143610
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001931607623651127,
+      "loss": 0.4831,
+      "step": 143620
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019311836163582016,
+      "loss": 0.3488,
+      "step": 143630
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001930759609065276,
+      "loss": 0.3522,
+      "step": 143640
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019303356017723506,
+      "loss": 0.4414,
+      "step": 143650
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001929911594479425,
+      "loss": 0.4216,
+      "step": 143660
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019294875871864997,
+      "loss": 0.4154,
+      "step": 143670
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019290635798935744,
+      "loss": 0.4816,
+      "step": 143680
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019286395726006488,
+      "loss": 0.4875,
+      "step": 143690
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019282155653077234,
+      "loss": 0.4278,
+      "step": 143700
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001927791558014798,
+      "loss": 0.4589,
+      "step": 143710
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019273675507218725,
+      "loss": 0.4597,
+      "step": 143720
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019269435434289472,
+      "loss": 0.3599,
+      "step": 143730
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019265195361360216,
+      "loss": 0.3937,
+      "step": 143740
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019260955288430962,
+      "loss": 0.3842,
+      "step": 143750
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001925671521550171,
+      "loss": 0.3533,
+      "step": 143760
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019252475142572453,
+      "loss": 0.4035,
+      "step": 143770
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.000192482350696432,
+      "loss": 0.377,
+      "step": 143780
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019243994996713946,
+      "loss": 0.4049,
+      "step": 143790
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001923975492378469,
+      "loss": 0.3367,
+      "step": 143800
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019235514850855437,
+      "loss": 0.3727,
+      "step": 143810
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001923127477792618,
+      "loss": 0.4536,
+      "step": 143820
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019227034704996928,
+      "loss": 0.4978,
+      "step": 143830
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019222794632067674,
+      "loss": 0.3582,
+      "step": 143840
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019218554559138418,
+      "loss": 0.3485,
+      "step": 143850
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019214314486209165,
+      "loss": 0.4139,
+      "step": 143860
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019210074413279912,
+      "loss": 0.4689,
+      "step": 143870
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019205834340350656,
+      "loss": 0.4615,
+      "step": 143880
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019201594267421402,
+      "loss": 0.3773,
+      "step": 143890
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019197354194492146,
+      "loss": 0.4557,
+      "step": 143900
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019193114121562893,
+      "loss": 0.4496,
+      "step": 143910
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001918887404863364,
+      "loss": 0.3796,
+      "step": 143920
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019184633975704383,
+      "loss": 0.4902,
+      "step": 143930
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001918039390277513,
+      "loss": 0.4156,
+      "step": 143940
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019176153829845874,
+      "loss": 0.4085,
+      "step": 143950
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001917191375691662,
+      "loss": 0.3926,
+      "step": 143960
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019167673683987365,
+      "loss": 0.4005,
+      "step": 143970
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019163433611058111,
+      "loss": 0.3337,
+      "step": 143980
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019159193538128855,
+      "loss": 0.4086,
+      "step": 143990
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019154953465199602,
+      "loss": 0.3994,
+      "step": 144000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001915071339227035,
+      "loss": 0.4026,
+      "step": 144010
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019146473319341093,
+      "loss": 0.3769,
+      "step": 144020
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001914223324641184,
+      "loss": 0.4473,
+      "step": 144030
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019137993173482583,
+      "loss": 0.4743,
+      "step": 144040
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001913375310055333,
+      "loss": 0.4121,
+      "step": 144050
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019129513027624074,
+      "loss": 0.395,
+      "step": 144060
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001912527295469482,
+      "loss": 0.4219,
+      "step": 144070
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019121032881765567,
+      "loss": 0.3769,
+      "step": 144080
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019116792808836311,
+      "loss": 0.4029,
+      "step": 144090
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019112552735907058,
+      "loss": 0.4074,
+      "step": 144100
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019108312662977802,
+      "loss": 0.3609,
+      "step": 144110
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001910407259004855,
+      "loss": 0.4157,
+      "step": 144120
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019099832517119295,
+      "loss": 0.3623,
+      "step": 144130
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0001909559244419004,
+      "loss": 0.3637,
+      "step": 144140
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019091352371260786,
+      "loss": 0.4978,
+      "step": 144150
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019087112298331533,
+      "loss": 0.4247,
+      "step": 144160
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019082872225402277,
+      "loss": 0.4548,
+      "step": 144170
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019078632152473023,
+      "loss": 0.4758,
+      "step": 144180
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019074392079543767,
+      "loss": 0.4737,
+      "step": 144190
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019070152006614514,
+      "loss": 0.3734,
+      "step": 144200
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0001906591193368526,
+      "loss": 0.4585,
+      "step": 144210
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019061671860756005,
+      "loss": 0.4382,
+      "step": 144220
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0001905743178782675,
+      "loss": 0.3986,
+      "step": 144230
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019053191714897498,
+      "loss": 0.408,
+      "step": 144240
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019048951641968242,
+      "loss": 0.4543,
+      "step": 144250
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0001904471156903899,
+      "loss": 0.4397,
+      "step": 144260
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019040471496109733,
+      "loss": 0.4121,
+      "step": 144270
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0001903623142318048,
+      "loss": 0.4448,
+      "step": 144280
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019031991350251226,
+      "loss": 0.3833,
+      "step": 144290
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0001902775127732197,
+      "loss": 0.3725,
+      "step": 144300
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019023511204392717,
+      "loss": 0.3544,
+      "step": 144310
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019019271131463463,
+      "loss": 0.3441,
+      "step": 144320
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019015031058534207,
+      "loss": 0.3425,
+      "step": 144330
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019010790985604954,
+      "loss": 0.4337,
+      "step": 144340
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019006550912675698,
+      "loss": 0.3828,
+      "step": 144350
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019002310839746445,
+      "loss": 0.4379,
+      "step": 144360
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001899807076681719,
+      "loss": 0.4271,
+      "step": 144370
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018993830693887935,
+      "loss": 0.3987,
+      "step": 144380
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018989590620958682,
+      "loss": 0.4131,
+      "step": 144390
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018985350548029429,
+      "loss": 0.3887,
+      "step": 144400
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018981110475100173,
+      "loss": 0.4537,
+      "step": 144410
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001897687040217092,
+      "loss": 0.4587,
+      "step": 144420
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018972630329241663,
+      "loss": 0.3811,
+      "step": 144430
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001896839025631241,
+      "loss": 0.4048,
+      "step": 144440
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018964150183383157,
+      "loss": 0.3693,
+      "step": 144450
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.000189599101104539,
+      "loss": 0.3903,
+      "step": 144460
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018955670037524647,
+      "loss": 0.3941,
+      "step": 144470
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018951429964595394,
+      "loss": 0.5154,
+      "step": 144480
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018947189891666138,
+      "loss": 0.3756,
+      "step": 144490
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018942949818736885,
+      "loss": 0.4005,
+      "step": 144500
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018938709745807629,
+      "loss": 0.4222,
+      "step": 144510
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018934469672878375,
+      "loss": 0.4088,
+      "step": 144520
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018930229599949122,
+      "loss": 0.3773,
+      "step": 144530
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018925989527019866,
+      "loss": 0.36,
+      "step": 144540
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018921749454090613,
+      "loss": 0.3481,
+      "step": 144550
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001891750938116136,
+      "loss": 0.4042,
+      "step": 144560
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018913269308232103,
+      "loss": 0.419,
+      "step": 144570
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001890902923530285,
+      "loss": 0.412,
+      "step": 144580
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018904789162373594,
+      "loss": 0.413,
+      "step": 144590
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001890054908944434,
+      "loss": 0.3914,
+      "step": 144600
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018896309016515084,
+      "loss": 0.457,
+      "step": 144610
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001889206894358583,
+      "loss": 0.4394,
+      "step": 144620
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018887828870656578,
+      "loss": 0.4843,
+      "step": 144630
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018883588797727322,
+      "loss": 0.3387,
+      "step": 144640
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018879348724798068,
+      "loss": 0.3744,
+      "step": 144650
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018875108651868812,
+      "loss": 0.4289,
+      "step": 144660
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001887086857893956,
+      "loss": 0.3899,
+      "step": 144670
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018866628506010303,
+      "loss": 0.4657,
+      "step": 144680
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001886238843308105,
+      "loss": 0.4419,
+      "step": 144690
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018858148360151794,
+      "loss": 0.4014,
+      "step": 144700
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001885390828722254,
+      "loss": 0.4231,
+      "step": 144710
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018849668214293284,
+      "loss": 0.372,
+      "step": 144720
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001884542814136403,
+      "loss": 0.4059,
+      "step": 144730
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018841188068434778,
+      "loss": 0.3875,
+      "step": 144740
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018836947995505522,
+      "loss": 0.3717,
+      "step": 144750
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018832707922576268,
+      "loss": 0.4321,
+      "step": 144760
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018828467849647015,
+      "loss": 0.3869,
+      "step": 144770
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001882422777671776,
+      "loss": 0.4149,
+      "step": 144780
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018819987703788506,
+      "loss": 0.4253,
+      "step": 144790
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001881574763085925,
+      "loss": 0.376,
+      "step": 144800
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018811507557929996,
+      "loss": 0.3834,
+      "step": 144810
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018807267485000743,
+      "loss": 0.3722,
+      "step": 144820
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018803027412071487,
+      "loss": 0.4209,
+      "step": 144830
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018798787339142234,
+      "loss": 0.4268,
+      "step": 144840
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001879454726621298,
+      "loss": 0.4235,
+      "step": 144850
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018790307193283724,
+      "loss": 0.3711,
+      "step": 144860
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001878606712035447,
+      "loss": 0.3582,
+      "step": 144870
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018781827047425215,
+      "loss": 0.4053,
+      "step": 144880
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018777586974495962,
+      "loss": 0.4469,
+      "step": 144890
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018773346901566708,
+      "loss": 0.4452,
+      "step": 144900
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018769106828637452,
+      "loss": 0.4117,
+      "step": 144910
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.000187648667557082,
+      "loss": 0.4189,
+      "step": 144920
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018760626682778946,
+      "loss": 0.3463,
+      "step": 144930
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001875638660984969,
+      "loss": 0.3739,
+      "step": 144940
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018752146536920436,
+      "loss": 0.4439,
+      "step": 144950
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001874790646399118,
+      "loss": 0.4493,
+      "step": 144960
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018743666391061927,
+      "loss": 0.4878,
+      "step": 144970
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018739426318132674,
+      "loss": 0.4139,
+      "step": 144980
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018735186245203418,
+      "loss": 0.3811,
+      "step": 144990
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018730946172274164,
+      "loss": 0.3376,
+      "step": 145000
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001872670609934491,
+      "loss": 0.4493,
+      "step": 145010
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018722466026415655,
+      "loss": 0.4143,
+      "step": 145020
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018718225953486402,
+      "loss": 0.4372,
+      "step": 145030
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018713985880557146,
+      "loss": 0.3637,
+      "step": 145040
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018709745807627892,
+      "loss": 0.3817,
+      "step": 145050
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001870550573469864,
+      "loss": 0.4344,
+      "step": 145060
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018701265661769383,
+      "loss": 0.3871,
+      "step": 145070
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001869702558884013,
+      "loss": 0.4058,
+      "step": 145080
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018692785515910876,
+      "loss": 0.3962,
+      "step": 145090
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001868854544298162,
+      "loss": 0.4249,
+      "step": 145100
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018684305370052367,
+      "loss": 0.4257,
+      "step": 145110
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001868006529712311,
+      "loss": 0.366,
+      "step": 145120
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018675825224193858,
+      "loss": 0.38,
+      "step": 145130
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018671585151264604,
+      "loss": 0.3706,
+      "step": 145140
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018667345078335348,
+      "loss": 0.4063,
+      "step": 145150
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018663105005406095,
+      "loss": 0.354,
+      "step": 145160
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018658864932476842,
+      "loss": 0.4309,
+      "step": 145170
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018654624859547586,
+      "loss": 0.4472,
+      "step": 145180
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018650384786618332,
+      "loss": 0.3967,
+      "step": 145190
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018646144713689076,
+      "loss": 0.4212,
+      "step": 145200
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018641904640759823,
+      "loss": 0.4234,
+      "step": 145210
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001863766456783057,
+      "loss": 0.3429,
+      "step": 145220
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018633424494901314,
+      "loss": 0.3706,
+      "step": 145230
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001862918442197206,
+      "loss": 0.3842,
+      "step": 145240
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018624944349042804,
+      "loss": 0.4368,
+      "step": 145250
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001862070427611355,
+      "loss": 0.445,
+      "step": 145260
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018616464203184298,
+      "loss": 0.4927,
+      "step": 145270
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018612224130255041,
+      "loss": 0.4809,
+      "step": 145280
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018607984057325788,
+      "loss": 0.4594,
+      "step": 145290
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018603743984396532,
+      "loss": 0.4302,
+      "step": 145300
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001859950391146728,
+      "loss": 0.4424,
+      "step": 145310
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018595263838538023,
+      "loss": 0.4369,
+      "step": 145320
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001859102376560877,
+      "loss": 0.3561,
+      "step": 145330
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018586783692679513,
+      "loss": 0.436,
+      "step": 145340
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001858254361975026,
+      "loss": 0.4107,
+      "step": 145350
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018578303546821004,
+      "loss": 0.4635,
+      "step": 145360
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001857406347389175,
+      "loss": 0.3527,
+      "step": 145370
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018569823400962497,
+      "loss": 0.4151,
+      "step": 145380
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018565583328033241,
+      "loss": 0.3704,
+      "step": 145390
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018561343255103988,
+      "loss": 0.3829,
+      "step": 145400
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018557103182174732,
+      "loss": 0.4721,
+      "step": 145410
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001855286310924548,
+      "loss": 0.3691,
+      "step": 145420
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018548623036316225,
+      "loss": 0.4487,
+      "step": 145430
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001854438296338697,
+      "loss": 0.466,
+      "step": 145440
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018540142890457716,
+      "loss": 0.3853,
+      "step": 145450
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018535902817528463,
+      "loss": 0.3458,
+      "step": 145460
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018531662744599207,
+      "loss": 0.3893,
+      "step": 145470
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018527422671669953,
+      "loss": 0.4818,
+      "step": 145480
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018523182598740697,
+      "loss": 0.4613,
+      "step": 145490
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018518942525811444,
+      "loss": 0.3547,
+      "step": 145500
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001851470245288219,
+      "loss": 0.4367,
+      "step": 145510
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018510462379952935,
+      "loss": 0.4658,
+      "step": 145520
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001850622230702368,
+      "loss": 0.4301,
+      "step": 145530
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018501982234094428,
+      "loss": 0.3794,
+      "step": 145540
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018497742161165172,
+      "loss": 0.4776,
+      "step": 145550
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001849350208823592,
+      "loss": 0.431,
+      "step": 145560
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018489262015306663,
+      "loss": 0.3359,
+      "step": 145570
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001848502194237741,
+      "loss": 0.3601,
+      "step": 145580
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018480781869448156,
+      "loss": 0.4329,
+      "step": 145590
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.000184765417965189,
+      "loss": 0.4361,
+      "step": 145600
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018472301723589647,
+      "loss": 0.4023,
+      "step": 145610
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018468061650660393,
+      "loss": 0.4225,
+      "step": 145620
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018463821577731137,
+      "loss": 0.3832,
+      "step": 145630
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018459581504801884,
+      "loss": 0.4427,
+      "step": 145640
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018455341431872628,
+      "loss": 0.4582,
+      "step": 145650
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018451101358943375,
+      "loss": 0.514,
+      "step": 145660
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001844686128601412,
+      "loss": 0.3773,
+      "step": 145670
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018442621213084865,
+      "loss": 0.3924,
+      "step": 145680
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018438381140155612,
+      "loss": 0.394,
+      "step": 145690
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018434141067226359,
+      "loss": 0.4187,
+      "step": 145700
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018429900994297103,
+      "loss": 0.4234,
+      "step": 145710
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001842566092136785,
+      "loss": 0.3529,
+      "step": 145720
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018421420848438593,
+      "loss": 0.4189,
+      "step": 145730
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001841718077550934,
+      "loss": 0.4345,
+      "step": 145740
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018412940702580087,
+      "loss": 0.3774,
+      "step": 145750
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001840870062965083,
+      "loss": 0.454,
+      "step": 145760
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018404460556721577,
+      "loss": 0.418,
+      "step": 145770
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018400220483792324,
+      "loss": 0.4786,
+      "step": 145780
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018395980410863068,
+      "loss": 0.399,
+      "step": 145790
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018391740337933815,
+      "loss": 0.4947,
+      "step": 145800
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018387500265004559,
+      "loss": 0.4316,
+      "step": 145810
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018383260192075305,
+      "loss": 0.4373,
+      "step": 145820
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018379020119146052,
+      "loss": 0.4448,
+      "step": 145830
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018374780046216796,
+      "loss": 0.4345,
+      "step": 145840
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018370539973287543,
+      "loss": 0.4362,
+      "step": 145850
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001836629990035829,
+      "loss": 0.3256,
+      "step": 145860
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018362059827429033,
+      "loss": 0.461,
+      "step": 145870
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001835781975449978,
+      "loss": 0.4742,
+      "step": 145880
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018353579681570524,
+      "loss": 0.4306,
+      "step": 145890
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001834933960864127,
+      "loss": 0.4,
+      "step": 145900
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018345099535712017,
+      "loss": 0.4743,
+      "step": 145910
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001834085946278276,
+      "loss": 0.3929,
+      "step": 145920
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018336619389853508,
+      "loss": 0.3555,
+      "step": 145930
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018332379316924252,
+      "loss": 0.4241,
+      "step": 145940
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018328139243994998,
+      "loss": 0.4161,
+      "step": 145950
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018323899171065742,
+      "loss": 0.4161,
+      "step": 145960
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001831965909813649,
+      "loss": 0.4022,
+      "step": 145970
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018315419025207233,
+      "loss": 0.4496,
+      "step": 145980
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001831117895227798,
+      "loss": 0.4749,
+      "step": 145990
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018306938879348724,
+      "loss": 0.4791,
+      "step": 146000
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001830269880641947,
+      "loss": 0.4379,
+      "step": 146010
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018298458733490217,
+      "loss": 0.4554,
+      "step": 146020
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001829421866056096,
+      "loss": 0.3981,
+      "step": 146030
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018289978587631708,
+      "loss": 0.3735,
+      "step": 146040
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018285738514702452,
+      "loss": 0.479,
+      "step": 146050
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018281498441773198,
+      "loss": 0.4228,
+      "step": 146060
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018277258368843945,
+      "loss": 0.4178,
+      "step": 146070
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001827301829591469,
+      "loss": 0.3783,
+      "step": 146080
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018268778222985436,
+      "loss": 0.4075,
+      "step": 146090
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001826453815005618,
+      "loss": 0.384,
+      "step": 146100
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018260298077126926,
+      "loss": 0.3865,
+      "step": 146110
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018256058004197673,
+      "loss": 0.4648,
+      "step": 146120
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018251817931268417,
+      "loss": 0.4227,
+      "step": 146130
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018247577858339164,
+      "loss": 0.5446,
+      "step": 146140
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001824333778540991,
+      "loss": 0.4242,
+      "step": 146150
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018239097712480654,
+      "loss": 0.3752,
+      "step": 146160
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.000182348576395514,
+      "loss": 0.419,
+      "step": 146170
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018230617566622145,
+      "loss": 0.378,
+      "step": 146180
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018226377493692892,
+      "loss": 0.3827,
+      "step": 146190
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018222137420763638,
+      "loss": 0.4113,
+      "step": 146200
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018217897347834382,
+      "loss": 0.4134,
+      "step": 146210
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001821365727490513,
+      "loss": 0.3815,
+      "step": 146220
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018209417201975876,
+      "loss": 0.3556,
+      "step": 146230
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001820517712904662,
+      "loss": 0.3789,
+      "step": 146240
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018200937056117366,
+      "loss": 0.3808,
+      "step": 146250
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001819669698318811,
+      "loss": 0.4059,
+      "step": 146260
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018192456910258857,
+      "loss": 0.4246,
+      "step": 146270
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018188216837329604,
+      "loss": 0.405,
+      "step": 146280
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018183976764400348,
+      "loss": 0.4559,
+      "step": 146290
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018179736691471094,
+      "loss": 0.4096,
+      "step": 146300
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001817549661854184,
+      "loss": 0.3847,
+      "step": 146310
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018171256545612585,
+      "loss": 0.412,
+      "step": 146320
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018167016472683332,
+      "loss": 0.4103,
+      "step": 146330
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018162776399754076,
+      "loss": 0.4903,
+      "step": 146340
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018158536326824822,
+      "loss": 0.4608,
+      "step": 146350
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001815429625389557,
+      "loss": 0.3919,
+      "step": 146360
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018150056180966313,
+      "loss": 0.4385,
+      "step": 146370
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001814581610803706,
+      "loss": 0.4011,
+      "step": 146380
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018141576035107806,
+      "loss": 0.3867,
+      "step": 146390
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001813733596217855,
+      "loss": 0.4377,
+      "step": 146400
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018133095889249297,
+      "loss": 0.4357,
+      "step": 146410
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001812885581632004,
+      "loss": 0.4293,
+      "step": 146420
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018124615743390788,
+      "loss": 0.4687,
+      "step": 146430
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018120375670461534,
+      "loss": 0.3649,
+      "step": 146440
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018116135597532278,
+      "loss": 0.3803,
+      "step": 146450
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018111895524603025,
+      "loss": 0.3431,
+      "step": 146460
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018107655451673772,
+      "loss": 0.3762,
+      "step": 146470
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018103415378744516,
+      "loss": 0.4243,
+      "step": 146480
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018099175305815262,
+      "loss": 0.451,
+      "step": 146490
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018094935232886006,
+      "loss": 0.468,
+      "step": 146500
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018090695159956753,
+      "loss": 0.4076,
+      "step": 146510
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.000180864550870275,
+      "loss": 0.4111,
+      "step": 146520
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018082215014098244,
+      "loss": 0.4225,
+      "step": 146530
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001807797494116899,
+      "loss": 0.489,
+      "step": 146540
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018073734868239737,
+      "loss": 0.4969,
+      "step": 146550
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001806949479531048,
+      "loss": 0.4233,
+      "step": 146560
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018065254722381228,
+      "loss": 0.4511,
+      "step": 146570
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018061014649451971,
+      "loss": 0.3949,
+      "step": 146580
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018056774576522718,
+      "loss": 0.3591,
+      "step": 146590
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018052534503593462,
+      "loss": 0.4517,
+      "step": 146600
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001804829443066421,
+      "loss": 0.5151,
+      "step": 146610
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018044054357734953,
+      "loss": 0.3692,
+      "step": 146620
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.000180398142848057,
+      "loss": 0.4301,
+      "step": 146630
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018035574211876446,
+      "loss": 0.397,
+      "step": 146640
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001803133413894719,
+      "loss": 0.4201,
+      "step": 146650
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018027094066017937,
+      "loss": 0.4148,
+      "step": 146660
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001802285399308868,
+      "loss": 0.3897,
+      "step": 146670
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018018613920159427,
+      "loss": 0.44,
+      "step": 146680
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018014373847230171,
+      "loss": 0.4989,
+      "step": 146690
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018010133774300918,
+      "loss": 0.4121,
+      "step": 146700
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018005893701371665,
+      "loss": 0.3807,
+      "step": 146710
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001800165362844241,
+      "loss": 0.4286,
+      "step": 146720
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017997413555513155,
+      "loss": 0.3665,
+      "step": 146730
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.000179931734825839,
+      "loss": 0.4794,
+      "step": 146740
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017988933409654646,
+      "loss": 0.5037,
+      "step": 146750
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017984693336725393,
+      "loss": 0.4559,
+      "step": 146760
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017980453263796137,
+      "loss": 0.4002,
+      "step": 146770
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017976213190866883,
+      "loss": 0.3469,
+      "step": 146780
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017971973117937627,
+      "loss": 0.5095,
+      "step": 146790
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017967733045008374,
+      "loss": 0.4599,
+      "step": 146800
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001796349297207912,
+      "loss": 0.3891,
+      "step": 146810
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017959252899149865,
+      "loss": 0.3963,
+      "step": 146820
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017955012826220611,
+      "loss": 0.4786,
+      "step": 146830
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017950772753291358,
+      "loss": 0.3538,
+      "step": 146840
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017946532680362102,
+      "loss": 0.4592,
+      "step": 146850
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001794229260743285,
+      "loss": 0.3861,
+      "step": 146860
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017938052534503593,
+      "loss": 0.4106,
+      "step": 146870
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001793381246157434,
+      "loss": 0.4359,
+      "step": 146880
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017929572388645086,
+      "loss": 0.4042,
+      "step": 146890
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001792533231571583,
+      "loss": 0.3885,
+      "step": 146900
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017921092242786577,
+      "loss": 0.382,
+      "step": 146910
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017916852169857323,
+      "loss": 0.4324,
+      "step": 146920
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017912612096928067,
+      "loss": 0.4234,
+      "step": 146930
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017908372023998814,
+      "loss": 0.3727,
+      "step": 146940
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017904131951069558,
+      "loss": 0.4776,
+      "step": 146950
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017899891878140305,
+      "loss": 0.414,
+      "step": 146960
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0001789565180521105,
+      "loss": 0.3763,
+      "step": 146970
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017891411732281795,
+      "loss": 0.3977,
+      "step": 146980
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017887171659352542,
+      "loss": 0.5221,
+      "step": 146990
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017882931586423289,
+      "loss": 0.4471,
+      "step": 147000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017878691513494033,
+      "loss": 0.4043,
+      "step": 147010
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0001787445144056478,
+      "loss": 0.3915,
+      "step": 147020
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017870211367635523,
+      "loss": 0.4404,
+      "step": 147030
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0001786597129470627,
+      "loss": 0.3471,
+      "step": 147040
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017861731221777017,
+      "loss": 0.3604,
+      "step": 147050
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0001785749114884776,
+      "loss": 0.406,
+      "step": 147060
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017853251075918507,
+      "loss": 0.3686,
+      "step": 147070
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017849011002989254,
+      "loss": 0.3571,
+      "step": 147080
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017844770930059998,
+      "loss": 0.3328,
+      "step": 147090
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017840530857130745,
+      "loss": 0.4292,
+      "step": 147100
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017836290784201489,
+      "loss": 0.4613,
+      "step": 147110
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017832050711272235,
+      "loss": 0.4661,
+      "step": 147120
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017827810638342982,
+      "loss": 0.4361,
+      "step": 147130
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017823570565413726,
+      "loss": 0.3788,
+      "step": 147140
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017819330492484473,
+      "loss": 0.4478,
+      "step": 147150
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0001781509041955522,
+      "loss": 0.4474,
+      "step": 147160
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017810850346625963,
+      "loss": 0.383,
+      "step": 147170
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0001780661027369671,
+      "loss": 0.424,
+      "step": 147180
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017802370200767454,
+      "loss": 0.3824,
+      "step": 147190
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.000177981301278382,
+      "loss": 0.4404,
+      "step": 147200
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017793890054908947,
+      "loss": 0.4427,
+      "step": 147210
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001778964998197969,
+      "loss": 0.3679,
+      "step": 147220
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017785409909050438,
+      "loss": 0.3997,
+      "step": 147230
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017781169836121182,
+      "loss": 0.3983,
+      "step": 147240
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017776929763191929,
+      "loss": 0.4177,
+      "step": 147250
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017772689690262672,
+      "loss": 0.4338,
+      "step": 147260
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001776844961733342,
+      "loss": 0.3515,
+      "step": 147270
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017764209544404166,
+      "loss": 0.4394,
+      "step": 147280
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001775996947147491,
+      "loss": 0.3203,
+      "step": 147290
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017755729398545656,
+      "loss": 0.4528,
+      "step": 147300
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.000177514893256164,
+      "loss": 0.4052,
+      "step": 147310
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017747249252687147,
+      "loss": 0.3966,
+      "step": 147320
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001774300917975789,
+      "loss": 0.4124,
+      "step": 147330
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017738769106828638,
+      "loss": 0.4337,
+      "step": 147340
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017734529033899382,
+      "loss": 0.4638,
+      "step": 147350
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017730288960970128,
+      "loss": 0.3944,
+      "step": 147360
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017726048888040875,
+      "loss": 0.4178,
+      "step": 147370
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001772180881511162,
+      "loss": 0.4081,
+      "step": 147380
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017717568742182366,
+      "loss": 0.3966,
+      "step": 147390
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017713328669253112,
+      "loss": 0.3427,
+      "step": 147400
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017709088596323856,
+      "loss": 0.4021,
+      "step": 147410
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017704848523394603,
+      "loss": 0.3807,
+      "step": 147420
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017700608450465347,
+      "loss": 0.359,
+      "step": 147430
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017696368377536094,
+      "loss": 0.4458,
+      "step": 147440
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001769212830460684,
+      "loss": 0.4334,
+      "step": 147450
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017687888231677584,
+      "loss": 0.4663,
+      "step": 147460
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001768364815874833,
+      "loss": 0.3677,
+      "step": 147470
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017679408085819078,
+      "loss": 0.4231,
+      "step": 147480
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017675168012889822,
+      "loss": 0.4202,
+      "step": 147490
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017670927939960568,
+      "loss": 0.3875,
+      "step": 147500
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017666687867031312,
+      "loss": 0.4612,
+      "step": 147510
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001766244779410206,
+      "loss": 0.4056,
+      "step": 147520
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017658207721172806,
+      "loss": 0.407,
+      "step": 147530
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001765396764824355,
+      "loss": 0.4166,
+      "step": 147540
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017649727575314296,
+      "loss": 0.4638,
+      "step": 147550
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001764548750238504,
+      "loss": 0.447,
+      "step": 147560
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017641247429455787,
+      "loss": 0.466,
+      "step": 147570
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017637007356526534,
+      "loss": 0.437,
+      "step": 147580
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017632767283597278,
+      "loss": 0.4592,
+      "step": 147590
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017628527210668024,
+      "loss": 0.3855,
+      "step": 147600
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001762428713773877,
+      "loss": 0.3937,
+      "step": 147610
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017620047064809515,
+      "loss": 0.397,
+      "step": 147620
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017615806991880262,
+      "loss": 0.4127,
+      "step": 147630
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017611566918951006,
+      "loss": 0.4594,
+      "step": 147640
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017607326846021752,
+      "loss": 0.4462,
+      "step": 147650
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.000176030867730925,
+      "loss": 0.414,
+      "step": 147660
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017598846700163243,
+      "loss": 0.4581,
+      "step": 147670
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001759460662723399,
+      "loss": 0.4493,
+      "step": 147680
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017590366554304736,
+      "loss": 0.3802,
+      "step": 147690
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001758612648137548,
+      "loss": 0.4374,
+      "step": 147700
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017581886408446227,
+      "loss": 0.3717,
+      "step": 147710
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001757764633551697,
+      "loss": 0.4221,
+      "step": 147720
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017573406262587718,
+      "loss": 0.4211,
+      "step": 147730
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017569166189658464,
+      "loss": 0.4184,
+      "step": 147740
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017564926116729208,
+      "loss": 0.4402,
+      "step": 147750
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017560686043799955,
+      "loss": 0.3711,
+      "step": 147760
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017556445970870702,
+      "loss": 0.3952,
+      "step": 147770
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017552205897941446,
+      "loss": 0.4248,
+      "step": 147780
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017547965825012192,
+      "loss": 0.4354,
+      "step": 147790
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017543725752082936,
+      "loss": 0.4048,
+      "step": 147800
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017539485679153683,
+      "loss": 0.3771,
+      "step": 147810
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001753524560622443,
+      "loss": 0.3701,
+      "step": 147820
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017531005533295174,
+      "loss": 0.4507,
+      "step": 147830
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001752676546036592,
+      "loss": 0.3975,
+      "step": 147840
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017522525387436667,
+      "loss": 0.4382,
+      "step": 147850
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001751828531450741,
+      "loss": 0.4444,
+      "step": 147860
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017514045241578158,
+      "loss": 0.3519,
+      "step": 147870
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017509805168648902,
+      "loss": 0.4855,
+      "step": 147880
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017505565095719648,
+      "loss": 0.3703,
+      "step": 147890
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017501325022790395,
+      "loss": 0.4397,
+      "step": 147900
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001749708494986114,
+      "loss": 0.4141,
+      "step": 147910
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017492844876931886,
+      "loss": 0.3538,
+      "step": 147920
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0001748860480400263,
+      "loss": 0.4333,
+      "step": 147930
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017484364731073376,
+      "loss": 0.4269,
+      "step": 147940
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0001748012465814412,
+      "loss": 0.3769,
+      "step": 147950
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017475884585214867,
+      "loss": 0.4156,
+      "step": 147960
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0001747164451228561,
+      "loss": 0.3675,
+      "step": 147970
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017467404439356357,
+      "loss": 0.381,
+      "step": 147980
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017463164366427101,
+      "loss": 0.4458,
+      "step": 147990
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017458924293497848,
+      "loss": 0.3588,
+      "step": 148000
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017454684220568595,
+      "loss": 0.4374,
+      "step": 148010
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0001745044414763934,
+      "loss": 0.4285,
+      "step": 148020
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017446204074710085,
+      "loss": 0.3682,
+      "step": 148030
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0001744196400178083,
+      "loss": 0.4618,
+      "step": 148040
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017437723928851576,
+      "loss": 0.4327,
+      "step": 148050
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017433483855922323,
+      "loss": 0.3827,
+      "step": 148060
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017429243782993067,
+      "loss": 0.4408,
+      "step": 148070
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017425003710063813,
+      "loss": 0.353,
+      "step": 148080
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0001742076363713456,
+      "loss": 0.3618,
+      "step": 148090
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017416523564205304,
+      "loss": 0.4135,
+      "step": 148100
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0001741228349127605,
+      "loss": 0.3933,
+      "step": 148110
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017408043418346795,
+      "loss": 0.3555,
+      "step": 148120
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017403803345417541,
+      "loss": 0.4045,
+      "step": 148130
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017399563272488288,
+      "loss": 0.3698,
+      "step": 148140
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017395323199559032,
+      "loss": 0.6143,
+      "step": 148150
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001739108312662978,
+      "loss": 0.4554,
+      "step": 148160
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017386843053700525,
+      "loss": 0.4295,
+      "step": 148170
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001738260298077127,
+      "loss": 0.4077,
+      "step": 148180
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017378362907842016,
+      "loss": 0.4251,
+      "step": 148190
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001737412283491276,
+      "loss": 0.4576,
+      "step": 148200
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017369882761983507,
+      "loss": 0.4749,
+      "step": 148210
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017365642689054253,
+      "loss": 0.4167,
+      "step": 148220
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017361402616124997,
+      "loss": 0.4246,
+      "step": 148230
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017357162543195744,
+      "loss": 0.4588,
+      "step": 148240
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017352922470266488,
+      "loss": 0.4145,
+      "step": 148250
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017348682397337235,
+      "loss": 0.3726,
+      "step": 148260
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001734444232440798,
+      "loss": 0.3717,
+      "step": 148270
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017340202251478725,
+      "loss": 0.3899,
+      "step": 148280
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017335962178549472,
+      "loss": 0.3918,
+      "step": 148290
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001733172210562022,
+      "loss": 0.4908,
+      "step": 148300
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017327482032690963,
+      "loss": 0.4318,
+      "step": 148310
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001732324195976171,
+      "loss": 0.3898,
+      "step": 148320
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017319001886832453,
+      "loss": 0.4109,
+      "step": 148330
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.000173147618139032,
+      "loss": 0.4344,
+      "step": 148340
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017310521740973947,
+      "loss": 0.4131,
+      "step": 148350
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001730628166804469,
+      "loss": 0.4608,
+      "step": 148360
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017302041595115437,
+      "loss": 0.4144,
+      "step": 148370
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017297801522186184,
+      "loss": 0.3894,
+      "step": 148380
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017293561449256928,
+      "loss": 0.3999,
+      "step": 148390
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017289321376327675,
+      "loss": 0.4242,
+      "step": 148400
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017285081303398419,
+      "loss": 0.4314,
+      "step": 148410
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017280841230469165,
+      "loss": 0.4684,
+      "step": 148420
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017276601157539912,
+      "loss": 0.2868,
+      "step": 148430
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017272361084610656,
+      "loss": 0.4535,
+      "step": 148440
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017268121011681403,
+      "loss": 0.4571,
+      "step": 148450
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001726388093875215,
+      "loss": 0.3412,
+      "step": 148460
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017259640865822893,
+      "loss": 0.5075,
+      "step": 148470
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001725540079289364,
+      "loss": 0.4627,
+      "step": 148480
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017251160719964384,
+      "loss": 0.3632,
+      "step": 148490
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001724692064703513,
+      "loss": 0.3812,
+      "step": 148500
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017242680574105877,
+      "loss": 0.3858,
+      "step": 148510
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001723844050117662,
+      "loss": 0.391,
+      "step": 148520
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017234200428247368,
+      "loss": 0.3904,
+      "step": 148530
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017229960355318115,
+      "loss": 0.4135,
+      "step": 148540
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017225720282388859,
+      "loss": 0.4198,
+      "step": 148550
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017221480209459605,
+      "loss": 0.3907,
+      "step": 148560
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001721724013653035,
+      "loss": 0.3899,
+      "step": 148570
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017213000063601096,
+      "loss": 0.4637,
+      "step": 148580
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001720875999067184,
+      "loss": 0.375,
+      "step": 148590
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017204519917742587,
+      "loss": 0.4699,
+      "step": 148600
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001720027984481333,
+      "loss": 0.376,
+      "step": 148610
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017196039771884077,
+      "loss": 0.4271,
+      "step": 148620
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001719179969895482,
+      "loss": 0.408,
+      "step": 148630
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017187559626025568,
+      "loss": 0.4013,
+      "step": 148640
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017183319553096314,
+      "loss": 0.392,
+      "step": 148650
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017179079480167058,
+      "loss": 0.4231,
+      "step": 148660
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017174839407237805,
+      "loss": 0.4223,
+      "step": 148670
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001717059933430855,
+      "loss": 0.4981,
+      "step": 148680
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017166359261379296,
+      "loss": 0.3927,
+      "step": 148690
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017162119188450042,
+      "loss": 0.406,
+      "step": 148700
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017157879115520786,
+      "loss": 0.4278,
+      "step": 148710
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017153639042591533,
+      "loss": 0.433,
+      "step": 148720
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017149398969662277,
+      "loss": 0.385,
+      "step": 148730
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017145158896733024,
+      "loss": 0.4408,
+      "step": 148740
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001714091882380377,
+      "loss": 0.4873,
+      "step": 148750
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017136678750874514,
+      "loss": 0.4124,
+      "step": 148760
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001713243867794526,
+      "loss": 0.4302,
+      "step": 148770
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017128198605016008,
+      "loss": 0.4301,
+      "step": 148780
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017123958532086752,
+      "loss": 0.4528,
+      "step": 148790
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017119718459157498,
+      "loss": 0.3964,
+      "step": 148800
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017115478386228242,
+      "loss": 0.4126,
+      "step": 148810
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001711123831329899,
+      "loss": 0.363,
+      "step": 148820
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017106998240369736,
+      "loss": 0.3766,
+      "step": 148830
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001710275816744048,
+      "loss": 0.4382,
+      "step": 148840
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017098518094511226,
+      "loss": 0.4468,
+      "step": 148850
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017094278021581973,
+      "loss": 0.4049,
+      "step": 148860
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017090037948652717,
+      "loss": 0.4063,
+      "step": 148870
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017085797875723464,
+      "loss": 0.3596,
+      "step": 148880
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017081557802794208,
+      "loss": 0.4508,
+      "step": 148890
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017077317729864954,
+      "loss": 0.4823,
+      "step": 148900
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.000170730776569357,
+      "loss": 0.3928,
+      "step": 148910
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017068837584006445,
+      "loss": 0.4226,
+      "step": 148920
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017064597511077192,
+      "loss": 0.4728,
+      "step": 148930
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017060357438147936,
+      "loss": 0.3924,
+      "step": 148940
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017056117365218682,
+      "loss": 0.3808,
+      "step": 148950
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0001705187729228943,
+      "loss": 0.4142,
+      "step": 148960
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017047637219360173,
+      "loss": 0.4363,
+      "step": 148970
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0001704339714643092,
+      "loss": 0.5135,
+      "step": 148980
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017039157073501666,
+      "loss": 0.3809,
+      "step": 148990
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0001703491700057241,
+      "loss": 0.4514,
+      "step": 149000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017030676927643157,
+      "loss": 0.4754,
+      "step": 149010
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.000170264368547139,
+      "loss": 0.4643,
+      "step": 149020
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017022196781784648,
+      "loss": 0.4273,
+      "step": 149030
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017017956708855394,
+      "loss": 0.403,
+      "step": 149040
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017013716635926138,
+      "loss": 0.392,
+      "step": 149050
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017009476562996885,
+      "loss": 0.4083,
+      "step": 149060
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017005236490067632,
+      "loss": 0.4095,
+      "step": 149070
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017000996417138376,
+      "loss": 0.4127,
+      "step": 149080
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00016996756344209122,
+      "loss": 0.4037,
+      "step": 149090
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016992516271279866,
+      "loss": 0.403,
+      "step": 149100
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016988276198350613,
+      "loss": 0.3797,
+      "step": 149110
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001698403612542136,
+      "loss": 0.4228,
+      "step": 149120
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016979796052492104,
+      "loss": 0.3669,
+      "step": 149130
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001697555597956285,
+      "loss": 0.403,
+      "step": 149140
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016971315906633597,
+      "loss": 0.3899,
+      "step": 149150
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001696707583370434,
+      "loss": 0.4175,
+      "step": 149160
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016962835760775088,
+      "loss": 0.4095,
+      "step": 149170
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016958595687845832,
+      "loss": 0.4742,
+      "step": 149180
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016954355614916578,
+      "loss": 0.4137,
+      "step": 149190
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016950115541987325,
+      "loss": 0.4807,
+      "step": 149200
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001694587546905807,
+      "loss": 0.4503,
+      "step": 149210
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016941635396128816,
+      "loss": 0.3842,
+      "step": 149220
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001693739532319956,
+      "loss": 0.3936,
+      "step": 149230
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016933155250270306,
+      "loss": 0.3672,
+      "step": 149240
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001692891517734105,
+      "loss": 0.3701,
+      "step": 149250
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016924675104411797,
+      "loss": 0.4404,
+      "step": 149260
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016920435031482544,
+      "loss": 0.4373,
+      "step": 149270
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016916194958553287,
+      "loss": 0.479,
+      "step": 149280
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016911954885624034,
+      "loss": 0.419,
+      "step": 149290
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016907714812694778,
+      "loss": 0.3587,
+      "step": 149300
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016903474739765525,
+      "loss": 0.4173,
+      "step": 149310
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001689923466683627,
+      "loss": 0.3782,
+      "step": 149320
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016894994593907015,
+      "loss": 0.4247,
+      "step": 149330
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001689075452097776,
+      "loss": 0.6197,
+      "step": 149340
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016886514448048506,
