diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,26032 @@
+{
+  "best_metric": 0.8022026419639587,
+  "best_model_checkpoint": "./wikiart/checkpoint-43270",
+  "epoch": 5.0,
+  "global_step": 43270,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9995377859949158e-05,
+      "loss": 2.3889,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9990755719898315e-05,
+      "loss": 2.3964,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.998613357984747e-05,
+      "loss": 2.3587,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9981511439796628e-05,
+      "loss": 2.3379,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997688929974578e-05,
+      "loss": 2.2528,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997226715969494e-05,
+      "loss": 2.3131,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9967645019644098e-05,
+      "loss": 2.2779,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.996302287959325e-05,
+      "loss": 2.2319,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9958400739542408e-05,
+      "loss": 2.2564,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9953778599491568e-05,
+      "loss": 2.2565,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.994915645944072e-05,
+      "loss": 2.2063,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9944534319389878e-05,
+      "loss": 2.1082,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9939912179339035e-05,
+      "loss": 2.1605,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.993529003928819e-05,
+      "loss": 2.0891,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9930667899237348e-05,
+      "loss": 2.1729,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9926045759186505e-05,
+      "loss": 2.1034,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.992142361913566e-05,
+      "loss": 2.1315,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9916801479084818e-05,
+      "loss": 2.0766,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9912179339033975e-05,
+      "loss": 1.993,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.990755719898313e-05,
+      "loss": 1.8563,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9902935058932288e-05,
+      "loss": 1.8833,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9898312918881445e-05,
+      "loss": 2.0582,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.98936907788306e-05,
+      "loss": 1.9494,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9889068638779755e-05,
+      "loss": 1.9579,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9884446498728915e-05,
+      "loss": 1.8937,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.987982435867807e-05,
+      "loss": 1.8906,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9875202218627225e-05,
+      "loss": 1.9543,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.987058007857638e-05,
+      "loss": 1.745,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.986595793852554e-05,
+      "loss": 1.7936,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9861335798474695e-05,
+      "loss": 1.7922,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.985671365842385e-05,
+      "loss": 1.8234,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9852091518373008e-05,
+      "loss": 1.8136,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9847469378322165e-05,
+      "loss": 1.7941,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.984284723827132e-05,
+      "loss": 1.8309,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9838225098220478e-05,
+      "loss": 1.7168,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9833602958169635e-05,
+      "loss": 1.7522,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.982898081811879e-05,
+      "loss": 1.803,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9824358678067948e-05,
+      "loss": 1.7825,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98197365380171e-05,
+      "loss": 1.7336,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.981511439796626e-05,
+      "loss": 1.8995,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9810492257915418e-05,
+      "loss": 1.759,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.980587011786457e-05,
+      "loss": 1.7984,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9801247977813728e-05,
+      "loss": 1.7064,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9796625837762888e-05,
+      "loss": 1.71,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9792003697712044e-05,
+      "loss": 1.7663,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9787381557661198e-05,
+      "loss": 1.7312,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9782759417610354e-05,
+      "loss": 1.6678,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.977813727755951e-05,
+      "loss": 1.6362,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9773515137508668e-05,
+      "loss": 1.6622,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9768892997457824e-05,
+      "loss": 1.5505,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.976427085740698e-05,
+      "loss": 1.5995,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9759648717356138e-05,
+      "loss": 1.437,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9755026577305294e-05,
+      "loss": 1.7087,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.975040443725445e-05,
+      "loss": 1.6499,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9745782297203608e-05,
+      "loss": 1.6227,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9741160157152764e-05,
+      "loss": 1.5741,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973653801710192e-05,
+      "loss": 1.5967,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9731915877051074e-05,
+      "loss": 1.5143,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9727293737000234e-05,
+      "loss": 1.5463,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972267159694939e-05,
+      "loss": 1.6656,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9718049456898544e-05,
+      "loss": 1.4812,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97134273168477e-05,
+      "loss": 1.5896,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9708805176796857e-05,
+      "loss": 1.5029,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9704183036746014e-05,
+      "loss": 1.5234,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969956089669517e-05,
+      "loss": 1.6496,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9694938756644327e-05,
+      "loss": 1.5465,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9690316616593484e-05,
+      "loss": 1.4895,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968569447654264e-05,
+      "loss": 1.6183,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9681072336491797e-05,
+      "loss": 1.5208,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9676450196440954e-05,
+      "loss": 1.307,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967182805639011e-05,
+      "loss": 1.4443,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9667205916339267e-05,
+      "loss": 1.4331,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9662583776288424e-05,
+      "loss": 1.3673,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965796163623758e-05,
+      "loss": 1.4723,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9653339496186737e-05,
+      "loss": 1.2713,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9648717356135894e-05,
+      "loss": 1.5305,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9644095216085047e-05,
+      "loss": 1.5498,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9639473076034204e-05,
+      "loss": 1.5565,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9634850935983364e-05,
+      "loss": 1.3136,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9630228795932517e-05,
+      "loss": 1.3937,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9625606655881674e-05,
+      "loss": 1.506,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962098451583083e-05,
+      "loss": 1.4627,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9616362375779987e-05,
+      "loss": 1.4754,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9611740235729144e-05,
+      "loss": 1.3804,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.96071180956783e-05,
+      "loss": 1.4877,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9602495955627457e-05,
+      "loss": 1.4336,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9597873815576614e-05,
+      "loss": 1.4304,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959325167552577e-05,
+      "loss": 1.4311,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9588629535474927e-05,
+      "loss": 1.422,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9584007395424084e-05,
+      "loss": 1.4656,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.957938525537324e-05,
+      "loss": 1.3024,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9574763115322394e-05,
+      "loss": 1.3572,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.957014097527155e-05,
+      "loss": 1.3245,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.956551883522071e-05,
+      "loss": 1.486,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9560896695169864e-05,
+      "loss": 1.3922,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.955627455511902e-05,
+      "loss": 1.3372,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9551652415068177e-05,
+      "loss": 1.4009,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9547030275017334e-05,
+      "loss": 1.3173,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.954240813496649e-05,
+      "loss": 1.3584,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537785994915647e-05,
+      "loss": 1.2217,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533163854864804e-05,
+      "loss": 1.6893,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.952854171481396e-05,
+      "loss": 1.3965,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9523919574763117e-05,
+      "loss": 1.2936,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9519297434712274e-05,
+      "loss": 1.3456,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.951467529466143e-05,
+      "loss": 1.5588,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9510053154610587e-05,
+      "loss": 1.3207,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9505431014559744e-05,
+      "loss": 1.2455,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95008088745089e-05,
+      "loss": 1.3793,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496186734458057e-05,
+      "loss": 1.3548,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491564594407214e-05,
+      "loss": 1.501,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486942454356367e-05,
+      "loss": 1.4357,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482320314305524e-05,
+      "loss": 1.2892,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477698174254684e-05,
+      "loss": 1.3757,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473076034203837e-05,
+      "loss": 1.383,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468453894152993e-05,
+      "loss": 1.3751,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946383175410215e-05,
+      "loss": 1.4045,
+      "step": 1160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9459209614051307e-05,
+      "loss": 1.3845,
+      "step": 1170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9454587474000463e-05,
+      "loss": 1.348,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.944996533394962e-05,
+      "loss": 1.423,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9445343193898777e-05,
+      "loss": 1.3571,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9440721053847933e-05,
+      "loss": 1.3323,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.943609891379709e-05,
+      "loss": 1.3282,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9431476773746247e-05,
+      "loss": 1.2862,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426854633695403e-05,
+      "loss": 1.5374,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942223249364456e-05,
+      "loss": 1.3002,
+      "step": 1250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9417610353593713e-05,
+      "loss": 1.2503,
+      "step": 1260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.941298821354287e-05,
+      "loss": 1.4021,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.940836607349203e-05,
+      "loss": 1.2019,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9403743933441187e-05,
+      "loss": 1.443,
+      "step": 1290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.939912179339034e-05,
+      "loss": 1.402,
+      "step": 1300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9394499653339497e-05,
+      "loss": 1.4088,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9389877513288657e-05,
+      "loss": 1.3539,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.938525537323781e-05,
+      "loss": 1.269,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9380633233186967e-05,
+      "loss": 1.3832,
+      "step": 1340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9376011093136123e-05,
+      "loss": 1.2932,
+      "step": 1350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.937138895308528e-05,
+      "loss": 1.2389,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9366766813034437e-05,
+      "loss": 1.4409,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9362144672983593e-05,
+      "loss": 1.1224,
+      "step": 1380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.935752253293275e-05,
+      "loss": 1.2164,
+      "step": 1390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9352900392881907e-05,
+      "loss": 1.2708,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9348278252831063e-05,
+      "loss": 1.1321,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9343656112780216e-05,
+      "loss": 1.2621,
+      "step": 1420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9339033972729377e-05,
+      "loss": 1.3742,
+      "step": 1430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9334411832678533e-05,
+      "loss": 1.2819,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9329789692627686e-05,
+      "loss": 1.2043,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9325167552576843e-05,
+      "loss": 1.3292,
+      "step": 1460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9320545412526003e-05,
+      "loss": 1.3365,
+      "step": 1470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9315923272475156e-05,
+      "loss": 1.2925,
+      "step": 1480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9311301132424313e-05,
+      "loss": 1.4563,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.930667899237347e-05,
+      "loss": 1.4431,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9302056852322626e-05,
+      "loss": 1.2683,
+      "step": 1510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9297434712271783e-05,
+      "loss": 1.2307,
+      "step": 1520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.929281257222094e-05,
+      "loss": 1.521,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9288190432170096e-05,
+      "loss": 1.3066,
+      "step": 1540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9283568292119253e-05,
+      "loss": 1.3608,
+      "step": 1550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.927894615206841e-05,
+      "loss": 1.398,
+      "step": 1560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9274324012017566e-05,
+      "loss": 1.2869,
+      "step": 1570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9269701871966723e-05,
+      "loss": 1.3803,
+      "step": 1580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.926507973191588e-05,
+      "loss": 1.4582,
+      "step": 1590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9260457591865036e-05,
+      "loss": 1.4873,
+      "step": 1600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.925583545181419e-05,
+      "loss": 1.193,
+      "step": 1610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.925121331176335e-05,
+      "loss": 1.2259,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9246591171712506e-05,
+      "loss": 1.325,
+      "step": 1630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.924196903166166e-05,
+      "loss": 1.1706,
+      "step": 1640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9237346891610816e-05,
+      "loss": 1.2486,
+      "step": 1650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9232724751559976e-05,
+      "loss": 1.3872,
+      "step": 1660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.922810261150913e-05,
+      "loss": 1.3436,
+      "step": 1670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9223480471458286e-05,
+      "loss": 1.2404,
+      "step": 1680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9218858331407443e-05,
+      "loss": 1.2883,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.92142361913566e-05,
+      "loss": 1.2494,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9209614051305756e-05,
+      "loss": 1.267,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9204991911254913e-05,
+      "loss": 1.016,
+      "step": 1720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.920036977120407e-05,
+      "loss": 1.3929,
+      "step": 1730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9195747631153226e-05,
+      "loss": 1.1148,
+      "step": 1740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9191125491102383e-05,
+      "loss": 1.3616,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9186503351051536e-05,
+      "loss": 1.3019,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9181881211000696e-05,
+      "loss": 1.2434,
+      "step": 1770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9177259070949853e-05,
+      "loss": 1.1432,
+      "step": 1780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9172636930899006e-05,
+      "loss": 1.3384,
+      "step": 1790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9168014790848163e-05,
+      "loss": 1.1971,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9163392650797323e-05,
+      "loss": 1.1846,
+      "step": 1810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9158770510746476e-05,
+      "loss": 1.2765,
+      "step": 1820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9154148370695633e-05,
+      "loss": 1.102,
+      "step": 1830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.914952623064479e-05,
+      "loss": 1.4218,
+      "step": 1840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9144904090593946e-05,
+      "loss": 1.1716,
+      "step": 1850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9140281950543103e-05,
+      "loss": 1.1829,
+      "step": 1860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.913565981049226e-05,
+      "loss": 1.3906,
+      "step": 1870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9131037670441416e-05,
+      "loss": 1.1364,
+      "step": 1880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9126415530390573e-05,
+      "loss": 1.2781,
+      "step": 1890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.912179339033973e-05,
+      "loss": 1.3724,
+      "step": 1900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9117171250288886e-05,
+      "loss": 1.2422,
+      "step": 1910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9112549110238043e-05,
+      "loss": 1.3088,
+      "step": 1920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.91079269701872e-05,
+      "loss": 1.2479,
+      "step": 1930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9103304830136356e-05,
+      "loss": 1.2599,
+      "step": 1940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.909868269008551e-05,
+      "loss": 1.2685,
+      "step": 1950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.909406055003467e-05,
+      "loss": 1.3681,
+      "step": 1960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9089438409983826e-05,
+      "loss": 1.1765,
+      "step": 1970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.908481626993298e-05,
+      "loss": 1.3525,
+      "step": 1980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9080194129882136e-05,
+      "loss": 1.2632,
+      "step": 1990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9075571989831292e-05,
+      "loss": 1.2889,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.907094984978045e-05,
+      "loss": 1.2279,
+      "step": 2010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9066327709729606e-05,
+      "loss": 1.372,
+      "step": 2020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9061705569678762e-05,
+      "loss": 1.3015,
+      "step": 2030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.905708342962792e-05,
+      "loss": 1.275,
+      "step": 2040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9052461289577076e-05,
+      "loss": 1.3691,
+      "step": 2050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9047839149526232e-05,
+      "loss": 1.4017,
+      "step": 2060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.904321700947539e-05,
+      "loss": 1.2366,
+      "step": 2070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9038594869424546e-05,
+      "loss": 1.172,
+      "step": 2080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9033972729373702e-05,
+      "loss": 1.0791,
+      "step": 2090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9029350589322856e-05,
+      "loss": 1.1896,
+      "step": 2100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9024728449272016e-05,
+      "loss": 1.1924,
+      "step": 2110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9020106309221172e-05,
+      "loss": 1.156,
+      "step": 2120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.901548416917033e-05,
+      "loss": 1.3714,
+      "step": 2130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9010862029119482e-05,
+      "loss": 1.2248,
+      "step": 2140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9006239889068642e-05,
+      "loss": 1.0685,
+      "step": 2150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.90016177490178e-05,
+      "loss": 1.1972,
+      "step": 2160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8996995608966952e-05,
+      "loss": 1.3127,
+      "step": 2170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.899237346891611e-05,
+      "loss": 1.1867,
+      "step": 2180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8987751328865266e-05,
+      "loss": 1.2504,
+      "step": 2190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8983129188814422e-05,
+      "loss": 1.5356,
+      "step": 2200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.897850704876358e-05,
+      "loss": 1.2071,
+      "step": 2210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8973884908712735e-05,
+      "loss": 1.2274,
+      "step": 2220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8969262768661892e-05,
+      "loss": 1.1609,
+      "step": 2230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.896464062861105e-05,
+      "loss": 1.1606,
+      "step": 2240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8960018488560205e-05,
+      "loss": 1.1231,
+      "step": 2250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8955396348509362e-05,
+      "loss": 1.1841,
+      "step": 2260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.895077420845852e-05,
+      "loss": 0.9819,
+      "step": 2270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8946152068407675e-05,
+      "loss": 1.0814,
+      "step": 2280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.894152992835683e-05,
+      "loss": 1.1365,
+      "step": 2290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.893690778830599e-05,
+      "loss": 1.5023,
+      "step": 2300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8932285648255145e-05,
+      "loss": 1.1865,
+      "step": 2310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.89276635082043e-05,
+      "loss": 1.2644,
+      "step": 2320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8923041368153455e-05,
+      "loss": 1.3072,
+      "step": 2330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8918419228102612e-05,
+      "loss": 1.1206,
+      "step": 2340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.891379708805177e-05,
+      "loss": 1.2632,
+      "step": 2350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8909174948000925e-05,
+      "loss": 1.292,
+      "step": 2360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8904552807950082e-05,
+      "loss": 1.1591,
+      "step": 2370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.889993066789924e-05,
+      "loss": 1.0314,
+      "step": 2380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8895308527848395e-05,
+      "loss": 1.1041,
+      "step": 2390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8890686387797552e-05,
+      "loss": 1.206,
+      "step": 2400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.888606424774671e-05,
+      "loss": 1.1292,
+      "step": 2410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8881442107695865e-05,
+      "loss": 1.382,
+      "step": 2420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8876819967645022e-05,
+      "loss": 1.2121,
+      "step": 2430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.887219782759418e-05,
+      "loss": 1.2414,
+      "step": 2440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8867575687543335e-05,
+      "loss": 1.084,
+      "step": 2450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8862953547492492e-05,
+      "loss": 1.1946,
+      "step": 2460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.885833140744165e-05,
+      "loss": 1.1969,
+      "step": 2470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8853709267390802e-05,
+      "loss": 1.0586,
+      "step": 2480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.884908712733996e-05,
+      "loss": 1.0634,
+      "step": 2490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.884446498728912e-05,
+      "loss": 1.1831,
+      "step": 2500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8839842847238272e-05,
+      "loss": 1.1763,
+      "step": 2510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.883522070718743e-05,
+      "loss": 1.3915,
+      "step": 2520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8830598567136585e-05,
+      "loss": 1.3104,
+      "step": 2530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8825976427085742e-05,
+      "loss": 0.988,
+      "step": 2540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.88213542870349e-05,
+      "loss": 1.0718,
+      "step": 2550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8816732146984055e-05,
+      "loss": 1.1107,
+      "step": 2560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8812110006933212e-05,
+      "loss": 1.2246,
+      "step": 2570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.880748786688237e-05,
+      "loss": 1.095,
+      "step": 2580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8802865726831525e-05,
+      "loss": 1.0693,
+      "step": 2590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.879824358678068e-05,
+      "loss": 1.3382,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.879362144672984e-05,
+      "loss": 1.1433,
+      "step": 2610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8788999306678995e-05,
+      "loss": 1.0758,
+      "step": 2620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8784377166628148e-05,
+      "loss": 1.1923,
+      "step": 2630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8779755026577305e-05,
+      "loss": 1.0192,
+      "step": 2640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8775132886526465e-05,
+      "loss": 1.0242,
+      "step": 2650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8770510746475618e-05,
+      "loss": 0.9666,
+      "step": 2660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8765888606424775e-05,
+      "loss": 1.3071,
+      "step": 2670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.876126646637393e-05,
+      "loss": 1.0935,
+      "step": 2680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.875664432632309e-05,
+      "loss": 1.1457,
+      "step": 2690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8752022186272245e-05,
+      "loss": 1.2322,
+      "step": 2700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.87474000462214e-05,
+      "loss": 1.1762,
+      "step": 2710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8742777906170558e-05,
+      "loss": 1.2718,
+      "step": 2720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8738155766119715e-05,
+      "loss": 1.2882,
+      "step": 2730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.873353362606887e-05,
+      "loss": 1.3582,
+      "step": 2740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8728911486018028e-05,
+      "loss": 1.0482,
+      "step": 2750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8724289345967185e-05,
+      "loss": 0.9126,
+      "step": 2760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.871966720591634e-05,
+      "loss": 1.1649,
+      "step": 2770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8715045065865498e-05,
+      "loss": 1.1562,
+      "step": 2780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.871042292581465e-05,
+      "loss": 1.1534,
+      "step": 2790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.870580078576381e-05,
+      "loss": 1.1332,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8701178645712968e-05,
+      "loss": 1.0996,
+      "step": 2810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.869655650566212e-05,
+      "loss": 1.2614,
+      "step": 2820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8691934365611278e-05,
+      "loss": 1.2897,
+      "step": 2830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8687312225560438e-05,
+      "loss": 1.3399,
+      "step": 2840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.868269008550959e-05,
+      "loss": 1.2899,
+      "step": 2850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8678067945458748e-05,
+      "loss": 1.213,
+      "step": 2860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8673445805407905e-05,
+      "loss": 1.1954,
+      "step": 2870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.866882366535706e-05,
+      "loss": 1.1798,
+      "step": 2880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8664201525306218e-05,
+      "loss": 1.026,
+      "step": 2890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8659579385255375e-05,
+      "loss": 1.3567,
+      "step": 2900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.865495724520453e-05,
+      "loss": 1.0325,
+      "step": 2910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8650335105153688e-05,
+      "loss": 1.2255,
+      "step": 2920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8645712965102845e-05,
+      "loss": 1.2002,
+      "step": 2930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8641090825051998e-05,
+      "loss": 1.1264,
+      "step": 2940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8636468685001158e-05,
+      "loss": 1.4016,