+      "loss": 0.3552,
+      "step": 149350
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016882274375119253,
+      "loss": 0.4788,
+      "step": 149360
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016878034302189997,
+      "loss": 0.4158,
+      "step": 149370
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016873794229260743,
+      "loss": 0.4653,
+      "step": 149380
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001686955415633149,
+      "loss": 0.4088,
+      "step": 149390
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016865314083402234,
+      "loss": 0.4587,
+      "step": 149400
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001686107401047298,
+      "loss": 0.4681,
+      "step": 149410
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016856833937543725,
+      "loss": 0.3771,
+      "step": 149420
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016852593864614471,
+      "loss": 0.4414,
+      "step": 149430
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016848353791685218,
+      "loss": 0.4895,
+      "step": 149440
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016844113718755962,
+      "loss": 0.4233,
+      "step": 149450
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001683987364582671,
+      "loss": 0.3459,
+      "step": 149460
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016835633572897455,
+      "loss": 0.3918,
+      "step": 149470
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.000168313934999682,
+      "loss": 0.4583,
+      "step": 149480
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016827153427038946,
+      "loss": 0.3916,
+      "step": 149490
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001682291335410969,
+      "loss": 0.3955,
+      "step": 149500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016818673281180437,
+      "loss": 0.4913,
+      "step": 149510
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016814433208251183,
+      "loss": 0.4011,
+      "step": 149520
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016810193135321927,
+      "loss": 0.4563,
+      "step": 149530
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016805953062392674,
+      "loss": 0.4679,
+      "step": 149540
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001680171298946342,
+      "loss": 0.4781,
+      "step": 149550
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016797472916534165,
+      "loss": 0.4308,
+      "step": 149560
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016793232843604911,
+      "loss": 0.4228,
+      "step": 149570
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016788992770675655,
+      "loss": 0.4473,
+      "step": 149580
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016784752697746402,
+      "loss": 0.4784,
+      "step": 149590
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001678051262481715,
+      "loss": 0.4267,
+      "step": 149600
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016776272551887893,
+      "loss": 0.4034,
+      "step": 149610
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001677203247895864,
+      "loss": 0.3861,
+      "step": 149620
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016767792406029386,
+      "loss": 0.479,
+      "step": 149630
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001676355233310013,
+      "loss": 0.391,
+      "step": 149640
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016759312260170877,
+      "loss": 0.4248,
+      "step": 149650
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001675507218724162,
+      "loss": 0.333,
+      "step": 149660
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016750832114312367,
+      "loss": 0.4662,
+      "step": 149670
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016746592041383114,
+      "loss": 0.3612,
+      "step": 149680
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016742351968453858,
+      "loss": 0.43,
+      "step": 149690
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016738111895524605,
+      "loss": 0.3623,
+      "step": 149700
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016733871822595349,
+      "loss": 0.4173,
+      "step": 149710
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016729631749666095,
+      "loss": 0.3944,
+      "step": 149720
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016725391676736842,
+      "loss": 0.4291,
+      "step": 149730
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016721151603807586,
+      "loss": 0.4213,
+      "step": 149740
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016716911530878333,
+      "loss": 0.4218,
+      "step": 149750
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001671267145794908,
+      "loss": 0.3202,
+      "step": 149760
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016708431385019823,
+      "loss": 0.3829,
+      "step": 149770
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001670419131209057,
+      "loss": 0.4195,
+      "step": 149780
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016699951239161314,
+      "loss": 0.4307,
+      "step": 149790
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001669571116623206,
+      "loss": 0.3513,
+      "step": 149800
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016691471093302807,
+      "loss": 0.3721,
+      "step": 149810
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0001668723102037355,
+      "loss": 0.4445,
+      "step": 149820
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016682990947444298,
+      "loss": 0.4334,
+      "step": 149830
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016678750874515045,
+      "loss": 0.4046,
+      "step": 149840
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016674510801585789,
+      "loss": 0.4297,
+      "step": 149850
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016670270728656535,
+      "loss": 0.3865,
+      "step": 149860
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0001666603065572728,
+      "loss": 0.4855,
+      "step": 149870
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016661790582798026,
+      "loss": 0.3764,
+      "step": 149880
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0001665755050986877,
+      "loss": 0.3701,
+      "step": 149890
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016653310436939517,
+      "loss": 0.4425,
+      "step": 149900
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016649070364010263,
+      "loss": 0.4233,
+      "step": 149910
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016644830291081007,
+      "loss": 0.4632,
+      "step": 149920
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016640590218151754,
+      "loss": 0.383,
+      "step": 149930
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016636350145222498,
+      "loss": 0.3903,
+      "step": 149940
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016632110072293244,
+      "loss": 0.4477,
+      "step": 149950
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016627869999363988,
+      "loss": 0.3624,
+      "step": 149960
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016623629926434735,
+      "loss": 0.3793,
+      "step": 149970
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0001661938985350548,
+      "loss": 0.4434,
+      "step": 149980
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016615149780576226,
+      "loss": 0.4944,
+      "step": 149990
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016610909707646972,
+      "loss": 0.4031,
+      "step": 150000
+    },
+    {
+      "epoch": 6.34,
+      "eval_loss": 0.6249366998672485,
+      "eval_runtime": 337.5836,
+      "eval_samples_per_second": 15.567,
+      "eval_steps_per_second": 3.892,
+      "step": 150000
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016606669634717716,
+      "loss": 0.3895,
+      "step": 150010
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016602429561788463,
+      "loss": 0.4483,
+      "step": 150020
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016598189488859207,
+      "loss": 0.3734,
+      "step": 150030
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016593949415929954,
+      "loss": 0.4779,
+      "step": 150040
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.000165897093430007,
+      "loss": 0.4544,
+      "step": 150050
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016585469270071444,
+      "loss": 0.4851,
+      "step": 150060
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0001658122919714219,
+      "loss": 0.4594,
+      "step": 150070
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016576989124212938,
+      "loss": 0.359,
+      "step": 150080
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016572749051283682,
+      "loss": 0.4762,
+      "step": 150090
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016568508978354428,
+      "loss": 0.4376,
+      "step": 150100
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016564268905425172,
+      "loss": 0.3709,
+      "step": 150110
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0001656002883249592,
+      "loss": 0.4145,
+      "step": 150120
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016555788759566666,
+      "loss": 0.364,
+      "step": 150130
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0001655154868663741,
+      "loss": 0.4267,
+      "step": 150140
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016547308613708156,
+      "loss": 0.4239,
+      "step": 150150
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016543068540778903,
+      "loss": 0.3614,
+      "step": 150160
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016538828467849647,
+      "loss": 0.3792,
+      "step": 150170
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016534588394920394,
+      "loss": 0.4179,
+      "step": 150180
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016530348321991138,
+      "loss": 0.3971,
+      "step": 150190
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016526108249061884,
+      "loss": 0.4234,
+      "step": 150200
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0001652186817613263,
+      "loss": 0.3982,
+      "step": 150210
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016517628103203375,
+      "loss": 0.4357,
+      "step": 150220
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016513388030274122,
+      "loss": 0.3829,
+      "step": 150230
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016509147957344868,
+      "loss": 0.3748,
+      "step": 150240
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016504907884415612,
+      "loss": 0.4179,
+      "step": 150250
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0001650066781148636,
+      "loss": 0.4115,
+      "step": 150260
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016496427738557103,
+      "loss": 0.3821,
+      "step": 150270
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0001649218766562785,
+      "loss": 0.399,
+      "step": 150280
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016487947592698596,
+      "loss": 0.3966,
+      "step": 150290
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0001648370751976934,
+      "loss": 0.4367,
+      "step": 150300
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016479467446840087,
+      "loss": 0.4425,
+      "step": 150310
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016475227373910834,
+      "loss": 0.4246,
+      "step": 150320
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016470987300981578,
+      "loss": 0.3847,
+      "step": 150330
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016466747228052324,
+      "loss": 0.3874,
+      "step": 150340
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016462507155123068,
+      "loss": 0.4303,
+      "step": 150350
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016458267082193815,
+      "loss": 0.48,
+      "step": 150360
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016454027009264562,
+      "loss": 0.4324,
+      "step": 150370
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016449786936335306,
+      "loss": 0.373,
+      "step": 150380
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016445546863406052,
+      "loss": 0.4895,
+      "step": 150390
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016441306790476796,
+      "loss": 0.4179,
+      "step": 150400
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016437066717547543,
+      "loss": 0.4882,
+      "step": 150410
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0001643282664461829,
+      "loss": 0.3627,
+      "step": 150420
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016428586571689034,
+      "loss": 0.4683,
+      "step": 150430
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0001642434649875978,
+      "loss": 0.49,
+      "step": 150440
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016420106425830527,
+      "loss": 0.4712,
+      "step": 150450
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0001641586635290127,
+      "loss": 0.4057,
+      "step": 150460
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016411626279972018,
+      "loss": 0.424,
+      "step": 150470
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016407386207042762,
+      "loss": 0.3612,
+      "step": 150480
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016403146134113508,
+      "loss": 0.4253,
+      "step": 150490
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016398906061184255,
+      "loss": 0.3882,
+      "step": 150500
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016394665988255,
+      "loss": 0.4262,
+      "step": 150510
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016390425915325746,
+      "loss": 0.3863,
+      "step": 150520
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016386185842396492,
+      "loss": 0.4432,
+      "step": 150530
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016381945769467236,
+      "loss": 0.4083,
+      "step": 150540
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016377705696537983,
+      "loss": 0.4122,
+      "step": 150550
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016373465623608727,
+      "loss": 0.4523,
+      "step": 150560
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016369225550679474,
+      "loss": 0.3884,
+      "step": 150570
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016364985477750218,
+      "loss": 0.3428,
+      "step": 150580
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016360745404820964,
+      "loss": 0.421,
+      "step": 150590
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016356505331891708,
+      "loss": 0.4618,
+      "step": 150600
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016352265258962455,
+      "loss": 0.4178,
+      "step": 150610
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.000163480251860332,
+      "loss": 0.3935,
+      "step": 150620
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016343785113103945,
+      "loss": 0.3658,
+      "step": 150630
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0001633954504017469,
+      "loss": 0.4078,
+      "step": 150640
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016335304967245436,
+      "loss": 0.3357,
+      "step": 150650
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016331064894316183,
+      "loss": 0.4275,
+      "step": 150660
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016326824821386927,
+      "loss": 0.427,
+      "step": 150670
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016322584748457673,
+      "loss": 0.3687,
+      "step": 150680
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0001631834467552842,
+      "loss": 0.4328,
+      "step": 150690
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016314104602599164,
+      "loss": 0.4064,
+      "step": 150700
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0001630986452966991,
+      "loss": 0.4135,
+      "step": 150710
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016305624456740655,
+      "loss": 0.3168,
+      "step": 150720
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016301384383811401,
+      "loss": 0.3937,
+      "step": 150730
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016297144310882148,
+      "loss": 0.3963,
+      "step": 150740
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016292904237952892,
+      "loss": 0.4083,
+      "step": 150750
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0001628866416502364,
+      "loss": 0.3808,
+      "step": 150760
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016284424092094385,
+      "loss": 0.4177,
+      "step": 150770
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0001628018401916513,
+      "loss": 0.4028,
+      "step": 150780
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016275943946235876,
+      "loss": 0.433,
+      "step": 150790
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0001627170387330662,
+      "loss": 0.467,
+      "step": 150800
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016267463800377367,
+      "loss": 0.4543,
+      "step": 150810
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016263223727448113,
+      "loss": 0.3426,
+      "step": 150820
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016258983654518857,
+      "loss": 0.3924,
+      "step": 150830
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016254743581589604,
+      "loss": 0.4029,
+      "step": 150840
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0001625050350866035,
+      "loss": 0.3973,
+      "step": 150850
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016246263435731095,
+      "loss": 0.3794,
+      "step": 150860
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016242023362801841,
+      "loss": 0.4174,
+      "step": 150870
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016237783289872585,
+      "loss": 0.3387,
+      "step": 150880
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016233543216943332,
+      "loss": 0.3269,
+      "step": 150890
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0001622930314401408,
+      "loss": 0.359,
+      "step": 150900
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016225063071084823,
+      "loss": 0.4964,
+      "step": 150910
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0001622082299815557,
+      "loss": 0.3951,
+      "step": 150920
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016216582925226316,
+      "loss": 0.3614,
+      "step": 150930
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0001621234285229706,
+      "loss": 0.3896,
+      "step": 150940
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016208102779367807,
+      "loss": 0.4188,
+      "step": 150950
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0001620386270643855,
+      "loss": 0.4068,
+      "step": 150960
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016199622633509297,
+      "loss": 0.355,
+      "step": 150970
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016195382560580044,
+      "loss": 0.385,
+      "step": 150980
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016191142487650788,
+      "loss": 0.3681,
+      "step": 150990
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016186902414721535,
+      "loss": 0.3817,
+      "step": 151000
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001618266234179228,
+      "loss": 0.4426,
+      "step": 151010
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016178422268863025,
+      "loss": 0.4564,
+      "step": 151020
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016174182195933772,
+      "loss": 0.3323,
+      "step": 151030
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016169942123004516,
+      "loss": 0.3175,
+      "step": 151040
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016165702050075263,
+      "loss": 0.4285,
+      "step": 151050
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001616146197714601,
+      "loss": 0.5147,
+      "step": 151060
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016157221904216753,
+      "loss": 0.4106,
+      "step": 151070
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.000161529818312875,
+      "loss": 0.4123,
+      "step": 151080
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016148741758358244,
+      "loss": 0.4185,
+      "step": 151090
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001614450168542899,
+      "loss": 0.3875,
+      "step": 151100
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016140261612499737,
+      "loss": 0.4155,
+      "step": 151110
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001613602153957048,
+      "loss": 0.4242,
+      "step": 151120
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016131781466641228,
+      "loss": 0.4106,
+      "step": 151130
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016127541393711975,
+      "loss": 0.4172,
+      "step": 151140
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016123301320782719,
+      "loss": 0.4805,
+      "step": 151150
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016119061247853465,
+      "loss": 0.4552,
+      "step": 151160
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001611482117492421,
+      "loss": 0.4333,
+      "step": 151170
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016110581101994956,
+      "loss": 0.3812,
+      "step": 151180
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016106341029065703,
+      "loss": 0.4355,
+      "step": 151190
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016102100956136447,
+      "loss": 0.4061,
+      "step": 151200
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016097860883207193,
+      "loss": 0.4372,
+      "step": 151210
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016093620810277937,
+      "loss": 0.4157,
+      "step": 151220
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016089380737348684,
+      "loss": 0.4504,
+      "step": 151230
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016085140664419428,
+      "loss": 0.4392,
+      "step": 151240
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016080900591490175,
+      "loss": 0.3766,
+      "step": 151250
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016076660518560918,
+      "loss": 0.3798,
+      "step": 151260
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016072420445631665,
+      "loss": 0.4652,
+      "step": 151270
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016068180372702412,
+      "loss": 0.4565,
+      "step": 151280
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016063940299773156,
+      "loss": 0.3943,
+      "step": 151290
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016059700226843902,
+      "loss": 0.4209,
+      "step": 151300
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016055460153914646,
+      "loss": 0.3977,
+      "step": 151310
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016051220080985393,
+      "loss": 0.4083,
+      "step": 151320
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016046980008056137,
+      "loss": 0.4728,
+      "step": 151330
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016042739935126884,
+      "loss": 0.4214,
+      "step": 151340
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001603849986219763,
+      "loss": 0.4493,
+      "step": 151350
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016034259789268374,
+      "loss": 0.4123,
+      "step": 151360
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001603001971633912,
+      "loss": 0.4344,
+      "step": 151370
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016025779643409868,
+      "loss": 0.3829,
+      "step": 151380
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016021539570480612,
+      "loss": 0.4158,
+      "step": 151390
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016017299497551358,
+      "loss": 0.4484,
+      "step": 151400
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016013059424622102,
+      "loss": 0.4172,
+      "step": 151410
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001600881935169285,
+      "loss": 0.3934,
+      "step": 151420
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016004579278763596,
+      "loss": 0.4016,
+      "step": 151430
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001600033920583434,
+      "loss": 0.4995,
+      "step": 151440
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00015996099132905086,
+      "loss": 0.4238,
+      "step": 151450
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015991859059975833,
+      "loss": 0.4599,
+      "step": 151460
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015987618987046577,
+      "loss": 0.4421,
+      "step": 151470
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015983378914117324,
+      "loss": 0.3996,
+      "step": 151480
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015979138841188068,
+      "loss": 0.4511,
+      "step": 151490
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015974898768258814,
+      "loss": 0.4413,
+      "step": 151500
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001597065869532956,
+      "loss": 0.344,
+      "step": 151510
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015966418622400305,
+      "loss": 0.4413,
+      "step": 151520
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015962178549471052,
+      "loss": 0.4279,
+      "step": 151530
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015957938476541798,
+      "loss": 0.4211,
+      "step": 151540
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015953698403612542,
+      "loss": 0.44,
+      "step": 151550
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001594945833068329,
+      "loss": 0.3862,
+      "step": 151560
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015945218257754033,
+      "loss": 0.4193,
+      "step": 151570
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001594097818482478,
+      "loss": 0.4102,
+      "step": 151580
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015936738111895526,
+      "loss": 0.3941,
+      "step": 151590
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001593249803896627,
+      "loss": 0.4784,
+      "step": 151600
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015928257966037017,
+      "loss": 0.393,
+      "step": 151610
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015924017893107764,
+      "loss": 0.3827,
+      "step": 151620
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015919777820178508,
+      "loss": 0.4172,
+      "step": 151630
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015915537747249254,
+      "loss": 0.4122,
+      "step": 151640
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015911297674319998,
+      "loss": 0.4817,
+      "step": 151650
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015907057601390745,
+      "loss": 0.4364,
+      "step": 151660
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015902817528461492,
+      "loss": 0.4187,
+      "step": 151670
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015898577455532236,
+      "loss": 0.4517,
+      "step": 151680
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015894337382602982,
+      "loss": 0.4456,
+      "step": 151690
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0001589009730967373,
+      "loss": 0.3248,
+      "step": 151700
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015885857236744473,
+      "loss": 0.3664,
+      "step": 151710
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0001588161716381522,
+      "loss": 0.5148,
+      "step": 151720
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015877377090885964,
+      "loss": 0.437,
+      "step": 151730
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0001587313701795671,
+      "loss": 0.4816,
+      "step": 151740
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015868896945027457,
+      "loss": 0.3516,
+      "step": 151750
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.000158646568720982,
+      "loss": 0.4266,
+      "step": 151760
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015860416799168948,
+      "loss": 0.3786,
+      "step": 151770
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015856176726239694,
+      "loss": 0.4289,
+      "step": 151780
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015851936653310438,
+      "loss": 0.3995,
+      "step": 151790
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015847696580381185,
+      "loss": 0.3583,
+      "step": 151800
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0001584345650745193,
+      "loss": 0.3835,
+      "step": 151810
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015839216434522676,
+      "loss": 0.3831,
+      "step": 151820
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015834976361593422,
+      "loss": 0.4797,
+      "step": 151830
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015830736288664166,
+      "loss": 0.386,
+      "step": 151840
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015826496215734913,
+      "loss": 0.4323,
+      "step": 151850
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015822256142805657,
+      "loss": 0.4556,
+      "step": 151860
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015818016069876404,
+      "loss": 0.4268,
+      "step": 151870
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015813775996947148,
+      "loss": 0.3987,
+      "step": 151880
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015809535924017894,
+      "loss": 0.3079,
+      "step": 151890
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0001580529585108864,
+      "loss": 0.4228,
+      "step": 151900
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015801055778159385,
+      "loss": 0.3922,
+      "step": 151910
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015796815705230132,
+      "loss": 0.4522,
+      "step": 151920
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015792575632300875,
+      "loss": 0.4056,
+      "step": 151930
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015788335559371622,
+      "loss": 0.4161,
+      "step": 151940
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015784095486442366,
+      "loss": 0.404,
+      "step": 151950
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015779855413513113,
+      "loss": 0.3697,
+      "step": 151960
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015775615340583857,
+      "loss": 0.357,
+      "step": 151970
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015771375267654603,
+      "loss": 0.439,
+      "step": 151980
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001576713519472535,
+      "loss": 0.4395,
+      "step": 151990
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015762895121796094,
+      "loss": 0.4022,
+      "step": 152000
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001575865504886684,
+      "loss": 0.3744,
+      "step": 152010
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015754414975937585,
+      "loss": 0.4094,
+      "step": 152020
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015750174903008331,
+      "loss": 0.3225,
+      "step": 152030
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015745934830079078,
+      "loss": 0.4295,
+      "step": 152040
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015741694757149822,
+      "loss": 0.452,
+      "step": 152050
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001573745468422057,
+      "loss": 0.4311,
+      "step": 152060
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015733214611291315,
+      "loss": 0.3545,
+      "step": 152070
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001572897453836206,
+      "loss": 0.4389,
+      "step": 152080
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015724734465432806,
+      "loss": 0.3916,
+      "step": 152090
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001572049439250355,
+      "loss": 0.3855,
+      "step": 152100
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015716254319574297,
+      "loss": 0.4006,
+      "step": 152110
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015712014246645043,
+      "loss": 0.4898,
+      "step": 152120
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015707774173715787,
+      "loss": 0.3661,
+      "step": 152130
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015703534100786534,
+      "loss": 0.3761,
+      "step": 152140
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001569929402785728,
+      "loss": 0.4819,
+      "step": 152150
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015695053954928025,
+      "loss": 0.3641,
+      "step": 152160
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015690813881998771,
+      "loss": 0.3792,
+      "step": 152170
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015686573809069515,
+      "loss": 0.4497,
+      "step": 152180
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015682333736140262,
+      "loss": 0.4825,
+      "step": 152190
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001567809366321101,
+      "loss": 0.3937,
+      "step": 152200
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015673853590281753,
+      "loss": 0.4413,
+      "step": 152210
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.000156696135173525,
+      "loss": 0.4402,
+      "step": 152220
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015665373444423246,
+      "loss": 0.4194,
+      "step": 152230
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001566113337149399,
+      "loss": 0.4641,
+      "step": 152240
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015656893298564737,
+      "loss": 0.3776,
+      "step": 152250
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001565265322563548,
+      "loss": 0.4253,
+      "step": 152260
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015648413152706227,
+      "loss": 0.4108,
+      "step": 152270
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015644173079776974,
+      "loss": 0.4138,
+      "step": 152280
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015639933006847718,
+      "loss": 0.4931,
+      "step": 152290
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015635692933918465,
+      "loss": 0.3982,
+      "step": 152300
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001563145286098921,
+      "loss": 0.4446,
+      "step": 152310
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015627212788059955,
+      "loss": 0.4441,
+      "step": 152320
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015622972715130702,
+      "loss": 0.3652,
+      "step": 152330
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015618732642201446,
+      "loss": 0.4904,
+      "step": 152340
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015614492569272193,
+      "loss": 0.4766,
+      "step": 152350
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001561025249634294,
+      "loss": 0.431,
+      "step": 152360
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015606012423413683,
+      "loss": 0.4174,
+      "step": 152370
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001560177235048443,
+      "loss": 0.4113,
+      "step": 152380
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015597532277555177,
+      "loss": 0.4291,
+      "step": 152390
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001559329220462592,
+      "loss": 0.3874,
+      "step": 152400
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015589052131696667,
+      "loss": 0.3958,
+      "step": 152410
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0001558481205876741,
+      "loss": 0.405,
+      "step": 152420
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015580571985838158,
+      "loss": 0.4053,
+      "step": 152430
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015576331912908905,
+      "loss": 0.3916,
+      "step": 152440
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015572091839979649,
+      "loss": 0.3717,
+      "step": 152450
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015567851767050395,
+      "loss": 0.4219,
+      "step": 152460
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015563611694121142,
+      "loss": 0.3797,
+      "step": 152470
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015559371621191886,
+      "loss": 0.4713,
+      "step": 152480
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015555131548262633,
+      "loss": 0.4502,
+      "step": 152490
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015550891475333377,
+      "loss": 0.4485,
+      "step": 152500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015546651402404123,
+      "loss": 0.4677,
+      "step": 152510
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015542411329474867,
+      "loss": 0.4576,
+      "step": 152520
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015538171256545614,
+      "loss": 0.4313,
+      "step": 152530
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0001553393118361636,
+      "loss": 0.4679,
+      "step": 152540
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015529691110687105,
+      "loss": 0.4352,
+      "step": 152550
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0001552545103775785,
+      "loss": 0.4051,
+      "step": 152560
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015521210964828595,
+      "loss": 0.4146,
+      "step": 152570
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015516970891899342,
+      "loss": 0.3984,
+      "step": 152580
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015512730818970086,
+      "loss": 0.3957,
+      "step": 152590
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015508490746040833,
+      "loss": 0.4286,
+      "step": 152600
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015504250673111576,
+      "loss": 0.4235,
+      "step": 152610
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015500010600182323,
+      "loss": 0.3969,
+      "step": 152620
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015495770527253067,
+      "loss": 0.3559,
+      "step": 152630
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015491530454323814,
+      "loss": 0.4208,
+      "step": 152640
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0001548729038139456,
+      "loss": 0.4163,
+      "step": 152650
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015483050308465304,
+      "loss": 0.4257,
+      "step": 152660
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0001547881023553605,
+      "loss": 0.3815,
+      "step": 152670
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015474570162606798,
+      "loss": 0.3603,
+      "step": 152680
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015470330089677542,
+      "loss": 0.4414,
+      "step": 152690
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015466090016748288,
+      "loss": 0.3837,
+      "step": 152700
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015461849943819032,
+      "loss": 0.3884,
+      "step": 152710
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0001545760987088978,
+      "loss": 0.4546,
+      "step": 152720
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015453369797960526,
+      "loss": 0.3745,
+      "step": 152730
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0001544912972503127,
+      "loss": 0.4314,
+      "step": 152740
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015444889652102016,
+      "loss": 0.3314,
+      "step": 152750
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015440649579172763,
+      "loss": 0.3408,
+      "step": 152760
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015436409506243507,
+      "loss": 0.4093,
+      "step": 152770
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015432169433314254,
+      "loss": 0.4076,
+      "step": 152780
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015427929360384998,
+      "loss": 0.4339,
+      "step": 152790
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015423689287455744,
+      "loss": 0.4486,
+      "step": 152800
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0001541944921452649,
+      "loss": 0.457,
+      "step": 152810
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015415209141597235,
+      "loss": 0.4782,
+      "step": 152820
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015410969068667982,
+      "loss": 0.3989,
+      "step": 152830
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015406728995738728,
+      "loss": 0.3932,
+      "step": 152840
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015402488922809472,
+      "loss": 0.3803,
+      "step": 152850
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0001539824884988022,
+      "loss": 0.4187,
+      "step": 152860
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015394008776950963,
+      "loss": 0.4,
+      "step": 152870
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001538976870402171,
+      "loss": 0.428,
+      "step": 152880
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015385528631092456,
+      "loss": 0.4025,
+      "step": 152890
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.000153812885581632,
+      "loss": 0.3999,
+      "step": 152900
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015377048485233947,
+      "loss": 0.3863,
+      "step": 152910
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015372808412304694,
+      "loss": 0.4502,
+      "step": 152920
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015368568339375438,
+      "loss": 0.4458,
+      "step": 152930
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015364328266446184,
+      "loss": 0.3726,
+      "step": 152940
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015360088193516928,
+      "loss": 0.4704,
+      "step": 152950
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015355848120587675,
+      "loss": 0.414,
+      "step": 152960
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015351608047658422,
+      "loss": 0.4238,
+      "step": 152970
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015347367974729166,
+      "loss": 0.4284,
+      "step": 152980
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015343127901799912,
+      "loss": 0.3967,
+      "step": 152990
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001533888782887066,
+      "loss": 0.4244,
+      "step": 153000
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015334647755941403,
+      "loss": 0.4053,
+      "step": 153010
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001533040768301215,
+      "loss": 0.4067,
+      "step": 153020
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015326167610082894,
+      "loss": 0.4316,
+      "step": 153030
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001532192753715364,
+      "loss": 0.4119,
+      "step": 153040
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015317687464224387,
+      "loss": 0.3908,
+      "step": 153050
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001531344739129513,
+      "loss": 0.4314,
+      "step": 153060
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015309207318365878,
+      "loss": 0.4131,
+      "step": 153070
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015304967245436624,
+      "loss": 0.4003,
+      "step": 153080
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015300727172507368,
+      "loss": 0.3409,
+      "step": 153090
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015296487099578115,
+      "loss": 0.3564,
+      "step": 153100
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001529224702664886,
+      "loss": 0.4007,
+      "step": 153110
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015288006953719606,
+      "loss": 0.3842,
+      "step": 153120
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015283766880790352,
+      "loss": 0.3858,
+      "step": 153130
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015279526807861096,
+      "loss": 0.4296,
+      "step": 153140
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015275286734931843,
+      "loss": 0.4134,
+      "step": 153150
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001527104666200259,
+      "loss": 0.4054,
+      "step": 153160
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015266806589073334,
+      "loss": 0.4091,
+      "step": 153170
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001526256651614408,
+      "loss": 0.4443,
+      "step": 153180
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015258326443214824,
+      "loss": 0.411,
+      "step": 153190
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001525408637028557,
+      "loss": 0.4238,
+      "step": 153200
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015249846297356315,
+      "loss": 0.4557,
+      "step": 153210
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015245606224427062,
+      "loss": 0.4489,
+      "step": 153220
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015241366151497806,
+      "loss": 0.4473,
+      "step": 153230
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015237126078568552,
+      "loss": 0.4224,
+      "step": 153240
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015232886005639296,
+      "loss": 0.3966,
+      "step": 153250
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015228645932710043,
+      "loss": 0.3376,
+      "step": 153260
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015224405859780787,
+      "loss": 0.4348,
+      "step": 153270
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015220165786851533,
+      "loss": 0.4279,
+      "step": 153280
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001521592571392228,
+      "loss": 0.3613,
+      "step": 153290
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015211685640993024,
+      "loss": 0.4158,
+      "step": 153300
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001520744556806377,
+      "loss": 0.4067,
+      "step": 153310
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015203205495134515,
+      "loss": 0.3876,
+      "step": 153320
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015198965422205261,
+      "loss": 0.458,
+      "step": 153330
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015194725349276008,
+      "loss": 0.3794,
+      "step": 153340
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015190485276346752,
+      "loss": 0.4486,
+      "step": 153350
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.000151862452034175,
+      "loss": 0.4532,
+      "step": 153360
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015182005130488245,
+      "loss": 0.4171,
+      "step": 153370
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001517776505755899,
+      "loss": 0.4,
+      "step": 153380
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015173524984629736,
+      "loss": 0.4297,
+      "step": 153390
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001516928491170048,
+      "loss": 0.4979,
+      "step": 153400
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015165044838771227,
+      "loss": 0.4537,
+      "step": 153410
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015160804765841973,
+      "loss": 0.381,
+      "step": 153420
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015156564692912717,
+      "loss": 0.4284,
+      "step": 153430
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015152324619983464,
+      "loss": 0.3918,
+      "step": 153440
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001514808454705421,
+      "loss": 0.3888,
+      "step": 153450
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015143844474124955,
+      "loss": 0.4466,
+      "step": 153460
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015139604401195701,
+      "loss": 0.4598,
+      "step": 153470
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015135364328266445,
+      "loss": 0.3538,
+      "step": 153480
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015131124255337192,
+      "loss": 0.4292,
+      "step": 153490
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001512688418240794,
+      "loss": 0.4457,
+      "step": 153500
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015122644109478683,
+      "loss": 0.3925,
+      "step": 153510
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001511840403654943,
+      "loss": 0.3724,
+      "step": 153520
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015114163963620176,
+      "loss": 0.4726,
+      "step": 153530
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001510992389069092,
+      "loss": 0.3939,
+      "step": 153540
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015105683817761667,
+      "loss": 0.3785,
+      "step": 153550
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001510144374483241,
+      "loss": 0.3633,
+      "step": 153560
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015097203671903157,
+      "loss": 0.4695,
+      "step": 153570
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015092963598973904,
+      "loss": 0.5212,
+      "step": 153580
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015088723526044648,
+      "loss": 0.4202,
+      "step": 153590
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015084483453115395,
+      "loss": 0.3625,
+      "step": 153600
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015080243380186141,
+      "loss": 0.3812,
+      "step": 153610
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015076003307256885,
+      "loss": 0.393,
+      "step": 153620
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015071763234327632,
+      "loss": 0.4378,
+      "step": 153630
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015067523161398376,
+      "loss": 0.3793,
+      "step": 153640
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015063283088469123,
+      "loss": 0.4934,
+      "step": 153650
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001505904301553987,
+      "loss": 0.361,
+      "step": 153660
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015054802942610613,
+      "loss": 0.4316,
+      "step": 153670
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001505056286968136,
+      "loss": 0.4247,
+      "step": 153680
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015046322796752107,
+      "loss": 0.396,
+      "step": 153690
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001504208272382285,
+      "loss": 0.4327,
+      "step": 153700
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015037842650893597,
+      "loss": 0.3994,
+      "step": 153710
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001503360257796434,
+      "loss": 0.414,
+      "step": 153720
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015029362505035088,
+      "loss": 0.4266,
+      "step": 153730
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015025122432105835,
+      "loss": 0.3973,
+      "step": 153740
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015020882359176579,
+      "loss": 0.4018,
+      "step": 153750
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015016642286247325,
+      "loss": 0.4847,
+      "step": 153760
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015012402213318072,
+      "loss": 0.4152,
+      "step": 153770
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015008162140388816,
+      "loss": 0.4426,
+      "step": 153780
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015003922067459563,
+      "loss": 0.3902,
+      "step": 153790
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00014999681994530307,
+      "loss": 0.4551,
+      "step": 153800
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00014995441921601053,
+      "loss": 0.3904,
+      "step": 153810
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.000149912018486718,
+      "loss": 0.4214,
+      "step": 153820
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014986961775742544,
+      "loss": 0.4568,
+      "step": 153830
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001498272170281329,
+      "loss": 0.4283,
+      "step": 153840
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014978481629884035,
+      "loss": 0.4529,
+      "step": 153850
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001497424155695478,
+      "loss": 0.3253,
+      "step": 153860
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014970001484025525,
+      "loss": 0.4483,
+      "step": 153870
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014965761411096272,
+      "loss": 0.3925,
+      "step": 153880
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014961521338167016,
+      "loss": 0.4263,
+      "step": 153890
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014957281265237763,
+      "loss": 0.3742,
+      "step": 153900
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001495304119230851,
+      "loss": 0.426,
+      "step": 153910
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014948801119379253,
+      "loss": 0.3853,
+      "step": 153920
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001494456104645,
+      "loss": 0.3763,
+      "step": 153930
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014940320973520744,
+      "loss": 0.391,
+      "step": 153940
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001493608090059149,
+      "loss": 0.4035,
+      "step": 153950
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014931840827662234,
+      "loss": 0.4044,
+      "step": 153960
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001492760075473298,
+      "loss": 0.3674,
+      "step": 153970
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014923360681803728,
+      "loss": 0.5458,
+      "step": 153980
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014919120608874472,
+      "loss": 0.4002,
+      "step": 153990
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014914880535945218,
+      "loss": 0.5244,
+      "step": 154000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014910640463015962,
+      "loss": 0.4389,
+      "step": 154010
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001490640039008671,
+      "loss": 0.4293,
+      "step": 154020
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014902160317157456,
+      "loss": 0.4055,
+      "step": 154030
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.000148979202442282,
+      "loss": 0.4399,
+      "step": 154040
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014893680171298946,
+      "loss": 0.4213,
+      "step": 154050
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014889440098369693,
+      "loss": 0.3833,
+      "step": 154060
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014885200025440437,
+      "loss": 0.371,
+      "step": 154070
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014880959952511184,
+      "loss": 0.4711,
+      "step": 154080
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014876719879581928,
+      "loss": 0.3545,
+      "step": 154090
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014872479806652674,
+      "loss": 0.4106,
+      "step": 154100
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0001486823973372342,
+      "loss": 0.4528,
+      "step": 154110
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014863999660794165,
+      "loss": 0.4228,
+      "step": 154120
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014859759587864912,
+      "loss": 0.3715,
+      "step": 154130
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014855519514935658,
+      "loss": 0.418,
+      "step": 154140
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014851279442006402,
+      "loss": 0.355,
+      "step": 154150
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0001484703936907715,
+      "loss": 0.3995,
+      "step": 154160
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014842799296147893,
+      "loss": 0.4289,
+      "step": 154170
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0001483855922321864,
+      "loss": 0.4067,
+      "step": 154180
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014834319150289386,
+      "loss": 0.3546,
+      "step": 154190
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0001483007907736013,
+      "loss": 0.4044,
+      "step": 154200
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014825839004430877,
+      "loss": 0.4268,
+      "step": 154210
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014821598931501624,
+      "loss": 0.4267,
+      "step": 154220
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014817358858572368,
+      "loss": 0.455,
+      "step": 154230
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014813118785643114,
+      "loss": 0.434,
+      "step": 154240
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014808878712713858,
+      "loss": 0.3877,
+      "step": 154250
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014804638639784605,
+      "loss": 0.3736,
+      "step": 154260
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014800398566855352,
+      "loss": 0.4241,
+      "step": 154270
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014796158493926096,
+      "loss": 0.4276,
+      "step": 154280
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014791918420996842,
+      "loss": 0.3964,
+      "step": 154290
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001478767834806759,
+      "loss": 0.4713,
+      "step": 154300
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014783438275138333,
+      "loss": 0.5215,
+      "step": 154310
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001477919820220908,
+      "loss": 0.3523,
+      "step": 154320
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014774958129279824,
+      "loss": 0.4087,
+      "step": 154330
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001477071805635057,
+      "loss": 0.3954,
+      "step": 154340
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014766477983421317,
+      "loss": 0.3639,
+      "step": 154350
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001476223791049206,
+      "loss": 0.4001,
+      "step": 154360
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014757997837562808,
+      "loss": 0.4388,
+      "step": 154370
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014753757764633554,
+      "loss": 0.4551,
+      "step": 154380
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014749517691704298,
+      "loss": 0.389,
+      "step": 154390
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014745277618775045,
+      "loss": 0.4852,
+      "step": 154400
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001474103754584579,
+      "loss": 0.3706,
+      "step": 154410
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014736797472916536,
+      "loss": 0.4199,
+      "step": 154420
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014732557399987282,
+      "loss": 0.436,
+      "step": 154430
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014728317327058026,
+      "loss": 0.4347,
+      "step": 154440
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014724077254128773,
+      "loss": 0.3641,
+      "step": 154450
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001471983718119952,
+      "loss": 0.3587,
+      "step": 154460
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014715597108270264,
+      "loss": 0.4337,
+      "step": 154470
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001471135703534101,
+      "loss": 0.3921,
+      "step": 154480
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014707116962411754,
+      "loss": 0.3821,
+      "step": 154490
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.000147028768894825,
+      "loss": 0.429,
+      "step": 154500
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014698636816553245,
+      "loss": 0.3824,
+      "step": 154510
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014694396743623992,
+      "loss": 0.4578,
+      "step": 154520
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014690156670694736,
+      "loss": 0.3159,
+      "step": 154530
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014685916597765482,
+      "loss": 0.3999,
+      "step": 154540
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001468167652483623,
+      "loss": 0.4923,
+      "step": 154550
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014677436451906973,
+      "loss": 0.3966,
+      "step": 154560
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001467319637897772,
+      "loss": 0.3552,
+      "step": 154570
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014668956306048464,
+      "loss": 0.4286,
+      "step": 154580
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001466471623311921,
+      "loss": 0.4697,
+      "step": 154590
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014660476160189954,
+      "loss": 0.4442,
+      "step": 154600
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.000146562360872607,
+      "loss": 0.4347,
+      "step": 154610
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014651996014331448,
+      "loss": 0.3979,
+      "step": 154620
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014647755941402191,
+      "loss": 0.4005,
+      "step": 154630
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014643515868472938,
+      "loss": 0.3388,
+      "step": 154640
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014639275795543682,
+      "loss": 0.3966,
+      "step": 154650
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001463503572261443,
+      "loss": 0.3923,
+      "step": 154660
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014630795649685175,
+      "loss": 0.4557,
+      "step": 154670
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001462655557675592,
+      "loss": 0.454,
+      "step": 154680
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014622315503826666,
+      "loss": 0.4217,
+      "step": 154690
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001461807543089741,
+      "loss": 0.4072,
+      "step": 154700
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014613835357968157,
+      "loss": 0.4379,
+      "step": 154710
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014609595285038903,
+      "loss": 0.3844,
+      "step": 154720
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014605355212109647,
+      "loss": 0.4306,
+      "step": 154730
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014601115139180394,
+      "loss": 0.4822,
+      "step": 154740
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001459687506625114,
+      "loss": 0.4605,
+      "step": 154750
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014592634993321885,
+      "loss": 0.3928,
+      "step": 154760
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014588394920392631,
+      "loss": 0.3901,
+      "step": 154770
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014584154847463375,
+      "loss": 0.4102,
+      "step": 154780
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014579914774534122,
+      "loss": 0.4955,
+      "step": 154790
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001457567470160487,
+      "loss": 0.4518,
+      "step": 154800
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014571434628675613,
+      "loss": 0.5036,
+      "step": 154810
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001456719455574636,
+      "loss": 0.4731,
+      "step": 154820
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014562954482817106,
+      "loss": 0.4293,
+      "step": 154830
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001455871440988785,
+      "loss": 0.4143,
+      "step": 154840
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014554474336958597,
+      "loss": 0.4466,
+      "step": 154850
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001455023426402934,
+      "loss": 0.4447,
+      "step": 154860
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014545994191100087,
+      "loss": 0.4101,
+      "step": 154870
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014541754118170834,
+      "loss": 0.3885,
+      "step": 154880
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014537514045241578,
+      "loss": 0.4497,
+      "step": 154890
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014533273972312325,
+      "loss": 0.4248,
+      "step": 154900
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014529033899383071,
+      "loss": 0.415,
+      "step": 154910
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014524793826453815,
+      "loss": 0.4431,
+      "step": 154920
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014520553753524562,
+      "loss": 0.3836,
+      "step": 154930
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014516313680595306,
+      "loss": 0.4579,
+      "step": 154940
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014512073607666053,
+      "loss": 0.4084,
+      "step": 154950
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.000145078335347368,
+      "loss": 0.4567,
+      "step": 154960
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014503593461807543,
+      "loss": 0.3963,
+      "step": 154970
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001449935338887829,
+      "loss": 0.5191,
+      "step": 154980
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014495113315949037,
+      "loss": 0.4591,
+      "step": 154990
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001449087324301978,
+      "loss": 0.3978,
+      "step": 155000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014486633170090527,
+      "loss": 0.3645,
+      "step": 155010
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0001448239309716127,
+      "loss": 0.4246,
+      "step": 155020
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014478153024232018,
+      "loss": 0.4509,
+      "step": 155030
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014473912951302765,
+      "loss": 0.3781,
+      "step": 155040
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014469672878373509,
+      "loss": 0.3703,
+      "step": 155050
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014465432805444255,
+      "loss": 0.3775,
+      "step": 155060
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014461192732515002,
+      "loss": 0.4367,
+      "step": 155070
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014456952659585746,
+      "loss": 0.3645,
+      "step": 155080
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014452712586656493,
+      "loss": 0.3628,
+      "step": 155090
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014448472513727237,
+      "loss": 0.468,
+      "step": 155100
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014444232440797983,
+      "loss": 0.3547,
+      "step": 155110
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0001443999236786873,
+      "loss": 0.3932,
+      "step": 155120
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014435752294939474,
+      "loss": 0.4054,
+      "step": 155130
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0001443151222201022,
+      "loss": 0.3988,
+      "step": 155140
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014427272149080965,
+      "loss": 0.4163,
+      "step": 155150
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0001442303207615171,
+      "loss": 0.3936,
+      "step": 155160
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014418792003222458,
+      "loss": 0.3676,
+      "step": 155170
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014414551930293202,
+      "loss": 0.4842,
+      "step": 155180
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014410311857363949,
+      "loss": 0.3682,
+      "step": 155190
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014406071784434693,
+      "loss": 0.4326,
+      "step": 155200
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0001440183171150544,
+      "loss": 0.516,
+      "step": 155210
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014397591638576183,
+      "loss": 0.4506,
+      "step": 155220
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0001439335156564693,
+      "loss": 0.4902,
+      "step": 155230
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014389111492717674,
+      "loss": 0.3754,
+      "step": 155240
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0001438487141978842,
+      "loss": 0.4511,
+      "step": 155250
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014380631346859164,
+      "loss": 0.4753,
+      "step": 155260
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0001437639127392991,
+      "loss": 0.385,
+      "step": 155270
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014372151201000658,
+      "loss": 0.4015,
+      "step": 155280
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014367911128071402,
+      "loss": 0.403,
+      "step": 155290
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014363671055142148,
+      "loss": 0.3932,
+      "step": 155300
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014359430982212895,
+      "loss": 0.3865,
+      "step": 155310
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0001435519090928364,
+      "loss": 0.3902,
+      "step": 155320
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014350950836354386,
+      "loss": 0.4137,
+      "step": 155330
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0001434671076342513,
+      "loss": 0.4117,
+      "step": 155340
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014342470690495876,
+      "loss": 0.4472,
+      "step": 155350
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014338230617566623,
+      "loss": 0.4437,
+      "step": 155360
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014333990544637367,
+      "loss": 0.3608,
+      "step": 155370
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014329750471708114,
+      "loss": 0.3841,
+      "step": 155380
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014325510398778858,
+      "loss": 0.375,
+      "step": 155390
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014321270325849604,
+      "loss": 0.4609,
+      "step": 155400
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0001431703025292035,
+      "loss": 0.4521,
+      "step": 155410
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014312790179991095,
+      "loss": 0.4505,
+      "step": 155420
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014308550107061842,
+      "loss": 0.4108,
+      "step": 155430
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014304310034132588,
+      "loss": 0.4156,
+      "step": 155440
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014300069961203332,
+      "loss": 0.4207,
+      "step": 155450
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0001429582988827408,
+      "loss": 0.41,
+      "step": 155460
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014291589815344823,
+      "loss": 0.4404,
+      "step": 155470
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0001428734974241557,
+      "loss": 0.4142,
+      "step": 155480
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014283109669486316,
+      "loss": 0.445,
+      "step": 155490
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0001427886959655706,
+      "loss": 0.3727,
+      "step": 155500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014274629523627807,
+      "loss": 0.3979,
+      "step": 155510
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014270389450698554,
+      "loss": 0.497,
+      "step": 155520
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014266149377769298,
+      "loss": 0.4644,
+      "step": 155530
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014261909304840044,
+      "loss": 0.4732,
+      "step": 155540
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014257669231910788,
+      "loss": 0.4872,
+      "step": 155550
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014253429158981535,
+      "loss": 0.3759,
+      "step": 155560
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014249189086052282,
+      "loss": 0.4376,
+      "step": 155570
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014244949013123026,
+      "loss": 0.4252,
+      "step": 155580
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014240708940193772,
+      "loss": 0.384,
+      "step": 155590
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0001423646886726452,
+      "loss": 0.3531,
+      "step": 155600
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014232228794335263,
+      "loss": 0.3937,
+      "step": 155610
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0001422798872140601,
+      "loss": 0.4507,
+      "step": 155620
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014223748648476754,
+      "loss": 0.3388,
+      "step": 155630
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.000142195085755475,
+      "loss": 0.4306,
+      "step": 155640
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014215268502618247,
+      "loss": 0.379,
+      "step": 155650
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0001421102842968899,
+      "loss": 0.4185,
+      "step": 155660
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014206788356759738,
+      "loss": 0.3971,
+      "step": 155670
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014202548283830484,
+      "loss": 0.4256,
+      "step": 155680
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014198308210901228,
+      "loss": 0.4495,
+      "step": 155690
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014194068137971975,
+      "loss": 0.3932,
+      "step": 155700
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0001418982806504272,
+      "loss": 0.3911,
+      "step": 155710
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014185587992113466,
+      "loss": 0.4569,
+      "step": 155720
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014181347919184212,
+      "loss": 0.3629,
+      "step": 155730
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014177107846254956,
+      "loss": 0.4549,
+      "step": 155740
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014172867773325703,
+      "loss": 0.3505,
+      "step": 155750
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001416862770039645,
+      "loss": 0.4087,
+      "step": 155760
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014164387627467194,
+      "loss": 0.4262,
+      "step": 155770
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001416014755453794,
+      "loss": 0.446,
+      "step": 155780
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014155907481608684,
+      "loss": 0.3926,
+      "step": 155790
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001415166740867943,
+      "loss": 0.3797,
+      "step": 155800
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014147427335750178,
+      "loss": 0.3803,
+      "step": 155810
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014143187262820922,
+      "loss": 0.4817,
+      "step": 155820
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014138947189891668,
+      "loss": 0.4268,
+      "step": 155830
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014134707116962412,
+      "loss": 0.438,
+      "step": 155840
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001413046704403316,
+      "loss": 0.4293,
+      "step": 155850
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014126226971103903,
+      "loss": 0.417,
+      "step": 155860
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001412198689817465,
+      "loss": 0.4005,
+      "step": 155870
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014117746825245394,
+      "loss": 0.365,
+      "step": 155880
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001411350675231614,
+      "loss": 0.397,
+      "step": 155890
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014109266679386884,
+      "loss": 0.4522,
+      "step": 155900
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001410502660645763,
+      "loss": 0.3719,
+      "step": 155910
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014100786533528378,
+      "loss": 0.4545,
+      "step": 155920
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014096546460599122,
+      "loss": 0.4243,
+      "step": 155930
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014092306387669868,
+      "loss": 0.3802,
+      "step": 155940
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014088066314740612,
+      "loss": 0.3465,
+      "step": 155950
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001408382624181136,
+      "loss": 0.386,
+      "step": 155960
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014079586168882106,
+      "loss": 0.4301,
+      "step": 155970
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001407534609595285,
+      "loss": 0.4452,
+      "step": 155980
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014071106023023596,
+      "loss": 0.4097,
+      "step": 155990
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014066865950094343,
+      "loss": 0.3449,
+      "step": 156000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014062625877165087,
+      "loss": 0.4236,
+      "step": 156010
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014058385804235833,
+      "loss": 0.4394,
+      "step": 156020
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014054145731306577,
+      "loss": 0.4083,
+      "step": 156030
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014049905658377324,
+      "loss": 0.4244,
+      "step": 156040
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001404566558544807,
+      "loss": 0.4855,
+      "step": 156050
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014041425512518815,
+      "loss": 0.4104,
+      "step": 156060
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014037185439589561,
+      "loss": 0.3873,
+      "step": 156070
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014032945366660305,
+      "loss": 0.3976,
+      "step": 156080
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014028705293731052,
+      "loss": 0.3742,
+      "step": 156090
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.000140244652208018,
+      "loss": 0.4321,
+      "step": 156100
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014020225147872543,
+      "loss": 0.3523,
+      "step": 156110
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001401598507494329,
+      "loss": 0.3928,
+      "step": 156120
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014011745002014036,
+      "loss": 0.4285,
+      "step": 156130
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001400750492908478,
+      "loss": 0.4159,
+      "step": 156140
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014003264856155527,
+      "loss": 0.3922,
+      "step": 156150
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001399902478322627,
+      "loss": 0.4255,
+      "step": 156160
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00013994784710297017,
+      "loss": 0.3468,
+      "step": 156170
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00013990544637367764,
+      "loss": 0.3697,
+      "step": 156180
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013986304564438508,
+      "loss": 0.4387,
+      "step": 156190
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013982064491509255,
+      "loss": 0.5188,
+      "step": 156200
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013977824418580001,
+      "loss": 0.3661,
+      "step": 156210
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013973584345650745,
+      "loss": 0.3888,
+      "step": 156220
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013969344272721492,
+      "loss": 0.3922,
+      "step": 156230
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013965104199792236,
+      "loss": 0.3986,
+      "step": 156240
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013960864126862983,
+      "loss": 0.4038,
+      "step": 156250
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0001395662405393373,
+      "loss": 0.417,
+      "step": 156260
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013952383981004473,
+      "loss": 0.3803,
+      "step": 156270
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0001394814390807522,
+      "loss": 0.4705,
+      "step": 156280
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013943903835145967,
+      "loss": 0.4111,
+      "step": 156290
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0001393966376221671,
+      "loss": 0.381,
+      "step": 156300
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013935423689287457,
+      "loss": 0.4999,
+      "step": 156310
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.000139311836163582,
+      "loss": 0.3993,
+      "step": 156320
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013926943543428948,
+      "loss": 0.4185,
+      "step": 156330
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013922703470499695,
+      "loss": 0.4119,
+      "step": 156340
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0001391846339757044,
+      "loss": 0.4426,
+      "step": 156350
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013914223324641185,
+      "loss": 0.4188,
+      "step": 156360
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013909983251711932,
+      "loss": 0.392,
+      "step": 156370
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013905743178782676,
+      "loss": 0.4337,
+      "step": 156380
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013901503105853423,
+      "loss": 0.375,
+      "step": 156390
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013897263032924167,
+      "loss": 0.4631,
+      "step": 156400
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013893022959994913,
+      "loss": 0.3621,
+      "step": 156410
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0001388878288706566,
+      "loss": 0.3422,
+      "step": 156420
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013884542814136404,
+      "loss": 0.3714,
+      "step": 156430
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001388030274120715,
+      "loss": 0.4051,
+      "step": 156440
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013876062668277897,
+      "loss": 0.3895,
+      "step": 156450
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001387182259534864,
+      "loss": 0.4554,
+      "step": 156460
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013867582522419388,
+      "loss": 0.4039,
+      "step": 156470
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013863342449490132,
+      "loss": 0.4298,
+      "step": 156480
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013859102376560879,
+      "loss": 0.4213,
+      "step": 156490
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013854862303631623,
+      "loss": 0.4452,
+      "step": 156500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001385062223070237,
+      "loss": 0.3859,
+      "step": 156510
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013846382157773113,
+      "loss": 0.3814,
+      "step": 156520
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001384214208484386,
+      "loss": 0.4438,
+      "step": 156530
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013837902011914607,
+      "loss": 0.3728,
+      "step": 156540
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001383366193898535,
+      "loss": 0.416,
+      "step": 156550
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013829421866056097,
+      "loss": 0.4377,
+      "step": 156560
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001382518179312684,
+      "loss": 0.4262,
+      "step": 156570
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013820941720197588,
+      "loss": 0.3679,
+      "step": 156580
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013816701647268332,
+      "loss": 0.4706,
+      "step": 156590
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013812461574339079,
+      "loss": 0.4175,
+      "step": 156600
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013808221501409825,
+      "loss": 0.3575,
+      "step": 156610
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001380398142848057,
+      "loss": 0.382,
+      "step": 156620
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013799741355551316,
+      "loss": 0.4101,
+      "step": 156630
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001379550128262206,
+      "loss": 0.4137,
+      "step": 156640
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013791261209692806,
+      "loss": 0.3276,
+      "step": 156650
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013787021136763553,
+      "loss": 0.4347,
+      "step": 156660
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013782781063834297,
+      "loss": 0.4109,
+      "step": 156670
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013778540990905044,
+      "loss": 0.425,
+      "step": 156680
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0001377430091797579,
+      "loss": 0.3872,
+      "step": 156690
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013770060845046534,
+      "loss": 0.4144,
+      "step": 156700
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0001376582077211728,
+      "loss": 0.4498,
+      "step": 156710
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013761580699188025,
+      "loss": 0.4489,
+      "step": 156720
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013757340626258772,
+      "loss": 0.4037,
+      "step": 156730
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013753100553329518,
+      "loss": 0.4566,
+      "step": 156740
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013748860480400262,
+      "loss": 0.4779,
+      "step": 156750
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0001374462040747101,
+      "loss": 0.4171,
+      "step": 156760
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013740380334541756,
+      "loss": 0.4269,
+      "step": 156770
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.000137361402616125,
+      "loss": 0.4155,
+      "step": 156780
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013731900188683246,
+      "loss": 0.396,
+      "step": 156790
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0001372766011575399,
+      "loss": 0.4153,
+      "step": 156800
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013723420042824737,
+      "loss": 0.4422,
+      "step": 156810
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013719179969895484,
+      "loss": 0.3292,
+      "step": 156820
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013714939896966228,
+      "loss": 0.4496,
+      "step": 156830
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013710699824036974,
+      "loss": 0.4354,
+      "step": 156840
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013706459751107718,
+      "loss": 0.4034,
+      "step": 156850
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013702219678178465,
+      "loss": 0.3553,
+      "step": 156860
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013697979605249212,
+      "loss": 0.4574,
+      "step": 156870
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013693739532319956,
+      "loss": 0.3534,
+      "step": 156880
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013689499459390702,
+      "loss": 0.3818,
+      "step": 156890
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001368525938646145,
+      "loss": 0.3899,
+      "step": 156900
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013681019313532193,
+      "loss": 0.4219,
+      "step": 156910
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001367677924060294,
+      "loss": 0.4276,
+      "step": 156920
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013672539167673684,
+      "loss": 0.431,
+      "step": 156930
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001366829909474443,
+      "loss": 0.4465,
+      "step": 156940
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013664059021815177,
+      "loss": 0.5243,
+      "step": 156950
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001365981894888592,
+      "loss": 0.4504,
+      "step": 156960
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013655578875956668,
+      "loss": 0.3496,
+      "step": 156970
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013651338803027414,
+      "loss": 0.3399,
+      "step": 156980
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013647098730098158,
+      "loss": 0.4531,
+      "step": 156990
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013642858657168905,
+      "loss": 0.4082,
+      "step": 157000
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001363861858423965,
+      "loss": 0.3763,
+      "step": 157010
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013634378511310396,
+      "loss": 0.4638,
+      "step": 157020
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013630138438381142,
+      "loss": 0.3962,
+      "step": 157030
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013625898365451886,
+      "loss": 0.4232,
+      "step": 157040
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013621658292522633,
+      "loss": 0.432,
+      "step": 157050
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001361741821959338,
+      "loss": 0.3658,
+      "step": 157060
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013613178146664124,
+      "loss": 0.3691,
+      "step": 157070
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001360893807373487,
+      "loss": 0.465,
+      "step": 157080
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013604698000805614,
+      "loss": 0.449,
+      "step": 157090
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001360045792787636,
+      "loss": 0.3939,
+      "step": 157100
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013596217854947108,
+      "loss": 0.3934,
+      "step": 157110
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013591977782017852,
+      "loss": 0.4863,
+      "step": 157120
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013587737709088598,
+      "loss": 0.3602,
+      "step": 157130
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013583497636159342,
+      "loss": 0.4319,
+      "step": 157140
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001357925756323009,
+      "loss": 0.4623,
+      "step": 157150
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013575017490300833,
+      "loss": 0.4165,
+      "step": 157160
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001357077741737158,
+      "loss": 0.3893,
+      "step": 157170
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013566537344442326,
+      "loss": 0.3902,
+      "step": 157180
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001356229727151307,
+      "loss": 0.4786,
+      "step": 157190
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013558057198583817,
+      "loss": 0.4449,
+      "step": 157200
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001355381712565456,
+      "loss": 0.3804,
+      "step": 157210
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013549577052725308,
+      "loss": 0.4625,
+      "step": 157220
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013545336979796052,
+      "loss": 0.4768,
+      "step": 157230
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013541096906866798,
+      "loss": 0.4556,
+      "step": 157240
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013536856833937542,
+      "loss": 0.3901,
+      "step": 157250
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001353261676100829,
+      "loss": 0.4641,
+      "step": 157260
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013528376688079036,
+      "loss": 0.4532,
+      "step": 157270
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001352413661514978,
+      "loss": 0.4577,
+      "step": 157280
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013519896542220526,
+      "loss": 0.4378,
+      "step": 157290
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013515656469291273,
+      "loss": 0.4267,
+      "step": 157300
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013511416396362017,
+      "loss": 0.3574,
+      "step": 157310
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013507176323432764,
+      "loss": 0.4278,
+      "step": 157320
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013502936250503507,
+      "loss": 0.4179,
+      "step": 157330
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013498696177574254,
+      "loss": 0.4536,
+      "step": 157340
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013494456104645,
+      "loss": 0.3985,
+      "step": 157350
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013490216031715745,
+      "loss": 0.3994,
+      "step": 157360
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013485975958786491,
+      "loss": 0.3287,
+      "step": 157370
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013481735885857238,
+      "loss": 0.3811,
+      "step": 157380
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013477495812927982,
+      "loss": 0.4357,
+      "step": 157390
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001347325573999873,
+      "loss": 0.4021,
+      "step": 157400
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013469015667069473,
+      "loss": 0.4149,
+      "step": 157410
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001346477559414022,
+      "loss": 0.4983,
+      "step": 157420
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013460535521210966,
+      "loss": 0.3803,
+      "step": 157430
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001345629544828171,
+      "loss": 0.3834,
+      "step": 157440
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013452055375352457,
+      "loss": 0.414,
+      "step": 157450
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013447815302423203,
+      "loss": 0.4495,
+      "step": 157460
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013443575229493947,
+      "loss": 0.3874,
+      "step": 157470
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013439335156564694,
+      "loss": 0.4013,
+      "step": 157480
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013435095083635438,
+      "loss": 0.41,
+      "step": 157490
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013430855010706185,
+      "loss": 0.4104,
+      "step": 157500
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013426614937776931,
+      "loss": 0.4052,
+      "step": 157510
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013422374864847675,
+      "loss": 0.3576,
+      "step": 157520
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013418134791918422,
+      "loss": 0.4453,
+      "step": 157530
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013413894718989166,
+      "loss": 0.4459,
+      "step": 157540
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013409654646059913,
+      "loss": 0.4347,
+      "step": 157550
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001340541457313066,
+      "loss": 0.4077,
+      "step": 157560
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013401174500201403,
+      "loss": 0.3983,
+      "step": 157570
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001339693442727215,
+      "loss": 0.4659,
+      "step": 157580
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013392694354342897,
+      "loss": 0.4566,
+      "step": 157590
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001338845428141364,
+      "loss": 0.4533,
+      "step": 157600
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013384214208484387,
+      "loss": 0.4147,
+      "step": 157610
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001337997413555513,
+      "loss": 0.4302,
+      "step": 157620
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013375734062625878,
+      "loss": 0.4155,
+      "step": 157630
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013371493989696625,
+      "loss": 0.4178,
+      "step": 157640
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001336725391676737,
+      "loss": 0.4108,
+      "step": 157650
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013363013843838115,
+      "loss": 0.4093,
+      "step": 157660
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013358773770908862,
+      "loss": 0.3979,
+      "step": 157670
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013354533697979606,
+      "loss": 0.4611,
+      "step": 157680
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013350293625050353,
+      "loss": 0.4468,
+      "step": 157690
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013346053552121097,
+      "loss": 0.3595,
+      "step": 157700
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013341813479191843,
+      "loss": 0.4491,
+      "step": 157710
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001333757340626259,
+      "loss": 0.4017,
+      "step": 157720
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.3983,
+      "step": 157730
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001332909326040408,
+      "loss": 0.451,
+      "step": 157740
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013324853187474827,
+      "loss": 0.4129,
+      "step": 157750
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001332061311454557,
+      "loss": 0.4005,
+      "step": 157760
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013316373041616318,
+      "loss": 0.3446,
+      "step": 157770
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013312132968687062,
+      "loss": 0.4444,
+      "step": 157780
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013307892895757809,
+      "loss": 0.465,
+      "step": 157790
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013303652822828555,
+      "loss": 0.4865,
+      "step": 157800
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.000132994127498993,
+      "loss": 0.3962,
+      "step": 157810
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013295172676970046,
+      "loss": 0.3968,
+      "step": 157820
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001329093260404079,
+      "loss": 0.4275,
+      "step": 157830
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013286692531111537,
+      "loss": 0.4522,
+      "step": 157840
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001328245245818228,
+      "loss": 0.4037,
+      "step": 157850
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013278212385253027,
+      "loss": 0.3622,
+      "step": 157860
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001327397231232377,
+      "loss": 0.378,
+      "step": 157870
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013269732239394518,
+      "loss": 0.3909,
+      "step": 157880
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013265492166465262,
+      "loss": 0.4807,
+      "step": 157890
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013261252093536009,
+      "loss": 0.4552,
+      "step": 157900
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013257012020606755,
+      "loss": 0.416,
+      "step": 157910
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.000132527719476775,
+      "loss": 0.418,
+      "step": 157920
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013248531874748246,
+      "loss": 0.4083,
+      "step": 157930
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001324429180181899,
+      "loss": 0.4571,
+      "step": 157940
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013240051728889737,
+      "loss": 0.4415,
+      "step": 157950
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013235811655960483,
+      "loss": 0.4318,
+      "step": 157960
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013231571583031227,
+      "loss": 0.4797,
+      "step": 157970
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013227331510101974,
+      "loss": 0.3786,
+      "step": 157980
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001322309143717272,
+      "loss": 0.4037,
+      "step": 157990
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013218851364243464,
+      "loss": 0.4288,
+      "step": 158000
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001321461129131421,
+      "loss": 0.3957,
+      "step": 158010
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013210371218384955,
+      "loss": 0.4093,
+      "step": 158020
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013206131145455702,
+      "loss": 0.3776,
+      "step": 158030
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013201891072526448,
+      "loss": 0.3682,
+      "step": 158040
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013197650999597192,
+      "loss": 0.3536,
+      "step": 158050
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001319341092666794,
+      "loss": 0.4011,
+      "step": 158060
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013189170853738686,
+      "loss": 0.3939,
+      "step": 158070
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001318493078080943,
+      "loss": 0.3925,
+      "step": 158080
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013180690707880176,
+      "loss": 0.4163,
+      "step": 158090
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001317645063495092,
+      "loss": 0.407,
+      "step": 158100
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013172210562021667,
+      "loss": 0.3928,
+      "step": 158110
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013167970489092414,
+      "loss": 0.4379,
+      "step": 158120
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013163730416163158,
+      "loss": 0.3936,
+      "step": 158130
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013159490343233904,
+      "loss": 0.4168,
+      "step": 158140
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001315525027030465,
+      "loss": 0.3904,
+      "step": 158150
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013151010197375395,
+      "loss": 0.3685,
+      "step": 158160
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013146770124446142,
+      "loss": 0.5109,
+      "step": 158170
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013142530051516886,
+      "loss": 0.3505,
+      "step": 158180
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013138289978587632,
+      "loss": 0.4148,
+      "step": 158190
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001313404990565838,
+      "loss": 0.4117,
+      "step": 158200
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013129809832729123,
+      "loss": 0.3992,
+      "step": 158210
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001312556975979987,
+      "loss": 0.4807,
+      "step": 158220
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013121329686870614,
+      "loss": 0.4373,
+      "step": 158230
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001311708961394136,
+      "loss": 0.3259,
+      "step": 158240
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013112849541012107,
+      "loss": 0.3954,
+      "step": 158250
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001310860946808285,
+      "loss": 0.4347,
+      "step": 158260
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013104369395153598,
+      "loss": 0.4509,
+      "step": 158270
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013100129322224344,
+      "loss": 0.4512,
+      "step": 158280
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013095889249295088,
+      "loss": 0.435,
+      "step": 158290
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013091649176365835,
+      "loss": 0.4352,
+      "step": 158300
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001308740910343658,
+      "loss": 0.4161,
+      "step": 158310
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013083169030507326,
+      "loss": 0.3599,
+      "step": 158320
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013078928957578072,
+      "loss": 0.4158,
+      "step": 158330
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013074688884648816,
+      "loss": 0.4042,
+      "step": 158340
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013070448811719563,
+      "loss": 0.3857,
+      "step": 158350
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001306620873879031,
+      "loss": 0.4633,
+      "step": 158360
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013061968665861054,
+      "loss": 0.4538,
+      "step": 158370
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.000130577285929318,
+      "loss": 0.4081,
+      "step": 158380
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013053488520002544,
+      "loss": 0.4473,
+      "step": 158390
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001304924844707329,
+      "loss": 0.4251,
+      "step": 158400
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013045008374144038,
+      "loss": 0.4632,
+      "step": 158410
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013040768301214782,
+      "loss": 0.4083,
+      "step": 158420
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013036528228285528,
+      "loss": 0.4386,
+      "step": 158430
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013032288155356275,
+      "loss": 0.4343,
+      "step": 158440
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001302804808242702,
+      "loss": 0.3543,
+      "step": 158450
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013023808009497766,
+      "loss": 0.4665,
+      "step": 158460
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001301956793656851,
+      "loss": 0.452,
+      "step": 158470
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013015327863639256,
+      "loss": 0.3792,
+      "step": 158480
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001301108779071,
+      "loss": 0.4421,
+      "step": 158490
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013006847717780747,
+      "loss": 0.412,
+      "step": 158500
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001300260764485149,
+      "loss": 0.3561,
+      "step": 158510
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00012998367571922238,
+      "loss": 0.4143,
+      "step": 158520
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00012994127498992982,
+      "loss": 0.4288,
+      "step": 158530
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00012989887426063728,
+      "loss": 0.4681,
+      "step": 158540
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00012985647353134475,
+      "loss": 0.4353,
+      "step": 158550
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001298140728020522,
+      "loss": 0.4049,
+      "step": 158560
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012977167207275966,
+      "loss": 0.4624,
+      "step": 158570
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001297292713434671,
+      "loss": 0.3986,
+      "step": 158580
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012968687061417456,
+      "loss": 0.4676,
+      "step": 158590
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012964446988488203,
+      "loss": 0.4308,
+      "step": 158600
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012960206915558947,
+      "loss": 0.4365,
+      "step": 158610
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012955966842629694,
+      "loss": 0.4432,
+      "step": 158620
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012951726769700437,
+      "loss": 0.4392,
+      "step": 158630
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012947486696771184,
+      "loss": 0.3623,
+      "step": 158640
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001294324662384193,
+      "loss": 0.4553,
+      "step": 158650
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012939006550912675,
+      "loss": 0.3848,
+      "step": 158660
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012934766477983421,
+      "loss": 0.3446,
+      "step": 158670
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012930526405054168,
+      "loss": 0.4102,
+      "step": 158680
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012926286332124912,
+      "loss": 0.4517,
+      "step": 158690
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001292204625919566,
+      "loss": 0.3269,
+      "step": 158700
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012917806186266403,
+      "loss": 0.4255,
+      "step": 158710
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001291356611333715,
+      "loss": 0.4326,
+      "step": 158720
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012909326040407896,
+      "loss": 0.4231,
+      "step": 158730
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001290508596747864,
+      "loss": 0.4145,
+      "step": 158740
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012900845894549387,
+      "loss": 0.4713,
+      "step": 158750
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012896605821620133,
+      "loss": 0.4295,
+      "step": 158760
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012892365748690877,
+      "loss": 0.3763,
+      "step": 158770
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012888125675761624,
+      "loss": 0.3811,
+      "step": 158780
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012883885602832368,
+      "loss": 0.4104,
+      "step": 158790
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012879645529903115,
+      "loss": 0.4254,
+      "step": 158800
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012875405456973861,
+      "loss": 0.3958,
+      "step": 158810
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012871165384044605,
+      "loss": 0.4489,
+      "step": 158820
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012866925311115352,
+      "loss": 0.3966,
+      "step": 158830
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.000128626852381861,
+      "loss": 0.4117,
+      "step": 158840
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012858445165256843,
+      "loss": 0.4345,
+      "step": 158850
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0001285420509232759,
+      "loss": 0.3889,
+      "step": 158860
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012849965019398333,
+      "loss": 0.4425,
+      "step": 158870
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0001284572494646908,
+      "loss": 0.4487,
+      "step": 158880
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012841484873539827,
+      "loss": 0.4026,
+      "step": 158890
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0001283724480061057,
+      "loss": 0.4299,
+      "step": 158900
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012833004727681317,
+      "loss": 0.3529,
+      "step": 158910
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012828764654752061,
+      "loss": 0.4214,
+      "step": 158920
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012824524581822808,
+      "loss": 0.4637,
+      "step": 158930
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012820284508893555,
+      "loss": 0.396,
+      "step": 158940
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.000128160444359643,
+      "loss": 0.4552,
+      "step": 158950
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012811804363035045,
+      "loss": 0.4131,
+      "step": 158960
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012807564290105792,
+      "loss": 0.42,
+      "step": 158970
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012803324217176536,
+      "loss": 0.3663,
+      "step": 158980
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012799084144247283,
+      "loss": 0.4124,
+      "step": 158990
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012794844071318027,
+      "loss": 0.4342,
+      "step": 159000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012790603998388773,
+      "loss": 0.3737,
+      "step": 159010
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0001278636392545952,
+      "loss": 0.4225,
+      "step": 159020
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012782123852530264,
+      "loss": 0.4392,
+      "step": 159030
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001277788377960101,
+      "loss": 0.4131,
+      "step": 159040
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012773643706671757,
+      "loss": 0.4228,
+      "step": 159050
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.000127694036337425,
+      "loss": 0.4758,
+      "step": 159060
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012765163560813248,
+      "loss": 0.3479,
+      "step": 159070
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012760923487883992,
+      "loss": 0.4733,
+      "step": 159080
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001275668341495474,
+      "loss": 0.4288,
+      "step": 159090
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012752443342025485,
+      "loss": 0.4685,
+      "step": 159100
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001274820326909623,
+      "loss": 0.4455,
+      "step": 159110
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012743963196166976,
+      "loss": 0.4536,
+      "step": 159120
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001273972312323772,
+      "loss": 0.3953,
+      "step": 159130
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012735483050308467,
+      "loss": 0.4363,
+      "step": 159140
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001273124297737921,
+      "loss": 0.3678,
+      "step": 159150
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012727002904449957,
+      "loss": 0.4257,
+      "step": 159160
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.000127227628315207,
+      "loss": 0.4002,
+      "step": 159170
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012718522758591448,
+      "loss": 0.4228,
+      "step": 159180
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012714282685662195,
+      "loss": 0.3618,
+      "step": 159190
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012710042612732939,
+      "loss": 0.411,
+      "step": 159200
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012705802539803685,
+      "loss": 0.3611,
+      "step": 159210
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001270156246687443,
+      "loss": 0.4122,
+      "step": 159220
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012697322393945176,
+      "loss": 0.4197,
+      "step": 159230
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001269308232101592,
+      "loss": 0.4736,
+      "step": 159240
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012688842248086667,
+      "loss": 0.4356,
+      "step": 159250
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012684602175157413,
+      "loss": 0.342,
+      "step": 159260
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012680362102228157,
+      "loss": 0.3973,
+      "step": 159270
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012676122029298904,
+      "loss": 0.325,
+      "step": 159280
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001267188195636965,
+      "loss": 0.429,
+      "step": 159290
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012667641883440395,
+      "loss": 0.4751,
+      "step": 159300
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001266340181051114,
+      "loss": 0.4077,
+      "step": 159310
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012659161737581885,
+      "loss": 0.4727,
+      "step": 159320
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012654921664652632,
+      "loss": 0.4407,
+      "step": 159330
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012650681591723379,
+      "loss": 0.4166,
+      "step": 159340
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012646441518794122,
+      "loss": 0.4489,
+      "step": 159350
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001264220144586487,
+      "loss": 0.4584,
+      "step": 159360
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012637961372935616,
+      "loss": 0.4244,
+      "step": 159370
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001263372130000636,
+      "loss": 0.4072,
+      "step": 159380
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012629481227077106,
+      "loss": 0.4198,
+      "step": 159390
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001262524115414785,
+      "loss": 0.4104,
+      "step": 159400
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012621001081218597,
+      "loss": 0.4027,
+      "step": 159410
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012616761008289344,
+      "loss": 0.3846,
+      "step": 159420
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012612520935360088,
+      "loss": 0.3633,
+      "step": 159430
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012608280862430834,
+      "loss": 0.457,
+      "step": 159440
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001260404078950158,
+      "loss": 0.3168,
+      "step": 159450
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012599800716572325,
+      "loss": 0.4067,
+      "step": 159460
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012595560643643072,
+      "loss": 0.4226,
+      "step": 159470
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012591320570713816,
+      "loss": 0.4864,
+      "step": 159480
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012587080497784562,
+      "loss": 0.4166,
+      "step": 159490
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001258284042485531,
+      "loss": 0.3989,
+      "step": 159500
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012578600351926053,
+      "loss": 0.4028,
+      "step": 159510
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.000125743602789968,
+      "loss": 0.3738,
+      "step": 159520
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012570120206067546,
+      "loss": 0.424,
+      "step": 159530
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001256588013313829,
+      "loss": 0.4032,
+      "step": 159540
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012561640060209037,
+      "loss": 0.3911,
+      "step": 159550
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001255739998727978,
+      "loss": 0.4763,
+      "step": 159560
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012553159914350528,
+      "loss": 0.3946,
+      "step": 159570
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012548919841421274,
+      "loss": 0.4117,
+      "step": 159580
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012544679768492018,
+      "loss": 0.442,
+      "step": 159590
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012540439695562765,
+      "loss": 0.4974,
+      "step": 159600
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012536199622633512,
+      "loss": 0.3765,
+      "step": 159610
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012531959549704256,
+      "loss": 0.4534,
+      "step": 159620
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012527719476775002,
+      "loss": 0.3846,
+      "step": 159630
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012523479403845746,
+      "loss": 0.3766,
+      "step": 159640
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012519239330916493,
+      "loss": 0.4488,
+      "step": 159650
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001251499925798724,
+      "loss": 0.3595,
+      "step": 159660
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012510759185057984,
+      "loss": 0.4384,
+      "step": 159670
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001250651911212873,
+      "loss": 0.4275,
+      "step": 159680
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012502279039199474,
+      "loss": 0.3955,
+      "step": 159690
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001249803896627022,
+      "loss": 0.4127,
+      "step": 159700
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012493798893340968,
+      "loss": 0.4505,
+      "step": 159710
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012489558820411712,
+      "loss": 0.448,
+      "step": 159720
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012485318747482458,
+      "loss": 0.434,
+      "step": 159730
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012481078674553205,
+      "loss": 0.4885,
+      "step": 159740
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0001247683860162395,
+      "loss": 0.4095,
+      "step": 159750
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012472598528694696,
+      "loss": 0.418,
+      "step": 159760
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0001246835845576544,
+      "loss": 0.4174,
+      "step": 159770
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012464118382836186,
+      "loss": 0.4013,
+      "step": 159780
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0001245987830990693,
+      "loss": 0.3361,
+      "step": 159790
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012455638236977677,
+      "loss": 0.3713,
+      "step": 159800
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012451398164048424,
+      "loss": 0.4141,
+      "step": 159810
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012447158091119168,
+      "loss": 0.5273,
+      "step": 159820
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012442918018189914,
+      "loss": 0.395,
+      "step": 159830
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012438677945260658,
+      "loss": 0.4621,
+      "step": 159840
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012434437872331405,
+      "loss": 0.3713,
+      "step": 159850
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0001243019779940215,
+      "loss": 0.452,
+      "step": 159860
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012425957726472896,
+      "loss": 0.4523,
+      "step": 159870
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0001242171765354364,
+      "loss": 0.3627,
+      "step": 159880
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012417477580614386,
+      "loss": 0.3808,
+      "step": 159890
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012413237507685133,
+      "loss": 0.4035,
+      "step": 159900
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012408997434755877,
+      "loss": 0.3686,
+      "step": 159910
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012404757361826624,
+      "loss": 0.3698,
+      "step": 159920
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012400517288897368,
+      "loss": 0.414,
+      "step": 159930
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012396277215968114,
+      "loss": 0.3679,
+      "step": 159940
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0001239203714303886,
+      "loss": 0.3984,
+      "step": 159950
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012387797070109605,
+      "loss": 0.4771,
+      "step": 159960
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012383556997180352,
+      "loss": 0.3626,
+      "step": 159970
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012379316924251098,
+      "loss": 0.4223,
+      "step": 159980
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012375076851321842,
+      "loss": 0.3674,
+      "step": 159990
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0001237083677839259,
+      "loss": 0.4429,
+      "step": 160000
+    },
+    {
+      "epoch": 6.77,
+      "eval_loss": 0.6193305253982544,
+      "eval_runtime": 337.814,
+      "eval_samples_per_second": 15.556,
+      "eval_steps_per_second": 3.89,
+      "step": 160000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012366596705463333,
+      "loss": 0.3586,
+      "step": 160010
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0001236235663253408,
+      "loss": 0.4581,
+      "step": 160020
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012358116559604826,
+      "loss": 0.4582,
+      "step": 160030
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0001235387648667557,
+      "loss": 0.4384,
+      "step": 160040
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012349636413746317,
+      "loss": 0.4175,
+      "step": 160050
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012345396340817063,
+      "loss": 0.4511,
+      "step": 160060
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012341156267887807,
+      "loss": 0.4035,
+      "step": 160070
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012336916194958554,
+      "loss": 0.3402,
+      "step": 160080
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012332676122029298,
+      "loss": 0.4397,
+      "step": 160090
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012328436049100045,
+      "loss": 0.4377,
+      "step": 160100
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012324195976170791,
+      "loss": 0.3871,
+      "step": 160110
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012319955903241535,
+      "loss": 0.4212,
+      "step": 160120
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012315715830312282,
+      "loss": 0.407,
+      "step": 160130
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0001231147575738303,
+      "loss": 0.4141,
+      "step": 160140
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012307235684453773,
+      "loss": 0.3954,
+      "step": 160150
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0001230299561152452,
+      "loss": 0.448,
+      "step": 160160
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012298755538595263,
+      "loss": 0.4154,
+      "step": 160170
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0001229451546566601,
+      "loss": 0.4147,
+      "step": 160180
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012290275392736757,
+      "loss": 0.3833,
+      "step": 160190
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.000122860353198075,
+      "loss": 0.4365,
+      "step": 160200
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012281795246878247,
+      "loss": 0.4368,
+      "step": 160210
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012277555173948994,
+      "loss": 0.4194,
+      "step": 160220
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012273315101019738,
+      "loss": 0.432,
+      "step": 160230
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012269075028090485,
+      "loss": 0.4702,
+      "step": 160240
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0001226483495516123,
+      "loss": 0.3777,
+      "step": 160250
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012260594882231975,
+      "loss": 0.4414,
+      "step": 160260
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012256354809302722,
+      "loss": 0.3737,
+      "step": 160270
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012252114736373466,
+      "loss": 0.5201,
+      "step": 160280
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012247874663444213,
+      "loss": 0.4046,
+      "step": 160290
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0001224363459051496,
+      "loss": 0.4428,
+      "step": 160300
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012239394517585703,
+      "loss": 0.4452,
+      "step": 160310
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0001223515444465645,
+      "loss": 0.4505,
+      "step": 160320
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012230914371727194,
+      "loss": 0.355,
+      "step": 160330
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0001222667429879794,
+      "loss": 0.3732,
+      "step": 160340
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012222434225868687,
+      "loss": 0.4491,
+      "step": 160350
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0001221819415293943,
+      "loss": 0.4507,
+      "step": 160360
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012213954080010178,
+      "loss": 0.4283,
+      "step": 160370
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012209714007080922,
+      "loss": 0.4181,
+      "step": 160380
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012205473934151667,
+      "loss": 0.415,
+      "step": 160390
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012201233861222414,
+      "loss": 0.4026,
+      "step": 160400
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012196993788293158,
+      "loss": 0.5036,
+      "step": 160410
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012192753715363905,
+      "loss": 0.4076,
+      "step": 160420
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012188513642434651,
+      "loss": 0.3875,
+      "step": 160430
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012184273569505395,
+      "loss": 0.4771,
+      "step": 160440
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012180033496576142,
+      "loss": 0.4242,
+      "step": 160450
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012175793423646886,
+      "loss": 0.367,
+      "step": 160460
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012171553350717633,
+      "loss": 0.4615,
+      "step": 160470
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012167313277788379,
+      "loss": 0.4065,
+      "step": 160480
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012163073204859123,
+      "loss": 0.4618,
+      "step": 160490
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001215883313192987,
+      "loss": 0.3895,
+      "step": 160500
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012154593059000617,
+      "loss": 0.3979,
+      "step": 160510
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001215035298607136,
+      "loss": 0.4776,
+      "step": 160520
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012146112913142107,
+      "loss": 0.4243,
+      "step": 160530
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012141872840212851,
+      "loss": 0.3659,
+      "step": 160540
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012137632767283598,
+      "loss": 0.3813,
+      "step": 160550
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012133392694354345,
+      "loss": 0.3998,
+      "step": 160560
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012129152621425089,
+      "loss": 0.3773,
+      "step": 160570
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012124912548495835,
+      "loss": 0.3941,
+      "step": 160580
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012120672475566582,
+      "loss": 0.4783,
+      "step": 160590
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012116432402637326,
+      "loss": 0.4185,
+      "step": 160600
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012112192329708073,
+      "loss": 0.3467,
+      "step": 160610
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012107952256778817,
+      "loss": 0.4063,
+      "step": 160620
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012103712183849563,
+      "loss": 0.3242,
+      "step": 160630
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012099472110920309,
+      "loss": 0.4444,
+      "step": 160640
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012095232037991054,
+      "loss": 0.3551,
+      "step": 160650
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012090991965061799,
+      "loss": 0.392,
+      "step": 160660
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012086751892132546,
+      "loss": 0.3785,
+      "step": 160670
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001208251181920329,
+      "loss": 0.486,
+      "step": 160680
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012078271746274037,
+      "loss": 0.4681,
+      "step": 160690
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001207403167334478,
+      "loss": 0.4396,
+      "step": 160700
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012069791600415527,
+      "loss": 0.3874,
+      "step": 160710
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012065551527486274,
+      "loss": 0.4349,
+      "step": 160720
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012061311454557018,
+      "loss": 0.4447,
+      "step": 160730
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012057071381627764,
+      "loss": 0.435,
+      "step": 160740
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012052831308698511,
+      "loss": 0.4351,
+      "step": 160750
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012048591235769255,
+      "loss": 0.4847,
+      "step": 160760
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012044351162840002,
+      "loss": 0.4171,
+      "step": 160770
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012040111089910746,