+      "step": 2950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8631846544950315e-05,
+      "loss": 1.1077,
+      "step": 2960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.862722440489947e-05,
+      "loss": 1.5143,
+      "step": 2970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8622602264848624e-05,
+      "loss": 1.0623,
+      "step": 2980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8617980124797785e-05,
+      "loss": 1.0831,
+      "step": 2990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.861335798474694e-05,
+      "loss": 1.2574,
+      "step": 3000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8608735844696094e-05,
+      "loss": 1.1072,
+      "step": 3010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.860411370464525e-05,
+      "loss": 1.0831,
+      "step": 3020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.859949156459441e-05,
+      "loss": 1.3604,
+      "step": 3030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8594869424543564e-05,
+      "loss": 1.2112,
+      "step": 3040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.859024728449272e-05,
+      "loss": 1.1375,
+      "step": 3050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8585625144441878e-05,
+      "loss": 1.1787,
+      "step": 3060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8581003004391034e-05,
+      "loss": 1.1487,
+      "step": 3070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.857638086434019e-05,
+      "loss": 1.1526,
+      "step": 3080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8571758724289348e-05,
+      "loss": 1.1473,
+      "step": 3090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8567136584238504e-05,
+      "loss": 1.0194,
+      "step": 3100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.856251444418766e-05,
+      "loss": 0.9566,
+      "step": 3110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8557892304136818e-05,
+      "loss": 1.3389,
+      "step": 3120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.855327016408597e-05,
+      "loss": 1.2753,
+      "step": 3130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.854864802403513e-05,
+      "loss": 1.1823,
+      "step": 3140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8544025883984288e-05,
+      "loss": 1.2134,
+      "step": 3150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.853940374393344e-05,
+      "loss": 1.0395,
+      "step": 3160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8534781603882598e-05,
+      "loss": 0.9284,
+      "step": 3170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8530159463831758e-05,
+      "loss": 1.0263,
+      "step": 3180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.852553732378091e-05,
+      "loss": 1.0749,
+      "step": 3190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8520915183730068e-05,
+      "loss": 1.0414,
+      "step": 3200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8516293043679224e-05,
+      "loss": 1.2108,
+      "step": 3210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.851167090362838e-05,
+      "loss": 1.264,
+      "step": 3220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8507048763577538e-05,
+      "loss": 1.1303,
+      "step": 3230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8502426623526694e-05,
+      "loss": 1.2682,
+      "step": 3240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.849780448347585e-05,
+      "loss": 1.0847,
+      "step": 3250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8493182343425007e-05,
+      "loss": 1.028,
+      "step": 3260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8488560203374164e-05,
+      "loss": 1.0998,
+      "step": 3270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.848393806332332e-05,
+      "loss": 1.2558,
+      "step": 3280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8479315923272477e-05,
+      "loss": 1.191,
+      "step": 3290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8474693783221634e-05,
+      "loss": 1.0555,
+      "step": 3300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.847007164317079e-05,
+      "loss": 1.2603,
+      "step": 3310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8465449503119944e-05,
+      "loss": 0.9417,
+      "step": 3320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8460827363069104e-05,
+      "loss": 1.1519,
+      "step": 3330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.845620522301826e-05,
+      "loss": 1.3679,
+      "step": 3340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8451583082967414e-05,
+      "loss": 1.3549,
+      "step": 3350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.844696094291657e-05,
+      "loss": 1.0626,
+      "step": 3360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.844233880286573e-05,
+      "loss": 1.0574,
+      "step": 3370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8437716662814884e-05,
+      "loss": 1.0951,
+      "step": 3380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.843309452276404e-05,
+      "loss": 1.2002,
+      "step": 3390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8428472382713197e-05,
+      "loss": 1.2621,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8423850242662354e-05,
+      "loss": 1.1487,
+      "step": 3410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.841922810261151e-05,
+      "loss": 1.2417,
+      "step": 3420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8414605962560667e-05,
+      "loss": 1.1242,
+      "step": 3430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8409983822509824e-05,
+      "loss": 1.1101,
+      "step": 3440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.840536168245898e-05,
+      "loss": 1.0401,
+      "step": 3450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8400739542408137e-05,
+      "loss": 1.1815,
+      "step": 3460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.839611740235729e-05,
+      "loss": 1.2077,
+      "step": 3470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.839149526230645e-05,
+      "loss": 1.0657,
+      "step": 3480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8386873122255607e-05,
+      "loss": 1.3498,
+      "step": 3490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.838225098220476e-05,
+      "loss": 1.1172,
+      "step": 3500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8377628842153917e-05,
+      "loss": 1.1106,
+      "step": 3510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8373006702103077e-05,
+      "loss": 0.9747,
+      "step": 3520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8368384562052234e-05,
+      "loss": 1.2039,
+      "step": 3530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8363762422001387e-05,
+      "loss": 1.27,
+      "step": 3540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8359140281950544e-05,
+      "loss": 1.3706,
+      "step": 3550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.83545181418997e-05,
+      "loss": 1.1361,
+      "step": 3560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8349896001848857e-05,
+      "loss": 1.2561,
+      "step": 3570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8345273861798014e-05,
+      "loss": 1.4084,
+      "step": 3580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.834065172174717e-05,
+      "loss": 1.3029,
+      "step": 3590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8336029581696327e-05,
+      "loss": 1.1308,
+      "step": 3600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8331407441645484e-05,
+      "loss": 0.9609,
+      "step": 3610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832678530159464e-05,
+      "loss": 1.1129,
+      "step": 3620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8322163161543797e-05,
+      "loss": 1.0129,
+      "step": 3630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8317541021492954e-05,
+      "loss": 0.9557,
+      "step": 3640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831291888144211e-05,
+      "loss": 1.0479,
+      "step": 3650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8308296741391264e-05,
+      "loss": 1.021,
+      "step": 3660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8303674601340424e-05,
+      "loss": 0.9833,
+      "step": 3670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829905246128958e-05,
+      "loss": 0.9384,
+      "step": 3680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8294430321238734e-05,
+      "loss": 0.9701,
+      "step": 3690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.828980818118789e-05,
+      "loss": 1.0501,
+      "step": 3700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8285186041137047e-05,
+      "loss": 1.1911,
+      "step": 3710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8280563901086204e-05,
+      "loss": 1.2042,
+      "step": 3720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.827594176103536e-05,
+      "loss": 1.2211,
+      "step": 3730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8271319620984517e-05,
+      "loss": 1.2117,
+      "step": 3740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8266697480933674e-05,
+      "loss": 1.0471,
+      "step": 3750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826207534088283e-05,
+      "loss": 1.2078,
+      "step": 3760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8257453200831987e-05,
+      "loss": 1.1027,
+      "step": 3770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8252831060781144e-05,
+      "loss": 1.1611,
+      "step": 3780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.82482089207303e-05,
+      "loss": 1.2885,
+      "step": 3790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8243586780679457e-05,
+      "loss": 1.2312,
+      "step": 3800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8238964640628613e-05,
+      "loss": 0.9937,
+      "step": 3810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.823434250057777e-05,
+      "loss": 1.2174,
+      "step": 3820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8229720360526927e-05,
+      "loss": 1.1851,
+      "step": 3830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8225098220476083e-05,
+      "loss": 1.1764,
+      "step": 3840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8220476080425237e-05,
+      "loss": 1.1064,
+      "step": 3850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8215853940374393e-05,
+      "loss": 0.9792,
+      "step": 3860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8211231800323553e-05,
+      "loss": 1.1031,
+      "step": 3870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8206609660272707e-05,
+      "loss": 1.294,
+      "step": 3880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8201987520221863e-05,
+      "loss": 1.2966,
+      "step": 3890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.819736538017102e-05,
+      "loss": 0.8264,
+      "step": 3900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8192743240120177e-05,
+      "loss": 1.2898,
+      "step": 3910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8188121100069333e-05,
+      "loss": 0.9579,
+      "step": 3920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.818349896001849e-05,
+      "loss": 1.0927,
+      "step": 3930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8178876819967647e-05,
+      "loss": 0.9784,
+      "step": 3940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8174254679916803e-05,
+      "loss": 1.2402,
+      "step": 3950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.816963253986596e-05,
+      "loss": 1.0832,
+      "step": 3960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8165010399815117e-05,
+      "loss": 1.4883,
+      "step": 3970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8160388259764273e-05,
+      "loss": 1.0722,
+      "step": 3980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.815576611971343e-05,
+      "loss": 1.2664,
+      "step": 3990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8151143979662583e-05,
+      "loss": 1.0371,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.814652183961174e-05,
+      "loss": 1.0832,
+      "step": 4010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.81418996995609e-05,
+      "loss": 1.1956,
+      "step": 4020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8137277559510053e-05,
+      "loss": 1.1934,
+      "step": 4030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.813265541945921e-05,
+      "loss": 1.1124,
+      "step": 4040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8128033279408366e-05,
+      "loss": 1.1333,
+      "step": 4050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8123411139357523e-05,
+      "loss": 0.9141,
+      "step": 4060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.811878899930668e-05,
+      "loss": 1.1591,
+      "step": 4070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8114166859255836e-05,
+      "loss": 1.0963,
+      "step": 4080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8109544719204993e-05,
+      "loss": 1.1133,
+      "step": 4090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.810492257915415e-05,
+      "loss": 1.1491,
+      "step": 4100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8100300439103306e-05,
+      "loss": 1.0735,
+      "step": 4110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8095678299052463e-05,
+      "loss": 1.2437,
+      "step": 4120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.809105615900162e-05,
+      "loss": 0.968,
+      "step": 4130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8086434018950776e-05,
+      "loss": 1.1369,
+      "step": 4140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8081811878899933e-05,
+      "loss": 1.1927,
+      "step": 4150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.807718973884909e-05,
+      "loss": 1.0901,
+      "step": 4160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8072567598798246e-05,
+      "loss": 1.2807,
+      "step": 4170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8067945458747403e-05,
+      "loss": 1.1302,
+      "step": 4180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8063323318696556e-05,
+      "loss": 1.0799,
+      "step": 4190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8058701178645713e-05,
+      "loss": 0.8996,
+      "step": 4200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8054079038594873e-05,
+      "loss": 1.1115,
+      "step": 4210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8049456898544026e-05,
+      "loss": 1.0048,
+      "step": 4220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8044834758493183e-05,
+      "loss": 0.8283,
+      "step": 4230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.804021261844234e-05,
+      "loss": 1.0445,
+      "step": 4240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8035590478391496e-05,
+      "loss": 0.9902,
+      "step": 4250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8030968338340653e-05,
+      "loss": 1.0734,
+      "step": 4260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.802634619828981e-05,
+      "loss": 1.0718,
+      "step": 4270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8021724058238966e-05,
+      "loss": 0.9862,
+      "step": 4280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8017101918188123e-05,
+      "loss": 1.1452,
+      "step": 4290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.801247977813728e-05,
+      "loss": 1.2783,
+      "step": 4300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8007857638086436e-05,
+      "loss": 1.1989,
+      "step": 4310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8003235498035593e-05,
+      "loss": 1.0156,
+      "step": 4320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.799861335798475e-05,
+      "loss": 1.0055,
+      "step": 4330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7993991217933903e-05,
+      "loss": 0.9824,
+      "step": 4340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.798936907788306e-05,
+      "loss": 1.1124,
+      "step": 4350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.798474693783222e-05,
+      "loss": 1.1933,
+      "step": 4360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7980124797781376e-05,
+      "loss": 1.1943,
+      "step": 4370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.797550265773053e-05,
+      "loss": 1.0805,
+      "step": 4380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7970880517679686e-05,
+      "loss": 1.0607,
+      "step": 4390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7966258377628846e-05,
+      "loss": 0.9657,
+      "step": 4400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7961636237578e-05,
+      "loss": 1.0422,
+      "step": 4410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7957014097527156e-05,
+      "loss": 1.1064,
+      "step": 4420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7952391957476313e-05,
+      "loss": 0.9067,
+      "step": 4430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.794776981742547e-05,
+      "loss": 1.1491,
+      "step": 4440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7943147677374626e-05,
+      "loss": 0.9649,
+      "step": 4450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7938525537323783e-05,
+      "loss": 1.0047,
+      "step": 4460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.793390339727294e-05,
+      "loss": 1.1943,
+      "step": 4470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7929281257222096e-05,
+      "loss": 1.2275,
+      "step": 4480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7924659117171253e-05,
+      "loss": 0.9264,
+      "step": 4490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7920036977120406e-05,
+      "loss": 1.4983,
+      "step": 4500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7915414837069566e-05,
+      "loss": 1.2524,
+      "step": 4510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7910792697018723e-05,
+      "loss": 1.0237,
+      "step": 4520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7906170556967876e-05,
+      "loss": 1.1045,
+      "step": 4530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7901548416917033e-05,
+      "loss": 0.9626,
+      "step": 4540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7896926276866193e-05,
+      "loss": 1.0998,
+      "step": 4550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7892304136815346e-05,
+      "loss": 1.1499,
+      "step": 4560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7887681996764502e-05,
+      "loss": 1.1892,
+      "step": 4570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.788305985671366e-05,
+      "loss": 0.8999,
+      "step": 4580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7878437716662816e-05,
+      "loss": 1.0934,
+      "step": 4590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7873815576611972e-05,
+      "loss": 0.9457,
+      "step": 4600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.786919343656113e-05,
+      "loss": 0.9881,
+      "step": 4610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7864571296510286e-05,
+      "loss": 1.2537,
+      "step": 4620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7859949156459442e-05,
+      "loss": 0.9357,
+      "step": 4630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.78553270164086e-05,
+      "loss": 1.0279,
+      "step": 4640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7850704876357756e-05,
+      "loss": 0.9582,
+      "step": 4650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7846082736306912e-05,
+      "loss": 1.2051,
+      "step": 4660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.784146059625607e-05,
+      "loss": 1.1061,
+      "step": 4670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7836838456205226e-05,
+      "loss": 1.0186,
+      "step": 4680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.783221631615438e-05,
+      "loss": 1.1878,
+      "step": 4690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.782759417610354e-05,
+      "loss": 1.2719,
+      "step": 4700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7822972036052696e-05,
+      "loss": 1.2793,
+      "step": 4710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.781834989600185e-05,
+      "loss": 1.3585,
+      "step": 4720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7813727755951006e-05,
+      "loss": 1.1022,
+      "step": 4730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7809105615900166e-05,
+      "loss": 1.0497,
+      "step": 4740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.780448347584932e-05,
+      "loss": 1.114,
+      "step": 4750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7799861335798476e-05,
+      "loss": 0.9732,
+      "step": 4760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7795239195747632e-05,
+      "loss": 1.1097,
+      "step": 4770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.779061705569679e-05,
+      "loss": 0.9982,
+      "step": 4780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7785994915645946e-05,
+      "loss": 1.0951,
+      "step": 4790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7781372775595102e-05,
+      "loss": 1.2558,
+      "step": 4800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.777675063554426e-05,
+      "loss": 1.042,
+      "step": 4810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7772128495493416e-05,
+      "loss": 1.3038,
+      "step": 4820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7767506355442572e-05,
+      "loss": 1.2353,
+      "step": 4830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7762884215391725e-05,
+      "loss": 1.1145,
+      "step": 4840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7758262075340885e-05,
+      "loss": 1.0235,
+      "step": 4850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7753639935290042e-05,
+      "loss": 0.9571,
+      "step": 4860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7749017795239195e-05,
+      "loss": 1.0169,
+      "step": 4870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7744395655188352e-05,
+      "loss": 1.038,
+      "step": 4880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7739773515137512e-05,
+      "loss": 1.2734,
+      "step": 4890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7735151375086665e-05,
+      "loss": 1.1351,
+      "step": 4900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7730529235035822e-05,
+      "loss": 1.2046,
+      "step": 4910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.772590709498498e-05,
+      "loss": 0.9644,
+      "step": 4920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7721284954934135e-05,
+      "loss": 1.1694,
+      "step": 4930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7716662814883292e-05,
+      "loss": 1.2234,
+      "step": 4940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.771204067483245e-05,
+      "loss": 1.1693,
+      "step": 4950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7707418534781605e-05,
+      "loss": 1.1737,
+      "step": 4960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7702796394730762e-05,
+      "loss": 1.0496,
+      "step": 4970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.769817425467992e-05,
+      "loss": 1.0657,
+      "step": 4980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7693552114629075e-05,
+      "loss": 1.0469,
+      "step": 4990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7688929974578232e-05,
+      "loss": 0.9483,
+      "step": 5000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.768430783452739e-05,
+      "loss": 1.0459,
+      "step": 5010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7679685694476545e-05,
+      "loss": 0.9762,
+      "step": 5020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.76750635544257e-05,
+      "loss": 0.9717,
+      "step": 5030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.767044141437486e-05,
+      "loss": 1.0767,
+      "step": 5040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7665819274324015e-05,
+      "loss": 1.0034,
+      "step": 5050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.766119713427317e-05,
+      "loss": 1.2245,
+      "step": 5060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7656574994222325e-05,
+      "loss": 1.1618,
+      "step": 5070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7651952854171482e-05,
+      "loss": 1.0739,
+      "step": 5080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.764733071412064e-05,
+      "loss": 1.0676,
+      "step": 5090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7642708574069795e-05,
+      "loss": 1.177,
+      "step": 5100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7638086434018952e-05,
+      "loss": 1.1787,
+      "step": 5110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.763346429396811e-05,
+      "loss": 1.1914,
+      "step": 5120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7628842153917265e-05,
+      "loss": 1.1025,
+      "step": 5130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7624220013866422e-05,
+      "loss": 1.4116,
+      "step": 5140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.761959787381558e-05,
+      "loss": 1.1818,
+      "step": 5150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7614975733764735e-05,
+      "loss": 1.073,
+      "step": 5160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7610353593713892e-05,
+      "loss": 0.9147,
+      "step": 5170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7605731453663045e-05,
+      "loss": 1.1985,
+      "step": 5180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7601109313612205e-05,
+      "loss": 1.0203,
+      "step": 5190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7596487173561362e-05,
+      "loss": 1.2276,
+      "step": 5200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.759186503351052e-05,
+      "loss": 1.1782,
+      "step": 5210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.758724289345967e-05,
+      "loss": 1.021,
+      "step": 5220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.758262075340883e-05,
+      "loss": 1.2512,
+      "step": 5230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.757799861335799e-05,
+      "loss": 0.8604,
+      "step": 5240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.757337647330714e-05,
+      "loss": 1.2508,
+      "step": 5250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7568754333256298e-05,
+      "loss": 0.9594,
+      "step": 5260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7564132193205455e-05,
+      "loss": 0.9819,
+      "step": 5270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.755951005315461e-05,
+      "loss": 1.0098,
+      "step": 5280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7554887913103768e-05,
+      "loss": 1.1188,
+      "step": 5290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7550265773052925e-05,
+      "loss": 1.0023,
+      "step": 5300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.754564363300208e-05,
+      "loss": 1.1674,
+      "step": 5310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7541021492951238e-05,
+      "loss": 1.3173,
+      "step": 5320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7536399352900395e-05,
+      "loss": 1.1943,
+      "step": 5330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.753177721284955e-05,
+      "loss": 1.2519,
+      "step": 5340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7527155072798708e-05,
+      "loss": 1.0812,
+      "step": 5350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7522532932747865e-05,
+      "loss": 1.0451,
+      "step": 5360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7517910792697018e-05,
+      "loss": 0.9264,
+      "step": 5370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7513288652646178e-05,
+      "loss": 1.0676,
+      "step": 5380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7508666512595335e-05,
+      "loss": 0.9933,
+      "step": 5390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7504044372544488e-05,
+      "loss": 1.0961,
+      "step": 5400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7499422232493645e-05,
+      "loss": 0.9239,
+      "step": 5410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.74948000924428e-05,
+      "loss": 1.23,
+      "step": 5420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7490177952391958e-05,
+      "loss": 1.0587,
+      "step": 5430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7485555812341115e-05,
+      "loss": 1.1866,
+      "step": 5440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.748093367229027e-05,
+      "loss": 0.9511,
+      "step": 5450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7476311532239428e-05,
+      "loss": 0.9471,
+      "step": 5460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7471689392188585e-05,
+      "loss": 0.9594,
+      "step": 5470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.746706725213774e-05,
+      "loss": 1.007,
+      "step": 5480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7462445112086898e-05,
+      "loss": 1.0517,
+      "step": 5490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7457822972036055e-05,
+      "loss": 1.0325,
+      "step": 5500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.745320083198521e-05,
+      "loss": 1.3203,
+      "step": 5510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7448578691934368e-05,
+      "loss": 0.9371,
+      "step": 5520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7443956551883525e-05,
+      "loss": 1.164,
+      "step": 5530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.743933441183268e-05,
+      "loss": 1.091,
+      "step": 5540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7434712271781838e-05,
+      "loss": 1.0484,
+      "step": 5550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.743009013173099e-05,
+      "loss": 1.1421,
+      "step": 5560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7425467991680148e-05,
+      "loss": 0.9402,
+      "step": 5570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7420845851629308e-05,
+      "loss": 1.2768,
+      "step": 5580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.741622371157846e-05,
+      "loss": 1.023,
+      "step": 5590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7411601571527618e-05,
+      "loss": 1.0531,
+      "step": 5600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7406979431476774e-05,
+      "loss": 0.9572,
+      "step": 5610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.740235729142593e-05,
+      "loss": 1.1002,
+      "step": 5620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7397735151375088e-05,
+      "loss": 1.1253,
+      "step": 5630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7393113011324244e-05,
+      "loss": 0.8883,
+      "step": 5640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.73884908712734e-05,
+      "loss": 1.0386,
+      "step": 5650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7383868731222558e-05,
+      "loss": 1.0399,
+      "step": 5660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7379246591171714e-05,
+      "loss": 1.1401,