+      "loss": 0.458,
+      "step": 160780
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012035871016981492,
+      "loss": 0.4465,
+      "step": 160790
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012031630944052239,
+      "loss": 0.4071,
+      "step": 160800
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012027390871122983,
+      "loss": 0.4501,
+      "step": 160810
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001202315079819373,
+      "loss": 0.4112,
+      "step": 160820
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012018910725264476,
+      "loss": 0.3814,
+      "step": 160830
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001201467065233522,
+      "loss": 0.4318,
+      "step": 160840
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012010430579405967,
+      "loss": 0.4259,
+      "step": 160850
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012006190506476711,
+      "loss": 0.4464,
+      "step": 160860
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012001950433547458,
+      "loss": 0.4427,
+      "step": 160870
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00011997710360618204,
+      "loss": 0.4449,
+      "step": 160880
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00011993470287688948,
+      "loss": 0.4828,
+      "step": 160890
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00011989230214759695,
+      "loss": 0.4887,
+      "step": 160900
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00011984990141830442,
+      "loss": 0.4053,
+      "step": 160910
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011980750068901186,
+      "loss": 0.4456,
+      "step": 160920
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011976509995971932,
+      "loss": 0.534,
+      "step": 160930
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011972269923042676,
+      "loss": 0.4499,
+      "step": 160940
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011968029850113423,
+      "loss": 0.4008,
+      "step": 160950
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011963789777184168,
+      "loss": 0.4225,
+      "step": 160960
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011959549704254914,
+      "loss": 0.3679,
+      "step": 160970
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011955309631325659,
+      "loss": 0.3583,
+      "step": 160980
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011951069558396406,
+      "loss": 0.4077,
+      "step": 160990
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0001194682948546715,
+      "loss": 0.4013,
+      "step": 161000
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011942589412537896,
+      "loss": 0.4033,
+      "step": 161010
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0001193834933960864,
+      "loss": 0.3432,
+      "step": 161020
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011934109266679387,
+      "loss": 0.5049,
+      "step": 161030
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011929869193750134,
+      "loss": 0.4097,
+      "step": 161040
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011925629120820878,
+      "loss": 0.4791,
+      "step": 161050
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011921389047891624,
+      "loss": 0.3714,
+      "step": 161060
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011917148974962368,
+      "loss": 0.3731,
+      "step": 161070
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011912908902033115,
+      "loss": 0.3103,
+      "step": 161080
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011908668829103862,
+      "loss": 0.4273,
+      "step": 161090
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011904428756174606,
+      "loss": 0.3363,
+      "step": 161100
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011900188683245352,
+      "loss": 0.3838,
+      "step": 161110
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011895948610316099,
+      "loss": 0.4587,
+      "step": 161120
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011891708537386843,
+      "loss": 0.4411,
+      "step": 161130
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0001188746846445759,
+      "loss": 0.3964,
+      "step": 161140
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011883228391528334,
+      "loss": 0.3941,
+      "step": 161150
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0001187898831859908,
+      "loss": 0.3977,
+      "step": 161160
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011874748245669827,
+      "loss": 0.3744,
+      "step": 161170
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011870508172740571,
+      "loss": 0.3732,
+      "step": 161180
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011866268099811318,
+      "loss": 0.3863,
+      "step": 161190
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011862028026882064,
+      "loss": 0.4746,
+      "step": 161200
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011857787953952808,
+      "loss": 0.4259,
+      "step": 161210
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011853547881023555,
+      "loss": 0.4235,
+      "step": 161220
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011849307808094299,
+      "loss": 0.4105,
+      "step": 161230
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011845067735165046,
+      "loss": 0.4454,
+      "step": 161240
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011840827662235792,
+      "loss": 0.3912,
+      "step": 161250
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011836587589306536,
+      "loss": 0.4487,
+      "step": 161260
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011832347516377283,
+      "loss": 0.4294,
+      "step": 161270
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011828107443448028,
+      "loss": 0.4226,
+      "step": 161280
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011823867370518774,
+      "loss": 0.4034,
+      "step": 161290
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011819627297589519,
+      "loss": 0.4225,
+      "step": 161300
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011815387224660264,
+      "loss": 0.4603,
+      "step": 161310
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0001181114715173101,
+      "loss": 0.4602,
+      "step": 161320
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011806907078801756,
+      "loss": 0.3774,
+      "step": 161330
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011802667005872502,
+      "loss": 0.3671,
+      "step": 161340
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011798426932943247,
+      "loss": 0.3734,
+      "step": 161350
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011794186860013994,
+      "loss": 0.4843,
+      "step": 161360
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011789946787084737,
+      "loss": 0.4343,
+      "step": 161370
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011785706714155484,
+      "loss": 0.4857,
+      "step": 161380
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011781466641226228,
+      "loss": 0.4503,
+      "step": 161390
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011777226568296975,
+      "loss": 0.4431,
+      "step": 161400
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011772986495367721,
+      "loss": 0.3792,
+      "step": 161410
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011768746422438465,
+      "loss": 0.4144,
+      "step": 161420
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011764506349509212,
+      "loss": 0.4078,
+      "step": 161430
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011760266276579959,
+      "loss": 0.459,
+      "step": 161440
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011756026203650703,
+      "loss": 0.4673,
+      "step": 161450
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001175178613072145,
+      "loss": 0.4341,
+      "step": 161460
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011747546057792193,
+      "loss": 0.4034,
+      "step": 161470
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001174330598486294,
+      "loss": 0.4057,
+      "step": 161480
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011739065911933687,
+      "loss": 0.4452,
+      "step": 161490
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011734825839004431,
+      "loss": 0.3836,
+      "step": 161500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011730585766075177,
+      "loss": 0.4964,
+      "step": 161510
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011726345693145924,
+      "loss": 0.436,
+      "step": 161520
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011722105620216668,
+      "loss": 0.4359,
+      "step": 161530
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011717865547287415,
+      "loss": 0.4204,
+      "step": 161540
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011713625474358159,
+      "loss": 0.4058,
+      "step": 161550
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011709385401428905,
+      "loss": 0.3547,
+      "step": 161560
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011705145328499652,
+      "loss": 0.3769,
+      "step": 161570
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011700905255570396,
+      "loss": 0.3594,
+      "step": 161580
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011696665182641143,
+      "loss": 0.3718,
+      "step": 161590
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011692425109711888,
+      "loss": 0.4571,
+      "step": 161600
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011688185036782633,
+      "loss": 0.4376,
+      "step": 161610
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011683944963853379,
+      "loss": 0.4902,
+      "step": 161620
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011679704890924124,
+      "loss": 0.4363,
+      "step": 161630
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.0001167546481799487,
+      "loss": 0.4485,
+      "step": 161640
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011671224745065616,
+      "loss": 0.4024,
+      "step": 161650
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011666984672136361,
+      "loss": 0.376,
+      "step": 161660
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011662744599207107,
+      "loss": 0.4457,
+      "step": 161670
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011658504526277853,
+      "loss": 0.4057,
+      "step": 161680
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011654264453348597,
+      "loss": 0.3863,
+      "step": 161690
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011650024380419344,
+      "loss": 0.3155,
+      "step": 161700
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011645784307490088,
+      "loss": 0.4464,
+      "step": 161710
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011641544234560835,
+      "loss": 0.3907,
+      "step": 161720
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011637304161631581,
+      "loss": 0.4546,
+      "step": 161730
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011633064088702325,
+      "loss": 0.4365,
+      "step": 161740
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011628824015773072,
+      "loss": 0.4225,
+      "step": 161750
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011624583942843819,
+      "loss": 0.4376,
+      "step": 161760
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011620343869914563,
+      "loss": 0.4492,
+      "step": 161770
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011616103796985309,
+      "loss": 0.374,
+      "step": 161780
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011611863724056053,
+      "loss": 0.381,
+      "step": 161790
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.000116076236511268,
+      "loss": 0.4006,
+      "step": 161800
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011603383578197547,
+      "loss": 0.3646,
+      "step": 161810
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.0001159914350526829,
+      "loss": 0.4144,
+      "step": 161820
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011594903432339037,
+      "loss": 0.5415,
+      "step": 161830
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011590663359409781,
+      "loss": 0.4381,
+      "step": 161840
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011586423286480528,
+      "loss": 0.3854,
+      "step": 161850
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011582183213551275,
+      "loss": 0.3917,
+      "step": 161860
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011577943140622019,
+      "loss": 0.423,
+      "step": 161870
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011573703067692765,
+      "loss": 0.3615,
+      "step": 161880
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011569462994763512,
+      "loss": 0.4412,
+      "step": 161890
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011565222921834256,
+      "loss": 0.4433,
+      "step": 161900
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011560982848905003,
+      "loss": 0.3654,
+      "step": 161910
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011556742775975747,
+      "loss": 0.3885,
+      "step": 161920
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011552502703046493,
+      "loss": 0.4587,
+      "step": 161930
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011548262630117239,
+      "loss": 0.4118,
+      "step": 161940
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011544022557187984,
+      "loss": 0.3761,
+      "step": 161950
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011539782484258729,
+      "loss": 0.4227,
+      "step": 161960
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011535542411329476,
+      "loss": 0.4481,
+      "step": 161970
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011531302338400221,
+      "loss": 0.4033,
+      "step": 161980
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011527062265470967,
+      "loss": 0.4975,
+      "step": 161990
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011522822192541712,
+      "loss": 0.4327,
+      "step": 162000
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011518582119612457,
+      "loss": 0.4076,
+      "step": 162010
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011514342046683204,
+      "loss": 0.4186,
+      "step": 162020
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011510101973753948,
+      "loss": 0.3624,
+      "step": 162030
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011505861900824695,
+      "loss": 0.3757,
+      "step": 162040
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011501621827895441,
+      "loss": 0.4108,
+      "step": 162050
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011497381754966185,
+      "loss": 0.397,
+      "step": 162060
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011493141682036932,
+      "loss": 0.4245,
+      "step": 162070
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011488901609107676,
+      "loss": 0.4892,
+      "step": 162080
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011484661536178422,
+      "loss": 0.4784,
+      "step": 162090
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011480421463249169,
+      "loss": 0.3936,
+      "step": 162100
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011476181390319913,
+      "loss": 0.3964,
+      "step": 162110
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0001147194131739066,
+      "loss": 0.2985,
+      "step": 162120
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011467701244461406,
+      "loss": 0.3979,
+      "step": 162130
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0001146346117153215,
+      "loss": 0.4363,
+      "step": 162140
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011459221098602897,
+      "loss": 0.4486,
+      "step": 162150
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011454981025673641,
+      "loss": 0.3953,
+      "step": 162160
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011450740952744388,
+      "loss": 0.3528,
+      "step": 162170
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011446500879815134,
+      "loss": 0.3992,
+      "step": 162180
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011442260806885878,
+      "loss": 0.4795,
+      "step": 162190
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011438020733956625,
+      "loss": 0.3902,
+      "step": 162200
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011433780661027372,
+      "loss": 0.394,
+      "step": 162210
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011429540588098116,
+      "loss": 0.3972,
+      "step": 162220
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011425300515168862,
+      "loss": 0.3891,
+      "step": 162230
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011421060442239606,
+      "loss": 0.4387,
+      "step": 162240
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011416820369310353,
+      "loss": 0.4759,
+      "step": 162250
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011412580296381098,
+      "loss": 0.4014,
+      "step": 162260
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011408340223451844,
+      "loss": 0.4274,
+      "step": 162270
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0001140410015052259,
+      "loss": 0.4149,
+      "step": 162280
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011399860077593336,
+      "loss": 0.4046,
+      "step": 162290
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011395620004664081,
+      "loss": 0.4047,
+      "step": 162300
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011391379931734826,
+      "loss": 0.4413,
+      "step": 162310
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011387139858805572,
+      "loss": 0.433,
+      "step": 162320
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011382899785876317,
+      "loss": 0.4586,
+      "step": 162330
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011378659712947064,
+      "loss": 0.4153,
+      "step": 162340
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011374419640017808,
+      "loss": 0.4736,
+      "step": 162350
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011370179567088554,
+      "loss": 0.4401,
+      "step": 162360
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011365939494159301,
+      "loss": 0.4231,
+      "step": 162370
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011361699421230045,
+      "loss": 0.4602,
+      "step": 162380
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011357459348300792,
+      "loss": 0.3744,
+      "step": 162390
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011353219275371536,
+      "loss": 0.441,
+      "step": 162400
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011348979202442282,
+      "loss": 0.466,
+      "step": 162410
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011344739129513029,
+      "loss": 0.3235,
+      "step": 162420
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011340499056583773,
+      "loss": 0.3576,
+      "step": 162430
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0001133625898365452,
+      "loss": 0.4398,
+      "step": 162440
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011332018910725266,
+      "loss": 0.4316,
+      "step": 162450
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0001132777883779601,
+      "loss": 0.3896,
+      "step": 162460
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011323538764866757,
+      "loss": 0.387,
+      "step": 162470
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011319298691937501,
+      "loss": 0.3852,
+      "step": 162480
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011315058619008248,
+      "loss": 0.3789,
+      "step": 162490
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011310818546078994,
+      "loss": 0.3927,
+      "step": 162500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011306578473149738,
+      "loss": 0.3967,
+      "step": 162510
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011302338400220485,
+      "loss": 0.4019,
+      "step": 162520
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011298098327291229,
+      "loss": 0.4174,
+      "step": 162530
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011293858254361976,
+      "loss": 0.4147,
+      "step": 162540
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011289618181432722,
+      "loss": 0.3213,
+      "step": 162550
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011285378108503466,
+      "loss": 0.4078,
+      "step": 162560
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011281138035574213,
+      "loss": 0.3978,
+      "step": 162570
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011276897962644958,
+      "loss": 0.4493,
+      "step": 162580
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011272657889715704,
+      "loss": 0.3932,
+      "step": 162590
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0001126841781678645,
+      "loss": 0.3711,
+      "step": 162600
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011264177743857194,
+      "loss": 0.4203,
+      "step": 162610
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011259937670927941,
+      "loss": 0.4329,
+      "step": 162620
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011255697597998686,
+      "loss": 0.3939,
+      "step": 162630
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011251457525069432,
+      "loss": 0.4899,
+      "step": 162640
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011247217452140177,
+      "loss": 0.4876,
+      "step": 162650
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011242977379210924,
+      "loss": 0.4664,
+      "step": 162660
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011238737306281668,
+      "loss": 0.4562,
+      "step": 162670
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011234497233352414,
+      "loss": 0.3758,
+      "step": 162680
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011230257160423158,
+      "loss": 0.4114,
+      "step": 162690
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011226017087493905,
+      "loss": 0.4234,
+      "step": 162700
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011221777014564652,
+      "loss": 0.4023,
+      "step": 162710
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011217536941635395,
+      "loss": 0.3744,
+      "step": 162720
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011213296868706142,
+      "loss": 0.4321,
+      "step": 162730
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011209056795776889,
+      "loss": 0.4304,
+      "step": 162740
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011204816722847633,
+      "loss": 0.4506,
+      "step": 162750
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0001120057664991838,
+      "loss": 0.4174,
+      "step": 162760
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011196336576989123,
+      "loss": 0.4715,
+      "step": 162770
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0001119209650405987,
+      "loss": 0.4083,
+      "step": 162780
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011187856431130617,
+      "loss": 0.4788,
+      "step": 162790
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011183616358201361,
+      "loss": 0.4185,
+      "step": 162800
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011179376285272107,
+      "loss": 0.4064,
+      "step": 162810
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011175136212342854,
+      "loss": 0.3524,
+      "step": 162820
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011170896139413598,
+      "loss": 0.3894,
+      "step": 162830
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011166656066484345,
+      "loss": 0.4624,
+      "step": 162840
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011162415993555089,
+      "loss": 0.3803,
+      "step": 162850
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011158175920625835,
+      "loss": 0.452,
+      "step": 162860
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011153935847696582,
+      "loss": 0.456,
+      "step": 162870
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011149695774767326,
+      "loss": 0.3846,
+      "step": 162880
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011145455701838073,
+      "loss": 0.4696,
+      "step": 162890
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011141215628908818,
+      "loss": 0.3727,
+      "step": 162900
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011136975555979563,
+      "loss": 0.3912,
+      "step": 162910
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001113273548305031,
+      "loss": 0.3799,
+      "step": 162920
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011128495410121054,
+      "loss": 0.3391,
+      "step": 162930
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011124255337191801,
+      "loss": 0.381,
+      "step": 162940
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011120015264262546,
+      "loss": 0.4316,
+      "step": 162950
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011115775191333291,
+      "loss": 0.4006,
+      "step": 162960
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011111535118404037,
+      "loss": 0.4757,
+      "step": 162970
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011107295045474783,
+      "loss": 0.4355,
+      "step": 162980
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011103054972545527,
+      "loss": 0.4215,
+      "step": 162990
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011098814899616274,
+      "loss": 0.4629,
+      "step": 163000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011094574826687018,
+      "loss": 0.4878,
+      "step": 163010
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011090334753757765,
+      "loss": 0.4175,
+      "step": 163020
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011086094680828511,
+      "loss": 0.3979,
+      "step": 163030
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011081854607899255,
+      "loss": 0.3632,
+      "step": 163040
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011077614534970002,
+      "loss": 0.411,
+      "step": 163050
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011073374462040749,
+      "loss": 0.4222,
+      "step": 163060
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011069134389111493,
+      "loss": 0.4223,
+      "step": 163070
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001106489431618224,
+      "loss": 0.4233,
+      "step": 163080
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011060654243252983,
+      "loss": 0.3864,
+      "step": 163090
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001105641417032373,
+      "loss": 0.3348,
+      "step": 163100
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011052174097394477,
+      "loss": 0.4167,
+      "step": 163110
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001104793402446522,
+      "loss": 0.4564,
+      "step": 163120
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011043693951535967,
+      "loss": 0.4776,
+      "step": 163130
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011039453878606714,
+      "loss": 0.4068,
+      "step": 163140
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011035213805677458,
+      "loss": 0.4089,
+      "step": 163150
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011030973732748205,
+      "loss": 0.3361,
+      "step": 163160
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011026733659818949,
+      "loss": 0.4712,
+      "step": 163170
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011022493586889695,
+      "loss": 0.4372,
+      "step": 163180
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011018253513960442,
+      "loss": 0.4144,
+      "step": 163190
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011014013441031186,
+      "loss": 0.4042,
+      "step": 163200
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011009773368101933,
+      "loss": 0.4799,
+      "step": 163210
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011005533295172677,
+      "loss": 0.4231,
+      "step": 163220
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011001293222243423,
+      "loss": 0.2971,
+      "step": 163230
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001099705314931417,
+      "loss": 0.4183,
+      "step": 163240
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010992813076384914,
+      "loss": 0.4522,
+      "step": 163250
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001098857300345566,
+      "loss": 0.4334,
+      "step": 163260
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010984332930526406,
+      "loss": 0.3781,
+      "step": 163270
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010980092857597151,
+      "loss": 0.3935,
+      "step": 163280
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010975852784667897,
+      "loss": 0.4039,
+      "step": 163290
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010971612711738642,
+      "loss": 0.4406,
+      "step": 163300
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010967372638809387,
+      "loss": 0.4856,
+      "step": 163310
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010963132565880134,
+      "loss": 0.4007,
+      "step": 163320
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010958892492950878,
+      "loss": 0.408,
+      "step": 163330
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010954652420021625,
+      "loss": 0.455,
+      "step": 163340
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010950412347092371,
+      "loss": 0.4144,
+      "step": 163350
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010946172274163115,
+      "loss": 0.4386,
+      "step": 163360
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010941932201233862,
+      "loss": 0.4312,
+      "step": 163370
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010937692128304606,
+      "loss": 0.381,
+      "step": 163380
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010933452055375352,
+      "loss": 0.3705,
+      "step": 163390
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010929211982446099,
+      "loss": 0.3503,
+      "step": 163400
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010924971909516843,
+      "loss": 0.4344,
+      "step": 163410
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0001092073183658759,
+      "loss": 0.4151,
+      "step": 163420
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010916491763658336,
+      "loss": 0.3877,
+      "step": 163430
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0001091225169072908,
+      "loss": 0.4096,
+      "step": 163440
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010908011617799827,
+      "loss": 0.3702,
+      "step": 163450
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010903771544870571,
+      "loss": 0.3954,
+      "step": 163460
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010899531471941318,
+      "loss": 0.4581,
+      "step": 163470
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010895291399012064,
+      "loss": 0.3823,
+      "step": 163480
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010891051326082808,
+      "loss": 0.4085,
+      "step": 163490
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010886811253153555,
+      "loss": 0.4403,
+      "step": 163500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010882571180224302,
+      "loss": 0.4557,
+      "step": 163510
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010878331107295046,
+      "loss": 0.3916,
+      "step": 163520
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010874091034365792,
+      "loss": 0.3904,
+      "step": 163530
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010869850961436536,
+      "loss": 0.4107,
+      "step": 163540
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010865610888507283,
+      "loss": 0.3764,
+      "step": 163550
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0001086137081557803,
+      "loss": 0.4031,
+      "step": 163560
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010857130742648774,
+      "loss": 0.4384,
+      "step": 163570
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0001085289066971952,
+      "loss": 0.4087,
+      "step": 163580
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010848650596790266,
+      "loss": 0.4064,
+      "step": 163590
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010844410523861011,
+      "loss": 0.4826,
+      "step": 163600
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010840170450931756,
+      "loss": 0.4048,
+      "step": 163610
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010835930378002502,
+      "loss": 0.4183,
+      "step": 163620
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010831690305073247,
+      "loss": 0.4324,
+      "step": 163630
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010827450232143994,
+      "loss": 0.3959,
+      "step": 163640
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010823210159214738,
+      "loss": 0.3588,
+      "step": 163650
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010818970086285484,
+      "loss": 0.4751,
+      "step": 163660
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010814730013356231,
+      "loss": 0.3934,
+      "step": 163670
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010810489940426975,
+      "loss": 0.4515,
+      "step": 163680
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010806249867497722,
+      "loss": 0.4145,
+      "step": 163690
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010802009794568466,
+      "loss": 0.3632,
+      "step": 163700
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010797769721639212,
+      "loss": 0.3606,
+      "step": 163710
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010793529648709959,
+      "loss": 0.4412,
+      "step": 163720
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010789289575780703,
+      "loss": 0.3875,
+      "step": 163730
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0001078504950285145,
+      "loss": 0.4931,
+      "step": 163740
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010780809429922196,
+      "loss": 0.4473,
+      "step": 163750
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0001077656935699294,
+      "loss": 0.3961,
+      "step": 163760
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010772329284063687,
+      "loss": 0.4438,
+      "step": 163770
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010768089211134431,
+      "loss": 0.4414,
+      "step": 163780
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010763849138205178,
+      "loss": 0.4168,
+      "step": 163790
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010759609065275924,
+      "loss": 0.4724,
+      "step": 163800
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010755368992346668,
+      "loss": 0.364,
+      "step": 163810
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010751128919417415,
+      "loss": 0.3987,
+      "step": 163820
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010746888846488162,
+      "loss": 0.3828,
+      "step": 163830
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010742648773558906,
+      "loss": 0.4406,
+      "step": 163840
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010738408700629652,
+      "loss": 0.4265,
+      "step": 163850
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010734168627700396,
+      "loss": 0.414,
+      "step": 163860
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010729928554771143,
+      "loss": 0.4672,
+      "step": 163870
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0001072568848184189,
+      "loss": 0.3902,
+      "step": 163880
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010721448408912634,
+      "loss": 0.4526,
+      "step": 163890
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0001071720833598338,
+      "loss": 0.4193,
+      "step": 163900
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010712968263054126,
+      "loss": 0.4014,
+      "step": 163910
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010708728190124871,
+      "loss": 0.4199,
+      "step": 163920
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010704488117195616,
+      "loss": 0.4005,
+      "step": 163930
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010700248044266362,
+      "loss": 0.4984,
+      "step": 163940
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010696007971337107,
+      "loss": 0.3383,
+      "step": 163950
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010691767898407854,
+      "loss": 0.4898,
+      "step": 163960
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010687527825478599,
+      "loss": 0.4532,
+      "step": 163970
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010683287752549344,
+      "loss": 0.4047,
+      "step": 163980
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0001067904767962009,
+      "loss": 0.4131,
+      "step": 163990
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010674807606690835,
+      "loss": 0.3795,
+      "step": 164000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010670567533761582,
+      "loss": 0.3664,
+      "step": 164010
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010666327460832326,
+      "loss": 0.3825,
+      "step": 164020
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010662087387903072,
+      "loss": 0.4579,
+      "step": 164030
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010657847314973819,
+      "loss": 0.4279,
+      "step": 164040
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010653607242044563,
+      "loss": 0.423,
+      "step": 164050
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.0001064936716911531,
+      "loss": 0.4133,
+      "step": 164060
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010645127096186053,
+      "loss": 0.3689,
+      "step": 164070
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.000106408870232568,
+      "loss": 0.4616,
+      "step": 164080
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010636646950327547,
+      "loss": 0.3983,
+      "step": 164090
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010632406877398291,
+      "loss": 0.4006,
+      "step": 164100
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010628166804469037,
+      "loss": 0.3724,
+      "step": 164110
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010623926731539784,
+      "loss": 0.354,
+      "step": 164120
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010619686658610528,
+      "loss": 0.4484,
+      "step": 164130
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010615446585681275,
+      "loss": 0.4153,
+      "step": 164140
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010611206512752019,
+      "loss": 0.4233,
+      "step": 164150
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010606966439822765,
+      "loss": 0.4312,
+      "step": 164160
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010602726366893512,
+      "loss": 0.4936,
+      "step": 164170
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010598486293964256,
+      "loss": 0.3604,
+      "step": 164180
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010594246221035003,
+      "loss": 0.4277,
+      "step": 164190
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.0001059000614810575,
+      "loss": 0.3834,
+      "step": 164200
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010585766075176493,
+      "loss": 0.3535,
+      "step": 164210
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.0001058152600224724,
+      "loss": 0.4771,
+      "step": 164220
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010577285929317984,
+      "loss": 0.4385,
+      "step": 164230
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010573045856388731,
+      "loss": 0.4364,
+      "step": 164240
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010568805783459476,
+      "loss": 0.41,
+      "step": 164250
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010564565710530221,
+      "loss": 0.3948,
+      "step": 164260
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010560325637600967,
+      "loss": 0.486,
+      "step": 164270
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010556085564671713,
+      "loss": 0.5081,
+      "step": 164280
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010551845491742459,
+      "loss": 0.4028,
+      "step": 164290
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010547605418813204,
+      "loss": 0.4751,
+      "step": 164300
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0001054336534588395,
+      "loss": 0.4355,
+      "step": 164310
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010539125272954695,
+      "loss": 0.3917,
+      "step": 164320
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010534885200025441,
+      "loss": 0.3667,
+      "step": 164330
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010530645127096185,
+      "loss": 0.4028,
+      "step": 164340
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010526405054166932,
+      "loss": 0.3937,
+      "step": 164350
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010522164981237679,
+      "loss": 0.4871,
+      "step": 164360
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010517924908308423,
+      "loss": 0.43,
+      "step": 164370
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0001051368483537917,
+      "loss": 0.4569,
+      "step": 164380
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010509444762449913,
+      "loss": 0.4184,
+      "step": 164390
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0001050520468952066,
+      "loss": 0.4332,
+      "step": 164400
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010500964616591407,
+      "loss": 0.3968,
+      "step": 164410
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0001049672454366215,
+      "loss": 0.3462,
+      "step": 164420
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010492484470732897,
+      "loss": 0.4177,
+      "step": 164430
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010488244397803644,
+      "loss": 0.4121,
+      "step": 164440
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010484004324874388,
+      "loss": 0.4459,
+      "step": 164450
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010479764251945135,
+      "loss": 0.42,
+      "step": 164460
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010475524179015879,
+      "loss": 0.5086,
+      "step": 164470
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010471284106086625,
+      "loss": 0.4314,
+      "step": 164480
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010467044033157372,
+      "loss": 0.3588,
+      "step": 164490
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010462803960228116,
+      "loss": 0.4637,
+      "step": 164500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010458563887298863,
+      "loss": 0.4369,
+      "step": 164510
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010454323814369609,
+      "loss": 0.4329,
+      "step": 164520
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010450083741440353,
+      "loss": 0.4363,
+      "step": 164530
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.000104458436685111,
+      "loss": 0.3831,
+      "step": 164540
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010441603595581844,
+      "loss": 0.4413,
+      "step": 164550
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0001043736352265259,
+      "loss": 0.4121,
+      "step": 164560
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010433123449723336,
+      "loss": 0.4134,
+      "step": 164570
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010428883376794081,
+      "loss": 0.4358,
+      "step": 164580
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010424643303864827,
+      "loss": 0.4656,
+      "step": 164590
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010420403230935573,
+      "loss": 0.4453,
+      "step": 164600
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010416163158006319,
+      "loss": 0.3493,
+      "step": 164610
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010411923085077064,
+      "loss": 0.426,
+      "step": 164620
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010407683012147809,
+      "loss": 0.4198,
+      "step": 164630
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010403442939218555,
+      "loss": 0.4355,
+      "step": 164640
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010399202866289301,
+      "loss": 0.4037,
+      "step": 164650
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010394962793360045,
+      "loss": 0.5075,
+      "step": 164660
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010390722720430792,
+      "loss": 0.3866,
+      "step": 164670
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010386482647501536,
+      "loss": 0.3899,
+      "step": 164680
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010382242574572283,
+      "loss": 0.4065,
+      "step": 164690
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010378002501643029,
+      "loss": 0.3803,
+      "step": 164700
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010373762428713773,
+      "loss": 0.4706,
+      "step": 164710
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0001036952235578452,
+      "loss": 0.5217,
+      "step": 164720
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010365282282855267,
+      "loss": 0.3844,
+      "step": 164730
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0001036104220992601,
+      "loss": 0.4103,
+      "step": 164740
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010356802136996757,
+      "loss": 0.4562,
+      "step": 164750
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010352562064067501,
+      "loss": 0.447,
+      "step": 164760
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010348321991138248,
+      "loss": 0.4476,
+      "step": 164770
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010344081918208994,
+      "loss": 0.496,
+      "step": 164780
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010339841845279738,
+      "loss": 0.3871,
+      "step": 164790
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010335601772350485,
+      "loss": 0.3957,
+      "step": 164800
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010331361699421232,
+      "loss": 0.4218,
+      "step": 164810
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010327121626491976,
+      "loss": 0.4064,
+      "step": 164820
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010322881553562722,
+      "loss": 0.3502,
+      "step": 164830
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010318641480633466,
+      "loss": 0.4369,
+      "step": 164840
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010314401407704213,
+      "loss": 0.4025,
+      "step": 164850
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0001031016133477496,
+      "loss": 0.3998,
+      "step": 164860
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010305921261845704,
+      "loss": 0.3435,
+      "step": 164870
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0001030168118891645,
+      "loss": 0.4234,
+      "step": 164880
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010297441115987196,
+      "loss": 0.4373,
+      "step": 164890
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010293201043057941,
+      "loss": 0.4138,
+      "step": 164900
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010288960970128688,
+      "loss": 0.4081,
+      "step": 164910
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010284720897199432,
+      "loss": 0.4506,
+      "step": 164920
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010280480824270178,
+      "loss": 0.4169,
+      "step": 164930
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010276240751340924,
+      "loss": 0.4653,
+      "step": 164940
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010272000678411669,
+      "loss": 0.4019,
+      "step": 164950
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010267760605482414,
+      "loss": 0.5007,
+      "step": 164960
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010263520532553161,
+      "loss": 0.4669,
+      "step": 164970
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010259280459623905,
+      "loss": 0.4867,
+      "step": 164980
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010255040386694652,
+      "loss": 0.3927,
+      "step": 164990
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010250800313765396,
+      "loss": 0.3938,
+      "step": 165000
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010246560240836142,
+      "loss": 0.4583,
+      "step": 165010
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010242320167906889,
+      "loss": 0.3938,
+      "step": 165020
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010238080094977633,
+      "loss": 0.4441,
+      "step": 165030
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0001023384002204838,
+      "loss": 0.4073,
+      "step": 165040
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010229599949119126,
+      "loss": 0.3919,
+      "step": 165050
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0001022535987618987,
+      "loss": 0.4077,
+      "step": 165060
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010221119803260617,
+      "loss": 0.3734,
+      "step": 165070
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010216879730331361,
+      "loss": 0.3839,
+      "step": 165080
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010212639657402108,
+      "loss": 0.4407,
+      "step": 165090
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010208399584472854,
+      "loss": 0.4247,
+      "step": 165100
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010204159511543598,
+      "loss": 0.4197,
+      "step": 165110
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010199919438614345,
+      "loss": 0.3818,
+      "step": 165120
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010195679365685092,
+      "loss": 0.3594,
+      "step": 165130
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010191439292755836,
+      "loss": 0.4552,
+      "step": 165140
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010187199219826582,
+      "loss": 0.4619,
+      "step": 165150
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010182959146897326,
+      "loss": 0.4532,
+      "step": 165160
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010178719073968073,
+      "loss": 0.4466,
+      "step": 165170
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0001017447900103882,
+      "loss": 0.4273,
+      "step": 165180
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010170238928109564,
+      "loss": 0.3453,
+      "step": 165190
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0001016599885518031,
+      "loss": 0.4605,
+      "step": 165200
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010161758782251056,
+      "loss": 0.396,
+      "step": 165210
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010157518709321801,
+      "loss": 0.419,
+      "step": 165220
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010153278636392548,
+      "loss": 0.4148,
+      "step": 165230
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010149038563463292,
+      "loss": 0.3956,
+      "step": 165240
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010144798490534038,
+      "loss": 0.3899,
+      "step": 165250
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010140558417604784,
+      "loss": 0.4139,
+      "step": 165260
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010136318344675529,
+      "loss": 0.4622,
+      "step": 165270
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010132078271746274,
+      "loss": 0.3957,
+      "step": 165280
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010127838198817021,
+      "loss": 0.3525,
+      "step": 165290
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010123598125887765,
+      "loss": 0.337,
+      "step": 165300
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010119358052958512,
+      "loss": 0.4205,
+      "step": 165310
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010115117980029256,
+      "loss": 0.4639,
+      "step": 165320
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010110877907100002,
+      "loss": 0.397,
+      "step": 165330
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010106637834170749,
+      "loss": 0.4336,
+      "step": 165340
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010102397761241493,
+      "loss": 0.4232,
+      "step": 165350
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0001009815768831224,
+      "loss": 0.3539,
+      "step": 165360
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010093917615382983,
+      "loss": 0.3643,
+      "step": 165370
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0001008967754245373,
+      "loss": 0.4238,
+      "step": 165380
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010085437469524477,
+      "loss": 0.408,
+      "step": 165390
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010081197396595221,
+      "loss": 0.4204,
+      "step": 165400
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010076957323665968,
+      "loss": 0.519,
+      "step": 165410
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010072717250736714,
+      "loss": 0.3761,
+      "step": 165420
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010068477177807458,
+      "loss": 0.3484,
+      "step": 165430
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010064237104878205,
+      "loss": 0.4608,
+      "step": 165440
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010059997031948949,
+      "loss": 0.4223,
+      "step": 165450
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010055756959019695,
+      "loss": 0.372,
+      "step": 165460
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010051516886090442,
+      "loss": 0.4247,
+      "step": 165470
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010047276813161186,
+      "loss": 0.41,
+      "step": 165480
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010043036740231933,
+      "loss": 0.3412,
+      "step": 165490
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0001003879666730268,
+      "loss": 0.4307,
+      "step": 165500
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010034556594373423,
+      "loss": 0.413,
+      "step": 165510
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0001003031652144417,
+      "loss": 0.3909,
+      "step": 165520
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010026076448514914,
+      "loss": 0.4391,
+      "step": 165530
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010021836375585661,
+      "loss": 0.4065,
+      "step": 165540
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010017596302656407,
+      "loss": 0.4402,
+      "step": 165550
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010013356229727151,
+      "loss": 0.3629,
+      "step": 165560
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010009116156797898,
+      "loss": 0.4048,
+      "step": 165570
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010004876083868643,
+      "loss": 0.2991,
+      "step": 165580
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010000636010939389,
+      "loss": 0.3781,
+      "step": 165590
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.996395938010134e-05,
+      "loss": 0.3547,
+      "step": 165600
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.99215586508088e-05,
+      "loss": 0.3889,
+      "step": 165610
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.987915792151625e-05,
+      "loss": 0.3725,
+      "step": 165620
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.98367571922237e-05,
+      "loss": 0.374,
+      "step": 165630
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.979435646293117e-05,
+      "loss": 0.3733,
+      "step": 165640
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.975195573363862e-05,
+      "loss": 0.344,
+      "step": 165650
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.970955500434607e-05,
+      "loss": 0.4012,
+      "step": 165660
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.966715427505353e-05,
+      "loss": 0.3826,
+      "step": 165670
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.9624753545761e-05,
+      "loss": 0.3475,
+      "step": 165680
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.958235281646845e-05,
+      "loss": 0.3278,
+      "step": 165690
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.95399520871759e-05,
+      "loss": 0.3751,
+      "step": 165700
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.949755135788335e-05,
+      "loss": 0.3684,
+      "step": 165710
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.945515062859082e-05,
+      "loss": 0.3428,
+      "step": 165720
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.941274989929827e-05,
+      "loss": 0.3092,
+      "step": 165730
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.937034917000573e-05,
+      "loss": 0.4498,
+      "step": 165740
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.932794844071318e-05,
+      "loss": 0.395,
+      "step": 165750
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.928554771142065e-05,
+      "loss": 0.4298,
+      "step": 165760
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.92431469821281e-05,
+      "loss": 0.3868,
+      "step": 165770
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.920074625283555e-05,
+      "loss": 0.3774,
+      "step": 165780
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.9158345523543e-05,
+      "loss": 0.4206,
+      "step": 165790
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.911594479425047e-05,
+      "loss": 0.3816,
+      "step": 165800
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.907354406495793e-05,
+      "loss": 0.3771,
+      "step": 165810
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.903114333566538e-05,
+      "loss": 0.3843,
+      "step": 165820
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.898874260637283e-05,
+      "loss": 0.3739,
+      "step": 165830
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.89463418770803e-05,
+      "loss": 0.353,
+      "step": 165840
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.890394114778775e-05,
+      "loss": 0.4054,
+      "step": 165850
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.88615404184952e-05,
+      "loss": 0.332,
+      "step": 165860
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.881913968920266e-05,
+      "loss": 0.3153,
+      "step": 165870
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.877673895991013e-05,
+      "loss": 0.341,
+      "step": 165880
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.873433823061758e-05,
+      "loss": 0.3992,
+      "step": 165890
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.869193750132503e-05,
+      "loss": 0.3512,
+      "step": 165900
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.864953677203249e-05,
+      "loss": 0.351,
+      "step": 165910
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.860713604273994e-05,
+      "loss": 0.4287,
+      "step": 165920
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.856473531344739e-05,
+      "loss": 0.3909,
+      "step": 165930
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.852233458415485e-05,
+      "loss": 0.4037,
+      "step": 165940
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.84799338548623e-05,
+      "loss": 0.3327,
+      "step": 165950
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.843753312556977e-05,
+      "loss": 0.3735,
+      "step": 165960
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.839513239627722e-05,
+      "loss": 0.3675,
+      "step": 165970
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.835273166698467e-05,
+      "loss": 0.3917,
+      "step": 165980
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.831033093769213e-05,
+      "loss": 0.3654,
+      "step": 165990
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.826793020839959e-05,
+      "loss": 0.3569,
+      "step": 166000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.822552947910705e-05,
+      "loss": 0.3674,
+      "step": 166010
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.81831287498145e-05,
+      "loss": 0.4262,
+      "step": 166020
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.814072802052195e-05,
+      "loss": 0.3666,
+      "step": 166030
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.809832729122942e-05,
+      "loss": 0.3668,
+      "step": 166040
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.805592656193687e-05,
+      "loss": 0.3908,
+      "step": 166050
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.801352583264433e-05,
+      "loss": 0.3611,
+      "step": 166060
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.797112510335178e-05,
+      "loss": 0.3639,
+      "step": 166070
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.792872437405923e-05,
+      "loss": 0.3405,
+      "step": 166080
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.78863236447667e-05,
+      "loss": 0.4063,
+      "step": 166090
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.784392291547415e-05,
+      "loss": 0.3968,
+      "step": 166100
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.78015221861816e-05,
+      "loss": 0.3375,
+      "step": 166110
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.775912145688906e-05,
+      "loss": 0.3439,
+      "step": 166120
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.771672072759652e-05,
+      "loss": 0.3975,
+      "step": 166130
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.767431999830398e-05,
+      "loss": 0.353,
+      "step": 166140
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.763191926901143e-05,
+      "loss": 0.3907,
+      "step": 166150
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.758951853971888e-05,
+      "loss": 0.3345,
+      "step": 166160
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.754711781042635e-05,
+      "loss": 0.3562,
+      "step": 166170
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.75047170811338e-05,
+      "loss": 0.3899,
+      "step": 166180
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.746231635184126e-05,
+      "loss": 0.4327,
+      "step": 166190
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.741991562254871e-05,
+      "loss": 0.3636,
+      "step": 166200
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.737751489325618e-05,
+      "loss": 0.3481,
+      "step": 166210
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.733511416396363e-05,
+      "loss": 0.3534,
+      "step": 166220
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.729271343467108e-05,
+      "loss": 0.3908,
+      "step": 166230
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.725031270537854e-05,
+      "loss": 0.3566,
+      "step": 166240
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.720791197608599e-05,
+      "loss": 0.3557,
+      "step": 166250
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.716551124679344e-05,
+      "loss": 0.3554,
+      "step": 166260
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.71231105175009e-05,
+      "loss": 0.3487,
+      "step": 166270
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.708070978820835e-05,
+      "loss": 0.3908,
+      "step": 166280
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.703830905891582e-05,
+      "loss": 0.3243,
+      "step": 166290
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.699590832962327e-05,
+      "loss": 0.4481,
+      "step": 166300
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.695350760033072e-05,
+      "loss": 0.3963,
+      "step": 166310
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.691110687103818e-05,
+      "loss": 0.3565,
+      "step": 166320
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.686870614174564e-05,
+      "loss": 0.4374,
+      "step": 166330
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.68263054124531e-05,
+      "loss": 0.3882,
+      "step": 166340
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.678390468316055e-05,
+      "loss": 0.4405,
+      "step": 166350
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.6741503953868e-05,
+      "loss": 0.3459,
+      "step": 166360
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.669910322457547e-05,
+      "loss": 0.4236,
+      "step": 166370
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.665670249528292e-05,
+      "loss": 0.3867,
+      "step": 166380
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.661430176599038e-05,
+      "loss": 0.3402,
+      "step": 166390
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.657190103669783e-05,
+      "loss": 0.3498,
+      "step": 166400
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.65295003074053e-05,
+      "loss": 0.405,
+      "step": 166410
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.648709957811275e-05,
+      "loss": 0.3577,
+      "step": 166420
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.64446988488202e-05,
+      "loss": 0.3593,
+      "step": 166430
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.640229811952766e-05,
+      "loss": 0.3405,
+      "step": 166440
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.635989739023512e-05,
+      "loss": 0.3189,
+      "step": 166450
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.631749666094258e-05,
+      "loss": 0.3951,
+      "step": 166460
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.627509593165003e-05,
+      "loss": 0.4063,
+      "step": 166470
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.623269520235748e-05,
+      "loss": 0.3434,
+      "step": 166480
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.619029447306495e-05,
+      "loss": 0.2953,
+      "step": 166490
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.61478937437724e-05,
+      "loss": 0.4415,
+      "step": 166500
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.610549301447986e-05,
+      "loss": 0.3631,
+      "step": 166510
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.606309228518731e-05,
+      "loss": 0.3715,
+      "step": 166520
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.602069155589478e-05,
+      "loss": 0.3486,
+      "step": 166530
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.597829082660223e-05,
+      "loss": 0.3978,
+      "step": 166540
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.593589009730968e-05,
+      "loss": 0.3321,
+      "step": 166550
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.589348936801714e-05,
+      "loss": 0.3787,
+      "step": 166560
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.585108863872459e-05,
+      "loss": 0.3611,
+      "step": 166570
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.580868790943204e-05,
+      "loss": 0.3458,
+      "step": 166580
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.57662871801395e-05,
+      "loss": 0.4109,
+      "step": 166590
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.572388645084695e-05,
+      "loss": 0.3457,
+      "step": 166600
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.568148572155442e-05,
+      "loss": 0.3636,
+      "step": 166610
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.563908499226187e-05,
+      "loss": 0.3644,
+      "step": 166620
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.559668426296932e-05,
+      "loss": 0.3363,
+      "step": 166630
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.555428353367678e-05,
+      "loss": 0.3285,
+      "step": 166640
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.551188280438424e-05,
+      "loss": 0.4555,
+      "step": 166650
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.54694820750917e-05,
+      "loss": 0.3886,
+      "step": 166660
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.542708134579915e-05,
+      "loss": 0.3483,
+      "step": 166670
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.53846806165066e-05,
+      "loss": 0.3525,
+      "step": 166680
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.534227988721407e-05,
+      "loss": 0.3553,
+      "step": 166690
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.529987915792152e-05,
+      "loss": 0.3999,
+      "step": 166700
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.525747842862898e-05,
+      "loss": 0.3618,
+      "step": 166710
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.521507769933643e-05,
+      "loss": 0.325,
+      "step": 166720
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.51726769700439e-05,
+      "loss": 0.3882,
+      "step": 166730
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.513027624075135e-05,
+      "loss": 0.3906,
+      "step": 166740
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.50878755114588e-05,
+      "loss": 0.4195,
+      "step": 166750
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.504547478216625e-05,
+      "loss": 0.4358,
+      "step": 166760
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.500307405287371e-05,
+      "loss": 0.3617,
+      "step": 166770
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.496067332358117e-05,
+      "loss": 0.465,
+      "step": 166780
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.491827259428863e-05,
+      "loss": 0.338,
+      "step": 166790
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.487587186499608e-05,
+      "loss": 0.3767,
+      "step": 166800
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.483347113570353e-05,
+      "loss": 0.3984,
+      "step": 166810
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.4791070406411e-05,
+      "loss": 0.3353,
+      "step": 166820
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.474866967711845e-05,
+      "loss": 0.412,
+      "step": 166830
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.470626894782591e-05,
+      "loss": 0.3567,
+      "step": 166840
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.466386821853336e-05,
+      "loss": 0.3249,
+      "step": 166850
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.462146748924083e-05,
+      "loss": 0.3803,
+      "step": 166860
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.457906675994828e-05,
+      "loss": 0.4376,
+      "step": 166870
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.453666603065573e-05,
+      "loss": 0.3578,
+      "step": 166880
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.449426530136319e-05,
+      "loss": 0.4502,
+      "step": 166890
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.445186457207064e-05,
+      "loss": 0.3847,
+      "step": 166900
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.44094638427781e-05,
+      "loss": 0.374,
+      "step": 166910
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.436706311348556e-05,
+      "loss": 0.3878,
+      "step": 166920
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.432466238419301e-05,
+      "loss": 0.3617,
+      "step": 166930
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.428226165490047e-05,
+      "loss": 0.3852,
+      "step": 166940
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.423986092560792e-05,
+      "loss": 0.3922,
+      "step": 166950
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.419746019631537e-05,
+      "loss": 0.4076,
+      "step": 166960
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.415505946702283e-05,
+      "loss": 0.4278,
+      "step": 166970
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.41126587377303e-05,
+      "loss": 0.401,
+      "step": 166980
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.407025800843775e-05,
+      "loss": 0.3484,
+      "step": 166990
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.40278572791452e-05,
+      "loss": 0.4709,
+      "step": 167000
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.398545654985265e-05,
+      "loss": 0.4524,
+      "step": 167010
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.394305582056012e-05,
+      "loss": 0.3449,
+      "step": 167020
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.390065509126757e-05,
+      "loss": 0.3999,
+      "step": 167030
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.385825436197503e-05,
+      "loss": 0.3916,
+      "step": 167040
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.381585363268248e-05,
+      "loss": 0.4179,
+      "step": 167050
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.377345290338995e-05,
+      "loss": 0.3997,
+      "step": 167060
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.37310521740974e-05,
+      "loss": 0.3252,
+      "step": 167070
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.368865144480485e-05,
+      "loss": 0.5115,
+      "step": 167080
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.36462507155123e-05,
+      "loss": 0.3878,
+      "step": 167090
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.360384998621977e-05,
+      "loss": 0.4041,
+      "step": 167100
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.356144925692723e-05,
+      "loss": 0.3912,
+      "step": 167110
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.351904852763468e-05,
+      "loss": 0.3961,
+      "step": 167120
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.347664779834213e-05,
+      "loss": 0.3886,
+      "step": 167130
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.34342470690496e-05,
+      "loss": 0.3309,
+      "step": 167140
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.339184633975705e-05,
+      "loss": 0.4054,
+      "step": 167150
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.33494456104645e-05,
+      "loss": 0.3756,
+      "step": 167160
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.330704488117196e-05,
+      "loss": 0.3525,
+      "step": 167170
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.326464415187943e-05,
+      "loss": 0.4158,
+      "step": 167180
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.322224342258688e-05,
+      "loss": 0.4022,
+      "step": 167190
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.317984269329433e-05,
+      "loss": 0.3686,
+      "step": 167200
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.313744196400179e-05,
+      "loss": 0.3599,
+      "step": 167210
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.309504123470924e-05,
+      "loss": 0.3984,
+      "step": 167220
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.30526405054167e-05,
+      "loss": 0.4048,
+      "step": 167230
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.301023977612416e-05,
+      "loss": 0.4043,
+      "step": 167240
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.296783904683161e-05,
+      "loss": 0.3575,
+      "step": 167250
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.292543831753907e-05,
+      "loss": 0.3955,
+      "step": 167260
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.288303758824652e-05,
+      "loss": 0.3579,
+      "step": 167270
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.284063685895397e-05,
+      "loss": 0.3982,
+      "step": 167280
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.279823612966143e-05,
+      "loss": 0.405,
+      "step": 167290
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.275583540036889e-05,
+      "loss": 0.3027,
+      "step": 167300
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.271343467107635e-05,
+      "loss": 0.3691,
+      "step": 167310
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.26710339417838e-05,
+      "loss": 0.3347,
+      "step": 167320
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.262863321249125e-05,
+      "loss": 0.3663,
+      "step": 167330
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.258623248319872e-05,
+      "loss": 0.3532,
+      "step": 167340
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.254383175390617e-05,
+      "loss": 0.3686,
+      "step": 167350
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.250143102461363e-05,
+      "loss": 0.3612,
+      "step": 167360
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.245903029532108e-05,
+      "loss": 0.4094,
+      "step": 167370
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.241662956602855e-05,
+      "loss": 0.4256,
+      "step": 167380
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.2374228836736e-05,
+      "loss": 0.3562,
+      "step": 167390
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.233182810744345e-05,
+      "loss": 0.3761,
+      "step": 167400
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.22894273781509e-05,
+      "loss": 0.3234,
+      "step": 167410
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.224702664885837e-05,
+      "loss": 0.3154,
+      "step": 167420
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.220462591956583e-05,
+      "loss": 0.3702,
+      "step": 167430
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.216222519027328e-05,
+      "loss": 0.3451,
+      "step": 167440
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.211982446098073e-05,
+      "loss": 0.3864,
+      "step": 167450
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.20774237316882e-05,
+      "loss": 0.3759,
+      "step": 167460
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.203502300239565e-05,
+      "loss": 0.4643,
+      "step": 167470
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.19926222731031e-05,
+      "loss": 0.3862,
+      "step": 167480
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.195022154381056e-05,
+      "loss": 0.3753,
+      "step": 167490
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.190782081451801e-05,
+      "loss": 0.3451,
+      "step": 167500
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.186542008522548e-05,
+      "loss": 0.3264,
+      "step": 167510
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.182301935593293e-05,
+      "loss": 0.3922,
+      "step": 167520
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.178061862664038e-05,
+      "loss": 0.3313,
+      "step": 167530
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.173821789734784e-05,
+      "loss": 0.3375,
+      "step": 167540
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.16958171680553e-05,
+      "loss": 0.3776,
+      "step": 167550
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.165341643876276e-05,
+      "loss": 0.4089,
+      "step": 167560
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.161101570947021e-05,
+      "loss": 0.37,
+      "step": 167570
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.156861498017766e-05,
+      "loss": 0.3863,
+      "step": 167580
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.152621425088512e-05,
+      "loss": 0.3978,
+      "step": 167590
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.148381352159257e-05,
+      "loss": 0.3614,
+      "step": 167600
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.144141279230002e-05,
+      "loss": 0.4322,
+      "step": 167610
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.139901206300748e-05,
+      "loss": 0.4248,
+      "step": 167620
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.135661133371494e-05,
+      "loss": 0.3555,
+      "step": 167630
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.13142106044224e-05,
+      "loss": 0.3591,
+      "step": 167640
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.127180987512985e-05,
+      "loss": 0.3873,
+      "step": 167650
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.12294091458373e-05,
+      "loss": 0.4038,
+      "step": 167660
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.118700841654477e-05,
+      "loss": 0.3806,
+      "step": 167670
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.114460768725222e-05,
+      "loss": 0.3879,
+      "step": 167680
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.110220695795968e-05,
+      "loss": 0.2949,
+      "step": 167690
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.105980622866713e-05,
+      "loss": 0.3481,
+      "step": 167700
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.10174054993746e-05,
+      "loss": 0.3394,
+      "step": 167710
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.097500477008205e-05,
+      "loss": 0.3667,
+      "step": 167720
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.09326040407895e-05,
+      "loss": 0.3549,
+      "step": 167730
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.089020331149696e-05,
+      "loss": 0.4024,
+      "step": 167740
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.084780258220442e-05,
+      "loss": 0.3958,
+      "step": 167750
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.080540185291188e-05,
+      "loss": 0.3945,
+      "step": 167760
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.076300112361933e-05,
+      "loss": 0.3769,
+      "step": 167770
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.072060039432678e-05,
+      "loss": 0.3694,
+      "step": 167780
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.067819966503425e-05,
+      "loss": 0.3555,
+      "step": 167790
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.06357989357417e-05,
+      "loss": 0.4114,
+      "step": 167800
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.059339820644916e-05,
+      "loss": 0.3526,
+      "step": 167810
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.055099747715661e-05,
+      "loss": 0.3704,
+      "step": 167820
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.050859674786408e-05,
+      "loss": 0.3461,
+      "step": 167830
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.046619601857153e-05,
+      "loss": 0.3496,
+      "step": 167840
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.042379528927898e-05,
+      "loss": 0.3703,
+      "step": 167850
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.038139455998644e-05,
+      "loss": 0.3709,
+      "step": 167860
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.03389938306939e-05,
+      "loss": 0.3827,
+      "step": 167870
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.029659310140136e-05,
+      "loss": 0.3801,
+      "step": 167880
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.025419237210881e-05,
+      "loss": 0.4944,
+      "step": 167890
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.021179164281626e-05,
+      "loss": 0.3096,
+      "step": 167900
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.016939091352372e-05,
+      "loss": 0.2842,
+      "step": 167910
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.012699018423117e-05,
+      "loss": 0.3721,
+      "step": 167920
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.008458945493862e-05,
+      "loss": 0.3918,
+      "step": 167930
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.004218872564608e-05,
+      "loss": 0.3498,
+      "step": 167940
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.999978799635354e-05,
+      "loss": 0.3286,
+      "step": 167950
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.9957387267061e-05,
+      "loss": 0.4289,
+      "step": 167960
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.991498653776845e-05,
+      "loss": 0.3721,
+      "step": 167970
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.98725858084759e-05,
+      "loss": 0.3721,
+      "step": 167980
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.983018507918337e-05,
+      "loss": 0.4148,
+      "step": 167990
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.978778434989082e-05,
+      "loss": 0.396,
+      "step": 168000
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.974538362059828e-05,
+      "loss": 0.3891,
+      "step": 168010
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.970298289130573e-05,
+      "loss": 0.395,
+      "step": 168020
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.96605821620132e-05,
+      "loss": 0.3075,
+      "step": 168030
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.961818143272065e-05,
+      "loss": 0.3434,
+      "step": 168040
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.95757807034281e-05,
+      "loss": 0.3561,
+      "step": 168050
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.953337997413556e-05,
+      "loss": 0.4042,
+      "step": 168060
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.949097924484302e-05,
+      "loss": 0.3442,
+      "step": 168070
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.944857851555048e-05,
+      "loss": 0.327,
+      "step": 168080
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.940617778625793e-05,
+      "loss": 0.391,
+      "step": 168090
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.936377705696538e-05,
+      "loss": 0.409,
+      "step": 168100
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.932137632767285e-05,
+      "loss": 0.3363,
+      "step": 168110
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.92789755983803e-05,
+      "loss": 0.4125,
+      "step": 168120
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.923657486908775e-05,
+      "loss": 0.4537,
+      "step": 168130
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.919417413979521e-05,
+      "loss": 0.3947,
+      "step": 168140
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.915177341050267e-05,
+      "loss": 0.3773,
+      "step": 168150
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.910937268121013e-05,
+      "loss": 0.4202,
+      "step": 168160
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.906697195191758e-05,
+      "loss": 0.403,
+      "step": 168170
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.902457122262503e-05,
+      "loss": 0.3263,
+      "step": 168180
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.89821704933325e-05,
+      "loss": 0.3624,
+      "step": 168190
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.893976976403995e-05,
+      "loss": 0.4079,
+      "step": 168200
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.889736903474741e-05,
+      "loss": 0.4057,
+      "step": 168210
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.885496830545486e-05,
+      "loss": 0.4123,
+      "step": 168220
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.881256757616231e-05,
+      "loss": 0.391,
+      "step": 168230
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.877016684686977e-05,
+      "loss": 0.3936,
+      "step": 168240
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.872776611757722e-05,
+      "loss": 0.3781,
+      "step": 168250
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.868536538828467e-05,
+      "loss": 0.4725,
+      "step": 168260
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.864296465899213e-05,
+      "loss": 0.3861,
+      "step": 168270
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.86005639296996e-05,
+      "loss": 0.4029,
+      "step": 168280
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.855816320040705e-05,
+      "loss": 0.397,
+      "step": 168290
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.85157624711145e-05,
+      "loss": 0.3692,
+      "step": 168300
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.847336174182195e-05,
+      "loss": 0.3412,
+      "step": 168310
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.843096101252942e-05,
+      "loss": 0.4182,
+      "step": 168320
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.838856028323687e-05,
+      "loss": 0.4095,
+      "step": 168330
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.834615955394433e-05,
+      "loss": 0.3638,
+      "step": 168340
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.830375882465178e-05,
+      "loss": 0.3586,
+      "step": 168350
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.826135809535925e-05,
+      "loss": 0.3838,
+      "step": 168360
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.82189573660667e-05,
+      "loss": 0.4291,
+      "step": 168370
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.817655663677415e-05,
+      "loss": 0.3544,
+      "step": 168380
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.813415590748161e-05,
+      "loss": 0.4241,
+      "step": 168390
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.809175517818907e-05,
+      "loss": 0.3191,
+      "step": 168400
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.804935444889653e-05,
+      "loss": 0.3628,
+      "step": 168410
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.800695371960398e-05,
+      "loss": 0.4156,
+      "step": 168420
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.796455299031143e-05,
+      "loss": 0.3337,
+      "step": 168430
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.79221522610189e-05,
+      "loss": 0.3319,
+      "step": 168440
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.787975153172635e-05,
+      "loss": 0.4399,
+      "step": 168450
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.78373508024338e-05,
+      "loss": 0.3778,
+      "step": 168460
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.779495007314126e-05,
+      "loss": 0.3469,
+      "step": 168470
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.775254934384873e-05,
+      "loss": 0.3435,
+      "step": 168480
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.771014861455618e-05,
+      "loss": 0.3623,
+      "step": 168490
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.766774788526363e-05,
+      "loss": 0.3656,
+      "step": 168500
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.762534715597109e-05,
+      "loss": 0.4326,
+      "step": 168510
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.758294642667855e-05,
+      "loss": 0.361,
+      "step": 168520
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.7540545697386e-05,
+      "loss": 0.3599,
+      "step": 168530
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.749814496809346e-05,
+      "loss": 0.3826,
+      "step": 168540
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.745574423880091e-05,
+      "loss": 0.4171,
+      "step": 168550
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.741334350950837e-05,
+      "loss": 0.4546,
+      "step": 168560
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.737094278021582e-05,
+      "loss": 0.3432,
+      "step": 168570
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.732854205092327e-05,
+      "loss": 0.3323,
+      "step": 168580
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.728614132163073e-05,
+      "loss": 0.3177,
+      "step": 168590
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.724374059233819e-05,
+      "loss": 0.3889,
+      "step": 168600
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.720133986304565e-05,
+      "loss": 0.3158,
+      "step": 168610
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.71589391337531e-05,
+      "loss": 0.4235,
+      "step": 168620
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.711653840446055e-05,
+      "loss": 0.3629,
+      "step": 168630
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.707413767516802e-05,
+      "loss": 0.4256,
+      "step": 168640
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.703173694587547e-05,
+      "loss": 0.3781,
+      "step": 168650
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.698933621658293e-05,
+      "loss": 0.3412,
+      "step": 168660
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.694693548729038e-05,
+      "loss": 0.3698,
+      "step": 168670
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.690453475799785e-05,
+      "loss": 0.4038,
+      "step": 168680
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.68621340287053e-05,
+      "loss": 0.4435,
+      "step": 168690
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.681973329941275e-05,
+      "loss": 0.3654,
+      "step": 168700
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.67773325701202e-05,
+      "loss": 0.4056,
+      "step": 168710
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.673493184082767e-05,
+      "loss": 0.404,
+      "step": 168720
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.669253111153513e-05,
+      "loss": 0.3634,
+      "step": 168730
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.665013038224258e-05,
+      "loss": 0.3846,
+      "step": 168740
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.660772965295003e-05,
+      "loss": 0.3637,
+      "step": 168750
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.65653289236575e-05,
+      "loss": 0.3653,
+      "step": 168760
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.652292819436495e-05,
+      "loss": 0.3549,
+      "step": 168770
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.64805274650724e-05,
+      "loss": 0.3724,
+      "step": 168780
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.643812673577986e-05,
+      "loss": 0.4418,
+      "step": 168790
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.639572600648733e-05,
+      "loss": 0.3779,
+      "step": 168800
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.635332527719478e-05,
+      "loss": 0.3678,
+      "step": 168810
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.631092454790223e-05,
+      "loss": 0.4291,
+      "step": 168820
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.626852381860968e-05,
+      "loss": 0.3617,
+      "step": 168830
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.622612308931715e-05,
+      "loss": 0.4082,
+      "step": 168840
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.61837223600246e-05,
+      "loss": 0.3544,
+      "step": 168850
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.614132163073206e-05,
+      "loss": 0.3772,
+      "step": 168860
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.609892090143951e-05,
+      "loss": 0.3742,
+      "step": 168870
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.605652017214696e-05,
+      "loss": 0.3768,
+      "step": 168880
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.601411944285442e-05,
+      "loss": 0.3792,
+      "step": 168890
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.597171871356187e-05,
+      "loss": 0.3988,
+      "step": 168900
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.592931798426932e-05,
+      "loss": 0.3788,
+      "step": 168910
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.588691725497678e-05,
+      "loss": 0.3726,
+      "step": 168920
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.584451652568424e-05,
+      "loss": 0.3605,
+      "step": 168930
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.58021157963917e-05,
+      "loss": 0.3591,
+      "step": 168940
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.575971506709915e-05,
+      "loss": 0.3571,
+      "step": 168950
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.57173143378066e-05,
+      "loss": 0.3068,
+      "step": 168960
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.567491360851407e-05,
+      "loss": 0.3259,
+      "step": 168970
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.563251287922152e-05,
+      "loss": 0.4198,
+      "step": 168980
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.559011214992898e-05,
+      "loss": 0.3973,
+      "step": 168990
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.554771142063643e-05,
+      "loss": 0.3951,
+      "step": 169000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.55053106913439e-05,
+      "loss": 0.4613,
+      "step": 169010
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.546290996205135e-05,
+      "loss": 0.3566,
+      "step": 169020
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.54205092327588e-05,
+      "loss": 0.3906,
+      "step": 169030
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.537810850346626e-05,
+      "loss": 0.3853,
+      "step": 169040
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.533570777417372e-05,
+      "loss": 0.3528,
+      "step": 169050
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.529330704488118e-05,
+      "loss": 0.3884,
+      "step": 169060
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.525090631558863e-05,
+      "loss": 0.4508,
+      "step": 169070
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.520850558629608e-05,
+      "loss": 0.3553,
+      "step": 169080
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.516610485700355e-05,
+      "loss": 0.4123,
+      "step": 169090
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.5123704127711e-05,
+      "loss": 0.3679,
+      "step": 169100
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.508130339841846e-05,
+      "loss": 0.4319,
+      "step": 169110
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.503890266912591e-05,
+      "loss": 0.3538,
+      "step": 169120
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.499650193983338e-05,
+      "loss": 0.3503,
+      "step": 169130
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.495410121054083e-05,
+      "loss": 0.4049,
+      "step": 169140
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.491170048124828e-05,
+      "loss": 0.3766,
+      "step": 169150
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.486929975195574e-05,
+      "loss": 0.3404,
+      "step": 169160
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.48268990226632e-05,
+      "loss": 0.4344,
+      "step": 169170
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.478449829337066e-05,
+      "loss": 0.3354,
+      "step": 169180
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.474209756407811e-05,
+      "loss": 0.3328,
+      "step": 169190
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.469969683478556e-05,
+      "loss": 0.3987,
+      "step": 169200
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.465729610549302e-05,
+      "loss": 0.3279,
+      "step": 169210
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.461489537620047e-05,
+      "loss": 0.4133,
+      "step": 169220
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.457249464690792e-05,
+      "loss": 0.3863,
+      "step": 169230
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.453009391761539e-05,
+      "loss": 0.4029,
+      "step": 169240
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.448769318832284e-05,
+      "loss": 0.3473,
+      "step": 169250
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.44452924590303e-05,
+      "loss": 0.3531,
+      "step": 169260
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.440289172973775e-05,
+      "loss": 0.379,
+      "step": 169270
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.43604910004452e-05,
+      "loss": 0.3671,
+      "step": 169280
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.431809027115267e-05,
+      "loss": 0.3688,
+      "step": 169290
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.427568954186012e-05,
+      "loss": 0.404,
+      "step": 169300
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.423328881256758e-05,
+      "loss": 0.352,
+      "step": 169310
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.419088808327503e-05,
+      "loss": 0.3647,
+      "step": 169320
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.41484873539825e-05,
+      "loss": 0.3826,
+      "step": 169330
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.410608662468995e-05,
+      "loss": 0.3898,
+      "step": 169340
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.40636858953974e-05,
+      "loss": 0.3766,
+      "step": 169350
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.402128516610486e-05,
+      "loss": 0.4087,
+      "step": 169360
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.397888443681232e-05,
+      "loss": 0.3444,
+      "step": 169370
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.393648370751978e-05,
+      "loss": 0.3529,
+      "step": 169380
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.389408297822723e-05,
+      "loss": 0.3495,
+      "step": 169390
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.385168224893468e-05,
+      "loss": 0.3483,
+      "step": 169400
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.380928151964215e-05,
+      "loss": 0.3632,
+      "step": 169410
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.37668807903496e-05,
+      "loss": 0.3955,
+      "step": 169420
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.372448006105706e-05,
+      "loss": 0.4494,
+      "step": 169430
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.368207933176451e-05,
+      "loss": 0.3963,
+      "step": 169440
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.363967860247198e-05,
+      "loss": 0.3791,
+      "step": 169450
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.359727787317943e-05,
+      "loss": 0.4079,
+      "step": 169460
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.355487714388688e-05,
+      "loss": 0.4172,
+      "step": 169470
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.351247641459433e-05,
+      "loss": 0.4115,
+      "step": 169480
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.34700756853018e-05,
+      "loss": 0.3809,
+      "step": 169490
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.342767495600925e-05,
+      "loss": 0.4046,
+      "step": 169500
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.338527422671671e-05,
+      "loss": 0.3889,
+      "step": 169510
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.334287349742416e-05,
+      "loss": 0.4262,
+      "step": 169520
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.330047276813161e-05,
+      "loss": 0.3818,
+      "step": 169530
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.325807203883907e-05,
+      "loss": 0.455,
+      "step": 169540
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.321567130954653e-05,
+      "loss": 0.4179,
+      "step": 169550
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.317327058025399e-05,
+      "loss": 0.4042,
+      "step": 169560
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.313086985096144e-05,
+      "loss": 0.3782,
+      "step": 169570
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.30884691216689e-05,
+      "loss": 0.3297,
+      "step": 169580
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.304606839237635e-05,
+      "loss": 0.3988,
+      "step": 169590
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.30036676630838e-05,
+      "loss": 0.3993,
+      "step": 169600
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.296126693379127e-05,
+      "loss": 0.3092,
+      "step": 169610
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.291886620449872e-05,
+      "loss": 0.3741,
+      "step": 169620
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.287646547520617e-05,
+      "loss": 0.3291,
+      "step": 169630
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.283406474591363e-05,
+      "loss": 0.3862,
+      "step": 169640
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.279166401662108e-05,
+      "loss": 0.3641,
+      "step": 169650
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.274926328732855e-05,
+      "loss": 0.3589,
+      "step": 169660
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.2706862558036e-05,
+      "loss": 0.451,
+      "step": 169670
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.266446182874345e-05,
+      "loss": 0.4004,
+      "step": 169680
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.262206109945091e-05,
+      "loss": 0.3653,
+      "step": 169690
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.257966037015837e-05,
+      "loss": 0.3105,
+      "step": 169700
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.253725964086583e-05,
+      "loss": 0.378,
+      "step": 169710
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.249485891157328e-05,
+      "loss": 0.3644,
+      "step": 169720
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.245245818228073e-05,
+      "loss": 0.3662,
+      "step": 169730
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.24100574529882e-05,
+      "loss": 0.386,
+      "step": 169740
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.236765672369565e-05,
+      "loss": 0.3114,
+      "step": 169750
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.232525599440311e-05,
+      "loss": 0.3947,
+      "step": 169760
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.228285526511056e-05,
+      "loss": 0.3149,
+      "step": 169770
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.224045453581803e-05,
+      "loss": 0.3703,
+      "step": 169780
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.219805380652548e-05,
+      "loss": 0.3626,
+      "step": 169790
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.215565307723293e-05,
+      "loss": 0.4188,
+      "step": 169800
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.211325234794039e-05,
+      "loss": 0.4035,
+      "step": 169810
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.207085161864785e-05,
+      "loss": 0.417,
+      "step": 169820
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.20284508893553e-05,
+      "loss": 0.3851,
+      "step": 169830
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.198605016006276e-05,
+      "loss": 0.3381,
+      "step": 169840
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.194364943077021e-05,
+      "loss": 0.3568,
+      "step": 169850
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.190124870147767e-05,
+      "loss": 0.3612,
+      "step": 169860
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.185884797218513e-05,
+      "loss": 0.4116,
+      "step": 169870
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.181644724289259e-05,
+      "loss": 0.3835,
+      "step": 169880
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.177404651360004e-05,
+      "loss": 0.4632,
+      "step": 169890
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.173164578430749e-05,
+      "loss": 0.368,
+      "step": 169900
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.168924505501495e-05,
+      "loss": 0.3429,
+      "step": 169910
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.16468443257224e-05,
+      "loss": 0.436,
+      "step": 169920
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.160444359642985e-05,
+      "loss": 0.3992,
+      "step": 169930
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.156204286713732e-05,
+      "loss": 0.3543,
+      "step": 169940
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.151964213784477e-05,
+      "loss": 0.3592,
+      "step": 169950
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.147724140855223e-05,
+      "loss": 0.2943,
+      "step": 169960
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.143484067925968e-05,
+      "loss": 0.3595,
+      "step": 169970
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.139243994996715e-05,
+      "loss": 0.4172,
+      "step": 169980
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.13500392206746e-05,
+      "loss": 0.3513,
+      "step": 169990
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.130763849138205e-05,
+      "loss": 0.4688,
+      "step": 170000
+    },
+    {
+      "epoch": 7.19,
+      "eval_loss": 0.6409904956817627,
+      "eval_runtime": 337.7795,
+      "eval_samples_per_second": 15.557,
+      "eval_steps_per_second": 3.89,
+      "step": 170000
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.12652377620895e-05,
+      "loss": 0.3743,
+      "step": 170010
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.122283703279697e-05,
+      "loss": 0.3994,
+      "step": 170020
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.118043630350443e-05,
+      "loss": 0.4012,
+      "step": 170030
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.113803557421188e-05,
+      "loss": 0.4289,
+      "step": 170040
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.109563484491933e-05,
+      "loss": 0.4133,
+      "step": 170050
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.10532341156268e-05,
+      "loss": 0.4181,
+      "step": 170060
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.101083338633425e-05,
+      "loss": 0.4194,
+      "step": 170070
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.09684326570417e-05,
+      "loss": 0.3584,
+      "step": 170080
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.092603192774916e-05,
+      "loss": 0.4197,
+      "step": 170090
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.088363119845663e-05,
+      "loss": 0.3714,
+      "step": 170100
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.084123046916408e-05,
+      "loss": 0.3981,
+      "step": 170110
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.079882973987153e-05,
+      "loss": 0.3119,
+      "step": 170120
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.075642901057898e-05,
+      "loss": 0.433,
+      "step": 170130
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.071402828128645e-05,
+      "loss": 0.4269,
+      "step": 170140
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.06716275519939e-05,
+      "loss": 0.409,
+      "step": 170150
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.062922682270136e-05,
+      "loss": 0.4083,
+      "step": 170160
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.058682609340881e-05,
+      "loss": 0.2996,
+      "step": 170170
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.054442536411628e-05,
+      "loss": 0.3121,
+      "step": 170180
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.050202463482373e-05,
+      "loss": 0.3983,
+      "step": 170190
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.045962390553118e-05,
+      "loss": 0.3894,
+      "step": 170200
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.041722317623864e-05,
+      "loss": 0.3806,
+      "step": 170210
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.037482244694609e-05,
+      "loss": 0.3836,
+      "step": 170220
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.033242171765354e-05,
+      "loss": 0.3592,
+      "step": 170230
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.0290020988361e-05,
+      "loss": 0.4927,
+      "step": 170240
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.024762025906845e-05,
+      "loss": 0.4103,
+      "step": 170250
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.020521952977592e-05,
+      "loss": 0.372,
+      "step": 170260
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.016281880048337e-05,
+      "loss": 0.3087,
+      "step": 170270
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.012041807119082e-05,
+      "loss": 0.3686,
+      "step": 170280
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.007801734189828e-05,
+      "loss": 0.4004,
+      "step": 170290
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.003561661260574e-05,
+      "loss": 0.3754,
+      "step": 170300
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 7.99932158833132e-05,
+      "loss": 0.3689,
+      "step": 170310
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 7.995081515402065e-05,
+      "loss": 0.3984,
+      "step": 170320
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 7.99084144247281e-05,
+      "loss": 0.3923,
+      "step": 170330
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 7.986601369543556e-05,
+      "loss": 0.3257,
+      "step": 170340
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 7.982361296614302e-05,
+      "loss": 0.3839,
+      "step": 170350
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 7.978121223685048e-05,
+      "loss": 0.3412,
+      "step": 170360
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 7.973881150755793e-05,
+      "loss": 0.3416,
+      "step": 170370
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.969641077826538e-05,
+      "loss": 0.4021,
+      "step": 170380
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.965401004897285e-05,
+      "loss": 0.4112,
+      "step": 170390
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.96116093196803e-05,
+      "loss": 0.3572,
+      "step": 170400
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.956920859038776e-05,
+      "loss": 0.3205,
+      "step": 170410
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.952680786109521e-05,
+      "loss": 0.3719,
+      "step": 170420
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.948440713180268e-05,
+      "loss": 0.3753,
+      "step": 170430
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.944200640251013e-05,
+      "loss": 0.2839,
+      "step": 170440
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.939960567321758e-05,
+      "loss": 0.3807,
+      "step": 170450
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.935720494392504e-05,
+      "loss": 0.3959,
+      "step": 170460
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.93148042146325e-05,
+      "loss": 0.4145,
+      "step": 170470
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.927240348533996e-05,
+      "loss": 0.4156,
+      "step": 170480
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.923000275604741e-05,
+      "loss": 0.3069,
+      "step": 170490
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.918760202675486e-05,
+      "loss": 0.416,
+      "step": 170500
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.914520129746233e-05,
+      "loss": 0.419,
+      "step": 170510
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.910280056816978e-05,
+      "loss": 0.3591,
+      "step": 170520
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.906039983887724e-05,
+      "loss": 0.4125,
+      "step": 170530
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.901799910958469e-05,
+      "loss": 0.3318,
+      "step": 170540
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.897559838029214e-05,
+      "loss": 0.356,
+      "step": 170550
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.89331976509996e-05,
+      "loss": 0.3338,
+      "step": 170560
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.889079692170705e-05,
+      "loss": 0.4691,
+      "step": 170570
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.88483961924145e-05,
+      "loss": 0.4429,
+      "step": 170580
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.880599546312197e-05,
+      "loss": 0.4111,
+      "step": 170590
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.876359473382942e-05,
+      "loss": 0.4285,
+      "step": 170600
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.872119400453688e-05,
+      "loss": 0.3986,
+      "step": 170610
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.867879327524433e-05,
+      "loss": 0.4314,
+      "step": 170620
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.86363925459518e-05,
+      "loss": 0.3045,
+      "step": 170630
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.859399181665925e-05,
+      "loss": 0.4099,
+      "step": 170640
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.85515910873667e-05,
+      "loss": 0.387,
+      "step": 170650
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.850919035807416e-05,
+      "loss": 0.3686,
+      "step": 170660
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.846678962878162e-05,
+      "loss": 0.3773,
+      "step": 170670
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.842438889948908e-05,
+      "loss": 0.411,
+      "step": 170680
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.838198817019653e-05,
+      "loss": 0.3584,
+      "step": 170690
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.833958744090398e-05,
+      "loss": 0.3684,
+      "step": 170700
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.829718671161145e-05,
+      "loss": 0.3248,
+      "step": 170710
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.82547859823189e-05,
+      "loss": 0.3286,
+      "step": 170720
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.821238525302636e-05,
+      "loss": 0.3386,
+      "step": 170730
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.816998452373381e-05,
+      "loss": 0.3854,
+      "step": 170740
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.812758379444128e-05,
+      "loss": 0.3151,
+      "step": 170750
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.808518306514873e-05,
+      "loss": 0.3341,
+      "step": 170760
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.804278233585618e-05,
+      "loss": 0.3462,
+      "step": 170770
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.800038160656364e-05,
+      "loss": 0.378,
+      "step": 170780
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.79579808772711e-05,
+      "loss": 0.3625,
+      "step": 170790
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.791558014797856e-05,
+      "loss": 0.4656,
+      "step": 170800
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.787317941868601e-05,
+      "loss": 0.4071,
+      "step": 170810
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.783077868939346e-05,
+      "loss": 0.3687,
+      "step": 170820
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.778837796010093e-05,
+      "loss": 0.3901,
+      "step": 170830
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.774597723080838e-05,
+      "loss": 0.3837,
+      "step": 170840
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.770357650151583e-05,
+      "loss": 0.3887,
+      "step": 170850
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.766117577222329e-05,
+      "loss": 0.4009,
+      "step": 170860
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.761877504293074e-05,
+      "loss": 0.3983,
+      "step": 170870
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.75763743136382e-05,
+      "loss": 0.3567,
+      "step": 170880
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.753397358434565e-05,
+      "loss": 0.4064,
+      "step": 170890
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.74915728550531e-05,
+      "loss": 0.434,
+      "step": 170900
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.744917212576057e-05,
+      "loss": 0.3693,
+      "step": 170910
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.740677139646802e-05,
+      "loss": 0.3687,
+      "step": 170920
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.736437066717547e-05,
+      "loss": 0.3681,
+      "step": 170930
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.732196993788293e-05,
+      "loss": 0.4438,
+      "step": 170940
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.72795692085904e-05,
+      "loss": 0.3637,
+      "step": 170950
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.723716847929785e-05,
+      "loss": 0.352,
+      "step": 170960
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.71947677500053e-05,
+      "loss": 0.3578,
+      "step": 170970
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.715236702071275e-05,
+      "loss": 0.368,
+      "step": 170980
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.710996629142022e-05,
+      "loss": 0.3545,
+      "step": 170990
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.706756556212767e-05,
+      "loss": 0.3276,
+      "step": 171000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.702516483283513e-05,
+      "loss": 0.4462,
+      "step": 171010
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.698276410354258e-05,
+      "loss": 0.3985,
+      "step": 171020
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.694036337425005e-05,
+      "loss": 0.3256,
+      "step": 171030
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.68979626449575e-05,
+      "loss": 0.396,
+      "step": 171040
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.685556191566495e-05,
+      "loss": 0.3943,
+      "step": 171050
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.681316118637241e-05,
+      "loss": 0.3621,
+      "step": 171060
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.677076045707986e-05,
+      "loss": 0.3588,
+      "step": 171070
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.672835972778733e-05,
+      "loss": 0.3215,
+      "step": 171080
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.668595899849478e-05,
+      "loss": 0.3224,
+      "step": 171090
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.664355826920223e-05,
+      "loss": 0.3853,
+      "step": 171100
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.660115753990969e-05,
+      "loss": 0.3481,
+      "step": 171110
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.655875681061715e-05,
+      "loss": 0.419,
+      "step": 171120
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.65163560813246e-05,
+      "loss": 0.4104,
+      "step": 171130
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.647395535203206e-05,
+      "loss": 0.3409,
+      "step": 171140
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.643155462273951e-05,
+      "loss": 0.329,
+      "step": 171150
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.638915389344698e-05,
+      "loss": 0.3721,
+      "step": 171160
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.634675316415443e-05,
+      "loss": 0.2699,
+      "step": 171170
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.630435243486189e-05,
+      "loss": 0.3801,
+      "step": 171180
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.626195170556934e-05,
+      "loss": 0.3383,
+      "step": 171190
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.621955097627679e-05,
+      "loss": 0.465,
+      "step": 171200
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.617715024698425e-05,
+      "loss": 0.4145,
+      "step": 171210
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.61347495176917e-05,
+      "loss": 0.3599,
+      "step": 171220
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.609234878839915e-05,
+      "loss": 0.392,
+      "step": 171230
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.604994805910662e-05,
+      "loss": 0.3437,
+      "step": 171240
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.600754732981407e-05,
+      "loss": 0.4092,
+      "step": 171250
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.596514660052153e-05,
+      "loss": 0.3389,
+      "step": 171260
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.592274587122898e-05,
+      "loss": 0.3992,
+      "step": 171270
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.588034514193645e-05,
+      "loss": 0.361,
+      "step": 171280
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.58379444126439e-05,
+      "loss": 0.3174,
+      "step": 171290
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.579554368335135e-05,
+      "loss": 0.3782,
+      "step": 171300
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.57531429540588e-05,
+      "loss": 0.3276,
+      "step": 171310
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.571074222476627e-05,
+      "loss": 0.3997,
+      "step": 171320
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.566834149547373e-05,
+      "loss": 0.4079,
+      "step": 171330
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.562594076618118e-05,
+      "loss": 0.405,
+      "step": 171340
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.558354003688863e-05,
+      "loss": 0.4079,
+      "step": 171350
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.55411393075961e-05,
+      "loss": 0.3557,
+      "step": 171360
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.549873857830355e-05,
+      "loss": 0.374,
+      "step": 171370
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.5456337849011e-05,
+      "loss": 0.3766,
+      "step": 171380
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.541393711971846e-05,
+      "loss": 0.4272,
+      "step": 171390
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.537153639042593e-05,
+      "loss": 0.402,
+      "step": 171400
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.532913566113338e-05,
+      "loss": 0.3716,
+      "step": 171410
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.528673493184083e-05,
+      "loss": 0.3726,
+      "step": 171420
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.524433420254829e-05,
+      "loss": 0.413,
+      "step": 171430
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.520193347325575e-05,
+      "loss": 0.3826,
+      "step": 171440
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.51595327439632e-05,
+      "loss": 0.3967,
+      "step": 171450
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.511713201467066e-05,
+      "loss": 0.3142,
+      "step": 171460
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.507473128537811e-05,
+      "loss": 0.3711,
+      "step": 171470
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.503233055608558e-05,
+      "loss": 0.3387,
+      "step": 171480
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.498992982679303e-05,
+      "loss": 0.3564,
+      "step": 171490
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.494752909750048e-05,
+      "loss": 0.3998,
+      "step": 171500
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.490512836820794e-05,
+      "loss": 0.3978,
+      "step": 171510
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.486272763891539e-05,
+      "loss": 0.3881,
+      "step": 171520
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.482032690962284e-05,
+      "loss": 0.4302,
+      "step": 171530
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.47779261803303e-05,
+      "loss": 0.393,
+      "step": 171540
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.473552545103775e-05,
+      "loss": 0.4053,
+      "step": 171550
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.469312472174522e-05,
+      "loss": 0.4151,
+      "step": 171560
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.465072399245267e-05,
+      "loss": 0.3574,
+      "step": 171570
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.460832326316012e-05,
+      "loss": 0.3773,
+      "step": 171580
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.456592253386758e-05,
+      "loss": 0.3778,
+      "step": 171590
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.452352180457504e-05,
+      "loss": 0.3442,
+      "step": 171600
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.44811210752825e-05,
+      "loss": 0.3802,
+      "step": 171610
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.443872034598995e-05,
+      "loss": 0.3441,
+      "step": 171620
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.43963196166974e-05,
+      "loss": 0.3487,
+      "step": 171630
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.435391888740487e-05,
+      "loss": 0.3621,
+      "step": 171640
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.431151815811232e-05,
+      "loss": 0.3403,
+      "step": 171650
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.426911742881978e-05,
+      "loss": 0.3909,
+      "step": 171660
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.422671669952723e-05,
+      "loss": 0.3029,
+      "step": 171670
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.41843159702347e-05,
+      "loss": 0.382,
+      "step": 171680
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.414191524094215e-05,
+      "loss": 0.383,
+      "step": 171690
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.40995145116496e-05,
+      "loss": 0.3499,
+      "step": 171700
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.405711378235706e-05,
+      "loss": 0.4027,
+      "step": 171710
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.401471305306452e-05,
+      "loss": 0.3815,
+      "step": 171720
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.397231232377198e-05,
+      "loss": 0.3601,
+      "step": 171730
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.392991159447943e-05,
+      "loss": 0.3923,
+      "step": 171740
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.388751086518688e-05,
+      "loss": 0.3103,
+      "step": 171750
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.384511013589435e-05,
+      "loss": 0.3598,
+      "step": 171760
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.38027094066018e-05,
+      "loss": 0.3936,
+      "step": 171770
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.376030867730926e-05,
+      "loss": 0.3,
+      "step": 171780
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.371790794801671e-05,
+      "loss": 0.3332,
+      "step": 171790
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.367550721872416e-05,
+      "loss": 0.3595,
+      "step": 171800
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.363310648943163e-05,
+      "loss": 0.3371,
+      "step": 171810
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.359070576013908e-05,
+      "loss": 0.4136,
+      "step": 171820
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.354830503084654e-05,
+      "loss": 0.3537,
+      "step": 171830
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.350590430155399e-05,
+      "loss": 0.3894,
+      "step": 171840
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.346350357226144e-05,
+      "loss": 0.3667,
+      "step": 171850
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.34211028429689e-05,
+      "loss": 0.3133,
+      "step": 171860
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.337870211367636e-05,
+      "loss": 0.4134,
+      "step": 171870
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.333630138438382e-05,
+      "loss": 0.3117,
+      "step": 171880
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.329390065509127e-05,
+      "loss": 0.3367,
+      "step": 171890
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.325149992579872e-05,
+      "loss": 0.401,
+      "step": 171900
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.320909919650618e-05,
+      "loss": 0.3853,
+      "step": 171910
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.316669846721363e-05,
+      "loss": 0.3794,
+      "step": 171920
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.31242977379211e-05,
+      "loss": 0.324,
+      "step": 171930
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.308189700862855e-05,
+      "loss": 0.3347,
+      "step": 171940
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.3039496279336e-05,
+      "loss": 0.3859,
+      "step": 171950
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.299709555004346e-05,
+      "loss": 0.3902,
+      "step": 171960
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.295469482075092e-05,
+      "loss": 0.3681,
+      "step": 171970
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.291229409145838e-05,
+      "loss": 0.3484,
+      "step": 171980
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.286989336216583e-05,
+      "loss": 0.3982,
+      "step": 171990
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.282749263287328e-05,
+      "loss": 0.3683,
+      "step": 172000
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.278509190358075e-05,
+      "loss": 0.3707,
+      "step": 172010
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.27426911742882e-05,
+      "loss": 0.3684,
+      "step": 172020
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.270029044499566e-05,
+      "loss": 0.4124,
+      "step": 172030
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.265788971570311e-05,
+      "loss": 0.3506,
+      "step": 172040
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.261548898641058e-05,
+      "loss": 0.4148,
+      "step": 172050
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.257308825711803e-05,
+      "loss": 0.3499,
+      "step": 172060
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.253068752782548e-05,
+      "loss": 0.3862,
+      "step": 172070
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.248828679853294e-05,
+      "loss": 0.3426,
+      "step": 172080
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.24458860692404e-05,
+      "loss": 0.4725,
+      "step": 172090
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.240348533994786e-05,
+      "loss": 0.3597,
+      "step": 172100
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.236108461065531e-05,
+      "loss": 0.363,
+      "step": 172110
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.231868388136276e-05,
+      "loss": 0.4152,
+      "step": 172120
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.227628315207023e-05,