+      "step": 5670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.737462445112087e-05,
+      "loss": 1.0167,
+      "step": 5680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7370002311070028e-05,
+      "loss": 0.993,
+      "step": 5690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7365380171019184e-05,
+      "loss": 1.0234,
+      "step": 5700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7360758030968338e-05,
+      "loss": 1.2451,
+      "step": 5710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7356135890917494e-05,
+      "loss": 1.0098,
+      "step": 5720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7351513750866654e-05,
+      "loss": 0.8657,
+      "step": 5730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7346891610815808e-05,
+      "loss": 1.0989,
+      "step": 5740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7342269470764964e-05,
+      "loss": 0.9638,
+      "step": 5750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.733764733071412e-05,
+      "loss": 1.2626,
+      "step": 5760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.733302519066328e-05,
+      "loss": 0.9541,
+      "step": 5770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7328403050612434e-05,
+      "loss": 1.0848,
+      "step": 5780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.732378091056159e-05,
+      "loss": 1.0944,
+      "step": 5790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7319158770510748e-05,
+      "loss": 1.1235,
+      "step": 5800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7314536630459904e-05,
+      "loss": 1.0125,
+      "step": 5810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.730991449040906e-05,
+      "loss": 0.8293,
+      "step": 5820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7305292350358218e-05,
+      "loss": 1.05,
+      "step": 5830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7300670210307374e-05,
+      "loss": 1.1013,
+      "step": 5840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.729604807025653e-05,
+      "loss": 1.2373,
+      "step": 5850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7291425930205688e-05,
+      "loss": 0.9799,
+      "step": 5860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.728680379015484e-05,
+      "loss": 1.2017,
+      "step": 5870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7282181650104e-05,
+      "loss": 1.1122,
+      "step": 5880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7277559510053158e-05,
+      "loss": 1.1151,
+      "step": 5890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.727293737000231e-05,
+      "loss": 0.9771,
+      "step": 5900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7268315229951467e-05,
+      "loss": 1.2175,
+      "step": 5910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7263693089900627e-05,
+      "loss": 1.1378,
+      "step": 5920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.725907094984978e-05,
+      "loss": 0.8443,
+      "step": 5930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7254448809798937e-05,
+      "loss": 1.2818,
+      "step": 5940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7249826669748094e-05,
+      "loss": 0.9434,
+      "step": 5950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.724520452969725e-05,
+      "loss": 1.0996,
+      "step": 5960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7240582389646407e-05,
+      "loss": 1.138,
+      "step": 5970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7235960249595564e-05,
+      "loss": 0.8798,
+      "step": 5980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.723133810954472e-05,
+      "loss": 1.2328,
+      "step": 5990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7226715969493877e-05,
+      "loss": 1.1542,
+      "step": 6000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7222093829443034e-05,
+      "loss": 0.9816,
+      "step": 6010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7217471689392187e-05,
+      "loss": 0.9359,
+      "step": 6020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7212849549341347e-05,
+      "loss": 0.9092,
+      "step": 6030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7208227409290504e-05,
+      "loss": 1.1386,
+      "step": 6040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.720360526923966e-05,
+      "loss": 0.8923,
+      "step": 6050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7198983129188814e-05,
+      "loss": 0.9748,
+      "step": 6060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7194360989137974e-05,
+      "loss": 0.852,
+      "step": 6070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.718973884908713e-05,
+      "loss": 0.9772,
+      "step": 6080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7185116709036284e-05,
+      "loss": 0.9954,
+      "step": 6090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.718049456898544e-05,
+      "loss": 1.1308,
+      "step": 6100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.71758724289346e-05,
+      "loss": 0.8412,
+      "step": 6110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7171250288883754e-05,
+      "loss": 0.962,
+      "step": 6120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.716662814883291e-05,
+      "loss": 0.9478,
+      "step": 6130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7162006008782067e-05,
+      "loss": 0.9637,
+      "step": 6140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7157383868731224e-05,
+      "loss": 1.1045,
+      "step": 6150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.715276172868038e-05,
+      "loss": 1.0363,
+      "step": 6160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7148139588629537e-05,
+      "loss": 0.9901,
+      "step": 6170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7143517448578694e-05,
+      "loss": 0.8148,
+      "step": 6180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.713889530852785e-05,
+      "loss": 1.0114,
+      "step": 6190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7134273168477007e-05,
+      "loss": 1.0805,
+      "step": 6200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.712965102842616e-05,
+      "loss": 1.0605,
+      "step": 6210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.712502888837532e-05,
+      "loss": 0.8762,
+      "step": 6220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7120406748324477e-05,
+      "loss": 1.0604,
+      "step": 6230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.711578460827363e-05,
+      "loss": 1.0384,
+      "step": 6240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7111162468222787e-05,
+      "loss": 1.0305,
+      "step": 6250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7106540328171947e-05,
+      "loss": 1.2941,
+      "step": 6260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.71019181881211e-05,
+      "loss": 1.096,
+      "step": 6270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7097296048070257e-05,
+      "loss": 0.9669,
+      "step": 6280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7092673908019414e-05,
+      "loss": 1.068,
+      "step": 6290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.708805176796857e-05,
+      "loss": 1.1909,
+      "step": 6300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7083429627917727e-05,
+      "loss": 0.9538,
+      "step": 6310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7078807487866884e-05,
+      "loss": 1.1258,
+      "step": 6320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.707418534781604e-05,
+      "loss": 0.9283,
+      "step": 6330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7069563207765197e-05,
+      "loss": 1.086,
+      "step": 6340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7064941067714354e-05,
+      "loss": 1.1066,
+      "step": 6350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.706031892766351e-05,
+      "loss": 0.9729,
+      "step": 6360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7055696787612667e-05,
+      "loss": 1.2501,
+      "step": 6370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7051074647561824e-05,
+      "loss": 1.0994,
+      "step": 6380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.704645250751098e-05,
+      "loss": 1.0894,
+      "step": 6390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7041830367460133e-05,
+      "loss": 0.9294,
+      "step": 6400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7037208227409294e-05,
+      "loss": 1.0342,
+      "step": 6410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.703258608735845e-05,
+      "loss": 0.7911,
+      "step": 6420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7027963947307603e-05,
+      "loss": 1.3068,
+      "step": 6430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.702334180725676e-05,
+      "loss": 0.9894,
+      "step": 6440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.701871966720592e-05,
+      "loss": 1.1505,
+      "step": 6450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7014097527155073e-05,
+      "loss": 1.0107,
+      "step": 6460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.700947538710423e-05,
+      "loss": 1.0702,
+      "step": 6470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7004853247053387e-05,
+      "loss": 1.0967,
+      "step": 6480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7000231107002543e-05,
+      "loss": 1.1386,
+      "step": 6490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.69956089669517e-05,
+      "loss": 0.9393,
+      "step": 6500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6990986826900857e-05,
+      "loss": 1.1836,
+      "step": 6510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6986364686850013e-05,
+      "loss": 0.9144,
+      "step": 6520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.698174254679917e-05,
+      "loss": 1.1461,
+      "step": 6530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6977120406748327e-05,
+      "loss": 1.2101,
+      "step": 6540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.697249826669748e-05,
+      "loss": 1.1478,
+      "step": 6550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.696787612664664e-05,
+      "loss": 1.3228,
+      "step": 6560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6963253986595797e-05,
+      "loss": 1.1274,
+      "step": 6570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.695863184654495e-05,
+      "loss": 1.1506,
+      "step": 6580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6954009706494107e-05,
+      "loss": 0.9914,
+      "step": 6590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6949387566443267e-05,
+      "loss": 0.9621,
+      "step": 6600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6944765426392423e-05,
+      "loss": 1.1857,
+      "step": 6610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6940143286341577e-05,
+      "loss": 1.0759,
+      "step": 6620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6935521146290733e-05,
+      "loss": 1.039,
+      "step": 6630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.693089900623989e-05,
+      "loss": 0.9317,
+      "step": 6640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6926276866189047e-05,
+      "loss": 0.8067,
+      "step": 6650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6921654726138203e-05,
+      "loss": 0.975,
+      "step": 6660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.691703258608736e-05,
+      "loss": 1.3564,
+      "step": 6670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6912410446036516e-05,
+      "loss": 0.9628,
+      "step": 6680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6907788305985673e-05,
+      "loss": 1.1236,
+      "step": 6690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.690316616593483e-05,
+      "loss": 1.0621,
+      "step": 6700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6898544025883986e-05,
+      "loss": 0.9256,
+      "step": 6710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6893921885833143e-05,
+      "loss": 0.9744,
+      "step": 6720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.68892997457823e-05,
+      "loss": 1.0686,
+      "step": 6730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6884677605731453e-05,
+      "loss": 1.1468,
+      "step": 6740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6880055465680613e-05,
+      "loss": 1.1455,
+      "step": 6750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.687543332562977e-05,
+      "loss": 0.993,
+      "step": 6760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6870811185578923e-05,
+      "loss": 0.9076,
+      "step": 6770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.686618904552808e-05,
+      "loss": 1.2043,
+      "step": 6780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6861566905477236e-05,
+      "loss": 1.0956,
+      "step": 6790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6856944765426393e-05,
+      "loss": 1.1863,
+      "step": 6800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.685232262537555e-05,
+      "loss": 1.1048,
+      "step": 6810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6847700485324706e-05,
+      "loss": 0.9377,
+      "step": 6820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6843078345273863e-05,
+      "loss": 1.0707,
+      "step": 6830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.683845620522302e-05,
+      "loss": 0.9366,
+      "step": 6840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6833834065172176e-05,
+      "loss": 1.0371,
+      "step": 6850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6829211925121333e-05,
+      "loss": 0.9048,
+      "step": 6860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.682458978507049e-05,
+      "loss": 1.083,
+      "step": 6870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6819967645019646e-05,
+      "loss": 1.0081,
+      "step": 6880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6815345504968803e-05,
+      "loss": 1.1577,
+      "step": 6890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.681072336491796e-05,
+      "loss": 1.0077,
+      "step": 6900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6806101224867116e-05,
+      "loss": 0.9902,
+      "step": 6910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6801479084816273e-05,
+      "loss": 1.1284,
+      "step": 6920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6796856944765426e-05,
+      "loss": 1.0113,
+      "step": 6930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6792234804714583e-05,
+      "loss": 1.0946,
+      "step": 6940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6787612664663743e-05,
+      "loss": 0.9434,
+      "step": 6950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6782990524612896e-05,
+      "loss": 1.0813,
+      "step": 6960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6778368384562053e-05,
+      "loss": 1.0322,
+      "step": 6970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.677374624451121e-05,
+      "loss": 1.0166,
+      "step": 6980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6769124104460366e-05,
+      "loss": 1.0323,
+      "step": 6990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6764501964409523e-05,
+      "loss": 1.0458,
+      "step": 7000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.675987982435868e-05,
+      "loss": 0.9714,
+      "step": 7010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6755257684307836e-05,
+      "loss": 0.9385,
+      "step": 7020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6750635544256993e-05,
+      "loss": 1.1339,
+      "step": 7030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.674601340420615e-05,
+      "loss": 1.1222,
+      "step": 7040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6741391264155306e-05,
+      "loss": 0.9426,
+      "step": 7050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6736769124104463e-05,
+      "loss": 1.0367,
+      "step": 7060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.673214698405362e-05,
+      "loss": 0.9712,
+      "step": 7070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6727524844002773e-05,
+      "loss": 0.9698,
+      "step": 7080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.672290270395193e-05,
+      "loss": 1.068,
+      "step": 7090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.671828056390109e-05,
+      "loss": 1.1894,
+      "step": 7100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6713658423850243e-05,
+      "loss": 0.8129,
+      "step": 7110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.67090362837994e-05,
+      "loss": 0.9051,
+      "step": 7120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6704414143748556e-05,
+      "loss": 0.8062,
+      "step": 7130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6699792003697713e-05,
+      "loss": 1.0051,
+      "step": 7140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.669516986364687e-05,
+      "loss": 0.8871,
+      "step": 7150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6690547723596026e-05,
+      "loss": 1.0329,
+      "step": 7160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6685925583545183e-05,
+      "loss": 1.2144,
+      "step": 7170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.668130344349434e-05,
+      "loss": 1.1275,
+      "step": 7180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6676681303443496e-05,
+      "loss": 1.0718,
+      "step": 7190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6672059163392652e-05,
+      "loss": 0.9477,
+      "step": 7200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.666743702334181e-05,
+      "loss": 1.0615,
+      "step": 7210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6662814883290966e-05,
+      "loss": 0.9653,
+      "step": 7220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6658192743240122e-05,
+      "loss": 1.0649,
+      "step": 7230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.665357060318928e-05,
+      "loss": 0.9939,
+      "step": 7240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6648948463138436e-05,
+      "loss": 1.154,
+      "step": 7250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6644326323087592e-05,
+      "loss": 1.1241,
+      "step": 7260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6639704183036746e-05,
+      "loss": 1.0387,
+      "step": 7270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6635082042985902e-05,
+      "loss": 1.1142,
+      "step": 7280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6630459902935062e-05,
+      "loss": 1.1383,
+      "step": 7290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6625837762884216e-05,
+      "loss": 0.8875,
+      "step": 7300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6621215622833372e-05,
+      "loss": 0.8456,
+      "step": 7310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.661659348278253e-05,
+      "loss": 1.1513,
+      "step": 7320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6611971342731686e-05,
+      "loss": 1.0892,
+      "step": 7330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6607349202680842e-05,
+      "loss": 1.188,
+      "step": 7340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.660272706263e-05,
+      "loss": 1.1787,
+      "step": 7350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6598104922579156e-05,
+      "loss": 1.0492,
+      "step": 7360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6593482782528312e-05,
+      "loss": 1.2088,
+      "step": 7370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.658886064247747e-05,
+      "loss": 1.0627,
+      "step": 7380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6584238502426626e-05,
+      "loss": 1.1533,
+      "step": 7390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6579616362375782e-05,
+      "loss": 0.9313,
+      "step": 7400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.657499422232494e-05,
+      "loss": 1.0283,
+      "step": 7410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6570372082274092e-05,
+      "loss": 1.1645,
+      "step": 7420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.656574994222325e-05,
+      "loss": 1.1208,
+      "step": 7430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.656112780217241e-05,
+      "loss": 1.1293,
+      "step": 7440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6556505662121566e-05,
+      "loss": 1.2313,
+      "step": 7450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.655188352207072e-05,
+      "loss": 1.3014,
+      "step": 7460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6547261382019875e-05,
+      "loss": 1.0779,
+      "step": 7470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6542639241969036e-05,
+      "loss": 1.096,
+      "step": 7480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.653801710191819e-05,
+      "loss": 0.8675,
+      "step": 7490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6533394961867345e-05,
+      "loss": 1.0504,
+      "step": 7500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6528772821816502e-05,
+      "loss": 0.973,
+      "step": 7510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.652415068176566e-05,
+      "loss": 1.2366,
+      "step": 7520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6519528541714815e-05,
+      "loss": 1.0136,
+      "step": 7530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6514906401663972e-05,
+      "loss": 1.1407,
+      "step": 7540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.651028426161313e-05,
+      "loss": 0.9733,
+      "step": 7550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6505662121562285e-05,
+      "loss": 1.2175,
+      "step": 7560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6501039981511442e-05,
+      "loss": 0.9284,
+      "step": 7570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6496417841460595e-05,
+      "loss": 0.9219,
+      "step": 7580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6491795701409755e-05,
+      "loss": 1.1998,
+      "step": 7590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6487173561358912e-05,
+      "loss": 1.1432,
+      "step": 7600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6482551421308065e-05,
+      "loss": 0.8463,
+      "step": 7610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6477929281257222e-05,
+      "loss": 0.7653,
+      "step": 7620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6473307141206382e-05,
+      "loss": 1.1716,
+      "step": 7630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6468685001155535e-05,
+      "loss": 1.0687,
+      "step": 7640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6464062861104692e-05,
+      "loss": 1.1302,
+      "step": 7650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645944072105385e-05,
+      "loss": 1.0056,
+      "step": 7660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6454818581003005e-05,
+      "loss": 1.1402,
+      "step": 7670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6450196440952162e-05,
+      "loss": 1.1238,
+      "step": 7680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.644557430090132e-05,
+      "loss": 0.8988,
+      "step": 7690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6440952160850475e-05,
+      "loss": 0.8145,
+      "step": 7700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6436330020799632e-05,
+      "loss": 0.9824,
+      "step": 7710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.643170788074879e-05,
+      "loss": 1.0846,
+      "step": 7720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6427085740697945e-05,
+      "loss": 0.9707,
+      "step": 7730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6422463600647102e-05,
+      "loss": 1.024,
+      "step": 7740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641784146059626e-05,
+      "loss": 0.9732,
+      "step": 7750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6413219320545415e-05,
+      "loss": 0.9594,
+      "step": 7760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640859718049457e-05,
+      "loss": 0.8238,
+      "step": 7770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640397504044373e-05,
+      "loss": 1.0651,
+      "step": 7780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6399352900392885e-05,
+      "loss": 1.1187,
+      "step": 7790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639473076034204e-05,
+      "loss": 1.0225,
+      "step": 7800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6390108620291195e-05,
+      "loss": 0.9973,
+      "step": 7810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6385486480240355e-05,
+      "loss": 0.8651,
+      "step": 7820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638086434018951e-05,
+      "loss": 0.9785,
+      "step": 7830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6376242200138665e-05,
+      "loss": 1.2037,
+      "step": 7840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.637162006008782e-05,
+      "loss": 1.0257,
+      "step": 7850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636699792003698e-05,
+      "loss": 1.0339,
+      "step": 7860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362375779986135e-05,
+      "loss": 0.9993,
+      "step": 7870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635775363993529e-05,
+      "loss": 1.0234,
+      "step": 7880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6353131499884448e-05,
+      "loss": 1.1249,
+      "step": 7890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6348509359833605e-05,
+      "loss": 1.029,
+      "step": 7900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634388721978276e-05,
+      "loss": 1.0881,
+      "step": 7910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6339265079731915e-05,
+      "loss": 1.0293,
+      "step": 7920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6334642939681075e-05,
+      "loss": 1.1481,
+      "step": 7930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.633002079963023e-05,
+      "loss": 0.8906,
+      "step": 7940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6325398659579385e-05,
+      "loss": 0.9298,
+      "step": 7950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.632077651952854e-05,
+      "loss": 0.9433,
+      "step": 7960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.63161543794777e-05,
+      "loss": 1.0904,
+      "step": 7970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6311532239426855e-05,
+      "loss": 1.1747,
+      "step": 7980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.630691009937601e-05,
+      "loss": 0.922,
+      "step": 7990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6302287959325168e-05,
+      "loss": 1.0444,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6297665819274325e-05,
+      "loss": 1.103,
+      "step": 8010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.629304367922348e-05,
+      "loss": 0.982,
+      "step": 8020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6288421539172638e-05,
+      "loss": 1.0677,
+      "step": 8030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6283799399121795e-05,
+      "loss": 0.9627,
+      "step": 8040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.627917725907095e-05,
+      "loss": 0.9951,
+      "step": 8050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6274555119020108e-05,
+      "loss": 1.1865,
+      "step": 8060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6269932978969265e-05,
+      "loss": 0.8926,
+      "step": 8070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.626531083891842e-05,
+      "loss": 0.9363,
+      "step": 8080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6260688698867578e-05,
+      "loss": 1.0868,
+      "step": 8090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6256066558816735e-05,
+      "loss": 1.0776,
+      "step": 8100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6251444418765888e-05,
+      "loss": 0.9474,
+      "step": 8110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6246822278715048e-05,
+      "loss": 1.0268,
+      "step": 8120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6242200138664205e-05,
+      "loss": 1.018,
+      "step": 8130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6237577998613358e-05,
+      "loss": 0.7145,
+      "step": 8140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6232955858562515e-05,
+      "loss": 1.0552,
+      "step": 8150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.622833371851167e-05,
+      "loss": 1.2308,
+      "step": 8160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6223711578460828e-05,
+      "loss": 0.9904,
+      "step": 8170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6219089438409985e-05,
+      "loss": 1.0721,
+      "step": 8180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.621446729835914e-05,
+      "loss": 0.7966,
+      "step": 8190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6209845158308298e-05,
+      "loss": 0.9334,
+      "step": 8200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6205223018257455e-05,
+      "loss": 1.0431,
+      "step": 8210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.620060087820661e-05,
+      "loss": 1.0555,
+      "step": 8220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6195978738155768e-05,
+      "loss": 1.0423,
+      "step": 8230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6191356598104925e-05,
+      "loss": 1.035,
+      "step": 8240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.618673445805408e-05,
+      "loss": 0.948,
+      "step": 8250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6182112318003234e-05,
+      "loss": 0.9147,
+      "step": 8260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6177490177952394e-05,
+      "loss": 1.0039,
+      "step": 8270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.617286803790155e-05,
+      "loss": 0.8686,
+      "step": 8280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6168245897850708e-05,
+      "loss": 0.9829,
+      "step": 8290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.616362375779986e-05,
+      "loss": 0.9374,
+      "step": 8300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.615900161774902e-05,
+      "loss": 1.1601,
+      "step": 8310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6154379477698178e-05,
+      "loss": 1.1227,
+      "step": 8320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.614975733764733e-05,
+      "loss": 1.0984,
+      "step": 8330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6145135197596488e-05,
+      "loss": 1.0372,
+      "step": 8340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6140513057545644e-05,
+      "loss": 1.0146,
+      "step": 8350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.61358909174948e-05,
+      "loss": 1.0787,
+      "step": 8360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6131268777443958e-05,
+      "loss": 1.3111,
+      "step": 8370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6126646637393114e-05,
+      "loss": 1.1747,
+      "step": 8380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.612202449734227e-05,
+      "loss": 1.0599,
+      "step": 8390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6117402357291428e-05,
+      "loss": 1.1701,
+      "step": 8400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6112780217240584e-05,
+      "loss": 1.04,
+      "step": 8410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.610815807718974e-05,
+      "loss": 1.0704,
+      "step": 8420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6103535937138898e-05,
+      "loss": 1.0018,
+      "step": 8430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6098913797088054e-05,
+      "loss": 0.9551,
+      "step": 8440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6094291657037208e-05,
+      "loss": 0.9696,
+      "step": 8450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6089669516986368e-05,
+      "loss": 1.0889,
+      "step": 8460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6085047376935524e-05,
+      "loss": 1.2406,
+      "step": 8470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6080425236884678e-05,
+      "loss": 1.1869,
+      "step": 8480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6075803096833834e-05,
+      "loss": 1.022,
+      "step": 8490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.607118095678299e-05,
+      "loss": 0.9124,
+      "step": 8500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6066558816732147e-05,
+      "loss": 0.9698,
+      "step": 8510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6061936676681304e-05,
+      "loss": 0.9169,
+      "step": 8520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.605731453663046e-05,
+      "loss": 1.1044,
+      "step": 8530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6052692396579617e-05,
+      "loss": 0.9819,
+      "step": 8540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6048070256528774e-05,
+      "loss": 1.149,
+      "step": 8550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.604344811647793e-05,
+      "loss": 1.2384,
+      "step": 8560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6038825976427087e-05,
+      "loss": 1.0797,
+      "step": 8570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6034203836376244e-05,
+      "loss": 1.1958,
+      "step": 8580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.60295816963254e-05,
+      "loss": 1.2108,
+      "step": 8590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6024959556274557e-05,
+      "loss": 0.7952,
+      "step": 8600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6020337416223714e-05,
+      "loss": 0.9755,
+      "step": 8610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.601571527617287e-05,
+      "loss": 1.0416,
+      "step": 8620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6011093136122027e-05,
+      "loss": 1.2034,
+      "step": 8630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.600647099607118e-05,
+      "loss": 0.9255,
+      "step": 8640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6001848856020337e-05,
+      "loss": 0.9779,
+      "step": 8650
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.6743062945076532,
+      "eval_loss": 0.9191309213638306,
+      "eval_runtime": 364.7598,
+      "eval_samples_per_second": 33.493,
+      "eval_steps_per_second": 4.189,
+      "step": 8654
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5997226715969497e-05,
+      "loss": 1.0775,
+      "step": 8660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.599260457591865e-05,
+      "loss": 1.0962,
+      "step": 8670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5987982435867807e-05,
+      "loss": 0.8082,
+      "step": 8680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5983360295816964e-05,
+      "loss": 1.1367,
+      "step": 8690
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.597873815576612e-05,
+      "loss": 0.9168,
+      "step": 8700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5974116015715277e-05,
+      "loss": 1.0296,
+      "step": 8710
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5969493875664434e-05,
+      "loss": 0.7765,
+      "step": 8720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.596487173561359e-05,
+      "loss": 1.0985,
+      "step": 8730
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5960249595562747e-05,
+      "loss": 0.8216,
+      "step": 8740
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5955627455511904e-05,
+      "loss": 1.0874,
+      "step": 8750
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.595100531546106e-05,
+      "loss": 0.8438,
+      "step": 8760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5946383175410217e-05,
+      "loss": 0.8097,
+      "step": 8770
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5941761035359374e-05,
+      "loss": 1.1491,
+      "step": 8780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5937138895308527e-05,
+      "loss": 0.8727,
+      "step": 8790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5932516755257684e-05,
+      "loss": 0.8972,
+      "step": 8800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5927894615206844e-05,
+      "loss": 0.8487,
+      "step": 8810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5923272475155997e-05,
+      "loss": 1.0183,
+      "step": 8820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5918650335105154e-05,
+      "loss": 0.8829,
+      "step": 8830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.591402819505431e-05,
+      "loss": 0.9424,
+      "step": 8840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5909406055003467e-05,
+      "loss": 0.9508,
+      "step": 8850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5904783914952624e-05,
+      "loss": 0.9266,
+      "step": 8860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.590016177490178e-05,
+      "loss": 0.8597,
+      "step": 8870
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5895539634850937e-05,
+      "loss": 0.8776,
+      "step": 8880
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5890917494800094e-05,
+      "loss": 1.0067,
+      "step": 8890
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.588629535474925e-05,
+      "loss": 1.1206,
+      "step": 8900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5881673214698407e-05,
+      "loss": 0.8182,
+      "step": 8910
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5877051074647564e-05,
+      "loss": 1.069,
+      "step": 8920
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.587242893459672e-05,
+      "loss": 1.0043,
+      "step": 8930
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5867806794545877e-05,
+      "loss": 1.1497,
+      "step": 8940
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.586318465449503e-05,
+      "loss": 0.8443,
+      "step": 8950
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.585856251444419e-05,
+      "loss": 0.7704,
+      "step": 8960
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5853940374393347e-05,
+      "loss": 1.0611,
+      "step": 8970
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.58493182343425e-05,
+      "loss": 1.0265,
+      "step": 8980
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5844696094291657e-05,
+      "loss": 0.887,
+      "step": 8990
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5840073954240817e-05,
+      "loss": 1.0237,
+      "step": 9000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.583545181418997e-05,
+      "loss": 0.9237,
+      "step": 9010
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5830829674139127e-05,
+      "loss": 0.9246,
+      "step": 9020
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5826207534088283e-05,
+      "loss": 1.0137,
+      "step": 9030
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.582158539403744e-05,
+      "loss": 0.8817,
+      "step": 9040
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5816963253986597e-05,
+      "loss": 1.1537,
+      "step": 9050
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5812341113935753e-05,
+      "loss": 0.8645,
+      "step": 9060
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.580771897388491e-05,
+      "loss": 1.0768,
+      "step": 9070
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5803096833834067e-05,
+      "loss": 0.9162,
+      "step": 9080
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5798474693783223e-05,
+      "loss": 0.827,
+      "step": 9090
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5793852553732377e-05,
+      "loss": 0.9145,
+      "step": 9100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5789230413681537e-05,
+      "loss": 0.7272,
+      "step": 9110
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5784608273630693e-05,
+      "loss": 1.0242,
+      "step": 9120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5779986133579847e-05,
+      "loss": 1.0197,
+      "step": 9130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5775363993529003e-05,
+      "loss": 1.1649,
+      "step": 9140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5770741853478163e-05,
+      "loss": 0.9219,
+      "step": 9150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.576611971342732e-05,
+      "loss": 1.0543,
+      "step": 9160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5761497573376473e-05,
+      "loss": 1.1068,
+      "step": 9170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.575687543332563e-05,
+      "loss": 1.0053,
+      "step": 9180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.575225329327479e-05,
+      "loss": 1.0719,
+      "step": 9190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5747631153223943e-05,
+      "loss": 0.8926,
+      "step": 9200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.57430090131731e-05,
+      "loss": 1.0943,
+      "step": 9210
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5738386873122257e-05,
+      "loss": 0.704,
+      "step": 9220
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5733764733071413e-05,
+      "loss": 1.1154,
+      "step": 9230
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.572914259302057e-05,
+      "loss": 1.0977,
+      "step": 9240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5724520452969727e-05,
+      "loss": 0.9968,
+      "step": 9250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5719898312918883e-05,
+      "loss": 0.9491,
+      "step": 9260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.571527617286804e-05,
+      "loss": 1.1942,
+      "step": 9270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5710654032817197e-05,
+      "loss": 0.9653,
+      "step": 9280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.570603189276635e-05,
+      "loss": 1.0442,
+      "step": 9290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.570140975271551e-05,
+      "loss": 1.1489,
+      "step": 9300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5696787612664666e-05,
+      "loss": 0.9836,
+      "step": 9310
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.569216547261382e-05,
+      "loss": 0.8024,
+      "step": 9320
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5687543332562976e-05,
+      "loss": 0.946,
+      "step": 9330
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5682921192512136e-05,
+      "loss": 1.037,
+      "step": 9340
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.567829905246129e-05,
+      "loss": 1.0684,
+      "step": 9350
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5673676912410446e-05,
+      "loss": 1.0922,
+      "step": 9360
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5669054772359603e-05,
+      "loss": 1.2742,
+      "step": 9370
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.566443263230876e-05,
+      "loss": 0.9406,
+      "step": 9380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5659810492257916e-05,
+      "loss": 0.7354,
+      "step": 9390
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5655188352207073e-05,
+      "loss": 0.9782,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.565056621215623e-05,
+      "loss": 1.0088,
+      "step": 9410
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5645944072105386e-05,
+      "loss": 1.1664,
+      "step": 9420
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5641321932054543e-05,
+      "loss": 0.9932,
+      "step": 9430
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.56366997920037e-05,
+      "loss": 0.866,
+      "step": 9440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5632077651952856e-05,
+      "loss": 0.8588,
+      "step": 9450
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5627455511902013e-05,
+      "loss": 0.8856,
+      "step": 9460
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.562283337185117e-05,
+      "loss": 1.1369,
+      "step": 9470
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5618211231800323e-05,
+      "loss": 1.0747,
+      "step": 9480
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5613589091749483e-05,
+      "loss": 1.0275,
+      "step": 9490
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.560896695169864e-05,
+      "loss": 0.9178,
+      "step": 9500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5604344811647793e-05,
+      "loss": 1.062,
+      "step": 9510
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.559972267159695e-05,
+      "loss": 0.9609,
+      "step": 9520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.559510053154611e-05,
+      "loss": 1.08,
+      "step": 9530
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5590478391495263e-05,
+      "loss": 0.9288,
+      "step": 9540
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.558585625144442e-05,
+      "loss": 0.8439,
+      "step": 9550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5581234111393576e-05,
+      "loss": 0.945,
+      "step": 9560
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5576611971342733e-05,
+      "loss": 1.0991,
+      "step": 9570
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.557198983129189e-05,
+      "loss": 1.0151,
+      "step": 9580
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5567367691241046e-05,
+      "loss": 1.2526,
+      "step": 9590
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5562745551190203e-05,
+      "loss": 1.0941,
+      "step": 9600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.555812341113936e-05,
+      "loss": 1.2226,
+      "step": 9610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5553501271088516e-05,
+      "loss": 0.8771,
+      "step": 9620
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.554887913103767e-05,
+      "loss": 0.9892,
+      "step": 9630
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.554425699098683e-05,
+      "loss": 1.0761,
+      "step": 9640
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5539634850935986e-05,
+      "loss": 1.1765,
+      "step": 9650
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.553501271088514e-05,
+      "loss": 0.9622,
+      "step": 9660
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5530390570834296e-05,
+      "loss": 1.0357,
+      "step": 9670
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5525768430783456e-05,
+      "loss": 0.8497,
+      "step": 9680
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.552114629073261e-05,
+      "loss": 0.8819,
+      "step": 9690
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5516524150681766e-05,
+      "loss": 0.9315,
+      "step": 9700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5511902010630923e-05,
+      "loss": 1.0813,
+      "step": 9710
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.550727987058008e-05,
+      "loss": 1.0204,
+      "step": 9720
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5502657730529236e-05,
+      "loss": 1.0275,
+      "step": 9730
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5498035590478393e-05,
+      "loss": 1.0404,
+      "step": 9740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.549341345042755e-05,
+      "loss": 1.087,
+      "step": 9750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5488791310376706e-05,
+      "loss": 0.8796,
+      "step": 9760
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5484169170325863e-05,
+      "loss": 0.8488,
+      "step": 9770
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.547954703027502e-05,
+      "loss": 0.851,
+      "step": 9780
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5474924890224176e-05,
+      "loss": 0.9883,
+      "step": 9790
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5470302750173333e-05,
+      "loss": 0.8628,
+      "step": 9800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.546568061012249e-05,
+      "loss": 0.9253,
+      "step": 9810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5461058470071642e-05,
+      "loss": 1.0636,
+      "step": 9820
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5456436330020803e-05,
+      "loss": 1.0681,
+      "step": 9830
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.545181418996996e-05,
+      "loss": 0.978,
+      "step": 9840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5447192049919112e-05,
+      "loss": 1.0017,
+      "step": 9850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.544256990986827e-05,
+      "loss": 1.0162,
+      "step": 9860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5437947769817426e-05,
+      "loss": 0.8776,
+      "step": 9870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5433325629766582e-05,
+      "loss": 1.1119,
+      "step": 9880
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.542870348971574e-05,
+      "loss": 0.7713,
+      "step": 9890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5424081349664896e-05,
+      "loss": 0.8814,
+      "step": 9900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5419459209614052e-05,
+      "loss": 1.2732,
+      "step": 9910
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.541483706956321e-05,
+      "loss": 1.0034,
+      "step": 9920
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5410214929512366e-05,
+      "loss": 1.0679,
+      "step": 9930
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5405592789461522e-05,
+      "loss": 0.9341,
+      "step": 9940
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.540097064941068e-05,
+      "loss": 1.0323,
+      "step": 9950
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5396348509359836e-05,
+      "loss": 1.0457,
+      "step": 9960
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.539172636930899e-05,
+      "loss": 0.947,
+      "step": 9970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.538710422925815e-05,
+      "loss": 0.9611,
+      "step": 9980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5382482089207306e-05,
+      "loss": 0.8029,
+      "step": 9990
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5377859949156462e-05,
+      "loss": 0.9125,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5373237809105616e-05,
+      "loss": 1.084,
+      "step": 10010
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5368615669054772e-05,
+      "loss": 0.9954,
+      "step": 10020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5363993529003932e-05,
+      "loss": 1.1712,
+      "step": 10030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5359371388953086e-05,
+      "loss": 1.0646,
+      "step": 10040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5354749248902242e-05,
+      "loss": 0.9446,
+      "step": 10050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.53501271088514e-05,
+      "loss": 1.2355,
+      "step": 10060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5345504968800556e-05,
+      "loss": 0.9874,
+      "step": 10070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5340882828749712e-05,
+      "loss": 0.8092,
+      "step": 10080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.533626068869887e-05,
+      "loss": 1.1097,
+      "step": 10090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5331638548648025e-05,
+      "loss": 1.0249,
+      "step": 10100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5327016408597182e-05,
+      "loss": 0.729,
+      "step": 10110
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.532239426854634e-05,
+      "loss": 0.9743,
+      "step": 10120
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5317772128495495e-05,
+      "loss": 0.9493,
+      "step": 10130
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5313149988444652e-05,
+      "loss": 0.9619,
+      "step": 10140
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.530852784839381e-05,
+      "loss": 0.9141,
+      "step": 10150
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5303905708342962e-05,
+      "loss": 1.0066,
+      "step": 10160
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.529928356829212e-05,
+      "loss": 0.787,
+      "step": 10170
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.529466142824128e-05,
+      "loss": 1.0846,
+      "step": 10180
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5290039288190432e-05,
+      "loss": 0.897,
+      "step": 10190
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.528541714813959e-05,
+      "loss": 0.8513,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5280795008088745e-05,
+      "loss": 0.9586,
+      "step": 10210
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5276172868037902e-05,
+      "loss": 0.9526,
+      "step": 10220
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.527155072798706e-05,
+      "loss": 1.1779,
+      "step": 10230
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5266928587936215e-05,
+      "loss": 1.0145,
+      "step": 10240
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5262306447885372e-05,
+      "loss": 1.3035,
+      "step": 10250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5257684307834529e-05,
+      "loss": 1.054,
+      "step": 10260
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5253062167783685e-05,
+      "loss": 0.8731,
+      "step": 10270
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5248440027732842e-05,
+      "loss": 1.2764,
+      "step": 10280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5243817887681997e-05,
+      "loss": 0.9729,
+      "step": 10290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5239195747631155e-05,
+      "loss": 1.3819,
+      "step": 10300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5234573607580312e-05,
+      "loss": 0.9208,
+      "step": 10310
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5229951467529467e-05,
+      "loss": 1.0151,
+      "step": 10320
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5225329327478624e-05,
+      "loss": 0.9136,
+      "step": 10330
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5220707187427782e-05,
+      "loss": 0.8396,
+      "step": 10340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5216085047376937e-05,
+      "loss": 1.0763,
+      "step": 10350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5211462907326093e-05,
+      "loss": 1.0323,
+      "step": 10360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.520684076727525e-05,
+      "loss": 0.9772,
+      "step": 10370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5202218627224405e-05,
+      "loss": 1.053,
+      "step": 10380
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5197596487173563e-05,
+      "loss": 1.0057,
+      "step": 10390
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.519297434712272e-05,
+      "loss": 0.9191,
+      "step": 10400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5188352207071875e-05,
+      "loss": 0.7753,
+      "step": 10410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5183730067021032e-05,
+      "loss": 0.9107,
+      "step": 10420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5179107926970188e-05,
+      "loss": 0.8146,
+      "step": 10430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5174485786919343e-05,
+      "loss": 0.9178,
+      "step": 10440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5169863646868502e-05,
+      "loss": 0.9518,
+      "step": 10450
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5165241506817658e-05,
+      "loss": 1.2761,
+      "step": 10460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5160619366766813e-05,
+      "loss": 0.9215,
+      "step": 10470
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.515599722671597e-05,
+      "loss": 0.9759,
+      "step": 10480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5151375086665128e-05,
+      "loss": 0.985,
+      "step": 10490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5146752946614283e-05,
+      "loss": 0.9565,
+      "step": 10500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.514213080656344e-05,
+      "loss": 1.0607,
+      "step": 10510
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5137508666512597e-05,
+      "loss": 0.8761,
+      "step": 10520
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5132886526461752e-05,
+      "loss": 1.1379,
+      "step": 10530
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.512826438641091e-05,
+      "loss": 0.8514,
+      "step": 10540
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5123642246360067e-05,
+      "loss": 1.1785,
+      "step": 10550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5119020106309223e-05,
+      "loss": 1.1003,
+      "step": 10560
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5114397966258378e-05,
+      "loss": 1.0278,
+      "step": 10570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5109775826207535e-05,
+      "loss": 0.8454,
+      "step": 10580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5105153686156693e-05,
+      "loss": 1.0262,
+      "step": 10590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5100531546105848e-05,
+      "loss": 1.1937,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5095909406055005e-05,
+      "loss": 0.9366,
+      "step": 10610
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5091287266004161e-05,
+      "loss": 1.0213,
+      "step": 10620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5086665125953316e-05,
+      "loss": 0.9693,
+      "step": 10630
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5082042985902475e-05,
+      "loss": 1.0623,
+      "step": 10640
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5077420845851631e-05,
+      "loss": 0.9666,
+      "step": 10650
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5072798705800786e-05,
+      "loss": 0.7772,
+      "step": 10660
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5068176565749943e-05,
+      "loss": 0.9408,
+      "step": 10670
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5063554425699101e-05,
+      "loss": 0.9641,
+      "step": 10680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5058932285648256e-05,
+      "loss": 1.2113,
+      "step": 10690
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5054310145597413e-05,
+      "loss": 0.9688,
+      "step": 10700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.504968800554657e-05,
+      "loss": 0.9585,
+      "step": 10710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5045065865495725e-05,
+      "loss": 1.063,
+      "step": 10720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5040443725444881e-05,
+      "loss": 1.0069,
+      "step": 10730
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.503582158539404e-05,
+      "loss": 1.0296,
+      "step": 10740
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5031199445343195e-05,
+      "loss": 1.0953,
+      "step": 10750
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5026577305292351e-05,
+      "loss": 0.9676,
+      "step": 10760
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5021955165241508e-05,
+      "loss": 0.8617,
+      "step": 10770
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5017333025190663e-05,
+      "loss": 0.8331,
+      "step": 10780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5012710885139821e-05,
+      "loss": 0.9991,
+      "step": 10790
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5008088745088978e-05,
+      "loss": 0.8967,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5003466605038133e-05,
+      "loss": 1.1451,
+      "step": 10810
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.499884446498729e-05,
+      "loss": 0.9227,
+      "step": 10820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4994222324936448e-05,
+      "loss": 0.9758,
+      "step": 10830
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4989600184885605e-05,
+      "loss": 1.1239,
+      "step": 10840
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.498497804483476e-05,
+      "loss": 1.0926,
+      "step": 10850
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4980355904783916e-05,
+      "loss": 0.9059,
+      "step": 10860
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4975733764733073e-05,
+      "loss": 0.8824,
+      "step": 10870
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4971111624682228e-05,
+      "loss": 0.9551,
+      "step": 10880
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4966489484631386e-05,
+      "loss": 1.1899,
+      "step": 10890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4961867344580543e-05,
+      "loss": 0.8901,
+      "step": 10900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4957245204529698e-05,
+      "loss": 0.9535,
+      "step": 10910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4952623064478854e-05,
+      "loss": 0.9512,
+      "step": 10920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4948000924428013e-05,
+      "loss": 0.8077,
+      "step": 10930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4943378784377168e-05,
+      "loss": 0.8499,
+      "step": 10940
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4938756644326324e-05,
+      "loss": 1.2961,
+      "step": 10950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4934134504275481e-05,
+      "loss": 1.0772,
+      "step": 10960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4929512364224636e-05,
+      "loss": 1.0519,
+      "step": 10970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4924890224173794e-05,
+      "loss": 1.0236,
+      "step": 10980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4920268084122951e-05,
+      "loss": 1.0489,
+      "step": 10990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4915645944072106e-05,
+      "loss": 0.9161,
+      "step": 11000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4911023804021263e-05,
+      "loss": 0.9967,
+      "step": 11010
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.490640166397042e-05,
+      "loss": 0.9961,
+      "step": 11020
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4901779523919574e-05,
+      "loss": 0.9115,
+      "step": 11030
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4897157383868733e-05,
+      "loss": 0.9952,
+      "step": 11040
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.489253524381789e-05,
+      "loss": 0.9308,
+      "step": 11050
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4887913103767044e-05,
+      "loss": 1.0123,
+      "step": 11060
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4883290963716201e-05,
+      "loss": 1.0204,
+      "step": 11070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.487866882366536e-05,