+      "loss": 0.3483,
+      "step": 172130
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.223388242277768e-05,
+      "loss": 0.3556,
+      "step": 172140
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.219148169348514e-05,
+      "loss": 0.3508,
+      "step": 172150
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.214908096419259e-05,
+      "loss": 0.42,
+      "step": 172160
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.210668023490004e-05,
+      "loss": 0.3519,
+      "step": 172170
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.20642795056075e-05,
+      "loss": 0.3296,
+      "step": 172180
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.202187877631496e-05,
+      "loss": 0.378,
+      "step": 172190
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.197947804702241e-05,
+      "loss": 0.3099,
+      "step": 172200
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.193707731772987e-05,
+      "loss": 0.3531,
+      "step": 172210
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.189467658843732e-05,
+      "loss": 0.3529,
+      "step": 172220
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.185227585914477e-05,
+      "loss": 0.346,
+      "step": 172230
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.180987512985223e-05,
+      "loss": 0.3837,
+      "step": 172240
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.17674744005597e-05,
+      "loss": 0.4271,
+      "step": 172250
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.172507367126715e-05,
+      "loss": 0.35,
+      "step": 172260
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.16826729419746e-05,
+      "loss": 0.4909,
+      "step": 172270
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.164027221268205e-05,
+      "loss": 0.3565,
+      "step": 172280
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.159787148338952e-05,
+      "loss": 0.4062,
+      "step": 172290
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.155547075409697e-05,
+      "loss": 0.4308,
+      "step": 172300
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.151307002480443e-05,
+      "loss": 0.4223,
+      "step": 172310
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.147066929551188e-05,
+      "loss": 0.3274,
+      "step": 172320
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.142826856621935e-05,
+      "loss": 0.368,
+      "step": 172330
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.13858678369268e-05,
+      "loss": 0.4376,
+      "step": 172340
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.134346710763425e-05,
+      "loss": 0.4027,
+      "step": 172350
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.130106637834171e-05,
+      "loss": 0.3553,
+      "step": 172360
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.125866564904917e-05,
+      "loss": 0.3649,
+      "step": 172370
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.121626491975663e-05,
+      "loss": 0.3078,
+      "step": 172380
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.117386419046408e-05,
+      "loss": 0.4552,
+      "step": 172390
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.113146346117153e-05,
+      "loss": 0.4395,
+      "step": 172400
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.1089062731879e-05,
+      "loss": 0.4239,
+      "step": 172410
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.104666200258645e-05,
+      "loss": 0.3768,
+      "step": 172420
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.100426127329391e-05,
+      "loss": 0.3263,
+      "step": 172430
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.096186054400136e-05,
+      "loss": 0.3895,
+      "step": 172440
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.091945981470883e-05,
+      "loss": 0.3807,
+      "step": 172450
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.087705908541628e-05,
+      "loss": 0.4476,
+      "step": 172460
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.083465835612373e-05,
+      "loss": 0.3645,
+      "step": 172470
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.079225762683119e-05,
+      "loss": 0.4035,
+      "step": 172480
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.074985689753864e-05,
+      "loss": 0.3538,
+      "step": 172490
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.07074561682461e-05,
+      "loss": 0.3487,
+      "step": 172500
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.066505543895356e-05,
+      "loss": 0.3608,
+      "step": 172510
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.062265470966101e-05,
+      "loss": 0.3379,
+      "step": 172520
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.058025398036847e-05,
+      "loss": 0.4122,
+      "step": 172530
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.053785325107592e-05,
+      "loss": 0.3434,
+      "step": 172540
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.049545252178337e-05,
+      "loss": 0.4031,
+      "step": 172550
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.045305179249083e-05,
+      "loss": 0.3508,
+      "step": 172560
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.041065106319828e-05,
+      "loss": 0.4131,
+      "step": 172570
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.036825033390575e-05,
+      "loss": 0.3949,
+      "step": 172580
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.03258496046132e-05,
+      "loss": 0.3846,
+      "step": 172590
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.028344887532065e-05,
+      "loss": 0.4409,
+      "step": 172600
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.02410481460281e-05,
+      "loss": 0.3305,
+      "step": 172610
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.019864741673557e-05,
+      "loss": 0.4115,
+      "step": 172620
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.015624668744303e-05,
+      "loss": 0.3268,
+      "step": 172630
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.011384595815048e-05,
+      "loss": 0.3466,
+      "step": 172640
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.007144522885793e-05,
+      "loss": 0.352,
+      "step": 172650
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.00290444995654e-05,
+      "loss": 0.3175,
+      "step": 172660
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.998664377027285e-05,
+      "loss": 0.4346,
+      "step": 172670
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.99442430409803e-05,
+      "loss": 0.358,
+      "step": 172680
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.990184231168776e-05,
+      "loss": 0.3724,
+      "step": 172690
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.985944158239523e-05,
+      "loss": 0.3692,
+      "step": 172700
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.981704085310268e-05,
+      "loss": 0.3441,
+      "step": 172710
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.977464012381013e-05,
+      "loss": 0.4359,
+      "step": 172720
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.973223939451759e-05,
+      "loss": 0.315,
+      "step": 172730
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.968983866522505e-05,
+      "loss": 0.4402,
+      "step": 172740
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.96474379359325e-05,
+      "loss": 0.3975,
+      "step": 172750
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.960503720663996e-05,
+      "loss": 0.3618,
+      "step": 172760
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.956263647734741e-05,
+      "loss": 0.438,
+      "step": 172770
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.952023574805488e-05,
+      "loss": 0.3574,
+      "step": 172780
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.947783501876233e-05,
+      "loss": 0.3462,
+      "step": 172790
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.943543428946979e-05,
+      "loss": 0.3501,
+      "step": 172800
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.939303356017724e-05,
+      "loss": 0.3997,
+      "step": 172810
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.93506328308847e-05,
+      "loss": 0.3539,
+      "step": 172820
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.930823210159216e-05,
+      "loss": 0.3753,
+      "step": 172830
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.926583137229961e-05,
+      "loss": 0.401,
+      "step": 172840
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.922343064300706e-05,
+      "loss": 0.4588,
+      "step": 172850
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.918102991371452e-05,
+      "loss": 0.3881,
+      "step": 172860
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.913862918442197e-05,
+      "loss": 0.322,
+      "step": 172870
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.909622845512942e-05,
+      "loss": 0.4331,
+      "step": 172880
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.905382772583688e-05,
+      "loss": 0.3793,
+      "step": 172890
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.901142699654434e-05,
+      "loss": 0.381,
+      "step": 172900
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.89690262672518e-05,
+      "loss": 0.3423,
+      "step": 172910
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.892662553795925e-05,
+      "loss": 0.3895,
+      "step": 172920
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.88842248086667e-05,
+      "loss": 0.3675,
+      "step": 172930
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.884182407937417e-05,
+      "loss": 0.4274,
+      "step": 172940
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.879942335008162e-05,
+      "loss": 0.4417,
+      "step": 172950
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.875702262078908e-05,
+      "loss": 0.4222,
+      "step": 172960
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.871462189149653e-05,
+      "loss": 0.3852,
+      "step": 172970
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.8672221162204e-05,
+      "loss": 0.2801,
+      "step": 172980
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.862982043291145e-05,
+      "loss": 0.376,
+      "step": 172990
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.85874197036189e-05,
+      "loss": 0.4015,
+      "step": 173000
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.854501897432636e-05,
+      "loss": 0.3521,
+      "step": 173010
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.850261824503382e-05,
+      "loss": 0.3918,
+      "step": 173020
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.846021751574128e-05,
+      "loss": 0.4287,
+      "step": 173030
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.841781678644873e-05,
+      "loss": 0.366,
+      "step": 173040
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.837541605715618e-05,
+      "loss": 0.3727,
+      "step": 173050
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.833301532786365e-05,
+      "loss": 0.363,
+      "step": 173060
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.82906145985711e-05,
+      "loss": 0.3568,
+      "step": 173070
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.824821386927856e-05,
+      "loss": 0.4068,
+      "step": 173080
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.820581313998601e-05,
+      "loss": 0.3891,
+      "step": 173090
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.816341241069348e-05,
+      "loss": 0.3769,
+      "step": 173100
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.812101168140093e-05,
+      "loss": 0.3891,
+      "step": 173110
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.807861095210838e-05,
+      "loss": 0.3451,
+      "step": 173120
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.803621022281584e-05,
+      "loss": 0.3525,
+      "step": 173130
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.79938094935233e-05,
+      "loss": 0.3684,
+      "step": 173140
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.795140876423076e-05,
+      "loss": 0.4081,
+      "step": 173150
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.790900803493821e-05,
+      "loss": 0.3493,
+      "step": 173160
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.786660730564566e-05,
+      "loss": 0.3457,
+      "step": 173170
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.782420657635312e-05,
+      "loss": 0.399,
+      "step": 173180
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.778180584706057e-05,
+      "loss": 0.3855,
+      "step": 173190
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.773940511776802e-05,
+      "loss": 0.3548,
+      "step": 173200
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.769700438847548e-05,
+      "loss": 0.4051,
+      "step": 173210
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.765460365918293e-05,
+      "loss": 0.3812,
+      "step": 173220
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.76122029298904e-05,
+      "loss": 0.4145,
+      "step": 173230
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.756980220059785e-05,
+      "loss": 0.4225,
+      "step": 173240
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.75274014713053e-05,
+      "loss": 0.3765,
+      "step": 173250
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.748500074201276e-05,
+      "loss": 0.3869,
+      "step": 173260
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.744260001272022e-05,
+      "loss": 0.4071,
+      "step": 173270
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.740019928342768e-05,
+      "loss": 0.3194,
+      "step": 173280
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.735779855413513e-05,
+      "loss": 0.3864,
+      "step": 173290
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.731539782484258e-05,
+      "loss": 0.3122,
+      "step": 173300
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.727299709555005e-05,
+      "loss": 0.4194,
+      "step": 173310
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.72305963662575e-05,
+      "loss": 0.3644,
+      "step": 173320
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.718819563696496e-05,
+      "loss": 0.4265,
+      "step": 173330
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.714579490767241e-05,
+      "loss": 0.386,
+      "step": 173340
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.710339417837988e-05,
+      "loss": 0.4148,
+      "step": 173350
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.706099344908733e-05,
+      "loss": 0.3463,
+      "step": 173360
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.701859271979478e-05,
+      "loss": 0.3471,
+      "step": 173370
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.697619199050224e-05,
+      "loss": 0.3535,
+      "step": 173380
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.69337912612097e-05,
+      "loss": 0.3487,
+      "step": 173390
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.689139053191716e-05,
+      "loss": 0.4342,
+      "step": 173400
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.684898980262461e-05,
+      "loss": 0.4685,
+      "step": 173410
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.680658907333206e-05,
+      "loss": 0.4091,
+      "step": 173420
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.676418834403953e-05,
+      "loss": 0.4086,
+      "step": 173430
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.672178761474698e-05,
+      "loss": 0.3555,
+      "step": 173440
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.667938688545444e-05,
+      "loss": 0.367,
+      "step": 173450
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.663698615616189e-05,
+      "loss": 0.4097,
+      "step": 173460
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.659458542686936e-05,
+      "loss": 0.3858,
+      "step": 173470
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.655218469757681e-05,
+      "loss": 0.3607,
+      "step": 173480
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.650978396828426e-05,
+      "loss": 0.3548,
+      "step": 173490
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.646738323899171e-05,
+      "loss": 0.3247,
+      "step": 173500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.642498250969917e-05,
+      "loss": 0.3973,
+      "step": 173510
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.638258178040662e-05,
+      "loss": 0.3887,
+      "step": 173520
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.634018105111407e-05,
+      "loss": 0.3555,
+      "step": 173530
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.629778032182153e-05,
+      "loss": 0.3292,
+      "step": 173540
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.6255379592529e-05,
+      "loss": 0.3494,
+      "step": 173550
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.621297886323645e-05,
+      "loss": 0.3725,
+      "step": 173560
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.61705781339439e-05,
+      "loss": 0.383,
+      "step": 173570
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.612817740465135e-05,
+      "loss": 0.4238,
+      "step": 173580
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.608577667535882e-05,
+      "loss": 0.3906,
+      "step": 173590
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.604337594606627e-05,
+      "loss": 0.3269,
+      "step": 173600
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.600097521677373e-05,
+      "loss": 0.3748,
+      "step": 173610
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.595857448748118e-05,
+      "loss": 0.4058,
+      "step": 173620
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.591617375818865e-05,
+      "loss": 0.4392,
+      "step": 173630
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.58737730288961e-05,
+      "loss": 0.3932,
+      "step": 173640
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.583137229960355e-05,
+      "loss": 0.3973,
+      "step": 173650
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.578897157031101e-05,
+      "loss": 0.4611,
+      "step": 173660
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.574657084101847e-05,
+      "loss": 0.4586,
+      "step": 173670
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.570417011172593e-05,
+      "loss": 0.3624,
+      "step": 173680
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.566176938243338e-05,
+      "loss": 0.3708,
+      "step": 173690
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.561936865314083e-05,
+      "loss": 0.3922,
+      "step": 173700
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.55769679238483e-05,
+      "loss": 0.3983,
+      "step": 173710
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.553456719455575e-05,
+      "loss": 0.339,
+      "step": 173720
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.549216646526321e-05,
+      "loss": 0.3458,
+      "step": 173730
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.544976573597066e-05,
+      "loss": 0.4541,
+      "step": 173740
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.540736500667813e-05,
+      "loss": 0.3575,
+      "step": 173750
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.536496427738558e-05,
+      "loss": 0.3164,
+      "step": 173760
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.532256354809303e-05,
+      "loss": 0.3596,
+      "step": 173770
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.528016281880049e-05,
+      "loss": 0.3461,
+      "step": 173780
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.523776208950795e-05,
+      "loss": 0.4001,
+      "step": 173790
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.519536136021541e-05,
+      "loss": 0.4258,
+      "step": 173800
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.515296063092286e-05,
+      "loss": 0.4026,
+      "step": 173810
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.511055990163031e-05,
+      "loss": 0.4364,
+      "step": 173820
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.506815917233777e-05,
+      "loss": 0.3625,
+      "step": 173830
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.502575844304522e-05,
+      "loss": 0.299,
+      "step": 173840
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.498335771375267e-05,
+      "loss": 0.4057,
+      "step": 173850
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.494095698446013e-05,
+      "loss": 0.3528,
+      "step": 173860
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.489855625516759e-05,
+      "loss": 0.3375,
+      "step": 173870
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.485615552587505e-05,
+      "loss": 0.415,
+      "step": 173880
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.48137547965825e-05,
+      "loss": 0.412,
+      "step": 173890
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.477135406728995e-05,
+      "loss": 0.3714,
+      "step": 173900
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.47289533379974e-05,
+      "loss": 0.3782,
+      "step": 173910
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.468655260870487e-05,
+      "loss": 0.4251,
+      "step": 173920
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.464415187941233e-05,
+      "loss": 0.4202,
+      "step": 173930
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.460175115011978e-05,
+      "loss": 0.2971,
+      "step": 173940
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.455935042082723e-05,
+      "loss": 0.3705,
+      "step": 173950
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.45169496915347e-05,
+      "loss": 0.3384,
+      "step": 173960
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.447454896224215e-05,
+      "loss": 0.3753,
+      "step": 173970
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.44321482329496e-05,
+      "loss": 0.3352,
+      "step": 173980
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.438974750365706e-05,
+      "loss": 0.336,
+      "step": 173990
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.434734677436453e-05,
+      "loss": 0.3811,
+      "step": 174000
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.430494604507198e-05,
+      "loss": 0.3349,
+      "step": 174010
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.426254531577943e-05,
+      "loss": 0.4321,
+      "step": 174020
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.422014458648689e-05,
+      "loss": 0.4625,
+      "step": 174030
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.417774385719435e-05,
+      "loss": 0.383,
+      "step": 174040
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.41353431279018e-05,
+      "loss": 0.3544,
+      "step": 174050
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.409294239860926e-05,
+      "loss": 0.326,
+      "step": 174060
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.405054166931671e-05,
+      "loss": 0.3408,
+      "step": 174070
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.400814094002418e-05,
+      "loss": 0.421,
+      "step": 174080
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.396574021073163e-05,
+      "loss": 0.3765,
+      "step": 174090
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.392333948143909e-05,
+      "loss": 0.428,
+      "step": 174100
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.388093875214654e-05,
+      "loss": 0.4037,
+      "step": 174110
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.3838538022854e-05,
+      "loss": 0.4146,
+      "step": 174120
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.379613729356146e-05,
+      "loss": 0.3864,
+      "step": 174130
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.375373656426891e-05,
+      "loss": 0.409,
+      "step": 174140
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.371133583497637e-05,
+      "loss": 0.4444,
+      "step": 174150
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.366893510568382e-05,
+      "loss": 0.3463,
+      "step": 174160
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.362653437639127e-05,
+      "loss": 0.4287,
+      "step": 174170
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.358413364709872e-05,
+      "loss": 0.4066,
+      "step": 174180
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.354173291780619e-05,
+      "loss": 0.3587,
+      "step": 174190
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.349933218851364e-05,
+      "loss": 0.3499,
+      "step": 174200
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.34569314592211e-05,
+      "loss": 0.3972,
+      "step": 174210
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.341453072992855e-05,
+      "loss": 0.2979,
+      "step": 174220
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.3372130000636e-05,
+      "loss": 0.4268,
+      "step": 174230
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.332972927134347e-05,
+      "loss": 0.4125,
+      "step": 174240
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.328732854205092e-05,
+      "loss": 0.4404,
+      "step": 174250
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.324492781275838e-05,
+      "loss": 0.3742,
+      "step": 174260
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.320252708346583e-05,
+      "loss": 0.3958,
+      "step": 174270
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.31601263541733e-05,
+      "loss": 0.4461,
+      "step": 174280
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.311772562488075e-05,
+      "loss": 0.3608,
+      "step": 174290
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.30753248955882e-05,
+      "loss": 0.4369,
+      "step": 174300
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.303292416629566e-05,
+      "loss": 0.3674,
+      "step": 174310
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.299052343700312e-05,
+      "loss": 0.3832,
+      "step": 174320
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.294812270771058e-05,
+      "loss": 0.3463,
+      "step": 174330
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.290572197841803e-05,
+      "loss": 0.3489,
+      "step": 174340
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.286332124912548e-05,
+      "loss": 0.3796,
+      "step": 174350
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.282092051983295e-05,
+      "loss": 0.3685,
+      "step": 174360
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.27785197905404e-05,
+      "loss": 0.3854,
+      "step": 174370
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.273611906124786e-05,
+      "loss": 0.4135,
+      "step": 174380
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.269371833195531e-05,
+      "loss": 0.3676,
+      "step": 174390
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.265131760266278e-05,
+      "loss": 0.3841,
+      "step": 174400
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.260891687337023e-05,
+      "loss": 0.3186,
+      "step": 174410
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.256651614407768e-05,
+      "loss": 0.3675,
+      "step": 174420
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.252411541478514e-05,
+      "loss": 0.3773,
+      "step": 174430
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.24817146854926e-05,
+      "loss": 0.4095,
+      "step": 174440
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.243931395620006e-05,
+      "loss": 0.3566,
+      "step": 174450
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.239691322690751e-05,
+      "loss": 0.4223,
+      "step": 174460
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.235451249761496e-05,
+      "loss": 0.3743,
+      "step": 174470
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.231211176832242e-05,
+      "loss": 0.344,
+      "step": 174480
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.226971103902987e-05,
+      "loss": 0.3455,
+      "step": 174490
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.222731030973734e-05,
+      "loss": 0.3565,
+      "step": 174500
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.218490958044479e-05,
+      "loss": 0.3965,
+      "step": 174510
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.214250885115224e-05,
+      "loss": 0.3789,
+      "step": 174520
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.21001081218597e-05,
+      "loss": 0.4274,
+      "step": 174530
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.205770739256715e-05,
+      "loss": 0.3832,
+      "step": 174540
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.20153066632746e-05,
+      "loss": 0.3845,
+      "step": 174550
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.197290593398207e-05,
+      "loss": 0.3934,
+      "step": 174560
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.193050520468952e-05,
+      "loss": 0.3591,
+      "step": 174570
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.188810447539698e-05,
+      "loss": 0.4131,
+      "step": 174580
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.184570374610443e-05,
+      "loss": 0.3047,
+      "step": 174590
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.18033030168119e-05,
+      "loss": 0.381,
+      "step": 174600
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.176090228751935e-05,
+      "loss": 0.338,
+      "step": 174610
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.17185015582268e-05,
+      "loss": 0.3644,
+      "step": 174620
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.167610082893426e-05,
+      "loss": 0.3396,
+      "step": 174630
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.163370009964171e-05,
+      "loss": 0.3568,
+      "step": 174640
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.159129937034918e-05,
+      "loss": 0.4121,
+      "step": 174650
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.154889864105663e-05,
+      "loss": 0.3842,
+      "step": 174660
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.150649791176408e-05,
+      "loss": 0.3603,
+      "step": 174670
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.146409718247154e-05,
+      "loss": 0.2844,
+      "step": 174680
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.1421696453179e-05,
+      "loss": 0.4306,
+      "step": 174690
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.137929572388646e-05,
+      "loss": 0.3651,
+      "step": 174700
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.133689499459391e-05,
+      "loss": 0.3826,
+      "step": 174710
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.129449426530136e-05,
+      "loss": 0.3481,
+      "step": 174720
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.125209353600883e-05,
+      "loss": 0.3317,
+      "step": 174730
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.120969280671628e-05,
+      "loss": 0.3507,
+      "step": 174740
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.116729207742374e-05,
+      "loss": 0.4062,
+      "step": 174750
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.112489134813119e-05,
+      "loss": 0.313,
+      "step": 174760
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.108249061883866e-05,
+      "loss": 0.3911,
+      "step": 174770
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.104008988954611e-05,
+      "loss": 0.3823,
+      "step": 174780
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.0997689160253555e-05,
+      "loss": 0.348,
+      "step": 174790
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.095528843096101e-05,
+      "loss": 0.3332,
+      "step": 174800
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.0912887701668475e-05,
+      "loss": 0.3292,
+      "step": 174810
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.087048697237593e-05,
+      "loss": 0.3904,
+      "step": 174820
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.082808624308338e-05,
+      "loss": 0.4173,
+      "step": 174830
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.0785685513790835e-05,
+      "loss": 0.3525,
+      "step": 174840
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.07432847844983e-05,
+      "loss": 0.3782,
+      "step": 174850
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.0700884055205755e-05,
+      "loss": 0.3038,
+      "step": 174860
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.065848332591321e-05,
+      "loss": 0.3905,
+      "step": 174870
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.061608259662066e-05,
+      "loss": 0.349,
+      "step": 174880
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.057368186732812e-05,
+      "loss": 0.3189,
+      "step": 174890
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.053128113803558e-05,
+      "loss": 0.3492,
+      "step": 174900
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.0488880408743035e-05,
+      "loss": 0.4132,
+      "step": 174910
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.044647967945049e-05,
+      "loss": 0.3743,
+      "step": 174920
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.040407895015795e-05,
+      "loss": 0.3453,
+      "step": 174930
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.03616782208654e-05,
+      "loss": 0.3586,
+      "step": 174940
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.0319277491572854e-05,
+      "loss": 0.3046,
+      "step": 174950
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.027687676228031e-05,
+      "loss": 0.3693,
+      "step": 174960
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.0234476032987774e-05,
+      "loss": 0.4343,
+      "step": 174970
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.019207530369523e-05,
+      "loss": 0.3297,
+      "step": 174980
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.014967457440268e-05,
+      "loss": 0.3643,
+      "step": 174990
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.0107273845110134e-05,
+      "loss": 0.3618,
+      "step": 175000
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.00648731158176e-05,
+      "loss": 0.3853,
+      "step": 175010
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.0022472386525054e-05,
+      "loss": 0.3925,
+      "step": 175020
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.998007165723251e-05,
+      "loss": 0.3966,
+      "step": 175030
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.993767092793996e-05,
+      "loss": 0.4144,
+      "step": 175040
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.989527019864743e-05,
+      "loss": 0.4148,
+      "step": 175050
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.985286946935488e-05,
+      "loss": 0.3321,
+      "step": 175060
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.9810468740062334e-05,
+      "loss": 0.4209,
+      "step": 175070
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.976806801076979e-05,
+      "loss": 0.3828,
+      "step": 175080
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.972566728147725e-05,
+      "loss": 0.415,
+      "step": 175090
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.96832665521847e-05,
+      "loss": 0.3596,
+      "step": 175100
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.9640865822892154e-05,
+      "loss": 0.332,
+      "step": 175110
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.959846509359961e-05,
+      "loss": 0.3536,
+      "step": 175120
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.9556064364307074e-05,
+      "loss": 0.3176,
+      "step": 175130
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.951366363501453e-05,
+      "loss": 0.4183,
+      "step": 175140
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.947126290572198e-05,
+      "loss": 0.3716,
+      "step": 175150
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.942886217642943e-05,
+      "loss": 0.317,
+      "step": 175160
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.93864614471369e-05,
+      "loss": 0.3348,
+      "step": 175170
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.934406071784435e-05,
+      "loss": 0.4548,
+      "step": 175180
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.9301659988551807e-05,
+      "loss": 0.3574,
+      "step": 175190
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.925925925925926e-05,
+      "loss": 0.3478,
+      "step": 175200
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.921685852996673e-05,
+      "loss": 0.3579,
+      "step": 175210
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.917445780067418e-05,
+      "loss": 0.421,
+      "step": 175220
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.913205707138163e-05,
+      "loss": 0.34,
+      "step": 175230
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.9089656342089086e-05,
+      "loss": 0.3464,
+      "step": 175240
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.9047255612796546e-05,
+      "loss": 0.3571,
+      "step": 175250
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.9004854883504e-05,
+      "loss": 0.3104,
+      "step": 175260
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.896245415421145e-05,
+      "loss": 0.3454,
+      "step": 175270
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.8920053424918906e-05,
+      "loss": 0.3764,
+      "step": 175280
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.887765269562637e-05,
+      "loss": 0.352,
+      "step": 175290
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.8835251966333826e-05,
+      "loss": 0.384,
+      "step": 175300
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.879285123704128e-05,
+      "loss": 0.3834,
+      "step": 175310
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.875045050774873e-05,
+      "loss": 0.4073,
+      "step": 175320
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.87080497784562e-05,
+      "loss": 0.3307,
+      "step": 175330
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.866564904916365e-05,
+      "loss": 0.4457,
+      "step": 175340
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.8623248319871106e-05,
+      "loss": 0.3902,
+      "step": 175350
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.858084759057856e-05,
+      "loss": 0.3326,
+      "step": 175360
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.853844686128601e-05,
+      "loss": 0.3518,
+      "step": 175370
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.849604613199348e-05,
+      "loss": 0.4413,
+      "step": 175380
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.845364540270093e-05,
+      "loss": 0.4172,
+      "step": 175390
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.8411244673408386e-05,
+      "loss": 0.3694,
+      "step": 175400
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.836884394411584e-05,
+      "loss": 0.3862,
+      "step": 175410
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.83264432148233e-05,
+      "loss": 0.4081,
+      "step": 175420
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.828404248553075e-05,
+      "loss": 0.3461,
+      "step": 175430
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.8241641756238205e-05,
+      "loss": 0.3518,
+      "step": 175440
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.819924102694566e-05,
+      "loss": 0.3562,
+      "step": 175450
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.8156840297653125e-05,
+      "loss": 0.3649,
+      "step": 175460
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.811443956836058e-05,
+      "loss": 0.2926,
+      "step": 175470
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.807203883906803e-05,
+      "loss": 0.3788,
+      "step": 175480
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.8029638109775485e-05,
+      "loss": 0.3695,
+      "step": 175490
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.798723738048295e-05,
+      "loss": 0.3847,
+      "step": 175500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.7944836651190405e-05,
+      "loss": 0.3416,
+      "step": 175510
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.790243592189786e-05,
+      "loss": 0.3611,
+      "step": 175520
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.786003519260531e-05,
+      "loss": 0.3856,
+      "step": 175530
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.781763446331278e-05,
+      "loss": 0.3495,
+      "step": 175540
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.777523373402023e-05,
+      "loss": 0.3242,
+      "step": 175550
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.7732833004727685e-05,
+      "loss": 0.4048,
+      "step": 175560
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.769043227543514e-05,
+      "loss": 0.4478,
+      "step": 175570
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.76480315461426e-05,
+      "loss": 0.3036,
+      "step": 175580
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.760563081685005e-05,
+      "loss": 0.3859,
+      "step": 175590
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.7563230087557504e-05,
+      "loss": 0.3352,
+      "step": 175600
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.752082935826496e-05,
+      "loss": 0.2999,
+      "step": 175610
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.7478428628972425e-05,
+      "loss": 0.2795,
+      "step": 175620
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.743602789967988e-05,
+      "loss": 0.3459,
+      "step": 175630
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.739362717038733e-05,
+      "loss": 0.3902,
+      "step": 175640
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.7351226441094784e-05,
+      "loss": 0.3841,
+      "step": 175650
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.730882571180225e-05,
+      "loss": 0.3245,
+      "step": 175660
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.7266424982509704e-05,
+      "loss": 0.326,
+      "step": 175670
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.722402425321716e-05,
+      "loss": 0.3596,
+      "step": 175680
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.718162352392461e-05,
+      "loss": 0.3246,
+      "step": 175690
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.713922279463208e-05,
+      "loss": 0.3547,
+      "step": 175700
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.709682206533953e-05,
+      "loss": 0.3725,
+      "step": 175710
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.7054421336046984e-05,
+      "loss": 0.4036,
+      "step": 175720
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.701202060675444e-05,
+      "loss": 0.4171,
+      "step": 175730
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.69696198774619e-05,
+      "loss": 0.4039,
+      "step": 175740
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.692721914816935e-05,
+      "loss": 0.3858,
+      "step": 175750
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.6884818418876804e-05,
+      "loss": 0.4296,
+      "step": 175760
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.684241768958426e-05,
+      "loss": 0.3723,
+      "step": 175770
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.6800016960291724e-05,
+      "loss": 0.3846,
+      "step": 175780
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.675761623099918e-05,
+      "loss": 0.377,
+      "step": 175790
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.671521550170663e-05,
+      "loss": 0.4501,
+      "step": 175800
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.6672814772414083e-05,
+      "loss": 0.4178,
+      "step": 175810
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.663041404312155e-05,
+      "loss": 0.3466,
+      "step": 175820
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6588013313829003e-05,
+      "loss": 0.3654,
+      "step": 175830
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.654561258453646e-05,
+      "loss": 0.3272,
+      "step": 175840
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.650321185524391e-05,
+      "loss": 0.3593,
+      "step": 175850
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.646081112595138e-05,
+      "loss": 0.3666,
+      "step": 175860
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.641841039665883e-05,
+      "loss": 0.386,
+      "step": 175870
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.637600966736628e-05,
+      "loss": 0.4376,
+      "step": 175880
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6333608938073736e-05,
+      "loss": 0.3595,
+      "step": 175890
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6291208208781196e-05,
+      "loss": 0.3506,
+      "step": 175900
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.624880747948865e-05,
+      "loss": 0.3578,
+      "step": 175910
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.62064067501961e-05,
+      "loss": 0.4632,
+      "step": 175920
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6164006020903556e-05,
+      "loss": 0.3453,
+      "step": 175930
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.612160529161102e-05,
+      "loss": 0.3645,
+      "step": 175940
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6079204562318476e-05,
+      "loss": 0.3817,
+      "step": 175950
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.603680383302593e-05,
+      "loss": 0.3971,
+      "step": 175960
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.599440310373338e-05,
+      "loss": 0.3397,
+      "step": 175970
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.595200237444085e-05,
+      "loss": 0.3526,
+      "step": 175980
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.59096016451483e-05,
+      "loss": 0.3404,
+      "step": 175990
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.5867200915855756e-05,
+      "loss": 0.4214,
+      "step": 176000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.582480018656321e-05,
+      "loss": 0.3355,
+      "step": 176010
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.5782399457270676e-05,
+      "loss": 0.4348,
+      "step": 176020
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.573999872797813e-05,
+      "loss": 0.4481,
+      "step": 176030
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.569759799868558e-05,
+      "loss": 0.3806,
+      "step": 176040
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.5655197269393036e-05,
+      "loss": 0.3674,
+      "step": 176050
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.561279654010049e-05,
+      "loss": 0.3548,
+      "step": 176060
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.557039581080795e-05,
+      "loss": 0.3405,
+      "step": 176070
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.55279950815154e-05,
+      "loss": 0.4247,
+      "step": 176080
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.5485594352222855e-05,
+      "loss": 0.4044,
+      "step": 176090
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.544319362293031e-05,
+      "loss": 0.3834,
+      "step": 176100
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.5400792893637775e-05,
+      "loss": 0.4171,
+      "step": 176110
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.535839216434523e-05,
+      "loss": 0.4269,
+      "step": 176120
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.531599143505268e-05,
+      "loss": 0.3761,
+      "step": 176130
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.5273590705760135e-05,
+      "loss": 0.319,
+      "step": 176140
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.52311899764676e-05,
+      "loss": 0.3601,
+      "step": 176150
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.5188789247175055e-05,
+      "loss": 0.3993,
+      "step": 176160
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.514638851788251e-05,
+      "loss": 0.3788,
+      "step": 176170
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.510398778858996e-05,
+      "loss": 0.377,
+      "step": 176180
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.506158705929743e-05,
+      "loss": 0.3736,
+      "step": 176190
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.501918633000488e-05,
+      "loss": 0.3471,
+      "step": 176200
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.4976785600712335e-05,
+      "loss": 0.3423,
+      "step": 176210
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.493438487141979e-05,
+      "loss": 0.3802,
+      "step": 176220
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.489198414212725e-05,
+      "loss": 0.3765,
+      "step": 176230
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.48495834128347e-05,
+      "loss": 0.3609,
+      "step": 176240
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.4807182683542155e-05,
+      "loss": 0.3526,
+      "step": 176250
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.476478195424961e-05,
+      "loss": 0.3622,
+      "step": 176260
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.4722381224957075e-05,
+      "loss": 0.371,
+      "step": 176270
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.467998049566453e-05,
+      "loss": 0.3725,
+      "step": 176280
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.463757976637198e-05,
+      "loss": 0.3929,
+      "step": 176290
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.4595179037079434e-05,
+      "loss": 0.3034,
+      "step": 176300
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.45527783077869e-05,
+      "loss": 0.3299,
+      "step": 176310
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.4510377578494354e-05,
+      "loss": 0.3503,
+      "step": 176320
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.446797684920181e-05,
+      "loss": 0.4074,
+      "step": 176330
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.442557611990926e-05,
+      "loss": 0.3763,
+      "step": 176340
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.438317539061673e-05,
+      "loss": 0.3686,
+      "step": 176350
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.434077466132418e-05,
+      "loss": 0.3052,
+      "step": 176360
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.4298373932031634e-05,
+      "loss": 0.3402,
+      "step": 176370
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.425597320273909e-05,
+      "loss": 0.3552,
+      "step": 176380
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.421357247344655e-05,
+      "loss": 0.3635,
+      "step": 176390
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.4171171744154e-05,
+      "loss": 0.3628,
+      "step": 176400
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.4128771014861454e-05,
+      "loss": 0.411,
+      "step": 176410
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.408637028556891e-05,
+      "loss": 0.3293,
+      "step": 176420
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.4043969556276374e-05,
+      "loss": 0.3364,
+      "step": 176430
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.400156882698383e-05,
+      "loss": 0.4364,
+      "step": 176440
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.395916809769128e-05,
+      "loss": 0.4053,
+      "step": 176450
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.3916767368398734e-05,
+      "loss": 0.3674,
+      "step": 176460
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.38743666391062e-05,
+      "loss": 0.3821,
+      "step": 176470
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.3831965909813654e-05,
+      "loss": 0.3585,
+      "step": 176480
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.378956518052111e-05,
+      "loss": 0.4259,
+      "step": 176490
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.374716445122856e-05,
+      "loss": 0.3391,
+      "step": 176500
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.370476372193603e-05,
+      "loss": 0.3735,
+      "step": 176510
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.366236299264348e-05,
+      "loss": 0.361,
+      "step": 176520
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.361996226335093e-05,
+      "loss": 0.3635,
+      "step": 176530
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.3577561534058387e-05,
+      "loss": 0.3883,
+      "step": 176540
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.3535160804765847e-05,
+      "loss": 0.3756,
+      "step": 176550
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.34927600754733e-05,
+      "loss": 0.3928,
+      "step": 176560
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.345035934618075e-05,
+      "loss": 0.4315,
+      "step": 176570
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.3407958616888206e-05,
+      "loss": 0.3122,
+      "step": 176580
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.336555788759567e-05,
+      "loss": 0.3662,
+      "step": 176590
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.3323157158303126e-05,
+      "loss": 0.3335,
+      "step": 176600
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.328075642901058e-05,
+      "loss": 0.3843,
+      "step": 176610
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.323835569971803e-05,
+      "loss": 0.3528,
+      "step": 176620
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.31959549704255e-05,
+      "loss": 0.3934,
+      "step": 176630
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.315355424113295e-05,
+      "loss": 0.3595,
+      "step": 176640
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.3111153511840406e-05,
+      "loss": 0.3613,
+      "step": 176650
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.306875278254786e-05,
+      "loss": 0.4442,
+      "step": 176660
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.3026352053255326e-05,
+      "loss": 0.3459,
+      "step": 176670
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.298395132396278e-05,
+      "loss": 0.3446,
+      "step": 176680
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.294155059467023e-05,
+      "loss": 0.452,
+      "step": 176690
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.2899149865377686e-05,
+      "loss": 0.3363,
+      "step": 176700
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.2856749136085146e-05,
+      "loss": 0.3461,
+      "step": 176710
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.28143484067926e-05,
+      "loss": 0.3985,
+      "step": 176720
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.277194767750005e-05,
+      "loss": 0.3847,
+      "step": 176730
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.2729546948207505e-05,
+      "loss": 0.4289,
+      "step": 176740
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.268714621891497e-05,
+      "loss": 0.3596,
+      "step": 176750
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.2644745489622426e-05,
+      "loss": 0.3391,
+      "step": 176760
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.260234476032988e-05,
+      "loss": 0.4041,
+      "step": 176770
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.255994403103733e-05,
+      "loss": 0.3534,
+      "step": 176780
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.2517543301744785e-05,
+      "loss": 0.3483,
+      "step": 176790
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.247514257245225e-05,
+      "loss": 0.3358,
+      "step": 176800
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.2432741843159705e-05,
+      "loss": 0.3463,
+      "step": 176810
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.239034111386716e-05,
+      "loss": 0.3787,
+      "step": 176820
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.234794038457461e-05,
+      "loss": 0.3867,
+      "step": 176830
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.230553965528208e-05,
+      "loss": 0.385,
+      "step": 176840
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.226313892598953e-05,
+      "loss": 0.3834,
+      "step": 176850
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.2220738196696985e-05,
+      "loss": 0.3864,
+      "step": 176860
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.217833746740444e-05,
+      "loss": 0.4061,
+      "step": 176870
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.21359367381119e-05,
+      "loss": 0.347,
+      "step": 176880
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.209353600881935e-05,
+      "loss": 0.3571,
+      "step": 176890
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.205113527952681e-05,
+      "loss": 0.4263,
+      "step": 176900
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.2008734550234265e-05,
+      "loss": 0.4201,
+      "step": 176910
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.1966333820941725e-05,
+      "loss": 0.3858,
+      "step": 176920
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.192393309164918e-05,
+      "loss": 0.3914,
+      "step": 176930
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.188153236235663e-05,
+      "loss": 0.3384,
+      "step": 176940
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.1839131633064084e-05,
+      "loss": 0.3862,
+      "step": 176950
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.179673090377155e-05,
+      "loss": 0.3574,
+      "step": 176960
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.1754330174479004e-05,
+      "loss": 0.3561,
+      "step": 176970
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.171192944518646e-05,
+      "loss": 0.3892,
+      "step": 176980
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.166952871589391e-05,
+      "loss": 0.3934,
+      "step": 176990
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.162712798660138e-05,
+      "loss": 0.4013,
+      "step": 177000
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.158472725730883e-05,
+      "loss": 0.3351,
+      "step": 177010
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.1542326528016284e-05,
+      "loss": 0.3116,
+      "step": 177020
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.149992579872374e-05,
+      "loss": 0.3549,
+      "step": 177030
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.14575250694312e-05,
+      "loss": 0.3996,
+      "step": 177040
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.141512434013865e-05,
+      "loss": 0.3646,
+      "step": 177050
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.137272361084611e-05,
+      "loss": 0.3568,
+      "step": 177060
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.1330322881553564e-05,
+      "loss": 0.3754,
+      "step": 177070
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.1287922152261024e-05,
+      "loss": 0.3218,
+      "step": 177080
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.124552142296848e-05,
+      "loss": 0.3492,
+      "step": 177090
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.120312069367593e-05,
+      "loss": 0.4075,
+      "step": 177100
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.1160719964383384e-05,
+      "loss": 0.3514,
+      "step": 177110
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.111831923509085e-05,
+      "loss": 0.3478,
+      "step": 177120
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.1075918505798304e-05,
+      "loss": 0.3832,
+      "step": 177130
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.103351777650576e-05,
+      "loss": 0.4611,
+      "step": 177140
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.099111704721321e-05,
+      "loss": 0.3638,
+      "step": 177150
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.094871631792068e-05,
+      "loss": 0.3891,
+      "step": 177160
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.090631558862813e-05,
+      "loss": 0.3945,
+      "step": 177170
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.0863914859335583e-05,
+      "loss": 0.3863,
+      "step": 177180
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.082151413004304e-05,
+      "loss": 0.3595,
+      "step": 177190
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.07791134007505e-05,
+      "loss": 0.3559,
+      "step": 177200
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.073671267145795e-05,
+      "loss": 0.3648,
+      "step": 177210
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.069431194216541e-05,
+      "loss": 0.362,
+      "step": 177220
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.065191121287286e-05,
+      "loss": 0.3785,
+      "step": 177230
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.060951048358032e-05,
+      "loss": 0.372,
+      "step": 177240
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.0567109754287776e-05,
+      "loss": 0.4159,
+      "step": 177250
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.052470902499523e-05,
+      "loss": 0.3685,
+      "step": 177260
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.048230829570268e-05,
+      "loss": 0.3879,
+      "step": 177270
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.043990756641015e-05,
+      "loss": 0.3307,
+      "step": 177280
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.03975068371176e-05,
+      "loss": 0.3977,
+      "step": 177290
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.0355106107825056e-05,
+      "loss": 0.4097,
+      "step": 177300
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.031270537853251e-05,
+      "loss": 0.3374,
+      "step": 177310
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.0270304649239976e-05,
+      "loss": 0.4126,
+      "step": 177320
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.022790391994743e-05,
+      "loss": 0.4327,
+      "step": 177330
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.018550319065488e-05,
+      "loss": 0.3817,
+      "step": 177340
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.0143102461362336e-05,
+      "loss": 0.4152,
+      "step": 177350
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.0100701732069796e-05,
+      "loss": 0.3585,
+      "step": 177360
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.0058301002777256e-05,
+      "loss": 0.4104,
+      "step": 177370
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.001590027348471e-05,
+      "loss": 0.3844,
+      "step": 177380
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.997349954419216e-05,
+      "loss": 0.3476,
+      "step": 177390
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.9931098814899616e-05,
+      "loss": 0.3839,
+      "step": 177400
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.9888698085607076e-05,
+      "loss": 0.4208,
+      "step": 177410
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.984629735631453e-05,
+      "loss": 0.3254,
+      "step": 177420
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.980389662702199e-05,
+      "loss": 0.3538,
+      "step": 177430
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.976149589772944e-05,
+      "loss": 0.3768,
+      "step": 177440
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.97190951684369e-05,
+      "loss": 0.3438,
+      "step": 177450
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.9676694439144355e-05,
+      "loss": 0.3636,
+      "step": 177460
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.9634293709851815e-05,
+      "loss": 0.3079,
+      "step": 177470
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.959189298055927e-05,
+      "loss": 0.3713,
+      "step": 177480
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.954949225126672e-05,
+      "loss": 0.3978,
+      "step": 177490
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.950709152197418e-05,
+      "loss": 0.3946,
+      "step": 177500
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.9464690792681635e-05,
+      "loss": 0.4108,
+      "step": 177510
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.9422290063389095e-05,
+      "loss": 0.3699,
+      "step": 177520
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.937988933409655e-05,
+      "loss": 0.4176,
+      "step": 177530
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.933748860480401e-05,
+      "loss": 0.3334,
+      "step": 177540
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.929508787551146e-05,
+      "loss": 0.3863,
+      "step": 177550
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.9252687146218915e-05,
+      "loss": 0.436,
+      "step": 177560
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.921028641692637e-05,
+      "loss": 0.3383,
+      "step": 177570
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.916788568763383e-05,
+      "loss": 0.3618,
+      "step": 177580
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.912548495834128e-05,
+      "loss": 0.3666,
+      "step": 177590
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.908308422904874e-05,
+      "loss": 0.4362,
+      "step": 177600
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.9040683499756195e-05,
+      "loss": 0.39,
+      "step": 177610
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.8998282770463655e-05,
+      "loss": 0.3935,
+      "step": 177620
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.895588204117111e-05,
+      "loss": 0.4005,
+      "step": 177630
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.891348131187857e-05,
+      "loss": 0.4094,
+      "step": 177640
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.887108058258602e-05,
+      "loss": 0.3962,
+      "step": 177650
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.882867985329348e-05,
+      "loss": 0.3866,
+      "step": 177660
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.8786279124000934e-05,
+      "loss": 0.423,
+      "step": 177670
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.8743878394708394e-05,
+      "loss": 0.3751,
+      "step": 177680
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.870147766541585e-05,
+      "loss": 0.4473,
+      "step": 177690
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.865907693612331e-05,
+      "loss": 0.4267,
+      "step": 177700
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.861667620683076e-05,
+      "loss": 0.3404,
+      "step": 177710
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.8574275477538214e-05,
+      "loss": 0.3593,
+      "step": 177720
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.853187474824567e-05,
+      "loss": 0.3564,
+      "step": 177730
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.848947401895313e-05,
+      "loss": 0.4448,
+      "step": 177740
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.844707328966058e-05,
+      "loss": 0.3195,
+      "step": 177750
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.840467256036804e-05,
+      "loss": 0.3305,
+      "step": 177760
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.8362271831075494e-05,