+      "loss": 0.8076,
+      "step": 11080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4874046683614514e-05,
+      "loss": 0.8308,
+      "step": 11090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4869424543563671e-05,
+      "loss": 0.8124,
+      "step": 11100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4864802403512828e-05,
+      "loss": 0.8519,
+      "step": 11110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4860180263461986e-05,
+      "loss": 0.7837,
+      "step": 11120
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.485555812341114e-05,
+      "loss": 0.9414,
+      "step": 11130
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4850935983360297e-05,
+      "loss": 1.1679,
+      "step": 11140
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4846313843309454e-05,
+      "loss": 1.3595,
+      "step": 11150
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4841691703258609e-05,
+      "loss": 0.8255,
+      "step": 11160
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4837069563207766e-05,
+      "loss": 0.9988,
+      "step": 11170
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4832447423156924e-05,
+      "loss": 1.0168,
+      "step": 11180
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4827825283106079e-05,
+      "loss": 0.7425,
+      "step": 11190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4823203143055236e-05,
+      "loss": 1.1047,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4818581003004392e-05,
+      "loss": 0.8443,
+      "step": 11210
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4813958862953547e-05,
+      "loss": 1.1833,
+      "step": 11220
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4809336722902706e-05,
+      "loss": 0.8754,
+      "step": 11230
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4804714582851862e-05,
+      "loss": 0.7381,
+      "step": 11240
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4800092442801017e-05,
+      "loss": 1.0903,
+      "step": 11250
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4795470302750174e-05,
+      "loss": 0.7856,
+      "step": 11260
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4790848162699332e-05,
+      "loss": 0.9263,
+      "step": 11270
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4786226022648487e-05,
+      "loss": 1.052,
+      "step": 11280
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4781603882597644e-05,
+      "loss": 0.9682,
+      "step": 11290
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.47769817425468e-05,
+      "loss": 0.9812,
+      "step": 11300
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4772359602495956e-05,
+      "loss": 0.9551,
+      "step": 11310
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4767737462445112e-05,
+      "loss": 1.0078,
+      "step": 11320
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.476311532239427e-05,
+      "loss": 1.0054,
+      "step": 11330
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4758493182343426e-05,
+      "loss": 1.0079,
+      "step": 11340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4753871042292582e-05,
+      "loss": 1.042,
+      "step": 11350
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4749248902241739e-05,
+      "loss": 0.8029,
+      "step": 11360
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4744626762190894e-05,
+      "loss": 0.7557,
+      "step": 11370
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4740004622140052e-05,
+      "loss": 1.0472,
+      "step": 11380
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4735382482089209e-05,
+      "loss": 0.9712,
+      "step": 11390
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4730760342038365e-05,
+      "loss": 0.8649,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.472613820198752e-05,
+      "loss": 1.0485,
+      "step": 11410
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4721516061936679e-05,
+      "loss": 1.1004,
+      "step": 11420
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4716893921885835e-05,
+      "loss": 0.9052,
+      "step": 11430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.471227178183499e-05,
+      "loss": 0.9219,
+      "step": 11440
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4707649641784147e-05,
+      "loss": 1.0321,
+      "step": 11450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4703027501733305e-05,
+      "loss": 0.9662,
+      "step": 11460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.469840536168246e-05,
+      "loss": 1.0374,
+      "step": 11470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4693783221631617e-05,
+      "loss": 1.162,
+      "step": 11480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4689161081580774e-05,
+      "loss": 0.882,
+      "step": 11490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4684538941529929e-05,
+      "loss": 1.0923,
+      "step": 11500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4679916801479085e-05,
+      "loss": 0.9422,
+      "step": 11510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4675294661428244e-05,
+      "loss": 1.0684,
+      "step": 11520
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4670672521377399e-05,
+      "loss": 1.0826,
+      "step": 11530
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4666050381326555e-05,
+      "loss": 0.9905,
+      "step": 11540
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4661428241275712e-05,
+      "loss": 1.0433,
+      "step": 11550
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4656806101224867e-05,
+      "loss": 1.0038,
+      "step": 11560
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4652183961174025e-05,
+      "loss": 1.1181,
+      "step": 11570
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4647561821123182e-05,
+      "loss": 0.8923,
+      "step": 11580
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4642939681072337e-05,
+      "loss": 1.0653,
+      "step": 11590
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4638317541021494e-05,
+      "loss": 0.9434,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4633695400970652e-05,
+      "loss": 1.058,
+      "step": 11610
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4629073260919807e-05,
+      "loss": 1.0592,
+      "step": 11620
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4624451120868964e-05,
+      "loss": 1.0518,
+      "step": 11630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.461982898081812e-05,
+      "loss": 0.9088,
+      "step": 11640
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4615206840767275e-05,
+      "loss": 0.9816,
+      "step": 11650
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4610584700716432e-05,
+      "loss": 0.9416,
+      "step": 11660
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.460596256066559e-05,
+      "loss": 0.9548,
+      "step": 11670
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4601340420614747e-05,
+      "loss": 0.938,
+      "step": 11680
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4596718280563902e-05,
+      "loss": 1.0378,
+      "step": 11690
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4592096140513058e-05,
+      "loss": 1.0393,
+      "step": 11700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4587474000462217e-05,
+      "loss": 0.9142,
+      "step": 11710
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4582851860411372e-05,
+      "loss": 1.0219,
+      "step": 11720
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4578229720360528e-05,
+      "loss": 0.8601,
+      "step": 11730
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4573607580309685e-05,
+      "loss": 0.8268,
+      "step": 11740
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.456898544025884e-05,
+      "loss": 1.2248,
+      "step": 11750
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4564363300207998e-05,
+      "loss": 1.0301,
+      "step": 11760
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4559741160157155e-05,
+      "loss": 1.0645,
+      "step": 11770
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.455511902010631e-05,
+      "loss": 0.9062,
+      "step": 11780
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4550496880055467e-05,
+      "loss": 0.9872,
+      "step": 11790
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4545874740004623e-05,
+      "loss": 0.9582,
+      "step": 11800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4541252599953778e-05,
+      "loss": 1.0177,
+      "step": 11810
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4536630459902937e-05,
+      "loss": 0.9663,
+      "step": 11820
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4532008319852093e-05,
+      "loss": 0.8469,
+      "step": 11830
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4527386179801248e-05,
+      "loss": 1.0612,
+      "step": 11840
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4522764039750405e-05,
+      "loss": 0.8591,
+      "step": 11850
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4518141899699563e-05,
+      "loss": 0.9398,
+      "step": 11860
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4513519759648718e-05,
+      "loss": 1.0452,
+      "step": 11870
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4508897619597875e-05,
+      "loss": 1.0871,
+      "step": 11880
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4504275479547032e-05,
+      "loss": 0.8888,
+      "step": 11890
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4499653339496186e-05,
+      "loss": 1.0585,
+      "step": 11900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4495031199445345e-05,
+      "loss": 0.8526,
+      "step": 11910
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4490409059394502e-05,
+      "loss": 0.9001,
+      "step": 11920
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4485786919343656e-05,
+      "loss": 0.9917,
+      "step": 11930
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4481164779292813e-05,
+      "loss": 0.9708,
+      "step": 11940
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.447654263924197e-05,
+      "loss": 0.9513,
+      "step": 11950
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4471920499191128e-05,
+      "loss": 0.9889,
+      "step": 11960
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4467298359140283e-05,
+      "loss": 0.7781,
+      "step": 11970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.446267621908944e-05,
+      "loss": 0.9567,
+      "step": 11980
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4458054079038596e-05,
+      "loss": 0.762,
+      "step": 11990
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4453431938987751e-05,
+      "loss": 1.0431,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.444880979893691e-05,
+      "loss": 0.8857,
+      "step": 12010
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4444187658886066e-05,
+      "loss": 0.8688,
+      "step": 12020
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4439565518835221e-05,
+      "loss": 1.0304,
+      "step": 12030
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4434943378784378e-05,
+      "loss": 1.0701,
+      "step": 12040
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4430321238733536e-05,
+      "loss": 0.9386,
+      "step": 12050
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4425699098682691e-05,
+      "loss": 0.8658,
+      "step": 12060
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4421076958631848e-05,
+      "loss": 0.8696,
+      "step": 12070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4416454818581005e-05,
+      "loss": 0.9862,
+      "step": 12080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.441183267853016e-05,
+      "loss": 1.0049,
+      "step": 12090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4407210538479316e-05,
+      "loss": 0.9312,
+      "step": 12100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4402588398428475e-05,
+      "loss": 1.1013,
+      "step": 12110
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.439796625837763e-05,
+      "loss": 0.8546,
+      "step": 12120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4393344118326786e-05,
+      "loss": 1.1044,
+      "step": 12130
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4388721978275943e-05,
+      "loss": 1.1246,
+      "step": 12140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4384099838225098e-05,
+      "loss": 0.7461,
+      "step": 12150
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4379477698174256e-05,
+      "loss": 1.0628,
+      "step": 12160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4374855558123413e-05,
+      "loss": 1.1409,
+      "step": 12170
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4370233418072568e-05,
+      "loss": 0.8435,
+      "step": 12180
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4365611278021724e-05,
+      "loss": 0.977,
+      "step": 12190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4360989137970883e-05,
+      "loss": 1.0368,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4356366997920038e-05,
+      "loss": 0.8651,
+      "step": 12210
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4351744857869194e-05,
+      "loss": 0.8629,
+      "step": 12220
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4347122717818351e-05,
+      "loss": 1.1128,
+      "step": 12230
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4342500577767508e-05,
+      "loss": 1.0261,
+      "step": 12240
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4337878437716663e-05,
+      "loss": 1.0011,
+      "step": 12250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4333256297665821e-05,
+      "loss": 0.9979,
+      "step": 12260
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4328634157614978e-05,
+      "loss": 0.952,
+      "step": 12270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4324012017564133e-05,
+      "loss": 1.2788,
+      "step": 12280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.431938987751329e-05,
+      "loss": 0.9716,
+      "step": 12290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4314767737462448e-05,
+      "loss": 0.9811,
+      "step": 12300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4310145597411603e-05,
+      "loss": 0.9437,
+      "step": 12310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.430552345736076e-05,
+      "loss": 0.8387,
+      "step": 12320
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4300901317309916e-05,
+      "loss": 0.9143,
+      "step": 12330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4296279177259071e-05,
+      "loss": 0.6704,
+      "step": 12340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.429165703720823e-05,
+      "loss": 0.8004,
+      "step": 12350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4287034897157386e-05,
+      "loss": 0.8278,
+      "step": 12360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4282412757106541e-05,
+      "loss": 0.9079,
+      "step": 12370
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4277790617055698e-05,
+      "loss": 0.8253,
+      "step": 12380
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4273168477004856e-05,
+      "loss": 0.9644,
+      "step": 12390
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.426854633695401e-05,
+      "loss": 1.0033,
+      "step": 12400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4263924196903168e-05,
+      "loss": 1.0063,
+      "step": 12410
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4259302056852324e-05,
+      "loss": 1.2337,
+      "step": 12420
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.425467991680148e-05,
+      "loss": 0.9911,
+      "step": 12430
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4250057776750636e-05,
+      "loss": 0.9296,
+      "step": 12440
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4245435636699794e-05,
+      "loss": 1.053,
+      "step": 12450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4240813496648949e-05,
+      "loss": 1.0077,
+      "step": 12460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4236191356598106e-05,
+      "loss": 0.9427,
+      "step": 12470
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4231569216547262e-05,
+      "loss": 1.0463,
+      "step": 12480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4226947076496417e-05,
+      "loss": 0.8678,
+      "step": 12490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4222324936445576e-05,
+      "loss": 0.879,
+      "step": 12500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4217702796394732e-05,
+      "loss": 1.0596,
+      "step": 12510
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4213080656343889e-05,
+      "loss": 0.721,
+      "step": 12520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4208458516293044e-05,
+      "loss": 1.0289,
+      "step": 12530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4203836376242202e-05,
+      "loss": 0.8252,
+      "step": 12540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4199214236191359e-05,
+      "loss": 1.1602,
+      "step": 12550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4194592096140514e-05,
+      "loss": 0.7575,
+      "step": 12560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.418996995608967e-05,
+      "loss": 0.7874,
+      "step": 12570
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4185347816038827e-05,
+      "loss": 0.9707,
+      "step": 12580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4180725675987982e-05,
+      "loss": 1.0439,
+      "step": 12590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.417610353593714e-05,
+      "loss": 0.7911,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4171481395886297e-05,
+      "loss": 0.8945,
+      "step": 12610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4166859255835452e-05,
+      "loss": 0.9418,
+      "step": 12620
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4162237115784609e-05,
+      "loss": 1.0702,
+      "step": 12630
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4157614975733767e-05,
+      "loss": 0.8369,
+      "step": 12640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4152992835682922e-05,
+      "loss": 0.9746,
+      "step": 12650
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4148370695632079e-05,
+      "loss": 0.9298,
+      "step": 12660
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4143748555581236e-05,
+      "loss": 1.0635,
+      "step": 12670
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.413912641553039e-05,
+      "loss": 1.0415,
+      "step": 12680
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4134504275479549e-05,
+      "loss": 0.8033,
+      "step": 12690
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4129882135428706e-05,
+      "loss": 0.9983,
+      "step": 12700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.412525999537786e-05,
+      "loss": 1.0192,
+      "step": 12710
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4120637855327017e-05,
+      "loss": 0.9094,
+      "step": 12720
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4116015715276174e-05,
+      "loss": 1.0941,
+      "step": 12730
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4111393575225329e-05,
+      "loss": 0.9847,
+      "step": 12740
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4106771435174487e-05,
+      "loss": 0.7691,
+      "step": 12750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4102149295123644e-05,
+      "loss": 1.0179,
+      "step": 12760
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4097527155072799e-05,
+      "loss": 0.8985,
+      "step": 12770
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4092905015021955e-05,
+      "loss": 0.8682,
+      "step": 12780
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4088282874971114e-05,
+      "loss": 0.7668,
+      "step": 12790
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.408366073492027e-05,
+      "loss": 0.9481,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4079038594869425e-05,
+      "loss": 1.0551,
+      "step": 12810
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4074416454818582e-05,
+      "loss": 1.0996,
+      "step": 12820
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.406979431476774e-05,
+      "loss": 1.0142,
+      "step": 12830
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4065172174716895e-05,
+      "loss": 0.8044,
+      "step": 12840
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4060550034666052e-05,
+      "loss": 1.1479,
+      "step": 12850
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4055927894615209e-05,
+      "loss": 0.9294,
+      "step": 12860
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4051305754564364e-05,
+      "loss": 0.8662,
+      "step": 12870
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.404668361451352e-05,
+      "loss": 0.9874,
+      "step": 12880
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4042061474462679e-05,
+      "loss": 0.9288,
+      "step": 12890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4037439334411834e-05,
+      "loss": 1.0914,
+      "step": 12900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.403281719436099e-05,
+      "loss": 0.924,
+      "step": 12910
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4028195054310147e-05,
+      "loss": 1.0017,
+      "step": 12920
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4023572914259302e-05,
+      "loss": 1.0276,
+      "step": 12930
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.401895077420846e-05,
+      "loss": 1.253,
+      "step": 12940
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4014328634157617e-05,
+      "loss": 0.8551,
+      "step": 12950
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4009706494106772e-05,
+      "loss": 1.2152,
+      "step": 12960
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4005084354055928e-05,
+      "loss": 0.8197,
+      "step": 12970
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4000462214005087e-05,
+      "loss": 0.8638,
+      "step": 12980
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3995840073954242e-05,
+      "loss": 1.0175,
+      "step": 12990
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3991217933903398e-05,
+      "loss": 0.8506,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3986595793852555e-05,
+      "loss": 0.8363,
+      "step": 13010
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.398197365380171e-05,
+      "loss": 1.1258,
+      "step": 13020
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3977351513750867e-05,
+      "loss": 0.975,
+      "step": 13030
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3972729373700025e-05,
+      "loss": 1.0422,
+      "step": 13040
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.396810723364918e-05,
+      "loss": 0.958,
+      "step": 13050
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3963485093598337e-05,
+      "loss": 0.7382,
+      "step": 13060
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3958862953547493e-05,
+      "loss": 0.7985,
+      "step": 13070
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3954240813496652e-05,
+      "loss": 1.0844,
+      "step": 13080
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3949618673445807e-05,
+      "loss": 0.7485,
+      "step": 13090
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3944996533394963e-05,
+      "loss": 1.1303,
+      "step": 13100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.394037439334412e-05,
+      "loss": 1.0929,
+      "step": 13110
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3935752253293275e-05,
+      "loss": 0.8365,
+      "step": 13120
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3931130113242433e-05,
+      "loss": 0.9691,
+      "step": 13130
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.392650797319159e-05,
+      "loss": 1.0281,
+      "step": 13140
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3921885833140745e-05,
+      "loss": 1.1541,
+      "step": 13150
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3917263693089902e-05,
+      "loss": 1.0945,
+      "step": 13160
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3912641553039058e-05,
+      "loss": 0.8986,
+      "step": 13170
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3908019412988213e-05,
+      "loss": 0.8195,
+      "step": 13180
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3903397272937372e-05,
+      "loss": 0.9967,
+      "step": 13190
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3898775132886528e-05,
+      "loss": 0.8814,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3894152992835683e-05,
+      "loss": 1.037,
+      "step": 13210
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.388953085278484e-05,
+      "loss": 0.8045,
+      "step": 13220
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3884908712733998e-05,
+      "loss": 0.966,
+      "step": 13230
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3880286572683153e-05,
+      "loss": 1.0152,
+      "step": 13240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.387566443263231e-05,
+      "loss": 1.0469,
+      "step": 13250
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3871042292581466e-05,
+      "loss": 0.8645,
+      "step": 13260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3866420152530621e-05,
+      "loss": 0.9796,
+      "step": 13270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.386179801247978e-05,
+      "loss": 0.8448,
+      "step": 13280
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3857175872428936e-05,
+      "loss": 0.9861,
+      "step": 13290
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3852553732378091e-05,
+      "loss": 1.1515,
+      "step": 13300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3847931592327248e-05,
+      "loss": 1.1197,
+      "step": 13310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3843309452276405e-05,
+      "loss": 0.877,
+      "step": 13320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.383868731222556e-05,
+      "loss": 0.8558,
+      "step": 13330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3834065172174718e-05,
+      "loss": 0.8728,
+      "step": 13340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3829443032123875e-05,
+      "loss": 0.9648,
+      "step": 13350
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3824820892073031e-05,
+      "loss": 1.0466,
+      "step": 13360
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3820198752022186e-05,
+      "loss": 1.0684,
+      "step": 13370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3815576611971345e-05,
+      "loss": 0.7109,
+      "step": 13380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3810954471920501e-05,
+      "loss": 0.9178,
+      "step": 13390
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3806332331869656e-05,
+      "loss": 0.9577,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3801710191818813e-05,
+      "loss": 0.9842,
+      "step": 13410
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3797088051767971e-05,
+      "loss": 0.9871,
+      "step": 13420
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3792465911717126e-05,
+      "loss": 0.9904,
+      "step": 13430
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3787843771666283e-05,
+      "loss": 0.9244,
+      "step": 13440
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.378322163161544e-05,
+      "loss": 0.9307,
+      "step": 13450
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3778599491564595e-05,
+      "loss": 0.9572,
+      "step": 13460
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3773977351513753e-05,
+      "loss": 0.9783,
+      "step": 13470
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.376935521146291e-05,
+      "loss": 0.9214,
+      "step": 13480
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3764733071412064e-05,
+      "loss": 1.0805,
+      "step": 13490
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3760110931361221e-05,
+      "loss": 0.998,
+      "step": 13500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3755488791310378e-05,
+      "loss": 0.9469,
+      "step": 13510
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3750866651259533e-05,
+      "loss": 1.1012,
+      "step": 13520
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3746244511208691e-05,
+      "loss": 1.0108,
+      "step": 13530
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3741622371157848e-05,
+      "loss": 1.263,
+      "step": 13540
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3737000231107003e-05,
+      "loss": 0.9464,
+      "step": 13550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.373237809105616e-05,
+      "loss": 0.977,
+      "step": 13560
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3727755951005318e-05,
+      "loss": 0.9902,
+      "step": 13570
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3723133810954473e-05,
+      "loss": 0.9126,
+      "step": 13580
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.371851167090363e-05,
+      "loss": 1.0469,
+      "step": 13590
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3713889530852786e-05,
+      "loss": 1.1612,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3709267390801941e-05,
+      "loss": 0.7741,
+      "step": 13610
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.37046452507511e-05,
+      "loss": 0.8384,
+      "step": 13620
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3700023110700256e-05,
+      "loss": 0.9364,
+      "step": 13630
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3695400970649413e-05,
+      "loss": 1.118,
+      "step": 13640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3690778830598568e-05,
+      "loss": 1.0652,
+      "step": 13650
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3686156690547724e-05,
+      "loss": 1.1095,
+      "step": 13660
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3681534550496883e-05,
+      "loss": 1.0942,
+      "step": 13670
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3676912410446038e-05,
+      "loss": 1.0526,
+      "step": 13680
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3672290270395194e-05,
+      "loss": 0.9508,
+      "step": 13690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3667668130344351e-05,
+      "loss": 0.8237,
+      "step": 13700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3663045990293506e-05,
+      "loss": 0.898,
+      "step": 13710
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3658423850242664e-05,
+      "loss": 0.8493,
+      "step": 13720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3653801710191821e-05,
+      "loss": 0.8054,
+      "step": 13730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3649179570140976e-05,
+      "loss": 0.912,
+      "step": 13740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3644557430090132e-05,
+      "loss": 0.9517,
+      "step": 13750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3639935290039291e-05,
+      "loss": 1.0342,
+      "step": 13760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3635313149988446e-05,
+      "loss": 0.8068,
+      "step": 13770
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3630691009937602e-05,
+      "loss": 0.9485,
+      "step": 13780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3626068869886759e-05,
+      "loss": 0.788,
+      "step": 13790
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3621446729835914e-05,
+      "loss": 0.9772,
+      "step": 13800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.361682458978507e-05,
+      "loss": 0.8724,
+      "step": 13810
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3612202449734229e-05,
+      "loss": 1.1166,
+      "step": 13820
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3607580309683384e-05,
+      "loss": 0.9805,
+      "step": 13830
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.360295816963254e-05,
+      "loss": 1.0056,
+      "step": 13840
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3598336029581697e-05,
+      "loss": 0.9986,
+      "step": 13850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3593713889530852e-05,
+      "loss": 0.8805,
+      "step": 13860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.358909174948001e-05,
+      "loss": 0.8797,
+      "step": 13870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3584469609429167e-05,
+      "loss": 1.1309,
+      "step": 13880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3579847469378322e-05,
+      "loss": 1.0029,
+      "step": 13890
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3575225329327479e-05,
+      "loss": 0.8018,
+      "step": 13900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3570603189276637e-05,
+      "loss": 0.8749,
+      "step": 13910
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3565981049225794e-05,
+      "loss": 0.955,
+      "step": 13920
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3561358909174949e-05,
+      "loss": 1.0201,
+      "step": 13930
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3556736769124106e-05,
+      "loss": 0.9343,
+      "step": 13940
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3552114629073262e-05,
+      "loss": 0.9531,
+      "step": 13950
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3547492489022417e-05,
+      "loss": 1.0455,
+      "step": 13960
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3542870348971576e-05,
+      "loss": 0.9352,
+      "step": 13970
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3538248208920732e-05,
+      "loss": 0.8862,
+      "step": 13980
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3533626068869887e-05,
+      "loss": 0.8654,
+      "step": 13990
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3529003928819044e-05,
+      "loss": 0.8379,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3524381788768202e-05,
+      "loss": 0.9504,
+      "step": 14010
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3519759648717357e-05,
+      "loss": 0.8511,
+      "step": 14020
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3515137508666514e-05,
+      "loss": 1.196,
+      "step": 14030
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.351051536861567e-05,
+      "loss": 0.9663,
+      "step": 14040
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3505893228564825e-05,
+      "loss": 0.8993,
+      "step": 14050
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3501271088513984e-05,
+      "loss": 0.8275,
+      "step": 14060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.349664894846314e-05,
+      "loss": 0.9566,
+      "step": 14070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3492026808412295e-05,
+      "loss": 0.952,
+      "step": 14080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3487404668361452e-05,
+      "loss": 1.0835,
+      "step": 14090
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3482782528310609e-05,
+      "loss": 1.0976,
+      "step": 14100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3478160388259764e-05,
+      "loss": 0.8811,
+      "step": 14110
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3473538248208922e-05,
+      "loss": 0.8965,
+      "step": 14120
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3468916108158079e-05,
+      "loss": 1.1536,
+      "step": 14130
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3464293968107234e-05,
+      "loss": 0.8019,
+      "step": 14140
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.345967182805639e-05,
+      "loss": 1.0355,
+      "step": 14150
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3455049688005549e-05,
+      "loss": 0.8318,
+      "step": 14160
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3450427547954704e-05,
+      "loss": 0.9346,
+      "step": 14170
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.344580540790386e-05,
+      "loss": 0.7433,
+      "step": 14180
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3441183267853017e-05,
+      "loss": 0.647,
+      "step": 14190
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3436561127802175e-05,
+      "loss": 0.8202,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.343193898775133e-05,
+      "loss": 0.8467,
+      "step": 14210
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3427316847700487e-05,
+      "loss": 0.6931,
+      "step": 14220
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3422694707649644e-05,
+      "loss": 0.9951,
+      "step": 14230
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3418072567598799e-05,
+      "loss": 0.8928,
+      "step": 14240
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3413450427547955e-05,
+      "loss": 0.8897,
+      "step": 14250
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3408828287497114e-05,
+      "loss": 0.6943,
+      "step": 14260
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3404206147446269e-05,
+      "loss": 0.9858,
+      "step": 14270
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3399584007395425e-05,
+      "loss": 0.9546,
+      "step": 14280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3394961867344582e-05,
+      "loss": 1.1405,
+      "step": 14290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3390339727293737e-05,
+      "loss": 1.2061,
+      "step": 14300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3385717587242895e-05,
+      "loss": 1.1649,
+      "step": 14310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3381095447192052e-05,
+      "loss": 0.8602,
+      "step": 14320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3376473307141207e-05,
+      "loss": 0.9874,
+      "step": 14330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3371851167090363e-05,
+      "loss": 0.8943,
+      "step": 14340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3367229027039522e-05,
+      "loss": 1.0673,
+      "step": 14350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3362606886988677e-05,
+      "loss": 0.8266,
+      "step": 14360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3357984746937833e-05,
+      "loss": 1.0161,
+      "step": 14370
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.335336260688699e-05,
+      "loss": 0.988,
+      "step": 14380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3348740466836145e-05,
+      "loss": 0.826,
+      "step": 14390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3344118326785302e-05,
+      "loss": 0.9313,
+      "step": 14400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.333949618673446e-05,
+      "loss": 0.8041,
+      "step": 14410
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3334874046683615e-05,
+      "loss": 0.9973,
+      "step": 14420
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3330251906632772e-05,
+      "loss": 0.9554,
+      "step": 14430
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3325629766581928e-05,
+      "loss": 0.9547,
+      "step": 14440
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3321007626531083e-05,
+      "loss": 1.0297,
+      "step": 14450
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3316385486480242e-05,
+      "loss": 0.8907,
+      "step": 14460
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3311763346429398e-05,
+      "loss": 0.7738,
+      "step": 14470
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3307141206378555e-05,
+      "loss": 0.8127,
+      "step": 14480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.330251906632771e-05,
+      "loss": 1.0174,
+      "step": 14490
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3297896926276868e-05,
+      "loss": 0.8417,
+      "step": 14500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3293274786226025e-05,
+      "loss": 0.8948,
+      "step": 14510
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.328865264617518e-05,
+      "loss": 1.0244,
+      "step": 14520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3284030506124337e-05,
+      "loss": 0.9476,
+      "step": 14530
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3279408366073495e-05,
+      "loss": 0.8979,
+      "step": 14540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.327478622602265e-05,
+      "loss": 1.124,
+      "step": 14550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3270164085971806e-05,
+      "loss": 0.9694,
+      "step": 14560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3265541945920963e-05,
+      "loss": 1.0713,
+      "step": 14570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3260919805870118e-05,
+      "loss": 1.0593,
+      "step": 14580
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3256297665819275e-05,
+      "loss": 1.0752,
+      "step": 14590
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3251675525768433e-05,
+      "loss": 0.9087,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3247053385717588e-05,
+      "loss": 0.9233,
+      "step": 14610
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3242431245666745e-05,
+      "loss": 1.0916,
+      "step": 14620
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3237809105615901e-05,
+      "loss": 1.1941,
+      "step": 14630
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3233186965565056e-05,
+      "loss": 0.7972,
+      "step": 14640
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3228564825514215e-05,
+      "loss": 0.9516,
+      "step": 14650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3223942685463371e-05,
+      "loss": 0.9108,
+      "step": 14660
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3219320545412526e-05,
+      "loss": 0.7156,
+      "step": 14670
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3214698405361683e-05,
+      "loss": 1.0102,
+      "step": 14680
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3210076265310841e-05,
+      "loss": 1.1032,
+      "step": 14690
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3205454125259996e-05,
+      "loss": 0.8861,
+      "step": 14700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3200831985209153e-05,
+      "loss": 0.7843,
+      "step": 14710
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.319620984515831e-05,
+      "loss": 1.0489,
+      "step": 14720
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3191587705107465e-05,
+      "loss": 0.8377,
+      "step": 14730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3186965565056621e-05,
+      "loss": 0.9635,
+      "step": 14740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.318234342500578e-05,
+      "loss": 0.8916,
+      "step": 14750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3177721284954936e-05,
+      "loss": 1.038,
+      "step": 14760
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3173099144904091e-05,
+      "loss": 0.9797,
+      "step": 14770
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3168477004853248e-05,
+      "loss": 1.0368,
+      "step": 14780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3163854864802406e-05,
+      "loss": 0.9836,
+      "step": 14790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3159232724751561e-05,
+      "loss": 0.767,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3154610584700718e-05,
+      "loss": 0.9999,
+      "step": 14810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3149988444649874e-05,
+      "loss": 0.9058,
+      "step": 14820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.314536630459903e-05,
+      "loss": 1.1025,
+      "step": 14830
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3140744164548188e-05,
+      "loss": 0.9538,
+      "step": 14840
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3136122024497344e-05,
+      "loss": 0.8634,
+      "step": 14850
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.31314998844465e-05,
+      "loss": 1.1052,
+      "step": 14860
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3126877744395656e-05,
+      "loss": 0.9102,
+      "step": 14870
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3122255604344813e-05,
+      "loss": 1.122,
+      "step": 14880
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3117633464293968e-05,
+      "loss": 0.9868,
+      "step": 14890
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3113011324243126e-05,
+      "loss": 0.8592,
+      "step": 14900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3108389184192283e-05,
+      "loss": 0.8948,
+      "step": 14910
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3103767044141438e-05,
+      "loss": 0.9035,
+      "step": 14920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3099144904090594e-05,
+      "loss": 0.8172,
+      "step": 14930
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3094522764039753e-05,
+      "loss": 0.8732,
+      "step": 14940
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3089900623988908e-05,
+      "loss": 0.9213,
+      "step": 14950
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3085278483938064e-05,
+      "loss": 0.9462,
+      "step": 14960
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3080656343887221e-05,
+      "loss": 0.7915,
+      "step": 14970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3076034203836376e-05,
+      "loss": 0.9668,
+      "step": 14980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3071412063785534e-05,
+      "loss": 1.2192,
+      "step": 14990
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3066789923734691e-05,
+      "loss": 1.0358,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3062167783683846e-05,
+      "loss": 0.9366,
+      "step": 15010
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3057545643633003e-05,
+      "loss": 0.863,
+      "step": 15020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.305292350358216e-05,
+      "loss": 0.8242,
+      "step": 15030
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3048301363531318e-05,
+      "loss": 0.8349,
+      "step": 15040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3043679223480473e-05,
+      "loss": 0.8349,
+      "step": 15050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.303905708342963e-05,
+      "loss": 1.0068,
+      "step": 15060
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3034434943378786e-05,
+      "loss": 1.012,
+      "step": 15070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.302981280332794e-05,
+      "loss": 1.0006,
+      "step": 15080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3025190663277099e-05,
+      "loss": 0.9619,
+      "step": 15090
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3020568523226256e-05,
+      "loss": 1.1012,
+      "step": 15100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.301594638317541e-05,
+      "loss": 0.8684,
+      "step": 15110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3011324243124567e-05,
+      "loss": 0.8619,
+      "step": 15120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3006702103073726e-05,
+      "loss": 0.8861,
+      "step": 15130
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.300207996302288e-05,
+      "loss": 1.0799,
+      "step": 15140
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2997457822972037e-05,
+      "loss": 0.8722,
+      "step": 15150
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2992835682921194e-05,
+      "loss": 0.9542,
+      "step": 15160
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2988213542870349e-05,
+      "loss": 1.018,
+      "step": 15170
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2983591402819506e-05,
+      "loss": 0.7793,
+      "step": 15180
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2978969262768664e-05,
+      "loss": 1.2477,
+      "step": 15190
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2974347122717819e-05,
+      "loss": 1.0946,
+      "step": 15200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2969724982666976e-05,
+      "loss": 0.9283,
+      "step": 15210
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2965102842616132e-05,
+      "loss": 0.9561,
+      "step": 15220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2960480702565287e-05,
+      "loss": 0.9131,
+      "step": 15230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2955858562514446e-05,
+      "loss": 0.87,
+      "step": 15240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2951236422463602e-05,
+      "loss": 0.9529,
+      "step": 15250
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2946614282412757e-05,
+      "loss": 0.8795,
+      "step": 15260
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2941992142361914e-05,
+      "loss": 1.0589,
+      "step": 15270
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2937370002311072e-05,
+      "loss": 1.01,
+      "step": 15280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2932747862260227e-05,
+      "loss": 1.0467,
+      "step": 15290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2928125722209384e-05,
+      "loss": 0.9169,
+      "step": 15300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.292350358215854e-05,
+      "loss": 0.8819,
+      "step": 15310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2918881442107697e-05,
+      "loss": 0.9926,
+      "step": 15320
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2914259302056852e-05,
+      "loss": 0.9967,
+      "step": 15330
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.290963716200601e-05,
+      "loss": 0.9289,
+      "step": 15340
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2905015021955167e-05,
+      "loss": 0.9977,
+      "step": 15350
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2900392881904322e-05,
+      "loss": 1.0408,
+      "step": 15360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2895770741853479e-05,
+      "loss": 1.448,
+      "step": 15370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2891148601802637e-05,
+      "loss": 0.9106,
+      "step": 15380
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2886526461751792e-05,
+      "loss": 0.8868,
+      "step": 15390
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2881904321700949e-05,
+      "loss": 1.0192,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2877282181650105e-05,
+      "loss": 1.0353,
+      "step": 15410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.287266004159926e-05,
+      "loss": 0.7549,
+      "step": 15420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2868037901548419e-05,
+      "loss": 0.9725,
+      "step": 15430
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2863415761497575e-05,
+      "loss": 0.8139,
+      "step": 15440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.285879362144673e-05,
+      "loss": 0.927,
+      "step": 15450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2854171481395887e-05,
+      "loss": 1.0661,
+      "step": 15460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2849549341345045e-05,
+      "loss": 1.2262,
+      "step": 15470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2844927201294199e-05,
+      "loss": 0.9763,
+      "step": 15480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2840305061243357e-05,
+      "loss": 0.9127,
+      "step": 15490
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2835682921192514e-05,
+      "loss": 0.8658,
+      "step": 15500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2831060781141669e-05,
+      "loss": 1.0184,
+      "step": 15510
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2826438641090825e-05,
+      "loss": 1.0071,
+      "step": 15520
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2821816501039984e-05,
+      "loss": 0.9027,
+      "step": 15530
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2817194360989139e-05,
+      "loss": 1.0072,
+      "step": 15540
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2812572220938295e-05,
+      "loss": 1.0087,
+      "step": 15550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2807950080887452e-05,
+      "loss": 0.8297,
+      "step": 15560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2803327940836607e-05,
+      "loss": 0.8271,
+      "step": 15570
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2798705800785765e-05,
+      "loss": 0.9747,
+      "step": 15580
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2794083660734922e-05,
+      "loss": 0.9536,
+      "step": 15590
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2789461520684078e-05,
+      "loss": 0.9513,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2784839380633233e-05,
+      "loss": 0.8257,
+      "step": 15610
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2780217240582392e-05,
+      "loss": 0.8386,
+      "step": 15620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2775595100531548e-05,
+      "loss": 1.0339,
+      "step": 15630
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2770972960480703e-05,
+      "loss": 0.9644,
+      "step": 15640
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.276635082042986e-05,
+      "loss": 0.9494,
+      "step": 15650
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2761728680379017e-05,
+      "loss": 1.0225,
+      "step": 15660
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2757106540328172e-05,
+      "loss": 0.9467,
+      "step": 15670
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.275248440027733e-05,
+      "loss": 1.1014,
+      "step": 15680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2747862260226487e-05,
+      "loss": 0.951,
+      "step": 15690
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2743240120175642e-05,
+      "loss": 1.0269,
+      "step": 15700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2738617980124798e-05,
+      "loss": 1.0021,
+      "step": 15710
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2733995840073957e-05,
+      "loss": 1.0161,
+      "step": 15720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2729373700023112e-05,
+      "loss": 0.8741,
+      "step": 15730
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2724751559972268e-05,
+      "loss": 1.0828,
+      "step": 15740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2720129419921425e-05,
+      "loss": 0.987,
+      "step": 15750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.271550727987058e-05,
+      "loss": 0.8826,
+      "step": 15760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2710885139819738e-05,
+      "loss": 0.8753,
+      "step": 15770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2706262999768895e-05,
+      "loss": 0.8076,
+      "step": 15780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.270164085971805e-05,
+      "loss": 0.7232,
+      "step": 15790
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2697018719667207e-05,
+      "loss": 1.0209,
+      "step": 15800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2692396579616363e-05,
+      "loss": 1.0569,
+      "step": 15810
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2687774439565518e-05,
+      "loss": 0.9336,
+      "step": 15820
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2683152299514677e-05,
+      "loss": 1.1245,
+      "step": 15830
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2678530159463833e-05,
+      "loss": 0.9681,
+      "step": 15840
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2673908019412988e-05,
+      "loss": 0.8814,
+      "step": 15850
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2669285879362145e-05,
+      "loss": 1.1138,
+      "step": 15860
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2664663739311303e-05,
+      "loss": 0.8845,
+      "step": 15870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.266004159926046e-05,
+      "loss": 0.6793,
+      "step": 15880
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2655419459209615e-05,
+      "loss": 0.8102,
+      "step": 15890
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2650797319158771e-05,
+      "loss": 0.8293,
+      "step": 15900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.264617517910793e-05,
+      "loss": 1.0285,
+      "step": 15910
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2641553039057085e-05,
+      "loss": 0.9356,
+      "step": 15920
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2636930899006241e-05,
+      "loss": 0.9779,
+      "step": 15930
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2632308758955398e-05,
+      "loss": 0.8969,
+      "step": 15940
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2627686618904553e-05,
+      "loss": 0.9557,
+      "step": 15950
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.262306447885371e-05,
+      "loss": 0.8807,
+      "step": 15960
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2618442338802868e-05,
+      "loss": 0.8267,
+      "step": 15970
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2613820198752023e-05,
+      "loss": 1.0266,
+      "step": 15980
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.260919805870118e-05,
+      "loss": 0.9285,
+      "step": 15990
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2604575918650336e-05,
+      "loss": 1.1206,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2599953778599491e-05,
+      "loss": 0.9301,
+      "step": 16010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.259533163854865e-05,
+      "loss": 0.9762,
+      "step": 16020
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2590709498497806e-05,
+      "loss": 0.8708,
+      "step": 16030
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2586087358446961e-05,
+      "loss": 0.82,
+      "step": 16040
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2581465218396118e-05,
+      "loss": 0.8505,
+      "step": 16050
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2576843078345276e-05,
+      "loss": 1.0103,
+      "step": 16060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2572220938294431e-05,
+      "loss": 1.1762,
+      "step": 16070
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2567598798243588e-05,
+      "loss": 1.0845,
+      "step": 16080
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2562976658192745e-05,
+      "loss": 0.7893,
+      "step": 16090
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.25583545181419e-05,
+      "loss": 1.1017,
+      "step": 16100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2553732378091056e-05,
+      "loss": 0.9965,
+      "step": 16110
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2549110238040215e-05,
+      "loss": 1.0152,
+      "step": 16120
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.254448809798937e-05,
+      "loss": 1.0363,
+      "step": 16130
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2539865957938526e-05,
+      "loss": 0.7661,
+      "step": 16140
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2535243817887683e-05,
+      "loss": 0.8881,
+      "step": 16150
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2530621677836841e-05,
+      "loss": 0.8518,
+      "step": 16160
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2525999537785996e-05,
+      "loss": 1.1407,
+      "step": 16170
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2521377397735153e-05,
+      "loss": 1.136,
+      "step": 16180
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.251675525768431e-05,
+      "loss": 0.7408,
+      "step": 16190
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2512133117633464e-05,
+      "loss": 0.8149,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2507510977582623e-05,
+      "loss": 0.8844,
+      "step": 16210
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.250288883753178e-05,
+      "loss": 0.9429,
+      "step": 16220
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2498266697480934e-05,
+      "loss": 1.1482,
+      "step": 16230
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2493644557430091e-05,
+      "loss": 0.8589,
+      "step": 16240
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2489022417379248e-05,
+      "loss": 0.9854,
+      "step": 16250
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2484400277328403e-05,
+      "loss": 0.9536,
+      "step": 16260
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2479778137277561e-05,
+      "loss": 1.0079,
+      "step": 16270
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2475155997226718e-05,
+      "loss": 0.8624,
+      "step": 16280
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2470533857175873e-05,
+      "loss": 1.1549,
+      "step": 16290
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.246591171712503e-05,
+      "loss": 1.0317,
+      "step": 16300
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2461289577074188e-05,
+      "loss": 0.9475,
+      "step": 16310
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2456667437023343e-05,
+      "loss": 0.9946,
+      "step": 16320
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.24520452969725e-05,
+      "loss": 0.9693,
+      "step": 16330
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2447423156921656e-05,
+      "loss": 0.9215,
+      "step": 16340
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.244280101687081e-05,
+      "loss": 0.9813,
+      "step": 16350
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.243817887681997e-05,
+      "loss": 0.9166,
+      "step": 16360
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2433556736769126e-05,
+      "loss": 1.047,
+      "step": 16370
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.242893459671828e-05,
+      "loss": 0.7509,
+      "step": 16380
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2424312456667437e-05,
+      "loss": 0.8114,
+      "step": 16390
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2419690316616594e-05,
+      "loss": 0.9135,
+      "step": 16400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2415068176565749e-05,
+      "loss": 0.8332,
+      "step": 16410
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2410446036514907e-05,
+      "loss": 0.8657,
+      "step": 16420
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2405823896464064e-05,
+      "loss": 1.1066,
+      "step": 16430
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.240120175641322e-05,
+      "loss": 0.795,
+      "step": 16440
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2396579616362376e-05,
+      "loss": 1.0422,
+      "step": 16450
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2391957476311534e-05,
+      "loss": 1.2049,
+      "step": 16460