+      "loss": 0.3378,
+      "step": 177770
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.8319871101782954e-05,
+      "loss": 0.3551,
+      "step": 177780
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.827747037249041e-05,
+      "loss": 0.4005,
+      "step": 177790
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.823506964319787e-05,
+      "loss": 0.3447,
+      "step": 177800
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.819266891390532e-05,
+      "loss": 0.3841,
+      "step": 177810
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.815026818461278e-05,
+      "loss": 0.3985,
+      "step": 177820
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.8107867455320234e-05,
+      "loss": 0.3118,
+      "step": 177830
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.8065466726027694e-05,
+      "loss": 0.3412,
+      "step": 177840
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.802306599673515e-05,
+      "loss": 0.4339,
+      "step": 177850
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.798066526744261e-05,
+      "loss": 0.371,
+      "step": 177860
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.793826453815006e-05,
+      "loss": 0.3907,
+      "step": 177870
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.789586380885751e-05,
+      "loss": 0.3659,
+      "step": 177880
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.7853463079564966e-05,
+      "loss": 0.354,
+      "step": 177890
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.7811062350272426e-05,
+      "loss": 0.343,
+      "step": 177900
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.776866162097988e-05,
+      "loss": 0.3732,
+      "step": 177910
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.772626089168734e-05,
+      "loss": 0.3661,
+      "step": 177920
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.768386016239479e-05,
+      "loss": 0.3951,
+      "step": 177930
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.764145943310225e-05,
+      "loss": 0.379,
+      "step": 177940
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7599058703809706e-05,
+      "loss": 0.3893,
+      "step": 177950
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7556657974517166e-05,
+      "loss": 0.3487,
+      "step": 177960
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.751425724522462e-05,
+      "loss": 0.3407,
+      "step": 177970
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.747185651593208e-05,
+      "loss": 0.3892,
+      "step": 177980
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.742945578663953e-05,
+      "loss": 0.3416,
+      "step": 177990
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.738705505734699e-05,
+      "loss": 0.3175,
+      "step": 178000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7344654328054446e-05,
+      "loss": 0.3724,
+      "step": 178010
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7302253598761906e-05,
+      "loss": 0.3358,
+      "step": 178020
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.725985286946936e-05,
+      "loss": 0.4543,
+      "step": 178030
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.721745214017681e-05,
+      "loss": 0.4339,
+      "step": 178040
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7175051410884266e-05,
+      "loss": 0.361,
+      "step": 178050
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7132650681591726e-05,
+      "loss": 0.402,
+      "step": 178060
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.709024995229918e-05,
+      "loss": 0.3791,
+      "step": 178070
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.704784922300664e-05,
+      "loss": 0.3572,
+      "step": 178080
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.700544849371409e-05,
+      "loss": 0.4009,
+      "step": 178090
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.696304776442155e-05,
+      "loss": 0.3716,
+      "step": 178100
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.6920647035129005e-05,
+      "loss": 0.3875,
+      "step": 178110
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.6878246305836465e-05,
+      "loss": 0.3876,
+      "step": 178120
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.683584557654392e-05,
+      "loss": 0.3451,
+      "step": 178130
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.679344484725138e-05,
+      "loss": 0.3944,
+      "step": 178140
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.675104411795883e-05,
+      "loss": 0.3669,
+      "step": 178150
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.670864338866629e-05,
+      "loss": 0.3545,
+      "step": 178160
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.6666242659373745e-05,
+      "loss": 0.4213,
+      "step": 178170
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.6623841930081205e-05,
+      "loss": 0.374,
+      "step": 178180
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.658144120078866e-05,
+      "loss": 0.4139,
+      "step": 178190
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.653904047149611e-05,
+      "loss": 0.3336,
+      "step": 178200
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.6496639742203565e-05,
+      "loss": 0.3387,
+      "step": 178210
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.6454239012911025e-05,
+      "loss": 0.3357,
+      "step": 178220
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.641183828361848e-05,
+      "loss": 0.3914,
+      "step": 178230
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.636943755432593e-05,
+      "loss": 0.3595,
+      "step": 178240
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.632703682503339e-05,
+      "loss": 0.4255,
+      "step": 178250
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.6284636095740845e-05,
+      "loss": 0.397,
+      "step": 178260
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.6242235366448305e-05,
+      "loss": 0.4552,
+      "step": 178270
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.619983463715576e-05,
+      "loss": 0.381,
+      "step": 178280
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.615743390786322e-05,
+      "loss": 0.4404,
+      "step": 178290
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.611503317857067e-05,
+      "loss": 0.4421,
+      "step": 178300
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.607263244927813e-05,
+      "loss": 0.3205,
+      "step": 178310
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.6030231719985584e-05,
+      "loss": 0.3904,
+      "step": 178320
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.5987830990693044e-05,
+      "loss": 0.3961,
+      "step": 178330
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.59454302614005e-05,
+      "loss": 0.3389,
+      "step": 178340
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.590302953210796e-05,
+      "loss": 0.3585,
+      "step": 178350
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.586062880281541e-05,
+      "loss": 0.3281,
+      "step": 178360
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.5818228073522864e-05,
+      "loss": 0.4122,
+      "step": 178370
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.5775827344230324e-05,
+      "loss": 0.3352,
+      "step": 178380
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.573342661493778e-05,
+      "loss": 0.4145,
+      "step": 178390
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.569102588564523e-05,
+      "loss": 0.4218,
+      "step": 178400
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.564862515635269e-05,
+      "loss": 0.3875,
+      "step": 178410
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.5606224427060144e-05,
+      "loss": 0.3892,
+      "step": 178420
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.5563823697767604e-05,
+      "loss": 0.4021,
+      "step": 178430
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.552142296847506e-05,
+      "loss": 0.3915,
+      "step": 178440
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.547902223918252e-05,
+      "loss": 0.4285,
+      "step": 178450
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.543662150988997e-05,
+      "loss": 0.3606,
+      "step": 178460
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.539422078059743e-05,
+      "loss": 0.3625,
+      "step": 178470
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.5351820051304884e-05,
+      "loss": 0.4176,
+      "step": 178480
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.5309419322012344e-05,
+      "loss": 0.4132,
+      "step": 178490
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.52670185927198e-05,
+      "loss": 0.4508,
+      "step": 178500
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.522461786342726e-05,
+      "loss": 0.3895,
+      "step": 178510
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.518221713413471e-05,
+      "loss": 0.3388,
+      "step": 178520
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.513981640484217e-05,
+      "loss": 0.3944,
+      "step": 178530
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.509741567554962e-05,
+      "loss": 0.3509,
+      "step": 178540
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.5055014946257077e-05,
+      "loss": 0.3646,
+      "step": 178550
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.501261421696453e-05,
+      "loss": 0.3751,
+      "step": 178560
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.497021348767199e-05,
+      "loss": 0.3736,
+      "step": 178570
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.492781275837944e-05,
+      "loss": 0.3627,
+      "step": 178580
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.48854120290869e-05,
+      "loss": 0.4134,
+      "step": 178590
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.4843011299794356e-05,
+      "loss": 0.3227,
+      "step": 178600
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.4800610570501816e-05,
+      "loss": 0.3906,
+      "step": 178610
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.475820984120927e-05,
+      "loss": 0.3399,
+      "step": 178620
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.471580911191673e-05,
+      "loss": 0.425,
+      "step": 178630
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.467340838262418e-05,
+      "loss": 0.3898,
+      "step": 178640
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.463100765333164e-05,
+      "loss": 0.3421,
+      "step": 178650
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.4588606924039096e-05,
+      "loss": 0.3503,
+      "step": 178660
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.4546206194746556e-05,
+      "loss": 0.3146,
+      "step": 178670
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.450380546545401e-05,
+      "loss": 0.3686,
+      "step": 178680
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.446140473616147e-05,
+      "loss": 0.3639,
+      "step": 178690
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.441900400686892e-05,
+      "loss": 0.3721,
+      "step": 178700
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.4376603277576376e-05,
+      "loss": 0.39,
+      "step": 178710
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.433420254828383e-05,
+      "loss": 0.4112,
+      "step": 178720
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.429180181899129e-05,
+      "loss": 0.3495,
+      "step": 178730
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.424940108969874e-05,
+      "loss": 0.4141,
+      "step": 178740
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.42070003604062e-05,
+      "loss": 0.3076,
+      "step": 178750
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.4164599631113656e-05,
+      "loss": 0.3934,
+      "step": 178760
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.4122198901821116e-05,
+      "loss": 0.4204,
+      "step": 178770
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.407979817252857e-05,
+      "loss": 0.3942,
+      "step": 178780
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.403739744323603e-05,
+      "loss": 0.4555,
+      "step": 178790
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.399499671394348e-05,
+      "loss": 0.2976,
+      "step": 178800
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.395259598465094e-05,
+      "loss": 0.4542,
+      "step": 178810
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.3910195255358395e-05,
+      "loss": 0.3407,
+      "step": 178820
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.3867794526065855e-05,
+      "loss": 0.3681,
+      "step": 178830
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.382539379677331e-05,
+      "loss": 0.3662,
+      "step": 178840
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.378299306748077e-05,
+      "loss": 0.4305,
+      "step": 178850
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.374059233818822e-05,
+      "loss": 0.4427,
+      "step": 178860
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.3698191608895675e-05,
+      "loss": 0.3627,
+      "step": 178870
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.365579087960313e-05,
+      "loss": 0.3664,
+      "step": 178880
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.361339015031059e-05,
+      "loss": 0.3466,
+      "step": 178890
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.357098942101804e-05,
+      "loss": 0.376,
+      "step": 178900
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.3528588691725495e-05,
+      "loss": 0.3844,
+      "step": 178910
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.3486187962432955e-05,
+      "loss": 0.3174,
+      "step": 178920
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.344378723314041e-05,
+      "loss": 0.357,
+      "step": 178930
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.340138650384787e-05,
+      "loss": 0.3408,
+      "step": 178940
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.335898577455532e-05,
+      "loss": 0.3972,
+      "step": 178950
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.331658504526278e-05,
+      "loss": 0.3124,
+      "step": 178960
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.3274184315970235e-05,
+      "loss": 0.3292,
+      "step": 178970
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.3231783586677695e-05,
+      "loss": 0.3594,
+      "step": 178980
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.318938285738515e-05,
+      "loss": 0.3978,
+      "step": 178990
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.314698212809261e-05,
+      "loss": 0.3818,
+      "step": 179000
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.310458139880006e-05,
+      "loss": 0.3983,
+      "step": 179010
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.306218066950752e-05,
+      "loss": 0.3273,
+      "step": 179020
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.3019779940214974e-05,
+      "loss": 0.3672,
+      "step": 179030
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.297737921092243e-05,
+      "loss": 0.4191,
+      "step": 179040
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.293497848162988e-05,
+      "loss": 0.4039,
+      "step": 179050
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.289257775233734e-05,
+      "loss": 0.4022,
+      "step": 179060
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.2850177023044794e-05,
+      "loss": 0.3341,
+      "step": 179070
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.2807776293752254e-05,
+      "loss": 0.4387,
+      "step": 179080
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.276537556445971e-05,
+      "loss": 0.3251,
+      "step": 179090
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.272297483516717e-05,
+      "loss": 0.3433,
+      "step": 179100
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.268057410587462e-05,
+      "loss": 0.3918,
+      "step": 179110
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.263817337658208e-05,
+      "loss": 0.3941,
+      "step": 179120
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.2595772647289534e-05,
+      "loss": 0.4052,
+      "step": 179130
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.2553371917996994e-05,
+      "loss": 0.3662,
+      "step": 179140
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.251097118870445e-05,
+      "loss": 0.343,
+      "step": 179150
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.246857045941191e-05,
+      "loss": 0.3992,
+      "step": 179160
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.242616973011936e-05,
+      "loss": 0.3378,
+      "step": 179170
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.238376900082682e-05,
+      "loss": 0.4396,
+      "step": 179180
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.2341368271534273e-05,
+      "loss": 0.3548,
+      "step": 179190
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.229896754224173e-05,
+      "loss": 0.4095,
+      "step": 179200
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.225656681294918e-05,
+      "loss": 0.3603,
+      "step": 179210
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.221416608365664e-05,
+      "loss": 0.3897,
+      "step": 179220
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.217176535436409e-05,
+      "loss": 0.3906,
+      "step": 179230
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.212936462507155e-05,
+      "loss": 0.4147,
+      "step": 179240
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.2086963895779006e-05,
+      "loss": 0.3655,
+      "step": 179250
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.2044563166486466e-05,
+      "loss": 0.3884,
+      "step": 179260
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.200216243719392e-05,
+      "loss": 0.4055,
+      "step": 179270
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.195976170790138e-05,
+      "loss": 0.4126,
+      "step": 179280
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.191736097860883e-05,
+      "loss": 0.3597,
+      "step": 179290
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.187496024931629e-05,
+      "loss": 0.3108,
+      "step": 179300
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.1832559520023746e-05,
+      "loss": 0.38,
+      "step": 179310
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.1790158790731206e-05,
+      "loss": 0.3434,
+      "step": 179320
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.174775806143866e-05,
+      "loss": 0.3386,
+      "step": 179330
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.170535733214612e-05,
+      "loss": 0.3938,
+      "step": 179340
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.166295660285357e-05,
+      "loss": 0.3388,
+      "step": 179350
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.1620555873561026e-05,
+      "loss": 0.4119,
+      "step": 179360
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.157815514426848e-05,
+      "loss": 0.2934,
+      "step": 179370
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.153575441497594e-05,
+      "loss": 0.2932,
+      "step": 179380
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.149335368568339e-05,
+      "loss": 0.3877,
+      "step": 179390
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.145095295639085e-05,
+      "loss": 0.3148,
+      "step": 179400
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.1408552227098306e-05,
+      "loss": 0.3843,
+      "step": 179410
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.1366151497805766e-05,
+      "loss": 0.387,
+      "step": 179420
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.132375076851322e-05,
+      "loss": 0.3845,
+      "step": 179430
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.128135003922068e-05,
+      "loss": 0.3918,
+      "step": 179440
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.123894930992813e-05,
+      "loss": 0.3622,
+      "step": 179450
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.119654858063559e-05,
+      "loss": 0.3652,
+      "step": 179460
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.1154147851343045e-05,
+      "loss": 0.3982,
+      "step": 179470
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.1111747122050505e-05,
+      "loss": 0.3583,
+      "step": 179480
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.106934639275796e-05,
+      "loss": 0.3748,
+      "step": 179490
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.102694566346542e-05,
+      "loss": 0.3401,
+      "step": 179500
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.098454493417287e-05,
+      "loss": 0.3526,
+      "step": 179510
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.0942144204880325e-05,
+      "loss": 0.4288,
+      "step": 179520
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.089974347558778e-05,
+      "loss": 0.3416,
+      "step": 179530
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.085734274629524e-05,
+      "loss": 0.3578,
+      "step": 179540
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.081494201700269e-05,
+      "loss": 0.3617,
+      "step": 179550
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.077254128771015e-05,
+      "loss": 0.3703,
+      "step": 179560
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.0730140558417605e-05,
+      "loss": 0.3837,
+      "step": 179570
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.0687739829125065e-05,
+      "loss": 0.3072,
+      "step": 179580
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.064533909983252e-05,
+      "loss": 0.3505,
+      "step": 179590
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.060293837053998e-05,
+      "loss": 0.408,
+      "step": 179600
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.056053764124743e-05,
+      "loss": 0.3599,
+      "step": 179610
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.0518136911954885e-05,
+      "loss": 0.3681,
+      "step": 179620
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.0475736182662345e-05,
+      "loss": 0.3881,
+      "step": 179630
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.04333354533698e-05,
+      "loss": 0.4266,
+      "step": 179640
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.039093472407726e-05,
+      "loss": 0.342,
+      "step": 179650
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.034853399478471e-05,
+      "loss": 0.3934,
+      "step": 179660
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.030613326549217e-05,
+      "loss": 0.4083,
+      "step": 179670
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.0263732536199624e-05,
+      "loss": 0.3617,
+      "step": 179680
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.0221331806907084e-05,
+      "loss": 0.3543,
+      "step": 179690
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.017893107761454e-05,
+      "loss": 0.4502,
+      "step": 179700
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.013653034832199e-05,
+      "loss": 0.35,
+      "step": 179710
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.0094129619029444e-05,
+      "loss": 0.3396,
+      "step": 179720
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.0051728889736904e-05,
+      "loss": 0.351,
+      "step": 179730
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.000932816044436e-05,
+      "loss": 0.3412,
+      "step": 179740
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.996692743115182e-05,
+      "loss": 0.3846,
+      "step": 179750
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.992452670185927e-05,
+      "loss": 0.4162,
+      "step": 179760
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.988212597256673e-05,
+      "loss": 0.377,
+      "step": 179770
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.9839725243274184e-05,
+      "loss": 0.3058,
+      "step": 179780
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.9797324513981644e-05,
+      "loss": 0.3788,
+      "step": 179790
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.97549237846891e-05,
+      "loss": 0.3497,
+      "step": 179800
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.971252305539656e-05,
+      "loss": 0.4421,
+      "step": 179810
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.967012232610401e-05,
+      "loss": 0.4008,
+      "step": 179820
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.962772159681147e-05,
+      "loss": 0.3398,
+      "step": 179830
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.9585320867518924e-05,
+      "loss": 0.3406,
+      "step": 179840
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.9542920138226384e-05,
+      "loss": 0.4135,
+      "step": 179850
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.950051940893384e-05,
+      "loss": 0.3637,
+      "step": 179860
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.945811867964129e-05,
+      "loss": 0.3861,
+      "step": 179870
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.941571795034874e-05,
+      "loss": 0.3647,
+      "step": 179880
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.93733172210562e-05,
+      "loss": 0.4256,
+      "step": 179890
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.9330916491763657e-05,
+      "loss": 0.355,
+      "step": 179900
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.9288515762471117e-05,
+      "loss": 0.38,
+      "step": 179910
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.924611503317857e-05,
+      "loss": 0.4408,
+      "step": 179920
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.920371430388603e-05,
+      "loss": 0.4167,
+      "step": 179930
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.916131357459348e-05,
+      "loss": 0.2952,
+      "step": 179940
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.911891284530094e-05,
+      "loss": 0.3536,
+      "step": 179950
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.9076512116008396e-05,
+      "loss": 0.3394,
+      "step": 179960
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.9034111386715856e-05,
+      "loss": 0.4128,
+      "step": 179970
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.899171065742331e-05,
+      "loss": 0.3366,
+      "step": 179980
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.894930992813077e-05,
+      "loss": 0.4002,
+      "step": 179990
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.890690919883822e-05,
+      "loss": 0.3933,
+      "step": 180000
+    },
+    {
+      "epoch": 7.61,
+      "eval_loss": 0.6400695443153381,
+      "eval_runtime": 337.7542,
+      "eval_samples_per_second": 15.559,
+      "eval_steps_per_second": 3.89,
+      "step": 180000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.886450846954568e-05,
+      "loss": 0.4208,
+      "step": 180010
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.8822107740253136e-05,
+      "loss": 0.3394,
+      "step": 180020
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.877970701096059e-05,
+      "loss": 0.3709,
+      "step": 180030
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.873730628166804e-05,
+      "loss": 0.4343,
+      "step": 180040
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.86949055523755e-05,
+      "loss": 0.3522,
+      "step": 180050
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.8652504823082956e-05,
+      "loss": 0.3617,
+      "step": 180060
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.8610104093790416e-05,
+      "loss": 0.365,
+      "step": 180070
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.856770336449787e-05,
+      "loss": 0.3577,
+      "step": 180080
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.852530263520533e-05,
+      "loss": 0.4694,
+      "step": 180090
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.848290190591278e-05,
+      "loss": 0.3679,
+      "step": 180100
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.844050117662024e-05,
+      "loss": 0.3613,
+      "step": 180110
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.8398100447327696e-05,
+      "loss": 0.4121,
+      "step": 180120
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.8355699718035156e-05,
+      "loss": 0.3746,
+      "step": 180130
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.831329898874261e-05,
+      "loss": 0.4022,
+      "step": 180140
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.827089825945007e-05,
+      "loss": 0.386,
+      "step": 180150
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.822849753015752e-05,
+      "loss": 0.4199,
+      "step": 180160
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.818609680086498e-05,
+      "loss": 0.4033,
+      "step": 180170
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.8143696071572435e-05,
+      "loss": 0.3379,
+      "step": 180180
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.810129534227989e-05,
+      "loss": 0.4852,
+      "step": 180190
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.805889461298734e-05,
+      "loss": 0.3845,
+      "step": 180200
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.80164938836948e-05,
+      "loss": 0.3895,
+      "step": 180210
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.7974093154402255e-05,
+      "loss": 0.3831,
+      "step": 180220
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.7931692425109715e-05,
+      "loss": 0.3262,
+      "step": 180230
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.788929169581717e-05,
+      "loss": 0.3601,
+      "step": 180240
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.784689096652463e-05,
+      "loss": 0.3536,
+      "step": 180250
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.780449023723208e-05,
+      "loss": 0.2843,
+      "step": 180260
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.776208950793954e-05,
+      "loss": 0.4315,
+      "step": 180270
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.7719688778646995e-05,
+      "loss": 0.3922,
+      "step": 180280
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.7677288049354455e-05,
+      "loss": 0.3889,
+      "step": 180290
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.763488732006191e-05,
+      "loss": 0.3186,
+      "step": 180300
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.759248659076937e-05,
+      "loss": 0.3759,
+      "step": 180310
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.755008586147682e-05,
+      "loss": 0.3227,
+      "step": 180320
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7507685132184274e-05,
+      "loss": 0.4027,
+      "step": 180330
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7465284402891734e-05,
+      "loss": 0.4531,
+      "step": 180340
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.742288367359919e-05,
+      "loss": 0.3726,
+      "step": 180350
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.738048294430664e-05,
+      "loss": 0.3601,
+      "step": 180360
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7338082215014094e-05,
+      "loss": 0.4087,
+      "step": 180370
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7295681485721554e-05,
+      "loss": 0.3821,
+      "step": 180380
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.725328075642901e-05,
+      "loss": 0.397,
+      "step": 180390
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.721088002713647e-05,
+      "loss": 0.3892,
+      "step": 180400
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.716847929784392e-05,
+      "loss": 0.3634,
+      "step": 180410
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.712607856855138e-05,
+      "loss": 0.3577,
+      "step": 180420
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7083677839258834e-05,
+      "loss": 0.3735,
+      "step": 180430
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7041277109966294e-05,
+      "loss": 0.4313,
+      "step": 180440
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.699887638067375e-05,
+      "loss": 0.3971,
+      "step": 180450
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.695647565138121e-05,
+      "loss": 0.3582,
+      "step": 180460
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.691407492208866e-05,
+      "loss": 0.4024,
+      "step": 180470
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.687167419279612e-05,
+      "loss": 0.3732,
+      "step": 180480
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.6829273463503574e-05,
+      "loss": 0.3377,
+      "step": 180490
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.6786872734211034e-05,
+      "loss": 0.4495,
+      "step": 180500
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.674447200491849e-05,
+      "loss": 0.3528,
+      "step": 180510
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.670207127562594e-05,
+      "loss": 0.3054,
+      "step": 180520
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.6659670546333393e-05,
+      "loss": 0.3942,
+      "step": 180530
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.6617269817040853e-05,
+      "loss": 0.3301,
+      "step": 180540
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.657486908774831e-05,
+      "loss": 0.3718,
+      "step": 180550
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.653246835845577e-05,
+      "loss": 0.365,
+      "step": 180560
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.649006762916322e-05,
+      "loss": 0.3947,
+      "step": 180570
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.644766689987068e-05,
+      "loss": 0.4585,
+      "step": 180580
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.640526617057813e-05,
+      "loss": 0.399,
+      "step": 180590
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.636286544128559e-05,
+      "loss": 0.3233,
+      "step": 180600
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.6320464711993046e-05,
+      "loss": 0.3761,
+      "step": 180610
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.6278063982700506e-05,
+      "loss": 0.3624,
+      "step": 180620
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.623566325340796e-05,
+      "loss": 0.3181,
+      "step": 180630
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.619326252411542e-05,
+      "loss": 0.4137,
+      "step": 180640
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.615086179482287e-05,
+      "loss": 0.3353,
+      "step": 180650
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.610846106553033e-05,
+      "loss": 0.4408,
+      "step": 180660
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.6066060336237786e-05,
+      "loss": 0.4455,
+      "step": 180670
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.602365960694524e-05,
+      "loss": 0.385,
+      "step": 180680
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.598125887765269e-05,
+      "loss": 0.4412,
+      "step": 180690
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.593885814836015e-05,
+      "loss": 0.3809,
+      "step": 180700
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.5896457419067606e-05,
+      "loss": 0.3556,
+      "step": 180710
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.5854056689775066e-05,
+      "loss": 0.3696,
+      "step": 180720
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.581165596048252e-05,
+      "loss": 0.3562,
+      "step": 180730
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.576925523118998e-05,
+      "loss": 0.4093,
+      "step": 180740
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.572685450189743e-05,
+      "loss": 0.3859,
+      "step": 180750
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.568445377260489e-05,
+      "loss": 0.4329,
+      "step": 180760
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.5642053043312346e-05,
+      "loss": 0.3776,
+      "step": 180770
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.5599652314019806e-05,
+      "loss": 0.4454,
+      "step": 180780
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.555725158472726e-05,
+      "loss": 0.3829,
+      "step": 180790
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.551485085543472e-05,
+      "loss": 0.3394,
+      "step": 180800
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.547245012614217e-05,
+      "loss": 0.3746,
+      "step": 180810
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.543004939684963e-05,
+      "loss": 0.3673,
+      "step": 180820
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.5387648667557085e-05,
+      "loss": 0.3854,
+      "step": 180830
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.534524793826454e-05,
+      "loss": 0.3929,
+      "step": 180840
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.5302847208972e-05,
+      "loss": 0.4062,
+      "step": 180850
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.526044647967945e-05,
+      "loss": 0.3663,
+      "step": 180860
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.5218045750386905e-05,
+      "loss": 0.3839,
+      "step": 180870
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.5175645021094365e-05,
+      "loss": 0.3678,
+      "step": 180880
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.513324429180182e-05,
+      "loss": 0.3919,
+      "step": 180890
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.509084356250928e-05,
+      "loss": 0.3825,
+      "step": 180900
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.504844283321673e-05,
+      "loss": 0.3428,
+      "step": 180910
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.500604210392419e-05,
+      "loss": 0.396,
+      "step": 180920
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.4963641374631645e-05,
+      "loss": 0.3458,
+      "step": 180930
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.4921240645339105e-05,
+      "loss": 0.367,
+      "step": 180940
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.487883991604656e-05,
+      "loss": 0.3458,
+      "step": 180950
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.483643918675402e-05,
+      "loss": 0.3811,
+      "step": 180960
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.479403845746147e-05,
+      "loss": 0.3465,
+      "step": 180970
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.475163772816893e-05,
+      "loss": 0.3307,
+      "step": 180980
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.4709236998876385e-05,
+      "loss": 0.3629,
+      "step": 180990
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.466683626958384e-05,
+      "loss": 0.3769,
+      "step": 181000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.46244355402913e-05,
+      "loss": 0.3571,
+      "step": 181010
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.458203481099875e-05,
+      "loss": 0.3609,
+      "step": 181020
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4539634081706204e-05,
+      "loss": 0.4149,
+      "step": 181030
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4497233352413664e-05,
+      "loss": 0.4836,
+      "step": 181040
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.445483262312112e-05,
+      "loss": 0.3969,
+      "step": 181050
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.441243189382857e-05,
+      "loss": 0.2939,
+      "step": 181060
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.437003116453603e-05,
+      "loss": 0.3055,
+      "step": 181070
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4327630435243484e-05,
+      "loss": 0.4025,
+      "step": 181080
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4285229705950944e-05,
+      "loss": 0.3978,
+      "step": 181090
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.42428289766584e-05,
+      "loss": 0.3,
+      "step": 181100
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.420042824736586e-05,
+      "loss": 0.3492,
+      "step": 181110
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.415802751807331e-05,
+      "loss": 0.3746,
+      "step": 181120
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.411562678878077e-05,
+      "loss": 0.3734,
+      "step": 181130
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4073226059488224e-05,
+      "loss": 0.4154,
+      "step": 181140
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4030825330195684e-05,
+      "loss": 0.3692,
+      "step": 181150
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.398842460090314e-05,
+      "loss": 0.3865,
+      "step": 181160
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.39460238716106e-05,
+      "loss": 0.3091,
+      "step": 181170
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.390362314231805e-05,
+      "loss": 0.3165,
+      "step": 181180
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.3861222413025504e-05,
+      "loss": 0.3354,
+      "step": 181190
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.381882168373296e-05,
+      "loss": 0.3883,
+      "step": 181200
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.377642095444042e-05,
+      "loss": 0.4148,
+      "step": 181210
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.373402022514787e-05,
+      "loss": 0.4053,
+      "step": 181220
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.369161949585533e-05,
+      "loss": 0.3744,
+      "step": 181230
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.364921876656278e-05,
+      "loss": 0.3724,
+      "step": 181240
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.360681803727024e-05,
+      "loss": 0.4042,
+      "step": 181250
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.3564417307977696e-05,
+      "loss": 0.3867,
+      "step": 181260
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.3522016578685157e-05,
+      "loss": 0.3812,
+      "step": 181270
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.347961584939261e-05,
+      "loss": 0.3069,
+      "step": 181280
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.343721512010007e-05,
+      "loss": 0.3752,
+      "step": 181290
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.339481439080752e-05,
+      "loss": 0.3266,
+      "step": 181300
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.335241366151498e-05,
+      "loss": 0.2872,
+      "step": 181310
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.3310012932222436e-05,
+      "loss": 0.3462,
+      "step": 181320
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.3267612202929896e-05,
+      "loss": 0.3487,
+      "step": 181330
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.322521147363735e-05,
+      "loss": 0.3298,
+      "step": 181340
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.31828107443448e-05,
+      "loss": 0.346,
+      "step": 181350
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.3140410015052256e-05,
+      "loss": 0.4122,
+      "step": 181360
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.3098009285759716e-05,
+      "loss": 0.3593,
+      "step": 181370
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.305560855646717e-05,
+      "loss": 0.3296,
+      "step": 181380
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.301320782717463e-05,
+      "loss": 0.395,
+      "step": 181390
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.297080709788208e-05,
+      "loss": 0.4058,
+      "step": 181400
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.292840636858954e-05,
+      "loss": 0.4078,
+      "step": 181410
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.2886005639296996e-05,
+      "loss": 0.4161,
+      "step": 181420
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.2843604910004456e-05,
+      "loss": 0.3714,
+      "step": 181430
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.280120418071191e-05,
+      "loss": 0.3779,
+      "step": 181440
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.275880345141937e-05,
+      "loss": 0.3457,
+      "step": 181450
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.271640272212682e-05,
+      "loss": 0.3727,
+      "step": 181460
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.267400199283428e-05,
+      "loss": 0.311,
+      "step": 181470
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.2631601263541735e-05,
+      "loss": 0.3624,
+      "step": 181480
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.2589200534249195e-05,
+      "loss": 0.3822,
+      "step": 181490
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.254679980495665e-05,
+      "loss": 0.3704,
+      "step": 181500
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.25043990756641e-05,
+      "loss": 0.3609,
+      "step": 181510
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.2461998346371555e-05,
+      "loss": 0.4514,
+      "step": 181520
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.2419597617079015e-05,
+      "loss": 0.3456,
+      "step": 181530
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.237719688778647e-05,
+      "loss": 0.3665,
+      "step": 181540
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.233479615849393e-05,
+      "loss": 0.4226,
+      "step": 181550
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.229239542920138e-05,
+      "loss": 0.3868,
+      "step": 181560
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.224999469990884e-05,
+      "loss": 0.4243,
+      "step": 181570
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.2207593970616295e-05,
+      "loss": 0.3688,
+      "step": 181580
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.2165193241323755e-05,
+      "loss": 0.4084,
+      "step": 181590
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.212279251203121e-05,
+      "loss": 0.4025,
+      "step": 181600
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.208039178273867e-05,
+      "loss": 0.2797,
+      "step": 181610
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.203799105344612e-05,
+      "loss": 0.3384,
+      "step": 181620
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.199559032415358e-05,
+      "loss": 0.3842,
+      "step": 181630
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.1953189594861035e-05,
+      "loss": 0.3377,
+      "step": 181640
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.1910788865568495e-05,
+      "loss": 0.4121,
+      "step": 181650
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.186838813627595e-05,
+      "loss": 0.4085,
+      "step": 181660
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.18259874069834e-05,
+      "loss": 0.3427,
+      "step": 181670
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.1783586677690854e-05,
+      "loss": 0.3017,
+      "step": 181680
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.1741185948398314e-05,
+      "loss": 0.307,
+      "step": 181690
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.169878521910577e-05,
+      "loss": 0.4326,
+      "step": 181700
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.165638448981323e-05,
+      "loss": 0.3708,
+      "step": 181710
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.161398376052068e-05,
+      "loss": 0.4085,
+      "step": 181720
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.157158303122814e-05,
+      "loss": 0.3326,
+      "step": 181730
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.1529182301935594e-05,
+      "loss": 0.304,
+      "step": 181740
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.1486781572643054e-05,
+      "loss": 0.3431,
+      "step": 181750
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.144438084335051e-05,
+      "loss": 0.3844,
+      "step": 181760
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.140198011405796e-05,
+      "loss": 0.3735,
+      "step": 181770
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.135957938476542e-05,
+      "loss": 0.3927,
+      "step": 181780
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.1317178655472874e-05,
+      "loss": 0.3849,
+      "step": 181790
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.1274777926180334e-05,
+      "loss": 0.4756,
+      "step": 181800
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.123237719688779e-05,
+      "loss": 0.3566,
+      "step": 181810
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.118997646759525e-05,
+      "loss": 0.3783,
+      "step": 181820
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.11475757383027e-05,
+      "loss": 0.3641,
+      "step": 181830
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.1105175009010154e-05,
+      "loss": 0.4167,
+      "step": 181840
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.106277427971761e-05,
+      "loss": 0.4446,
+      "step": 181850
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.102037355042507e-05,
+      "loss": 0.3383,
+      "step": 181860
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.097797282113252e-05,
+      "loss": 0.4025,
+      "step": 181870
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.093557209183998e-05,
+      "loss": 0.3333,
+      "step": 181880
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.089317136254743e-05,
+      "loss": 0.3335,
+      "step": 181890
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.085077063325489e-05,
+      "loss": 0.327,
+      "step": 181900
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.0808369903962347e-05,
+      "loss": 0.3798,
+      "step": 181910
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.076596917466981e-05,
+      "loss": 0.3591,
+      "step": 181920
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.072356844537726e-05,
+      "loss": 0.3822,
+      "step": 181930
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.068116771608472e-05,
+      "loss": 0.3592,
+      "step": 181940
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.063876698679217e-05,
+      "loss": 0.3698,
+      "step": 181950
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.059636625749963e-05,
+      "loss": 0.4058,
+      "step": 181960
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0553965528207086e-05,
+      "loss": 0.3501,
+      "step": 181970
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0511564798914543e-05,
+      "loss": 0.3263,
+      "step": 181980
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0469164069621996e-05,
+      "loss": 0.3873,
+      "step": 181990
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0426763340329456e-05,
+      "loss": 0.4083,
+      "step": 182000
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.038436261103691e-05,
+      "loss": 0.3042,
+      "step": 182010
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.034196188174437e-05,
+      "loss": 0.3949,
+      "step": 182020
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0299561152451823e-05,
+      "loss": 0.3562,
+      "step": 182030
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.025716042315928e-05,
+      "loss": 0.3782,
+      "step": 182040
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0214759693866736e-05,
+      "loss": 0.4045,
+      "step": 182050
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0172358964574193e-05,
+      "loss": 0.3718,
+      "step": 182060
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0129958235281646e-05,
+      "loss": 0.3426,
+      "step": 182070
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0087557505989106e-05,
+      "loss": 0.3729,
+      "step": 182080
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.004515677669656e-05,
+      "loss": 0.3507,
+      "step": 182090
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.000275604740402e-05,
+      "loss": 0.3365,
+      "step": 182100
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.9960355318111472e-05,
+      "loss": 0.3382,
+      "step": 182110
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.991795458881893e-05,
+      "loss": 0.3762,
+      "step": 182120
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.9875553859526386e-05,
+      "loss": 0.364,
+      "step": 182130
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.9833153130233842e-05,
+      "loss": 0.4364,
+      "step": 182140
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.9790752400941295e-05,
+      "loss": 0.424,
+      "step": 182150
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.9748351671648755e-05,
+      "loss": 0.4052,
+      "step": 182160
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.970595094235621e-05,
+      "loss": 0.4235,
+      "step": 182170
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.966355021306367e-05,
+      "loss": 0.3885,
+      "step": 182180
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.9621149483771122e-05,
+      "loss": 0.3433,
+      "step": 182190
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.957874875447858e-05,
+      "loss": 0.3567,
+      "step": 182200
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9536348025186035e-05,
+      "loss": 0.3874,
+      "step": 182210
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9493947295893492e-05,
+      "loss": 0.3986,
+      "step": 182220
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9451546566600945e-05,
+      "loss": 0.3989,
+      "step": 182230
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9409145837308405e-05,
+      "loss": 0.3751,
+      "step": 182240
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9366745108015858e-05,
+      "loss": 0.3791,
+      "step": 182250
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9324344378723318e-05,
+      "loss": 0.3845,
+      "step": 182260
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.928194364943077e-05,
+      "loss": 0.4138,
+      "step": 182270
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9239542920138228e-05,
+      "loss": 0.4169,
+      "step": 182280
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9197142190845685e-05,
+      "loss": 0.3585,
+      "step": 182290
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.915474146155314e-05,
+      "loss": 0.3439,
+      "step": 182300
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9112340732260595e-05,
+      "loss": 0.385,
+      "step": 182310
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9069940002968055e-05,
+      "loss": 0.4092,
+      "step": 182320
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9027539273675508e-05,
+      "loss": 0.328,
+      "step": 182330
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.8985138544382968e-05,
+      "loss": 0.3645,
+      "step": 182340
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.894273781509042e-05,
+      "loss": 0.3626,
+      "step": 182350
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.890033708579788e-05,
+      "loss": 0.4505,
+      "step": 182360
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.8857936356505334e-05,
+      "loss": 0.3499,
+      "step": 182370
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.881553562721279e-05,
+      "loss": 0.4688,
+      "step": 182380
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.8773134897920244e-05,
+      "loss": 0.3437,
+      "step": 182390
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.8730734168627704e-05,
+      "loss": 0.4439,
+      "step": 182400
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.8688333439335157e-05,
+      "loss": 0.3735,
+      "step": 182410
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.8645932710042618e-05,
+      "loss": 0.3934,
+      "step": 182420
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.860353198075007e-05,
+      "loss": 0.364,
+      "step": 182430
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.856113125145753e-05,
+      "loss": 0.2971,
+      "step": 182440
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8518730522164984e-05,
+      "loss": 0.4174,
+      "step": 182450
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.847632979287244e-05,
+      "loss": 0.3494,
+      "step": 182460
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8433929063579894e-05,
+      "loss": 0.336,
+      "step": 182470
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8391528334287347e-05,
+      "loss": 0.3409,
+      "step": 182480
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8349127604994807e-05,
+      "loss": 0.3421,
+      "step": 182490
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.830672687570226e-05,
+      "loss": 0.3638,
+      "step": 182500
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.826432614640972e-05,
+      "loss": 0.4128,
+      "step": 182510
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8221925417117174e-05,
+      "loss": 0.2854,
+      "step": 182520
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8179524687824634e-05,
+      "loss": 0.3565,
+      "step": 182530
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8137123958532087e-05,
+      "loss": 0.3762,
+      "step": 182540
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8094723229239543e-05,
+      "loss": 0.3557,
+      "step": 182550
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8052322499946997e-05,
+      "loss": 0.3531,
+      "step": 182560
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8009921770654457e-05,
+      "loss": 0.3449,
+      "step": 182570
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.796752104136191e-05,
+      "loss": 0.4061,
+      "step": 182580
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.792512031206937e-05,
+      "loss": 0.4665,
+      "step": 182590
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7882719582776823e-05,
+      "loss": 0.3377,
+      "step": 182600
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7840318853484283e-05,
+      "loss": 0.4084,
+      "step": 182610
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7797918124191736e-05,
+      "loss": 0.4587,
+      "step": 182620
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7755517394899193e-05,
+      "loss": 0.3841,
+      "step": 182630
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7713116665606646e-05,
+      "loss": 0.2746,
+      "step": 182640
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7670715936314106e-05,
+      "loss": 0.3787,
+      "step": 182650
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.762831520702156e-05,
+      "loss": 0.3524,
+      "step": 182660
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.758591447772902e-05,
+      "loss": 0.3611,
+      "step": 182670
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7543513748436473e-05,
+      "loss": 0.3665,
+      "step": 182680
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7501113019143933e-05,
+      "loss": 0.3524,
+      "step": 182690
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7458712289851386e-05,
+      "loss": 0.3811,
+      "step": 182700
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7416311560558843e-05,
+      "loss": 0.4125,
+      "step": 182710
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7373910831266296e-05,
+      "loss": 0.4118,
+      "step": 182720
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7331510101973756e-05,
+      "loss": 0.3933,
+      "step": 182730
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.728910937268121e-05,
+      "loss": 0.3537,
+      "step": 182740
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.724670864338867e-05,
+      "loss": 0.3877,
+      "step": 182750
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7204307914096122e-05,
+      "loss": 0.4283,
+      "step": 182760
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7161907184803582e-05,
+      "loss": 0.339,
+      "step": 182770
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7119506455511036e-05,
+      "loss": 0.413,
+      "step": 182780
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7077105726218492e-05,
+      "loss": 0.3788,
+      "step": 182790
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7034704996925946e-05,
+      "loss": 0.3625,
+      "step": 182800
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.6992304267633406e-05,
+      "loss": 0.3874,
+      "step": 182810
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.694990353834086e-05,
+      "loss": 0.3317,
+      "step": 182820
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.690750280904832e-05,
+      "loss": 0.3454,
+      "step": 182830
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.6865102079755772e-05,
+      "loss": 0.3848,
+      "step": 182840
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.6822701350463232e-05,
+      "loss": 0.3201,
+      "step": 182850
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.6780300621170685e-05,
+      "loss": 0.4144,
+      "step": 182860
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.6737899891878142e-05,
+      "loss": 0.3589,
+      "step": 182870
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.6695499162585595e-05,
+      "loss": 0.3665,
+      "step": 182880
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.6653098433293055e-05,
+      "loss": 0.4048,
+      "step": 182890
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.661069770400051e-05,
+      "loss": 0.443,
+      "step": 182900
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.656829697470797e-05,
+      "loss": 0.3874,
+      "step": 182910
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.652589624541542e-05,
+      "loss": 0.4529,
+      "step": 182920
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.648349551612288e-05,
+      "loss": 0.3819,
+      "step": 182930
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.6441094786830335e-05,
+      "loss": 0.3623,
+      "step": 182940
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.639869405753779e-05,
+      "loss": 0.3992,
+      "step": 182950
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.6356293328245245e-05,
+      "loss": 0.404,
+      "step": 182960
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.6313892598952705e-05,
+      "loss": 0.3934,
+      "step": 182970
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.6271491869660158e-05,
+      "loss": 0.3584,
+      "step": 182980
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.6229091140367618e-05,
+      "loss": 0.3628,
+      "step": 182990
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.618669041107507e-05,
+      "loss": 0.4097,
+      "step": 183000
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.614428968178253e-05,
+      "loss": 0.3376,
+      "step": 183010
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.6101888952489984e-05,
+      "loss": 0.3635,
+      "step": 183020
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.605948822319744e-05,
+      "loss": 0.3744,
+      "step": 183030
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.6017087493904894e-05,
+      "loss": 0.386,
+      "step": 183040
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5974686764612354e-05,
+      "loss": 0.4421,
+      "step": 183050
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5932286035319808e-05,
+      "loss": 0.3538,
+      "step": 183060
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5889885306027268e-05,
+      "loss": 0.3459,
+      "step": 183070
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.584748457673472e-05,
+      "loss": 0.3453,
+      "step": 183080
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.580508384744218e-05,
+      "loss": 0.3149,
+      "step": 183090
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5762683118149634e-05,
+      "loss": 0.3306,
+      "step": 183100
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.572028238885709e-05,
+      "loss": 0.3347,
+      "step": 183110
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5677881659564544e-05,
+      "loss": 0.3902,
+      "step": 183120
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5635480930272004e-05,
+      "loss": 0.3292,
+      "step": 183130
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5593080200979457e-05,
+      "loss": 0.3657,
+      "step": 183140
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5550679471686917e-05,
+      "loss": 0.3692,
+      "step": 183150
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.550827874239437e-05,
+      "loss": 0.3675,
+      "step": 183160
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.546587801310183e-05,
+      "loss": 0.285,
+      "step": 183170
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5423477283809284e-05,
+      "loss": 0.3751,
+      "step": 183180
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5381076554516737e-05,
+      "loss": 0.3342,
+      "step": 183190
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5338675825224194e-05,
+      "loss": 0.4006,
+      "step": 183200
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.529627509593165e-05,
+      "loss": 0.3813,
+      "step": 183210
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5253874366639107e-05,
+      "loss": 0.3504,
+      "step": 183220
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.521147363734656e-05,
+      "loss": 0.3442,
+      "step": 183230
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.516907290805402e-05,
+      "loss": 0.3516,
+      "step": 183240
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5126672178761473e-05,
+      "loss": 0.4368,
+      "step": 183250
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5084271449468933e-05,
+      "loss": 0.3109,
+      "step": 183260
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5041870720176387e-05,
+      "loss": 0.3892,
+      "step": 183270
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4999469990883843e-05,
+      "loss": 0.3049,
+      "step": 183280
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.49570692615913e-05,
+      "loss": 0.3884,
+      "step": 183290
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4914668532298756e-05,
+      "loss": 0.3338,
+      "step": 183300
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4872267803006213e-05,
+      "loss": 0.3555,
+      "step": 183310
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.482986707371367e-05,
+      "loss": 0.3306,
+      "step": 183320
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4787466344421126e-05,
+      "loss": 0.4244,
+      "step": 183330
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4745065615128583e-05,
+      "loss": 0.3564,
+      "step": 183340
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.470266488583604e-05,
+      "loss": 0.3806,
+      "step": 183350
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4660264156543493e-05,
+      "loss": 0.3458,
+      "step": 183360
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.461786342725095e-05,
+      "loss": 0.3564,
+      "step": 183370
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4575462697958406e-05,
+      "loss": 0.432,