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.238733533626069e-05,
+      "loss": 0.7672,
+      "step": 16470
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2382713196209846e-05,
+      "loss": 0.9221,
+      "step": 16480
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2378091056159002e-05,
+      "loss": 1.0626,
+      "step": 16490
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.237346891610816e-05,
+      "loss": 0.8369,
+      "step": 16500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2368846776057316e-05,
+      "loss": 0.8742,
+      "step": 16510
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2364224636006472e-05,
+      "loss": 1.1064,
+      "step": 16520
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2359602495955629e-05,
+      "loss": 0.8728,
+      "step": 16530
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2354980355904784e-05,
+      "loss": 0.9981,
+      "step": 16540
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2350358215853942e-05,
+      "loss": 1.1017,
+      "step": 16550
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2345736075803099e-05,
+      "loss": 0.9693,
+      "step": 16560
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2341113935752254e-05,
+      "loss": 1.1115,
+      "step": 16570
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.233649179570141e-05,
+      "loss": 0.8382,
+      "step": 16580
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2331869655650567e-05,
+      "loss": 0.784,
+      "step": 16590
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2327247515599722e-05,
+      "loss": 0.9859,
+      "step": 16600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.232262537554888e-05,
+      "loss": 1.1296,
+      "step": 16610
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2318003235498037e-05,
+      "loss": 1.0938,
+      "step": 16620
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2313381095447192e-05,
+      "loss": 1.1174,
+      "step": 16630
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2308758955396349e-05,
+      "loss": 0.7618,
+      "step": 16640
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2304136815345507e-05,
+      "loss": 0.8632,
+      "step": 16650
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2299514675294662e-05,
+      "loss": 0.9334,
+      "step": 16660
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2294892535243819e-05,
+      "loss": 0.6448,
+      "step": 16670
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2290270395192975e-05,
+      "loss": 0.8398,
+      "step": 16680
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.228564825514213e-05,
+      "loss": 0.8781,
+      "step": 16690
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2281026115091289e-05,
+      "loss": 1.1231,
+      "step": 16700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2276403975040445e-05,
+      "loss": 0.906,
+      "step": 16710
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2271781834989602e-05,
+      "loss": 0.9467,
+      "step": 16720
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2267159694938757e-05,
+      "loss": 0.9074,
+      "step": 16730
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2262537554887914e-05,
+      "loss": 0.8402,
+      "step": 16740
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2257915414837072e-05,
+      "loss": 0.9162,
+      "step": 16750
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2253293274786227e-05,
+      "loss": 0.8457,
+      "step": 16760
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2248671134735384e-05,
+      "loss": 1.0204,
+      "step": 16770
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.224404899468454e-05,
+      "loss": 0.8309,
+      "step": 16780
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2239426854633695e-05,
+      "loss": 1.0883,
+      "step": 16790
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2234804714582854e-05,
+      "loss": 1.0621,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.223018257453201e-05,
+      "loss": 0.9601,
+      "step": 16810
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2225560434481165e-05,
+      "loss": 0.9289,
+      "step": 16820
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2220938294430322e-05,
+      "loss": 0.9418,
+      "step": 16830
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.221631615437948e-05,
+      "loss": 0.9748,
+      "step": 16840
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2211694014328635e-05,
+      "loss": 1.1388,
+      "step": 16850
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2207071874277792e-05,
+      "loss": 0.7877,
+      "step": 16860
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2202449734226949e-05,
+      "loss": 1.0999,
+      "step": 16870
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2197827594176104e-05,
+      "loss": 0.9263,
+      "step": 16880
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.219320545412526e-05,
+      "loss": 0.8368,
+      "step": 16890
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2188583314074419e-05,
+      "loss": 1.1056,
+      "step": 16900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2183961174023573e-05,
+      "loss": 0.8715,
+      "step": 16910
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.217933903397273e-05,
+      "loss": 0.9038,
+      "step": 16920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2174716893921887e-05,
+      "loss": 0.861,
+      "step": 16930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2170094753871042e-05,
+      "loss": 0.8111,
+      "step": 16940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.21654726138202e-05,
+      "loss": 1.0428,
+      "step": 16950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2160850473769357e-05,
+      "loss": 0.8901,
+      "step": 16960
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2156228333718512e-05,
+      "loss": 0.9122,
+      "step": 16970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2151606193667668e-05,
+      "loss": 0.7994,
+      "step": 16980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2146984053616827e-05,
+      "loss": 0.8432,
+      "step": 16990
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2142361913565983e-05,
+      "loss": 1.1026,
+      "step": 17000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2137739773515138e-05,
+      "loss": 0.8646,
+      "step": 17010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2133117633464295e-05,
+      "loss": 1.0822,
+      "step": 17020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2128495493413452e-05,
+      "loss": 1.0431,
+      "step": 17030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2123873353362607e-05,
+      "loss": 1.0974,
+      "step": 17040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2119251213311765e-05,
+      "loss": 0.9091,
+      "step": 17050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2114629073260922e-05,
+      "loss": 0.9877,
+      "step": 17060
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2110006933210077e-05,
+      "loss": 0.9514,
+      "step": 17070
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2105384793159233e-05,
+      "loss": 0.9849,
+      "step": 17080
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2100762653108392e-05,
+      "loss": 0.8109,
+      "step": 17090
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2096140513057547e-05,
+      "loss": 0.7389,
+      "step": 17100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2091518373006703e-05,
+      "loss": 0.8729,
+      "step": 17110
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.208689623295586e-05,
+      "loss": 0.8403,
+      "step": 17120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2082274092905015e-05,
+      "loss": 1.1238,
+      "step": 17130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2077651952854173e-05,
+      "loss": 1.1379,
+      "step": 17140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.207302981280333e-05,
+      "loss": 0.9131,
+      "step": 17150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2068407672752485e-05,
+      "loss": 0.8276,
+      "step": 17160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2063785532701641e-05,
+      "loss": 0.8535,
+      "step": 17170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2059163392650798e-05,
+      "loss": 0.9071,
+      "step": 17180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2054541252599953e-05,
+      "loss": 1.0166,
+      "step": 17190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2049919112549111e-05,
+      "loss": 1.022,
+      "step": 17200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2045296972498268e-05,
+      "loss": 0.9587,
+      "step": 17210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2040674832447423e-05,
+      "loss": 0.7878,
+      "step": 17220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.203605269239658e-05,
+      "loss": 0.8295,
+      "step": 17230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2031430552345738e-05,
+      "loss": 0.9604,
+      "step": 17240
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2026808412294893e-05,
+      "loss": 1.0321,
+      "step": 17250
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.202218627224405e-05,
+      "loss": 0.7573,
+      "step": 17260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2017564132193206e-05,
+      "loss": 0.9216,
+      "step": 17270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2012941992142365e-05,
+      "loss": 1.1622,
+      "step": 17280
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.200831985209152e-05,
+      "loss": 1.0677,
+      "step": 17290
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2003697712040676e-05,
+      "loss": 0.9959,
+      "step": 17300
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.6941147581239256,
+      "eval_loss": 0.8523356318473816,
+      "eval_runtime": 367.7187,
+      "eval_samples_per_second": 33.224,
+      "eval_steps_per_second": 4.155,
+      "step": 17308
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1999075571989833e-05,
+      "loss": 0.9337,
+      "step": 17310
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1994453431938988e-05,
+      "loss": 0.9174,
+      "step": 17320
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1989831291888145e-05,
+      "loss": 0.886,
+      "step": 17330
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1985209151837303e-05,
+      "loss": 0.8022,
+      "step": 17340
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1980587011786458e-05,
+      "loss": 0.9964,
+      "step": 17350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1975964871735615e-05,
+      "loss": 0.7045,
+      "step": 17360
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1971342731684771e-05,
+      "loss": 0.7849,
+      "step": 17370
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1966720591633926e-05,
+      "loss": 0.75,
+      "step": 17380
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1962098451583085e-05,
+      "loss": 0.6988,
+      "step": 17390
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1957476311532241e-05,
+      "loss": 0.8258,
+      "step": 17400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1952854171481396e-05,
+      "loss": 0.7548,
+      "step": 17410
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1948232031430553e-05,
+      "loss": 0.8017,
+      "step": 17420
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1943609891379711e-05,
+      "loss": 0.98,
+      "step": 17430
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1938987751328866e-05,
+      "loss": 0.7545,
+      "step": 17440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1934365611278023e-05,
+      "loss": 0.8506,
+      "step": 17450
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.192974347122718e-05,
+      "loss": 1.0343,
+      "step": 17460
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1925121331176334e-05,
+      "loss": 0.9846,
+      "step": 17470
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1920499191125491e-05,
+      "loss": 0.9844,
+      "step": 17480
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.191587705107465e-05,
+      "loss": 1.1832,
+      "step": 17490
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1911254911023804e-05,
+      "loss": 0.9722,
+      "step": 17500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1906632770972961e-05,
+      "loss": 0.906,
+      "step": 17510
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1902010630922118e-05,
+      "loss": 1.3125,
+      "step": 17520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1897388490871273e-05,
+      "loss": 0.6583,
+      "step": 17530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1892766350820431e-05,
+      "loss": 1.0258,
+      "step": 17540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1888144210769588e-05,
+      "loss": 0.8902,
+      "step": 17550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1883522070718743e-05,
+      "loss": 0.7855,
+      "step": 17560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.18788999306679e-05,
+      "loss": 1.005,
+      "step": 17570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1874277790617058e-05,
+      "loss": 0.8785,
+      "step": 17580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1869655650566214e-05,
+      "loss": 0.827,
+      "step": 17590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.186503351051537e-05,
+      "loss": 0.8907,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1860411370464526e-05,
+      "loss": 0.8739,
+      "step": 17610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1855789230413684e-05,
+      "loss": 0.8089,
+      "step": 17620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.185116709036284e-05,
+      "loss": 1.1024,
+      "step": 17630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1846544950311996e-05,
+      "loss": 0.7945,
+      "step": 17640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1841922810261153e-05,
+      "loss": 0.8869,
+      "step": 17650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1837300670210308e-05,
+      "loss": 0.9039,
+      "step": 17660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1832678530159464e-05,
+      "loss": 0.745,
+      "step": 17670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1828056390108623e-05,
+      "loss": 0.9384,
+      "step": 17680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1823434250057777e-05,
+      "loss": 0.942,
+      "step": 17690
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1818812110006934e-05,
+      "loss": 1.0015,
+      "step": 17700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.181418996995609e-05,
+      "loss": 1.0314,
+      "step": 17710
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1809567829905246e-05,
+      "loss": 1.0882,
+      "step": 17720
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1804945689854404e-05,
+      "loss": 0.8338,
+      "step": 17730
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.180032354980356e-05,
+      "loss": 0.7123,
+      "step": 17740
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1795701409752716e-05,
+      "loss": 1.1736,
+      "step": 17750
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1791079269701872e-05,
+      "loss": 0.9608,
+      "step": 17760
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.178645712965103e-05,
+      "loss": 0.8023,
+      "step": 17770
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1781834989600186e-05,
+      "loss": 0.8926,
+      "step": 17780
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1777212849549342e-05,
+      "loss": 1.0936,
+      "step": 17790
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1772590709498499e-05,
+      "loss": 0.8435,
+      "step": 17800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1767968569447654e-05,
+      "loss": 0.9201,
+      "step": 17810
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.176334642939681e-05,
+      "loss": 0.9303,
+      "step": 17820
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1758724289345969e-05,
+      "loss": 0.9556,
+      "step": 17830
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1754102149295124e-05,
+      "loss": 0.8094,
+      "step": 17840
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.174948000924428e-05,
+      "loss": 0.9807,
+      "step": 17850
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1744857869193437e-05,
+      "loss": 0.8264,
+      "step": 17860
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1740235729142596e-05,
+      "loss": 0.8501,
+      "step": 17870
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.173561358909175e-05,
+      "loss": 0.9986,
+      "step": 17880
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1730991449040907e-05,
+      "loss": 0.7975,
+      "step": 17890
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1726369308990064e-05,
+      "loss": 0.6342,
+      "step": 17900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1721747168939219e-05,
+      "loss": 0.8086,
+      "step": 17910
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1717125028888377e-05,
+      "loss": 0.9657,
+      "step": 17920
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1712502888837534e-05,
+      "loss": 0.9378,
+      "step": 17930
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1707880748786689e-05,
+      "loss": 0.8477,
+      "step": 17940
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1703258608735845e-05,
+      "loss": 0.7862,
+      "step": 17950
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1698636468685002e-05,
+      "loss": 0.7966,
+      "step": 17960
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1694014328634157e-05,
+      "loss": 0.885,
+      "step": 17970
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1689392188583315e-05,
+      "loss": 0.9679,
+      "step": 17980
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1684770048532472e-05,
+      "loss": 0.8629,
+      "step": 17990
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1680147908481627e-05,
+      "loss": 0.84,
+      "step": 18000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1675525768430784e-05,
+      "loss": 0.973,
+      "step": 18010
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1670903628379942e-05,
+      "loss": 1.0841,
+      "step": 18020
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1666281488329097e-05,
+      "loss": 0.8922,
+      "step": 18030
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1661659348278254e-05,
+      "loss": 0.8429,
+      "step": 18040
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.165703720822741e-05,
+      "loss": 0.9563,
+      "step": 18050
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1652415068176565e-05,
+      "loss": 0.8386,
+      "step": 18060
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1647792928125724e-05,
+      "loss": 1.058,
+      "step": 18070
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.164317078807488e-05,
+      "loss": 1.1503,
+      "step": 18080
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1638548648024035e-05,
+      "loss": 0.8895,
+      "step": 18090
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1633926507973192e-05,
+      "loss": 0.9213,
+      "step": 18100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1629304367922349e-05,
+      "loss": 0.8108,
+      "step": 18110
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1624682227871504e-05,
+      "loss": 0.9893,
+      "step": 18120
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1620060087820662e-05,
+      "loss": 0.8444,
+      "step": 18130
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1615437947769819e-05,
+      "loss": 0.7761,
+      "step": 18140
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1610815807718975e-05,
+      "loss": 0.8771,
+      "step": 18150
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.160619366766813e-05,
+      "loss": 0.9343,
+      "step": 18160
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1601571527617289e-05,
+      "loss": 1.075,
+      "step": 18170
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1596949387566445e-05,
+      "loss": 0.9296,
+      "step": 18180
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.15923272475156e-05,
+      "loss": 1.1034,
+      "step": 18190
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1587705107464757e-05,
+      "loss": 0.9955,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1583082967413915e-05,
+      "loss": 0.7911,
+      "step": 18210
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.157846082736307e-05,
+      "loss": 0.9218,
+      "step": 18220
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1573838687312227e-05,
+      "loss": 0.9925,
+      "step": 18230
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1569216547261383e-05,
+      "loss": 1.0658,
+      "step": 18240
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1564594407210538e-05,
+      "loss": 0.9982,
+      "step": 18250
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1559972267159695e-05,
+      "loss": 0.727,
+      "step": 18260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1555350127108853e-05,
+      "loss": 0.75,
+      "step": 18270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1550727987058008e-05,
+      "loss": 0.8871,
+      "step": 18280
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1546105847007165e-05,
+      "loss": 0.762,
+      "step": 18290
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1541483706956322e-05,
+      "loss": 0.7458,
+      "step": 18300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1536861566905477e-05,
+      "loss": 0.8941,
+      "step": 18310
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1532239426854635e-05,
+      "loss": 1.04,
+      "step": 18320
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1527617286803792e-05,
+      "loss": 0.7893,
+      "step": 18330
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1522995146752947e-05,
+      "loss": 0.8084,
+      "step": 18340
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1518373006702103e-05,
+      "loss": 0.6786,
+      "step": 18350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1513750866651262e-05,
+      "loss": 0.7869,
+      "step": 18360
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1509128726600417e-05,
+      "loss": 1.0144,
+      "step": 18370
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1504506586549573e-05,
+      "loss": 1.2085,
+      "step": 18380
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.149988444649873e-05,
+      "loss": 0.9414,
+      "step": 18390
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1495262306447885e-05,
+      "loss": 0.6261,
+      "step": 18400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1490640166397042e-05,
+      "loss": 0.8556,
+      "step": 18410
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.14860180263462e-05,
+      "loss": 0.9609,
+      "step": 18420
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1481395886295357e-05,
+      "loss": 0.7805,
+      "step": 18430
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1476773746244512e-05,
+      "loss": 0.8015,
+      "step": 18440
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1472151606193668e-05,
+      "loss": 0.8033,
+      "step": 18450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1467529466142827e-05,
+      "loss": 1.0236,
+      "step": 18460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1462907326091982e-05,
+      "loss": 0.6764,
+      "step": 18470
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1458285186041138e-05,
+      "loss": 0.8879,
+      "step": 18480
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1453663045990295e-05,
+      "loss": 0.8559,
+      "step": 18490
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.144904090593945e-05,
+      "loss": 0.8797,
+      "step": 18500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1444418765888608e-05,
+      "loss": 0.9126,
+      "step": 18510
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1439796625837765e-05,
+      "loss": 1.0996,
+      "step": 18520
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.143517448578692e-05,
+      "loss": 1.1183,
+      "step": 18530
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1430552345736076e-05,
+      "loss": 0.917,
+      "step": 18540
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1425930205685235e-05,
+      "loss": 1.0565,
+      "step": 18550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1421308065634388e-05,
+      "loss": 0.8611,
+      "step": 18560
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1416685925583546e-05,
+      "loss": 1.0068,
+      "step": 18570
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1412063785532703e-05,
+      "loss": 0.9017,
+      "step": 18580
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1407441645481858e-05,
+      "loss": 1.035,
+      "step": 18590
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1402819505431015e-05,
+      "loss": 1.0758,
+      "step": 18600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1398197365380173e-05,
+      "loss": 0.7765,
+      "step": 18610
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1393575225329328e-05,
+      "loss": 0.7286,
+      "step": 18620
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1388953085278485e-05,
+      "loss": 0.8632,
+      "step": 18630
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1384330945227641e-05,
+      "loss": 0.9885,
+      "step": 18640
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1379708805176796e-05,
+      "loss": 1.0964,
+      "step": 18650
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1375086665125955e-05,
+      "loss": 0.913,
+      "step": 18660
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1370464525075111e-05,
+      "loss": 1.1617,
+      "step": 18670
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1365842385024266e-05,
+      "loss": 0.9494,
+      "step": 18680
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1361220244973423e-05,
+      "loss": 0.9853,
+      "step": 18690
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1356598104922581e-05,
+      "loss": 1.0865,
+      "step": 18700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1351975964871738e-05,
+      "loss": 0.8846,
+      "step": 18710
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1347353824820893e-05,
+      "loss": 1.0057,
+      "step": 18720
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.134273168477005e-05,
+      "loss": 0.931,
+      "step": 18730
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1338109544719206e-05,
+      "loss": 0.8821,
+      "step": 18740
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1333487404668361e-05,
+      "loss": 0.9516,
+      "step": 18750
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.132886526461752e-05,
+      "loss": 0.8432,
+      "step": 18760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1324243124566676e-05,
+      "loss": 0.8922,
+      "step": 18770
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1319620984515831e-05,
+      "loss": 1.0529,
+      "step": 18780
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1314998844464988e-05,
+      "loss": 0.7998,
+      "step": 18790
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1310376704414146e-05,
+      "loss": 0.7976,
+      "step": 18800
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1305754564363301e-05,
+      "loss": 1.0546,
+      "step": 18810
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1301132424312458e-05,
+      "loss": 0.7807,
+      "step": 18820
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1296510284261614e-05,
+      "loss": 0.9085,
+      "step": 18830
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.129188814421077e-05,
+      "loss": 0.7798,
+      "step": 18840
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1287266004159928e-05,
+      "loss": 0.8445,
+      "step": 18850
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1282643864109084e-05,
+      "loss": 0.8001,
+      "step": 18860
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.127802172405824e-05,
+      "loss": 0.7526,
+      "step": 18870
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1273399584007396e-05,
+      "loss": 0.9032,
+      "step": 18880
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1268777443956553e-05,
+      "loss": 1.0327,
+      "step": 18890
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1264155303905708e-05,
+      "loss": 0.5758,
+      "step": 18900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1259533163854866e-05,
+      "loss": 0.9337,
+      "step": 18910
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1254911023804023e-05,
+      "loss": 0.7116,
+      "step": 18920
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1250288883753178e-05,
+      "loss": 0.8465,
+      "step": 18930
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1245666743702334e-05,
+      "loss": 0.6701,
+      "step": 18940
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1241044603651493e-05,
+      "loss": 0.8743,
+      "step": 18950
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1236422463600648e-05,
+      "loss": 0.9825,
+      "step": 18960
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1231800323549804e-05,
+      "loss": 0.8811,
+      "step": 18970
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1227178183498961e-05,
+      "loss": 1.1363,
+      "step": 18980
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.122255604344812e-05,
+      "loss": 1.024,
+      "step": 18990
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1217933903397274e-05,
+      "loss": 1.0409,
+      "step": 19000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.121331176334643e-05,
+      "loss": 1.009,
+      "step": 19010
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1208689623295587e-05,
+      "loss": 0.8639,
+      "step": 19020
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1204067483244742e-05,
+      "loss": 0.9773,
+      "step": 19030
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1199445343193899e-05,
+      "loss": 1.0419,
+      "step": 19040
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1194823203143057e-05,
+      "loss": 1.0523,
+      "step": 19050
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1190201063092212e-05,
+      "loss": 0.6958,
+      "step": 19060
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1185578923041369e-05,
+      "loss": 1.1942,
+      "step": 19070
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1180956782990526e-05,
+      "loss": 1.1829,
+      "step": 19080
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.117633464293968e-05,
+      "loss": 0.7812,
+      "step": 19090
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1171712502888839e-05,
+      "loss": 0.9334,
+      "step": 19100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1167090362837996e-05,
+      "loss": 1.0161,
+      "step": 19110
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.116246822278715e-05,
+      "loss": 0.9778,
+      "step": 19120
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1157846082736307e-05,