+      "step": 183380
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4533061968665863e-05,
+      "loss": 0.3419,
+      "step": 183390
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.449066123937332e-05,
+      "loss": 0.2882,
+      "step": 183400
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4448260510080776e-05,
+      "loss": 0.3424,
+      "step": 183410
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4405859780788233e-05,
+      "loss": 0.4403,
+      "step": 183420
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.436345905149569e-05,
+      "loss": 0.3997,
+      "step": 183430
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4321058322203142e-05,
+      "loss": 0.3191,
+      "step": 183440
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.42786575929106e-05,
+      "loss": 0.3822,
+      "step": 183450
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4236256863618056e-05,
+      "loss": 0.3781,
+      "step": 183460
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4193856134325512e-05,
+      "loss": 0.4212,
+      "step": 183470
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.415145540503297e-05,
+      "loss": 0.3141,
+      "step": 183480
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4109054675740426e-05,
+      "loss": 0.3492,
+      "step": 183490
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4066653946447882e-05,
+      "loss": 0.4218,
+      "step": 183500
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.402425321715534e-05,
+      "loss": 0.4447,
+      "step": 183510
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3981852487862795e-05,
+      "loss": 0.3339,
+      "step": 183520
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.393945175857025e-05,
+      "loss": 0.3729,
+      "step": 183530
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3897051029277702e-05,
+      "loss": 0.4137,
+      "step": 183540
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.385465029998516e-05,
+      "loss": 0.4112,
+      "step": 183550
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3812249570692615e-05,
+      "loss": 0.4026,
+      "step": 183560
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3769848841400072e-05,
+      "loss": 0.4015,
+      "step": 183570
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.372744811210753e-05,
+      "loss": 0.3402,
+      "step": 183580
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3685047382814985e-05,
+      "loss": 0.3767,
+      "step": 183590
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.364264665352244e-05,
+      "loss": 0.3169,
+      "step": 183600
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3600245924229898e-05,
+      "loss": 0.4019,
+      "step": 183610
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.355784519493735e-05,
+      "loss": 0.3686,
+      "step": 183620
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3515444465644808e-05,
+      "loss": 0.3945,
+      "step": 183630
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3473043736352265e-05,
+      "loss": 0.3539,
+      "step": 183640
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.343064300705972e-05,
+      "loss": 0.4016,
+      "step": 183650
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3388242277767178e-05,
+      "loss": 0.3874,
+      "step": 183660
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3345841548474635e-05,
+      "loss": 0.4051,
+      "step": 183670
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.330344081918209e-05,
+      "loss": 0.3978,
+      "step": 183680
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3261040089889548e-05,
+      "loss": 0.3903,
+      "step": 183690
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3218639360597e-05,
+      "loss": 0.3897,
+      "step": 183700
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3176238631304458e-05,
+      "loss": 0.3907,
+      "step": 183710
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3133837902011914e-05,
+      "loss": 0.3428,
+      "step": 183720
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.309143717271937e-05,
+      "loss": 0.3443,
+      "step": 183730
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3049036443426828e-05,
+      "loss": 0.3903,
+      "step": 183740
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3006635714134284e-05,
+      "loss": 0.3485,
+      "step": 183750
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.296423498484174e-05,
+      "loss": 0.3953,
+      "step": 183760
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.2921834255549197e-05,
+      "loss": 0.3635,
+      "step": 183770
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.287943352625665e-05,
+      "loss": 0.3239,
+      "step": 183780
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.2837032796964107e-05,
+      "loss": 0.3929,
+      "step": 183790
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.2794632067671564e-05,
+      "loss": 0.3812,
+      "step": 183800
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.275223133837902e-05,
+      "loss": 0.3755,
+      "step": 183810
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.2709830609086477e-05,
+      "loss": 0.3101,
+      "step": 183820
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.2667429879793934e-05,
+      "loss": 0.4312,
+      "step": 183830
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.262502915050139e-05,
+      "loss": 0.4142,
+      "step": 183840
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.2582628421208847e-05,
+      "loss": 0.3778,
+      "step": 183850
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.25402276919163e-05,
+      "loss": 0.3544,
+      "step": 183860
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.2497826962623757e-05,
+      "loss": 0.3691,
+      "step": 183870
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.2455426233331214e-05,
+      "loss": 0.464,
+      "step": 183880
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.241302550403867e-05,
+      "loss": 0.4208,
+      "step": 183890
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.2370624774746127e-05,
+      "loss": 0.3721,
+      "step": 183900
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.2328224045453583e-05,
+      "loss": 0.3458,
+      "step": 183910
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.228582331616104e-05,
+      "loss": 0.376,
+      "step": 183920
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.2243422586868497e-05,
+      "loss": 0.3222,
+      "step": 183930
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.220102185757595e-05,
+      "loss": 0.3822,
+      "step": 183940
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.2158621128283407e-05,
+      "loss": 0.3683,
+      "step": 183950
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.2116220398990863e-05,
+      "loss": 0.2676,
+      "step": 183960
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.207381966969832e-05,
+      "loss": 0.369,
+      "step": 183970
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.2031418940405776e-05,
+      "loss": 0.3758,
+      "step": 183980
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.1989018211113233e-05,
+      "loss": 0.3598,
+      "step": 183990
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.194661748182069e-05,
+      "loss": 0.3682,
+      "step": 184000
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.1904216752528146e-05,
+      "loss": 0.3553,
+      "step": 184010
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.18618160232356e-05,
+      "loss": 0.3728,
+      "step": 184020
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.1819415293943056e-05,
+      "loss": 0.3714,
+      "step": 184030
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.1777014564650513e-05,
+      "loss": 0.3882,
+      "step": 184040
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.173461383535797e-05,
+      "loss": 0.294,
+      "step": 184050
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.1692213106065426e-05,
+      "loss": 0.404,
+      "step": 184060
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.1649812376772883e-05,
+      "loss": 0.3445,
+      "step": 184070
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.160741164748034e-05,
+      "loss": 0.3456,
+      "step": 184080
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.1565010918187796e-05,
+      "loss": 0.4402,
+      "step": 184090
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1522610188895253e-05,
+      "loss": 0.3456,
+      "step": 184100
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1480209459602706e-05,
+      "loss": 0.3621,
+      "step": 184110
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1437808730310162e-05,
+      "loss": 0.3822,
+      "step": 184120
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.139540800101762e-05,
+      "loss": 0.4201,
+      "step": 184130
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1353007271725076e-05,
+      "loss": 0.4038,
+      "step": 184140
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1310606542432532e-05,
+      "loss": 0.326,
+      "step": 184150
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.126820581313999e-05,
+      "loss": 0.357,
+      "step": 184160
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1225805083847445e-05,
+      "loss": 0.4149,
+      "step": 184170
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1183404354554902e-05,
+      "loss": 0.3658,
+      "step": 184180
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1141003625262355e-05,
+      "loss": 0.2931,
+      "step": 184190
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1098602895969812e-05,
+      "loss": 0.4045,
+      "step": 184200
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.105620216667727e-05,
+      "loss": 0.3554,
+      "step": 184210
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1013801437384725e-05,
+      "loss": 0.375,
+      "step": 184220
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.0971400708092182e-05,
+      "loss": 0.4195,
+      "step": 184230
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.092899997879964e-05,
+      "loss": 0.3723,
+      "step": 184240
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.0886599249507095e-05,
+      "loss": 0.2803,
+      "step": 184250
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.084419852021455e-05,
+      "loss": 0.41,
+      "step": 184260
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.0801797790922005e-05,
+      "loss": 0.3553,
+      "step": 184270
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.0759397061629458e-05,
+      "loss": 0.3525,
+      "step": 184280
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.0716996332336915e-05,
+      "loss": 0.3692,
+      "step": 184290
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.067459560304437e-05,
+      "loss": 0.3556,
+      "step": 184300
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.0632194873751828e-05,
+      "loss": 0.3521,
+      "step": 184310
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.0589794144459285e-05,
+      "loss": 0.3495,
+      "step": 184320
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.054739341516674e-05,
+      "loss": 0.3358,
+      "step": 184330
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0504992685874198e-05,
+      "loss": 0.3917,
+      "step": 184340
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0462591956581655e-05,
+      "loss": 0.3652,
+      "step": 184350
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0420191227289108e-05,
+      "loss": 0.3577,
+      "step": 184360
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0377790497996564e-05,
+      "loss": 0.4198,
+      "step": 184370
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.033538976870402e-05,
+      "loss": 0.3253,
+      "step": 184380
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0292989039411478e-05,
+      "loss": 0.4207,
+      "step": 184390
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0250588310118934e-05,
+      "loss": 0.3946,
+      "step": 184400
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.020818758082639e-05,
+      "loss": 0.3663,
+      "step": 184410
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0165786851533848e-05,
+      "loss": 0.3623,
+      "step": 184420
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0123386122241304e-05,
+      "loss": 0.41,
+      "step": 184430
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0080985392948757e-05,
+      "loss": 0.3845,
+      "step": 184440
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0038584663656214e-05,
+      "loss": 0.3386,
+      "step": 184450
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.999618393436367e-05,
+      "loss": 0.3594,
+      "step": 184460
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.9953783205071127e-05,
+      "loss": 0.3848,
+      "step": 184470
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.9911382475778584e-05,
+      "loss": 0.3072,
+      "step": 184480
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.986898174648604e-05,
+      "loss": 0.3652,
+      "step": 184490
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.9826581017193497e-05,
+      "loss": 0.3684,
+      "step": 184500
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.9784180287900954e-05,
+      "loss": 0.3812,
+      "step": 184510
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.9741779558608407e-05,
+      "loss": 0.364,
+      "step": 184520
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.9699378829315864e-05,
+      "loss": 0.3583,
+      "step": 184530
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.965697810002332e-05,
+      "loss": 0.3665,
+      "step": 184540
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.9614577370730777e-05,
+      "loss": 0.4021,
+      "step": 184550
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.9572176641438234e-05,
+      "loss": 0.3641,
+      "step": 184560
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.952977591214569e-05,
+      "loss": 0.3798,
+      "step": 184570
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9487375182853147e-05,
+      "loss": 0.4578,
+      "step": 184580
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9444974453560603e-05,
+      "loss": 0.4162,
+      "step": 184590
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9402573724268057e-05,
+      "loss": 0.3748,
+      "step": 184600
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9360172994975513e-05,
+      "loss": 0.3441,
+      "step": 184610
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.931777226568297e-05,
+      "loss": 0.4127,
+      "step": 184620
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9275371536390427e-05,
+      "loss": 0.3466,
+      "step": 184630
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9232970807097883e-05,
+      "loss": 0.3582,
+      "step": 184640
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.919057007780534e-05,
+      "loss": 0.3218,
+      "step": 184650
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9148169348512796e-05,
+      "loss": 0.3771,
+      "step": 184660
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9105768619220253e-05,
+      "loss": 0.3632,
+      "step": 184670
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.906336788992771e-05,
+      "loss": 0.4084,
+      "step": 184680
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9020967160635163e-05,
+      "loss": 0.3025,
+      "step": 184690
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.897856643134262e-05,
+      "loss": 0.4001,
+      "step": 184700
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.8936165702050076e-05,
+      "loss": 0.3685,
+      "step": 184710
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.8893764972757533e-05,
+      "loss": 0.3924,
+      "step": 184720
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.885136424346499e-05,
+      "loss": 0.3888,
+      "step": 184730
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.8808963514172446e-05,
+      "loss": 0.3532,
+      "step": 184740
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.8766562784879903e-05,
+      "loss": 0.3608,
+      "step": 184750
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.872416205558736e-05,
+      "loss": 0.3787,
+      "step": 184760
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.8681761326294812e-05,
+      "loss": 0.4143,
+      "step": 184770
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.863936059700227e-05,
+      "loss": 0.3548,
+      "step": 184780
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.8596959867709726e-05,
+      "loss": 0.3699,
+      "step": 184790
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.8554559138417182e-05,
+      "loss": 0.3992,
+      "step": 184800
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.851215840912464e-05,
+      "loss": 0.4057,
+      "step": 184810
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.8469757679832096e-05,
+      "loss": 0.4115,
+      "step": 184820
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.8427356950539552e-05,
+      "loss": 0.37,
+      "step": 184830
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.838495622124701e-05,
+      "loss": 0.36,
+      "step": 184840
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.8342555491954462e-05,
+      "loss": 0.3133,
+      "step": 184850
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.830015476266192e-05,
+      "loss": 0.4425,
+      "step": 184860
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.8257754033369375e-05,
+      "loss": 0.3928,
+      "step": 184870
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.8215353304076832e-05,
+      "loss": 0.3717,
+      "step": 184880
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.817295257478429e-05,
+      "loss": 0.4305,
+      "step": 184890
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.8130551845491745e-05,
+      "loss": 0.415,
+      "step": 184900
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.8088151116199202e-05,
+      "loss": 0.3945,
+      "step": 184910
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.804575038690666e-05,
+      "loss": 0.4029,
+      "step": 184920
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.800334965761411e-05,
+      "loss": 0.3591,
+      "step": 184930
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.796094892832157e-05,
+      "loss": 0.3365,
+      "step": 184940
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.7918548199029025e-05,
+      "loss": 0.373,
+      "step": 184950
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.787614746973648e-05,
+      "loss": 0.4612,
+      "step": 184960
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.7833746740443935e-05,
+      "loss": 0.4433,
+      "step": 184970
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.779134601115139e-05,
+      "loss": 0.3811,
+      "step": 184980
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.7748945281858848e-05,
+      "loss": 0.4238,
+      "step": 184990
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.7706544552566305e-05,
+      "loss": 0.3846,
+      "step": 185000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.766414382327376e-05,
+      "loss": 0.3644,
+      "step": 185010
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.7621743093981215e-05,
+      "loss": 0.3907,
+      "step": 185020
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.757934236468867e-05,
+      "loss": 0.3514,
+      "step": 185030
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7536941635396128e-05,
+      "loss": 0.4321,
+      "step": 185040
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7494540906103584e-05,
+      "loss": 0.4335,
+      "step": 185050
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.745214017681104e-05,
+      "loss": 0.3566,
+      "step": 185060
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7409739447518498e-05,
+      "loss": 0.338,
+      "step": 185070
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7367338718225954e-05,
+      "loss": 0.3773,
+      "step": 185080
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.732493798893341e-05,
+      "loss": 0.386,
+      "step": 185090
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7282537259640864e-05,
+      "loss": 0.393,
+      "step": 185100
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.724013653034832e-05,
+      "loss": 0.3874,
+      "step": 185110
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7197735801055777e-05,
+      "loss": 0.3942,
+      "step": 185120
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7155335071763234e-05,
+      "loss": 0.4317,
+      "step": 185130
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.711293434247069e-05,
+      "loss": 0.3955,
+      "step": 185140
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7070533613178147e-05,
+      "loss": 0.3254,
+      "step": 185150
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7028132883885604e-05,
+      "loss": 0.3536,
+      "step": 185160
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.698573215459306e-05,
+      "loss": 0.4375,
+      "step": 185170
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.6943331425300514e-05,
+      "loss": 0.3472,
+      "step": 185180
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.690093069600797e-05,
+      "loss": 0.3742,
+      "step": 185190
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.6858529966715427e-05,
+      "loss": 0.3856,
+      "step": 185200
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.6816129237422884e-05,
+      "loss": 0.4201,
+      "step": 185210
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.677372850813034e-05,
+      "loss": 0.4182,
+      "step": 185220
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.6731327778837797e-05,
+      "loss": 0.3448,
+      "step": 185230
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.6688927049545253e-05,
+      "loss": 0.3576,
+      "step": 185240
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.664652632025271e-05,
+      "loss": 0.3505,
+      "step": 185250
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.6604125590960167e-05,
+      "loss": 0.5023,
+      "step": 185260
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.656172486166762e-05,
+      "loss": 0.3455,
+      "step": 185270
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6519324132375077e-05,
+      "loss": 0.3317,
+      "step": 185280
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6476923403082533e-05,
+      "loss": 0.4032,
+      "step": 185290
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.643452267378999e-05,
+      "loss": 0.3757,
+      "step": 185300
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6392121944497446e-05,
+      "loss": 0.3658,
+      "step": 185310
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6349721215204903e-05,
+      "loss": 0.3617,
+      "step": 185320
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.630732048591236e-05,
+      "loss": 0.3706,
+      "step": 185330
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6264919756619816e-05,
+      "loss": 0.4194,
+      "step": 185340
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.622251902732727e-05,
+      "loss": 0.3574,
+      "step": 185350
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6180118298034726e-05,
+      "loss": 0.4101,
+      "step": 185360
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6137717568742183e-05,
+      "loss": 0.3865,
+      "step": 185370
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.609531683944964e-05,
+      "loss": 0.3178,
+      "step": 185380
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6052916110157096e-05,
+      "loss": 0.3911,
+      "step": 185390
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6010515380864553e-05,
+      "loss": 0.3466,
+      "step": 185400
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.596811465157201e-05,
+      "loss": 0.3802,
+      "step": 185410
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.5925713922279466e-05,
+      "loss": 0.4018,
+      "step": 185420
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.588331319298692e-05,
+      "loss": 0.3965,
+      "step": 185430
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.5840912463694376e-05,
+      "loss": 0.4004,
+      "step": 185440
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.5798511734401832e-05,
+      "loss": 0.3426,
+      "step": 185450
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.575611100510929e-05,
+      "loss": 0.3918,
+      "step": 185460
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.5713710275816746e-05,
+      "loss": 0.3796,
+      "step": 185470
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.5671309546524202e-05,
+      "loss": 0.399,
+      "step": 185480
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.562890881723166e-05,
+      "loss": 0.4123,
+      "step": 185490
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.5586508087939116e-05,
+      "loss": 0.3148,
+      "step": 185500
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.554410735864657e-05,
+      "loss": 0.3669,
+      "step": 185510
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5501706629354025e-05,
+      "loss": 0.4057,
+      "step": 185520
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5459305900061482e-05,
+      "loss": 0.4121,
+      "step": 185530
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.541690517076894e-05,
+      "loss": 0.4053,
+      "step": 185540
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5374504441476395e-05,
+      "loss": 0.3636,
+      "step": 185550
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5332103712183852e-05,
+      "loss": 0.3092,
+      "step": 185560
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.528970298289131e-05,
+      "loss": 0.3915,
+      "step": 185570
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5247302253598763e-05,
+      "loss": 0.3744,
+      "step": 185580
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.520490152430622e-05,
+      "loss": 0.3883,
+      "step": 185590
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5162500795013677e-05,
+      "loss": 0.3589,
+      "step": 185600
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5120100065721132e-05,
+      "loss": 0.3103,
+      "step": 185610
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5077699336428588e-05,
+      "loss": 0.4398,
+      "step": 185620
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5035298607136045e-05,
+      "loss": 0.3487,
+      "step": 185630
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4992897877843502e-05,
+      "loss": 0.3662,
+      "step": 185640
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4950497148550956e-05,
+      "loss": 0.375,
+      "step": 185650
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4908096419258413e-05,
+      "loss": 0.3721,
+      "step": 185660
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.486569568996587e-05,
+      "loss": 0.3647,
+      "step": 185670
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4823294960673323e-05,
+      "loss": 0.3318,
+      "step": 185680
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.478089423138078e-05,
+      "loss": 0.4614,
+      "step": 185690
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4738493502088235e-05,
+      "loss": 0.3015,
+      "step": 185700
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4696092772795691e-05,
+      "loss": 0.4044,
+      "step": 185710
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4653692043503148e-05,
+      "loss": 0.4387,
+      "step": 185720
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4611291314210604e-05,
+      "loss": 0.3427,
+      "step": 185730
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.456889058491806e-05,
+      "loss": 0.3152,
+      "step": 185740
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4526489855625516e-05,
+      "loss": 0.3476,
+      "step": 185750
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4484089126332973e-05,
+      "loss": 0.3812,
+      "step": 185760
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.444168839704043e-05,
+      "loss": 0.3826,
+      "step": 185770
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4399287667747884e-05,
+      "loss": 0.4275,
+      "step": 185780
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.435688693845534e-05,
+      "loss": 0.3901,
+      "step": 185790
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4314486209162797e-05,
+      "loss": 0.3319,
+      "step": 185800
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4272085479870254e-05,
+      "loss": 0.3753,
+      "step": 185810
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.422968475057771e-05,
+      "loss": 0.4716,
+      "step": 185820
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4187284021285166e-05,
+      "loss": 0.4051,
+      "step": 185830
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4144883291992622e-05,
+      "loss": 0.3977,
+      "step": 185840
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4102482562700079e-05,
+      "loss": 0.4423,
+      "step": 185850
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4060081833407535e-05,
+      "loss": 0.3781,
+      "step": 185860
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.401768110411499e-05,
+      "loss": 0.3043,
+      "step": 185870
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3975280374822447e-05,
+      "loss": 0.3833,
+      "step": 185880
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3932879645529904e-05,
+      "loss": 0.4245,
+      "step": 185890
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.389047891623736e-05,
+      "loss": 0.4038,
+      "step": 185900
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3848078186944815e-05,
+      "loss": 0.4404,
+      "step": 185910
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3805677457652272e-05,
+      "loss": 0.3907,
+      "step": 185920
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3763276728359728e-05,
+      "loss": 0.4008,
+      "step": 185930
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3720875999067185e-05,
+      "loss": 0.3636,
+      "step": 185940
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.367847526977464e-05,
+      "loss": 0.3097,
+      "step": 185950
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3636074540482097e-05,
+      "loss": 0.3898,
+      "step": 185960
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3593673811189553e-05,
+      "loss": 0.3788,
+      "step": 185970
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.355127308189701e-05,
+      "loss": 0.3197,
+      "step": 185980
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3508872352604465e-05,
+      "loss": 0.329,
+      "step": 185990
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3466471623311921e-05,
+      "loss": 0.3515,
+      "step": 186000
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3424070894019378e-05,
+      "loss": 0.3704,
+      "step": 186010
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3381670164726835e-05,
+      "loss": 0.3703,
+      "step": 186020
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.333926943543429e-05,
+      "loss": 0.3936,
+      "step": 186030
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3296868706141746e-05,
+      "loss": 0.3891,
+      "step": 186040
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3254467976849203e-05,
+      "loss": 0.337,
+      "step": 186050
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.321206724755666e-05,
+      "loss": 0.4038,
+      "step": 186060
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3169666518264114e-05,
+      "loss": 0.4156,
+      "step": 186070
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3127265788971571e-05,
+      "loss": 0.4155,
+      "step": 186080
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3084865059679028e-05,
+      "loss": 0.3392,
+      "step": 186090
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3042464330386484e-05,
+      "loss": 0.4044,
+      "step": 186100
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.300006360109394e-05,
+      "loss": 0.386,
+      "step": 186110
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2957662871801396e-05,
+      "loss": 0.3983,
+      "step": 186120
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2915262142508852e-05,
+      "loss": 0.4207,
+      "step": 186130
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2872861413216309e-05,
+      "loss": 0.4145,
+      "step": 186140
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2830460683923764e-05,
+      "loss": 0.4018,
+      "step": 186150
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.278805995463122e-05,
+      "loss": 0.298,
+      "step": 186160
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2745659225338677e-05,
+      "loss": 0.328,
+      "step": 186170
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2703258496046134e-05,
+      "loss": 0.3534,
+      "step": 186180
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2660857766753589e-05,
+      "loss": 0.3538,
+      "step": 186190
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2618457037461045e-05,
+      "loss": 0.3785,
+      "step": 186200
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2576056308168502e-05,
+      "loss": 0.3375,
+      "step": 186210
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2533655578875959e-05,
+      "loss": 0.3753,
+      "step": 186220
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2491254849583414e-05,
+      "loss": 0.35,
+      "step": 186230
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2448854120290869e-05,
+      "loss": 0.4205,
+      "step": 186240
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2406453390998325e-05,
+      "loss": 0.3705,
+      "step": 186250
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2364052661705782e-05,
+      "loss": 0.3567,
+      "step": 186260
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2321651932413238e-05,
+      "loss": 0.3971,
+      "step": 186270
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2279251203120693e-05,
+      "loss": 0.384,
+      "step": 186280
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.223685047382815e-05,
+      "loss": 0.4279,
+      "step": 186290
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2194449744535607e-05,
+      "loss": 0.4351,
+      "step": 186300
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2152049015243063e-05,
+      "loss": 0.4061,
+      "step": 186310
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2109648285950518e-05,
+      "loss": 0.3652,
+      "step": 186320
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2067247556657975e-05,
+      "loss": 0.3425,
+      "step": 186330
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2024846827365431e-05,
+      "loss": 0.3397,
+      "step": 186340
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1982446098072888e-05,
+      "loss": 0.3757,
+      "step": 186350
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1940045368780343e-05,
+      "loss": 0.4406,
+      "step": 186360
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.18976446394878e-05,
+      "loss": 0.4133,
+      "step": 186370
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1855243910195256e-05,
+      "loss": 0.3724,
+      "step": 186380
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1812843180902713e-05,
+      "loss": 0.3731,
+      "step": 186390
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1770442451610168e-05,
+      "loss": 0.3044,
+      "step": 186400
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1728041722317624e-05,
+      "loss": 0.3422,
+      "step": 186410
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1685640993025081e-05,
+      "loss": 0.3862,
+      "step": 186420
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1643240263732538e-05,
+      "loss": 0.3238,
+      "step": 186430
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1600839534439993e-05,
+      "loss": 0.4351,
+      "step": 186440
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.155843880514745e-05,
+      "loss": 0.3973,
+      "step": 186450
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1516038075854906e-05,
+      "loss": 0.3736,
+      "step": 186460
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1473637346562362e-05,
+      "loss": 0.3503,
+      "step": 186470
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1431236617269817e-05,
+      "loss": 0.3079,
+      "step": 186480
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1388835887977274e-05,
+      "loss": 0.3765,
+      "step": 186490
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.134643515868473e-05,
+      "loss": 0.3824,
+      "step": 186500
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1304034429392187e-05,
+      "loss": 0.3941,
+      "step": 186510
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1261633700099642e-05,
+      "loss": 0.3593,
+      "step": 186520
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1219232970807099e-05,
+      "loss": 0.35,
+      "step": 186530
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1176832241514555e-05,
+      "loss": 0.2618,
+      "step": 186540
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1134431512222012e-05,
+      "loss": 0.3465,
+      "step": 186550
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1092030782929467e-05,
+      "loss": 0.4177,
+      "step": 186560
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1049630053636924e-05,
+      "loss": 0.3588,
+      "step": 186570
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1007229324344379e-05,
+      "loss": 0.3496,
+      "step": 186580
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0964828595051835e-05,
+      "loss": 0.3465,
+      "step": 186590
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0922427865759292e-05,
+      "loss": 0.3001,
+      "step": 186600
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0880027136466747e-05,
+      "loss": 0.4129,
+      "step": 186610
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0837626407174203e-05,
+      "loss": 0.3368,
+      "step": 186620
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.079522567788166e-05,
+      "loss": 0.3641,
+      "step": 186630
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0752824948589117e-05,
+      "loss": 0.4608,
+      "step": 186640
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0710424219296571e-05,
+      "loss": 0.3792,
+      "step": 186650
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0668023490004028e-05,
+      "loss": 0.3593,
+      "step": 186660
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0625622760711485e-05,
+      "loss": 0.4129,
+      "step": 186670
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0583222031418941e-05,
+      "loss": 0.366,
+      "step": 186680
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0540821302126396e-05,
+      "loss": 0.3457,
+      "step": 186690
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0498420572833853e-05,
+      "loss": 0.3944,
+      "step": 186700
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.045601984354131e-05,
+      "loss": 0.4162,
+      "step": 186710
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0413619114248766e-05,
+      "loss": 0.4099,
+      "step": 186720
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0371218384956221e-05,
+      "loss": 0.3969,
+      "step": 186730
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0328817655663678e-05,
+      "loss": 0.349,
+      "step": 186740
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0286416926371134e-05,
+      "loss": 0.346,
+      "step": 186750
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0244016197078591e-05,
+      "loss": 0.3775,
+      "step": 186760
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0201615467786046e-05,
+      "loss": 0.4394,
+      "step": 186770
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0159214738493503e-05,
+      "loss": 0.4033,
+      "step": 186780
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.011681400920096e-05,
+      "loss": 0.4326,
+      "step": 186790
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0074413279908416e-05,
+      "loss": 0.376,
+      "step": 186800
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.003201255061587e-05,
+      "loss": 0.4004,
+      "step": 186810
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.989611821323327e-06,
+      "loss": 0.3771,
+      "step": 186820
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.947211092030784e-06,
+      "loss": 0.3444,
+      "step": 186830
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.90481036273824e-06,
+      "loss": 0.3267,
+      "step": 186840
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.862409633445696e-06,
+      "loss": 0.3677,
+      "step": 186850
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.820008904153152e-06,
+      "loss": 0.358,
+      "step": 186860
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.777608174860609e-06,
+      "loss": 0.3816,
+      "step": 186870
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.735207445568065e-06,
+      "loss": 0.326,
+      "step": 186880
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.69280671627552e-06,
+      "loss": 0.3423,
+      "step": 186890
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.650405986982977e-06,
+      "loss": 0.2798,
+      "step": 186900
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.608005257690434e-06,
+      "loss": 0.3309,
+      "step": 186910
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.56560452839789e-06,
+      "loss": 0.4486,
+      "step": 186920
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.523203799105345e-06,
+      "loss": 0.3956,
+      "step": 186930
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.4808030698128e-06,
+      "loss": 0.4326,
+      "step": 186940
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.438402340520257e-06,
+      "loss": 0.3918,
+      "step": 186950
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.396001611227713e-06,
+      "loss": 0.4103,
+      "step": 186960
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.35360088193517e-06,
+      "loss": 0.3981,
+      "step": 186970
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.311200152642625e-06,
+      "loss": 0.383,
+      "step": 186980
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.268799423350081e-06,
+      "loss": 0.378,
+      "step": 186990
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.226398694057538e-06,
+      "loss": 0.3184,
+      "step": 187000
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.183997964764995e-06,
+      "loss": 0.3963,
+      "step": 187010
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.14159723547245e-06,
+      "loss": 0.4113,
+      "step": 187020
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.099196506179906e-06,
+      "loss": 0.3635,
+      "step": 187030
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.056795776887363e-06,
+      "loss": 0.3341,
+      "step": 187040
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.01439504759482e-06,
+      "loss": 0.3746,
+      "step": 187050
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.971994318302274e-06,
+      "loss": 0.4123,
+      "step": 187060
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.929593589009731e-06,
+      "loss": 0.3954,
+      "step": 187070
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.887192859717188e-06,
+      "loss": 0.3778,
+      "step": 187080
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.844792130424644e-06,
+      "loss": 0.3544,
+      "step": 187090
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.8023914011321e-06,
+      "loss": 0.3541,
+      "step": 187100
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.759990671839556e-06,
+      "loss": 0.4015,
+      "step": 187110
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.717589942547013e-06,
+      "loss": 0.4043,
+      "step": 187120
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.675189213254469e-06,
+      "loss": 0.3358,
+      "step": 187130
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.632788483961924e-06,
+      "loss": 0.3883,
+      "step": 187140
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.59038775466938e-06,
+      "loss": 0.3171,
+      "step": 187150
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.547987025376837e-06,
+      "loss": 0.4211,
+      "step": 187160
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.505586296084294e-06,
+      "loss": 0.3787,
+      "step": 187170
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.463185566791749e-06,
+      "loss": 0.3884,
+      "step": 187180
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.420784837499206e-06,
+      "loss": 0.3941,
+      "step": 187190
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.378384108206662e-06,
+      "loss": 0.3932,
+      "step": 187200
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.335983378914119e-06,
+      "loss": 0.3455,
+      "step": 187210
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.293582649621574e-06,
+      "loss": 0.4081,
+      "step": 187220
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.25118192032903e-06,
+      "loss": 0.4124,
+      "step": 187230
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.208781191036487e-06,
+      "loss": 0.3422,
+      "step": 187240
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.166380461743944e-06,
+      "loss": 0.3477,
+      "step": 187250
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.123979732451398e-06,
+      "loss": 0.2841,
+      "step": 187260
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.081579003158855e-06,
+      "loss": 0.4003,
+      "step": 187270
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.039178273866312e-06,
+      "loss": 0.3828,
+      "step": 187280
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.996777544573767e-06,
+      "loss": 0.4011,
+      "step": 187290
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.954376815281223e-06,
+      "loss": 0.4213,
+      "step": 187300
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.911976085988678e-06,
+      "loss": 0.3612,
+      "step": 187310
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.869575356696135e-06,
+      "loss": 0.3605,
+      "step": 187320
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.827174627403591e-06,
+      "loss": 0.4473,
+      "step": 187330
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.784773898111048e-06,
+      "loss": 0.384,
+      "step": 187340
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.742373168818503e-06,
+      "loss": 0.3554,
+      "step": 187350
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.69997243952596e-06,
+      "loss": 0.3842,
+      "step": 187360
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.657571710233416e-06,
+      "loss": 0.434,
+      "step": 187370
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.615170980940872e-06,
+      "loss": 0.3086,
+      "step": 187380
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.572770251648329e-06,
+      "loss": 0.3993,
+      "step": 187390
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.5303695223557845e-06,
+      "loss": 0.3495,
+      "step": 187400
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.487968793063241e-06,
+      "loss": 0.4184,
+      "step": 187410
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.445568063770697e-06,
+      "loss": 0.3354,
+      "step": 187420
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.4031673344781535e-06,
+      "loss": 0.3789,
+      "step": 187430
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.360766605185609e-06,
+      "loss": 0.3266,
+      "step": 187440
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.318365875893066e-06,
+      "loss": 0.3851,
+      "step": 187450
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.275965146600522e-06,
+      "loss": 0.3263,
+      "step": 187460
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.233564417307978e-06,
+      "loss": 0.3465,
+      "step": 187470
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.191163688015434e-06,
+      "loss": 0.3828,
+      "step": 187480
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.148762958722891e-06,
+      "loss": 0.3576,
+      "step": 187490
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.1063622294303465e-06,
+      "loss": 0.3423,
+      "step": 187500
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.063961500137803e-06,
+      "loss": 0.3403,
+      "step": 187510
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.021560770845259e-06,
+      "loss": 0.3606,
+      "step": 187520
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.9791600415527155e-06,
+      "loss": 0.3039,
+      "step": 187530
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.936759312260171e-06,
+      "loss": 0.3124,
+      "step": 187540
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.894358582967628e-06,
+      "loss": 0.3701,
+      "step": 187550
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.851957853675084e-06,
+      "loss": 0.3779,
+      "step": 187560
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.80955712438254e-06,
+      "loss": 0.4105,
+      "step": 187570
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.767156395089996e-06,
+      "loss": 0.3491,
+      "step": 187580
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.724755665797453e-06,
+      "loss": 0.3898,
+      "step": 187590
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.6823549365049085e-06,
+      "loss": 0.341,
+      "step": 187600
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.639954207212365e-06,
+      "loss": 0.352,
+      "step": 187610
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.597553477919821e-06,
+      "loss": 0.4399,
+      "step": 187620
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.5551527486272775e-06,
+      "loss": 0.362,
+      "step": 187630
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.512752019334733e-06,
+      "loss": 0.3749,
+      "step": 187640
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.470351290042188e-06,
+      "loss": 0.3234,
+      "step": 187650
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.427950560749645e-06,
+      "loss": 0.3269,
+      "step": 187660
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.385549831457101e-06,
+      "loss": 0.392,
+      "step": 187670
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.343149102164557e-06,
+      "loss": 0.394,
+      "step": 187680
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.300748372872013e-06,
+      "loss": 0.3628,
+      "step": 187690
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.25834764357947e-06,
+      "loss": 0.2963,
+      "step": 187700
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.215946914286926e-06,
+      "loss": 0.3753,
+      "step": 187710
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.173546184994383e-06,
+      "loss": 0.46,
+      "step": 187720
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.131145455701839e-06,
+      "loss": 0.3302,
+      "step": 187730
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.0887447264092944e-06,
+      "loss": 0.3223,
+      "step": 187740
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.04634399711675e-06,
+      "loss": 0.3791,
+      "step": 187750
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.003943267824207e-06,
+      "loss": 0.3155,
+      "step": 187760
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.961542538531663e-06,
+      "loss": 0.4258,
+      "step": 187770
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.919141809239119e-06,
+      "loss": 0.4531,
+      "step": 187780
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.876741079946575e-06,
+      "loss": 0.4223,
+      "step": 187790
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.834340350654032e-06,
+      "loss": 0.3574,
+      "step": 187800
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.7919396213614874e-06,
+      "loss": 0.3339,
+      "step": 187810
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.749538892068944e-06,
+      "loss": 0.3279,
+      "step": 187820
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.7071381627764e-06,
+      "loss": 0.318,
+      "step": 187830
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.6647374334838565e-06,
+      "loss": 0.3657,
+      "step": 187840
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.622336704191312e-06,
+      "loss": 0.388,
+      "step": 187850
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.579935974898769e-06,
+      "loss": 0.3398,
+      "step": 187860
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.537535245606225e-06,
+      "loss": 0.4048,
+      "step": 187870
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.495134516313681e-06,
+      "loss": 0.354,
+      "step": 187880
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.452733787021137e-06,
+      "loss": 0.3165,
+      "step": 187890
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.410333057728594e-06,
+      "loss": 0.4625,
+      "step": 187900
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.367932328436049e-06,
+      "loss": 0.332,
+      "step": 187910
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.325531599143505e-06,
+      "loss": 0.3661,
+      "step": 187920
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.283130869850962e-06,
+      "loss": 0.3643,
+      "step": 187930
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.240730140558418e-06,
+      "loss": 0.3534,
+      "step": 187940
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.198329411265874e-06,
+      "loss": 0.3512,
+      "step": 187950
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.15592868197333e-06,
+      "loss": 0.4055,
+      "step": 187960
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.113527952680787e-06,
+      "loss": 0.3783,
+      "step": 187970
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.071127223388242e-06,
+      "loss": 0.3498,
+      "step": 187980
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.028726494095699e-06,
+      "loss": 0.3665,
+      "step": 187990
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.986325764803155e-06,
+      "loss": 0.4156,
+      "step": 188000
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.9439250355106114e-06,
+      "loss": 0.3749,
+      "step": 188010
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.901524306218067e-06,
+      "loss": 0.4014,
+      "step": 188020
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.859123576925524e-06,
+      "loss": 0.4212,
+      "step": 188030
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.81672284763298e-06,
+      "loss": 0.3886,
+      "step": 188040
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.774322118340436e-06,
+      "loss": 0.363,
+      "step": 188050
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.731921389047892e-06,
+      "loss": 0.3829,
+      "step": 188060
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.689520659755349e-06,
+      "loss": 0.4149,
+      "step": 188070
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.6471199304628044e-06,
+      "loss": 0.4315,
+      "step": 188080
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.60471920117026e-06,
+      "loss": 0.3514,
+      "step": 188090
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.562318471877716e-06,
+      "loss": 0.3601,
+      "step": 188100
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.519917742585173e-06,
+      "loss": 0.3208,
+      "step": 188110
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.477517013292628e-06,
+      "loss": 0.3671,
+      "step": 188120
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.435116284000085e-06,
+      "loss": 0.347,
+      "step": 188130
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.392715554707541e-06,
+      "loss": 0.3588,
+      "step": 188140
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.350314825414997e-06,
+      "loss": 0.3675,
+      "step": 188150
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.307914096122453e-06,
+      "loss": 0.362,
+      "step": 188160
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.26551336682991e-06,
+      "loss": 0.3806,
+      "step": 188170
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.223112637537366e-06,
+      "loss": 0.399,
+      "step": 188180
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.180711908244822e-06,
+      "loss": 0.4158,
+      "step": 188190
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.138311178952278e-06,
+      "loss": 0.4081,
+      "step": 188200
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.095910449659735e-06,
+      "loss": 0.4062,
+      "step": 188210
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.05350972036719e-06,
+      "loss": 0.4085,
+      "step": 188220
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.011108991074647e-06,
+      "loss": 0.348,
+      "step": 188230
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.968708261782103e-06,
+      "loss": 0.3183,
+      "step": 188240
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.9263075324895594e-06,
+      "loss": 0.3781,
+      "step": 188250
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.883906803197015e-06,
+      "loss": 0.3701,
+      "step": 188260
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.841506073904471e-06,
+      "loss": 0.2834,
+      "step": 188270
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.799105344611927e-06,
+      "loss": 0.3667,
+      "step": 188280
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.7567046153193834e-06,
+      "loss": 0.3851,
+      "step": 188290
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.7143038860268396e-06,
+      "loss": 0.3611,
+      "step": 188300
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.6719031567342958e-06,
+      "loss": 0.3603,
+      "step": 188310
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.629502427441752e-06,
+      "loss": 0.3429,
+      "step": 188320
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.587101698149208e-06,
+      "loss": 0.3994,
+      "step": 188330
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.5447009688566644e-06,
+      "loss": 0.4298,
+      "step": 188340
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.5023002395641206e-06,
+      "loss": 0.3748,
+      "step": 188350
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.459899510271577e-06,
+      "loss": 0.3241,
+      "step": 188360
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.417498780979033e-06,
+      "loss": 0.3628,
+      "step": 188370
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.375098051686489e-06,
+      "loss": 0.3873,
+      "step": 188380
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.3326973223939454e-06,
+      "loss": 0.3908,
+      "step": 188390
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.2902965931014016e-06,
+      "loss": 0.3557,
+      "step": 188400
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.247895863808858e-06,
+      "loss": 0.3286,
+      "step": 188410
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.205495134516314e-06,
+      "loss": 0.3564,
+      "step": 188420
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.16309440522377e-06,
+      "loss": 0.3343,
+      "step": 188430
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.120693675931226e-06,
+      "loss": 0.4041,
+      "step": 188440
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.078292946638682e-06,
+      "loss": 0.3524,
+      "step": 188450
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.0358922173461384e-06,
+      "loss": 0.3915,
+      "step": 188460
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.9934914880535946e-06,
+      "loss": 0.3694,
+      "step": 188470
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.9510907587610508e-06,
+      "loss": 0.3221,
+      "step": 188480
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.908690029468507e-06,
+      "loss": 0.3741,
+      "step": 188490
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.866289300175963e-06,
+      "loss": 0.3468,
+      "step": 188500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.8238885708834194e-06,
+      "loss": 0.3717,
+      "step": 188510
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.7814878415908756e-06,
+      "loss": 0.3878,
+      "step": 188520
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.7390871122983318e-06,
+      "loss": 0.3663,
+      "step": 188530
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.696686383005788e-06,
+      "loss": 0.3775,
+      "step": 188540
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.654285653713244e-06,
+      "loss": 0.3357,
+      "step": 188550
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.6118849244207004e-06,
+      "loss": 0.3769,
+      "step": 188560
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.5694841951281566e-06,
+      "loss": 0.3984,
+      "step": 188570
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.5270834658356124e-06,
+      "loss": 0.3586,
+      "step": 188580
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.4846827365430686e-06,
+      "loss": 0.3351,
+      "step": 188590
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.4422820072505248e-06,
+      "loss": 0.3412,
+      "step": 188600
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.399881277957981e-06,
+      "loss": 0.3288,
+      "step": 188610
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.357480548665437e-06,
+      "loss": 0.3391,
+      "step": 188620
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.3150798193728934e-06,
+      "loss": 0.3556,
+      "step": 188630
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.2726790900803496e-06,
+      "loss": 0.3307,
+      "step": 188640
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.2302783607878058e-06,
+      "loss": 0.3969,
+      "step": 188650
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.187877631495262e-06,
+      "loss": 0.3858,
+      "step": 188660
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.1454769022027178e-06,
+      "loss": 0.386,
+      "step": 188670
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.103076172910174e-06,
+      "loss": 0.3921,
+      "step": 188680
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.06067544361763e-06,
+      "loss": 0.4552,
+      "step": 188690
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.0182747143250864e-06,
+      "loss": 0.4191,
+      "step": 188700
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.9758739850325426e-06,
+      "loss": 0.4667,
+      "step": 188710
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.9334732557399988e-06,
+      "loss": 0.4009,
+      "step": 188720
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.891072526447455e-06,
+      "loss": 0.3635,
+      "step": 188730
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.8486717971549112e-06,
+      "loss": 0.3378,
+      "step": 188740
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.8062710678623676e-06,
+      "loss": 0.4412,
+      "step": 188750
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.7638703385698233e-06,
+      "loss": 0.3824,
+      "step": 188760
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.7214696092772796e-06,
+      "loss": 0.4426,
+      "step": 188770
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.6790688799847358e-06,
+      "loss": 0.3922,
+      "step": 188780
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.636668150692192e-06,
+      "loss": 0.292,
+      "step": 188790
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.5942674213996482e-06,
+      "loss": 0.35,
+      "step": 188800
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.5518666921071044e-06,
+      "loss": 0.346,
+      "step": 188810
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.5094659628145606e-06,
+      "loss": 0.3384,
+      "step": 188820
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.4670652335220165e-06,
+      "loss": 0.4101,
+      "step": 188830
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.4246645042294727e-06,
+      "loss": 0.4087,
+      "step": 188840
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.382263774936929e-06,
+      "loss": 0.3239,
+      "step": 188850
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.3398630456443851e-06,
+      "loss": 0.36,
+      "step": 188860
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.2974623163518411e-06,
+      "loss": 0.3982,
+      "step": 188870
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.2550615870592973e-06,
+      "loss": 0.357,
+      "step": 188880
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.2126608577667535e-06,
+      "loss": 0.3256,
+      "step": 188890
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.17026012847421e-06,
+      "loss": 0.3523,
+      "step": 188900
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.1278593991816662e-06,
+      "loss": 0.3405,
+      "step": 188910
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0854586698891221e-06,
+      "loss": 0.3408,
+      "step": 188920
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0430579405965783e-06,
+      "loss": 0.3336,
+      "step": 188930
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0006572113040345e-06,
+      "loss": 0.3523,
+      "step": 188940
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.582564820114907e-07,
+      "loss": 0.4189,
+      "step": 188950
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.158557527189467e-07,
+      "loss": 0.3673,
+      "step": 188960
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 8.734550234264029e-07,
+      "loss": 0.3636,
+      "step": 188970
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 8.310542941338591e-07,
+      "loss": 0.3977,
+      "step": 188980
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 7.886535648413153e-07,
+      "loss": 0.3785,
+      "step": 188990
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 7.462528355487714e-07,
+      "loss": 0.4021,
+      "step": 189000
+    }
+  ],
+  "max_steps": 189176,
+  "num_train_epochs": 8,
+  "total_flos": 9.214706529557545e+17,
+  "trial_name": null,
+  "trial_params": null
+}