+      "loss": 1.0992,
+      "step": 19130
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1153223942685466e-05,
+      "loss": 0.8692,
+      "step": 19140
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.114860180263462e-05,
+      "loss": 0.9304,
+      "step": 19150
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1143979662583777e-05,
+      "loss": 0.8557,
+      "step": 19160
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1139357522532934e-05,
+      "loss": 0.8203,
+      "step": 19170
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1134735382482089e-05,
+      "loss": 0.9903,
+      "step": 19180
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1130113242431246e-05,
+      "loss": 0.8832,
+      "step": 19190
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1125491102380404e-05,
+      "loss": 0.7515,
+      "step": 19200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1120868962329559e-05,
+      "loss": 1.1099,
+      "step": 19210
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1116246822278716e-05,
+      "loss": 1.2256,
+      "step": 19220
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1111624682227872e-05,
+      "loss": 0.8131,
+      "step": 19230
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1107002542177027e-05,
+      "loss": 1.1555,
+      "step": 19240
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1102380402126186e-05,
+      "loss": 0.9623,
+      "step": 19250
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1097758262075342e-05,
+      "loss": 0.9426,
+      "step": 19260
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1093136122024499e-05,
+      "loss": 1.2023,
+      "step": 19270
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1088513981973654e-05,
+      "loss": 0.8817,
+      "step": 19280
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1083891841922812e-05,
+      "loss": 1.1819,
+      "step": 19290
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1079269701871969e-05,
+      "loss": 1.0753,
+      "step": 19300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1074647561821124e-05,
+      "loss": 0.9421,
+      "step": 19310
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.107002542177028e-05,
+      "loss": 0.9296,
+      "step": 19320
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1065403281719437e-05,
+      "loss": 1.0582,
+      "step": 19330
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1060781141668592e-05,
+      "loss": 0.9149,
+      "step": 19340
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.105615900161775e-05,
+      "loss": 0.7319,
+      "step": 19350
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1051536861566907e-05,
+      "loss": 0.8164,
+      "step": 19360
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1046914721516062e-05,
+      "loss": 0.8837,
+      "step": 19370
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1042292581465219e-05,
+      "loss": 1.0008,
+      "step": 19380
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1037670441414377e-05,
+      "loss": 0.8772,
+      "step": 19390
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1033048301363532e-05,
+      "loss": 1.0416,
+      "step": 19400
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1028426161312689e-05,
+      "loss": 0.7092,
+      "step": 19410
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1023804021261845e-05,
+      "loss": 0.9392,
+      "step": 19420
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1019181881211e-05,
+      "loss": 0.8311,
+      "step": 19430
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1014559741160159e-05,
+      "loss": 0.7308,
+      "step": 19440
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1009937601109315e-05,
+      "loss": 0.8804,
+      "step": 19450
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.100531546105847e-05,
+      "loss": 0.9505,
+      "step": 19460
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1000693321007627e-05,
+      "loss": 0.874,
+      "step": 19470
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.0996071180956784e-05,
+      "loss": 0.9941,
+      "step": 19480
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.0991449040905939e-05,
+      "loss": 0.655,
+      "step": 19490
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.0986826900855097e-05,
+      "loss": 0.9019,
+      "step": 19500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.0982204760804254e-05,
+      "loss": 0.7623,
+      "step": 19510
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0977582620753408e-05,
+      "loss": 0.9047,
+      "step": 19520
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0972960480702565e-05,
+      "loss": 0.9233,
+      "step": 19530
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0968338340651723e-05,
+      "loss": 0.9337,
+      "step": 19540
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.096371620060088e-05,
+      "loss": 0.9685,
+      "step": 19550
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0959094060550035e-05,
+      "loss": 1.0091,
+      "step": 19560
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0954471920499192e-05,
+      "loss": 0.8778,
+      "step": 19570
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.094984978044835e-05,
+      "loss": 0.9831,
+      "step": 19580
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0945227640397505e-05,
+      "loss": 0.877,
+      "step": 19590
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0940605500346662e-05,
+      "loss": 0.8429,
+      "step": 19600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0935983360295818e-05,
+      "loss": 1.0524,
+      "step": 19610
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0931361220244973e-05,
+      "loss": 0.7061,
+      "step": 19620
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0926739080194132e-05,
+      "loss": 0.8473,
+      "step": 19630
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0922116940143288e-05,
+      "loss": 1.0217,
+      "step": 19640
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0917494800092443e-05,
+      "loss": 0.9726,
+      "step": 19650
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.09128726600416e-05,
+      "loss": 1.0545,
+      "step": 19660
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0908250519990757e-05,
+      "loss": 0.8611,
+      "step": 19670
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0903628379939912e-05,
+      "loss": 0.7191,
+      "step": 19680
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.089900623988907e-05,
+      "loss": 0.9207,
+      "step": 19690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0894384099838227e-05,
+      "loss": 0.8135,
+      "step": 19700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0889761959787382e-05,
+      "loss": 0.9664,
+      "step": 19710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0885139819736538e-05,
+      "loss": 0.9663,
+      "step": 19720
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0880517679685697e-05,
+      "loss": 0.8896,
+      "step": 19730
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0875895539634852e-05,
+      "loss": 0.9769,
+      "step": 19740
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0871273399584008e-05,
+      "loss": 0.7921,
+      "step": 19750
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0866651259533165e-05,
+      "loss": 0.7135,
+      "step": 19760
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.086202911948232e-05,
+      "loss": 0.6762,
+      "step": 19770
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0857406979431478e-05,
+      "loss": 0.92,
+      "step": 19780
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0852784839380635e-05,
+      "loss": 0.7555,
+      "step": 19790
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.084816269932979e-05,
+      "loss": 0.818,
+      "step": 19800
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0843540559278946e-05,
+      "loss": 0.9247,
+      "step": 19810
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0838918419228103e-05,
+      "loss": 1.1512,
+      "step": 19820
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0834296279177261e-05,
+      "loss": 0.9596,
+      "step": 19830
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0829674139126416e-05,
+      "loss": 0.926,
+      "step": 19840
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0825051999075573e-05,
+      "loss": 0.7026,
+      "step": 19850
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.082042985902473e-05,
+      "loss": 0.7384,
+      "step": 19860
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0815807718973885e-05,
+      "loss": 1.0852,
+      "step": 19870
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0811185578923043e-05,
+      "loss": 0.9299,
+      "step": 19880
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.08065634388722e-05,
+      "loss": 0.9318,
+      "step": 19890
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0801941298821355e-05,
+      "loss": 0.9067,
+      "step": 19900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0797319158770511e-05,
+      "loss": 0.9674,
+      "step": 19910
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.079269701871967e-05,
+      "loss": 1.005,
+      "step": 19920
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0788074878668825e-05,
+      "loss": 0.925,
+      "step": 19930
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0783452738617981e-05,
+      "loss": 0.9356,
+      "step": 19940
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0778830598567138e-05,
+      "loss": 1.0745,
+      "step": 19950
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0774208458516293e-05,
+      "loss": 0.8014,
+      "step": 19960
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.076958631846545e-05,
+      "loss": 0.7784,
+      "step": 19970
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0764964178414608e-05,
+      "loss": 1.1419,
+      "step": 19980
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0760342038363763e-05,
+      "loss": 0.8068,
+      "step": 19990
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.075571989831292e-05,
+      "loss": 0.9346,
+      "step": 20000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0751097758262076e-05,
+      "loss": 0.7617,
+      "step": 20010
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0746475618211231e-05,
+      "loss": 0.8627,
+      "step": 20020
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.074185347816039e-05,
+      "loss": 0.8745,
+      "step": 20030
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0737231338109546e-05,
+      "loss": 0.9617,
+      "step": 20040
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0732609198058701e-05,
+      "loss": 0.6993,
+      "step": 20050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0727987058007858e-05,
+      "loss": 1.0069,
+      "step": 20060
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0723364917957016e-05,
+      "loss": 0.8178,
+      "step": 20070
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0718742777906171e-05,
+      "loss": 1.0069,
+      "step": 20080
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0714120637855328e-05,
+      "loss": 0.7952,
+      "step": 20090
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0709498497804484e-05,
+      "loss": 0.8581,
+      "step": 20100
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0704876357753641e-05,
+      "loss": 0.8083,
+      "step": 20110
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0700254217702796e-05,
+      "loss": 0.9221,
+      "step": 20120
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0695632077651954e-05,
+      "loss": 0.8767,
+      "step": 20130
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0691009937601111e-05,
+      "loss": 1.0205,
+      "step": 20140
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0686387797550266e-05,
+      "loss": 0.9777,
+      "step": 20150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0681765657499423e-05,
+      "loss": 0.9501,
+      "step": 20160
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0677143517448581e-05,
+      "loss": 0.9714,
+      "step": 20170
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0672521377397736e-05,
+      "loss": 0.9775,
+      "step": 20180
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0667899237346893e-05,
+      "loss": 0.9966,
+      "step": 20190
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.066327709729605e-05,
+      "loss": 0.9947,
+      "step": 20200
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0658654957245204e-05,
+      "loss": 1.1063,
+      "step": 20210
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0654032817194363e-05,
+      "loss": 0.797,
+      "step": 20220
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.064941067714352e-05,
+      "loss": 0.9074,
+      "step": 20230
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0644788537092674e-05,
+      "loss": 1.0695,
+      "step": 20240
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0640166397041831e-05,
+      "loss": 0.8265,
+      "step": 20250
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0635544256990988e-05,
+      "loss": 1.1112,
+      "step": 20260
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0630922116940143e-05,
+      "loss": 0.8189,
+      "step": 20270
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0626299976889301e-05,
+      "loss": 0.9308,
+      "step": 20280
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0621677836838458e-05,
+      "loss": 0.901,
+      "step": 20290
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0617055696787612e-05,
+      "loss": 1.0359,
+      "step": 20300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.061243355673677e-05,
+      "loss": 0.9075,
+      "step": 20310
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0607811416685928e-05,
+      "loss": 0.9664,
+      "step": 20320
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0603189276635082e-05,
+      "loss": 0.7992,
+      "step": 20330
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0598567136584239e-05,
+      "loss": 0.7321,
+      "step": 20340
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0593944996533396e-05,
+      "loss": 1.0145,
+      "step": 20350
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.058932285648255e-05,
+      "loss": 0.8895,
+      "step": 20360
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0584700716431709e-05,
+      "loss": 0.7823,
+      "step": 20370
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0580078576380866e-05,
+      "loss": 0.7709,
+      "step": 20380
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0575456436330022e-05,
+      "loss": 0.8899,
+      "step": 20390
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0570834296279177e-05,
+      "loss": 0.972,
+      "step": 20400
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0566212156228334e-05,
+      "loss": 0.7644,
+      "step": 20410
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0561590016177492e-05,
+      "loss": 0.7747,
+      "step": 20420
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0556967876126647e-05,
+      "loss": 0.8761,
+      "step": 20430
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0552345736075804e-05,
+      "loss": 0.6722,
+      "step": 20440
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.054772359602496e-05,
+      "loss": 0.8076,
+      "step": 20450
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0543101455974116e-05,
+      "loss": 0.997,
+      "step": 20460
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0538479315923274e-05,
+      "loss": 1.0217,
+      "step": 20470
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.053385717587243e-05,
+      "loss": 1.0737,
+      "step": 20480
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0529235035821586e-05,
+      "loss": 0.9404,
+      "step": 20490
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0524612895770742e-05,
+      "loss": 0.7826,
+      "step": 20500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.05199907557199e-05,
+      "loss": 0.7633,
+      "step": 20510
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0515368615669056e-05,
+      "loss": 1.1218,
+      "step": 20520
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0510746475618212e-05,
+      "loss": 1.1285,
+      "step": 20530
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0506124335567369e-05,
+      "loss": 1.0715,
+      "step": 20540
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0501502195516524e-05,
+      "loss": 0.7227,
+      "step": 20550
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.049688005546568e-05,
+      "loss": 0.8406,
+      "step": 20560
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0492257915414839e-05,
+      "loss": 1.0106,
+      "step": 20570
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0487635775363994e-05,
+      "loss": 0.9024,
+      "step": 20580
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.048301363531315e-05,
+      "loss": 0.9558,
+      "step": 20590
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0478391495262307e-05,
+      "loss": 0.8467,
+      "step": 20600
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0473769355211462e-05,
+      "loss": 0.9431,
+      "step": 20610
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.046914721516062e-05,
+      "loss": 0.8318,
+      "step": 20620
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0464525075109777e-05,
+      "loss": 0.9367,
+      "step": 20630
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0459902935058932e-05,
+      "loss": 0.9629,
+      "step": 20640
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0455280795008089e-05,
+      "loss": 0.8178,
+      "step": 20650
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0450658654957247e-05,
+      "loss": 0.7954,
+      "step": 20660
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0446036514906404e-05,
+      "loss": 0.9784,
+      "step": 20670
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0441414374855559e-05,
+      "loss": 1.1557,
+      "step": 20680
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0436792234804715e-05,
+      "loss": 1.1031,
+      "step": 20690
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0432170094753874e-05,
+      "loss": 0.9241,
+      "step": 20700
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0427547954703029e-05,
+      "loss": 0.9799,
+      "step": 20710
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0422925814652185e-05,
+      "loss": 0.9582,
+      "step": 20720
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0418303674601342e-05,
+      "loss": 1.0045,
+      "step": 20730
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0413681534550497e-05,
+      "loss": 1.0155,
+      "step": 20740
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0409059394499654e-05,
+      "loss": 0.8464,
+      "step": 20750
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0404437254448812e-05,
+      "loss": 0.7997,
+      "step": 20760
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0399815114397967e-05,
+      "loss": 0.7011,
+      "step": 20770
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0395192974347124e-05,
+      "loss": 0.7195,
+      "step": 20780
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.039057083429628e-05,
+      "loss": 1.0849,
+      "step": 20790
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0385948694245435e-05,
+      "loss": 0.7257,
+      "step": 20800
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0381326554194594e-05,
+      "loss": 0.9877,
+      "step": 20810
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.037670441414375e-05,
+      "loss": 0.7792,
+      "step": 20820
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0372082274092905e-05,
+      "loss": 0.9888,
+      "step": 20830
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0367460134042062e-05,
+      "loss": 0.77,
+      "step": 20840
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.036283799399122e-05,
+      "loss": 0.9518,
+      "step": 20850
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0358215853940375e-05,
+      "loss": 0.9827,
+      "step": 20860
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0353593713889532e-05,
+      "loss": 0.8232,
+      "step": 20870
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0348971573838688e-05,
+      "loss": 0.8785,
+      "step": 20880
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0344349433787843e-05,
+      "loss": 0.8621,
+      "step": 20890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0339727293737e-05,
+      "loss": 0.7914,
+      "step": 20900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0335105153686158e-05,
+      "loss": 0.8264,
+      "step": 20910
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0330483013635313e-05,
+      "loss": 1.0143,
+      "step": 20920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.032586087358447e-05,
+      "loss": 0.8254,
+      "step": 20930
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0321238733533627e-05,
+      "loss": 0.551,
+      "step": 20940
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0316616593482785e-05,
+      "loss": 0.9748,
+      "step": 20950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.031199445343194e-05,
+      "loss": 0.9774,
+      "step": 20960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0307372313381097e-05,
+      "loss": 1.0475,
+      "step": 20970
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0302750173330253e-05,
+      "loss": 0.872,
+      "step": 20980
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0298128033279408e-05,
+      "loss": 0.9561,
+      "step": 20990
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0293505893228567e-05,
+      "loss": 0.8278,
+      "step": 21000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0288883753177723e-05,
+      "loss": 0.8996,
+      "step": 21010
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0284261613126878e-05,
+      "loss": 0.8602,
+      "step": 21020
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0279639473076035e-05,
+      "loss": 1.0362,
+      "step": 21030
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0275017333025192e-05,
+      "loss": 0.8406,
+      "step": 21040
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0270395192974347e-05,
+      "loss": 0.6579,
+      "step": 21050
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0265773052923505e-05,
+      "loss": 0.8992,
+      "step": 21060
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0261150912872662e-05,
+      "loss": 0.9532,
+      "step": 21070
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0256528772821817e-05,
+      "loss": 0.7847,
+      "step": 21080
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0251906632770973e-05,
+      "loss": 1.0693,
+      "step": 21090
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0247284492720132e-05,
+      "loss": 0.7194,
+      "step": 21100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0242662352669286e-05,
+      "loss": 0.7797,
+      "step": 21110
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0238040212618443e-05,
+      "loss": 0.8814,
+      "step": 21120
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.02334180725676e-05,
+      "loss": 1.0122,
+      "step": 21130
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0228795932516755e-05,
+      "loss": 0.7688,
+      "step": 21140
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0224173792465913e-05,
+      "loss": 0.9158,
+      "step": 21150
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.021955165241507e-05,
+      "loss": 0.6633,
+      "step": 21160
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0214929512364225e-05,
+      "loss": 0.8078,
+      "step": 21170
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0210307372313381e-05,
+      "loss": 1.2094,
+      "step": 21180
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0205685232262538e-05,
+      "loss": 0.8986,
+      "step": 21190
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0201063092211693e-05,
+      "loss": 0.8862,
+      "step": 21200
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0196440952160851e-05,
+      "loss": 1.0572,
+      "step": 21210
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0191818812110008e-05,
+      "loss": 0.6935,
+      "step": 21220
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0187196672059165e-05,
+      "loss": 0.868,
+      "step": 21230
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.018257453200832e-05,
+      "loss": 0.9217,
+      "step": 21240
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0177952391957478e-05,
+      "loss": 0.9366,
+      "step": 21250
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0173330251906635e-05,
+      "loss": 0.8465,
+      "step": 21260
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.016870811185579e-05,
+      "loss": 0.85,
+      "step": 21270
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0164085971804946e-05,
+      "loss": 0.9342,
+      "step": 21280
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0159463831754105e-05,
+      "loss": 0.8743,
+      "step": 21290
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.015484169170326e-05,
+      "loss": 0.8613,
+      "step": 21300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0150219551652416e-05,
+      "loss": 0.7022,
+      "step": 21310
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0145597411601573e-05,
+      "loss": 1.0184,
+      "step": 21320
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0140975271550728e-05,
+      "loss": 0.6116,
+      "step": 21330
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0136353131499885e-05,
+      "loss": 0.9886,
+      "step": 21340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0131730991449043e-05,
+      "loss": 1.1962,
+      "step": 21350
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0127108851398198e-05,
+      "loss": 0.8802,
+      "step": 21360
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0122486711347354e-05,
+      "loss": 0.8685,
+      "step": 21370
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0117864571296511e-05,
+      "loss": 0.9768,
+      "step": 21380
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0113242431245666e-05,
+      "loss": 0.949,
+      "step": 21390
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0108620291194824e-05,
+      "loss": 1.0798,
+      "step": 21400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0103998151143981e-05,
+      "loss": 0.8029,
+      "step": 21410
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0099376011093136e-05,
+      "loss": 0.9573,
+      "step": 21420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0094753871042293e-05,
+      "loss": 0.9669,
+      "step": 21430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0090131730991451e-05,
+      "loss": 0.7605,
+      "step": 21440
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0085509590940606e-05,
+      "loss": 1.0217,
+      "step": 21450
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0080887450889763e-05,
+      "loss": 0.983,
+      "step": 21460
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.007626531083892e-05,
+      "loss": 0.7365,
+      "step": 21470
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0071643170788074e-05,
+      "loss": 0.9069,
+      "step": 21480
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0067021030737231e-05,
+      "loss": 0.9795,
+      "step": 21490
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.006239889068639e-05,
+      "loss": 0.9909,
+      "step": 21500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0057776750635546e-05,
+      "loss": 0.67,
+      "step": 21510
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0053154610584701e-05,
+      "loss": 0.9601,
+      "step": 21520
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0048532470533858e-05,
+      "loss": 0.9029,
+      "step": 21530
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0043910330483016e-05,
+      "loss": 0.8324,
+      "step": 21540
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0039288190432171e-05,
+      "loss": 1.0229,
+      "step": 21550
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0034666050381328e-05,
+      "loss": 0.8545,
+      "step": 21560
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0030043910330484e-05,
+      "loss": 0.9154,
+      "step": 21570
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.002542177027964e-05,
+      "loss": 1.0974,
+      "step": 21580
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0020799630228798e-05,
+      "loss": 0.7473,
+      "step": 21590
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0016177490177954e-05,
+      "loss": 0.8617,
+      "step": 21600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.001155535012711e-05,
+      "loss": 0.8909,
+      "step": 21610
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0006933210076266e-05,
+      "loss": 0.6992,
+      "step": 21620
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0002311070025424e-05,
+      "loss": 0.7797,
+      "step": 21630
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.997688929974579e-06,
+      "loss": 0.8549,
+      "step": 21640
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.993066789923736e-06,
+      "loss": 1.0307,
+      "step": 21650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.98844464987289e-06,
+      "loss": 0.7963,
+      "step": 21660
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.983822509822049e-06,
+      "loss": 0.9561,
+      "step": 21670
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.979200369771204e-06,
+      "loss": 0.8113,
+      "step": 21680
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.97457822972036e-06,
+      "loss": 1.1262,
+      "step": 21690
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.969956089669517e-06,
+      "loss": 1.0876,
+      "step": 21700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.965333949618674e-06,
+      "loss": 0.8263,
+      "step": 21710
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.96071180956783e-06,
+      "loss": 0.9587,
+      "step": 21720
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.956089669516987e-06,
+      "loss": 0.6608,
+      "step": 21730
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.951467529466144e-06,
+      "loss": 0.8889,
+      "step": 21740
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.9468453894153e-06,
+      "loss": 0.8929,
+      "step": 21750
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.942223249364457e-06,
+      "loss": 0.7387,
+      "step": 21760
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.937601109313612e-06,
+      "loss": 1.0162,
+      "step": 21770
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.93297896926277e-06,
+      "loss": 0.911,
+      "step": 21780
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.928356829211926e-06,
+      "loss": 0.9891,
+      "step": 21790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.923734689161082e-06,
+      "loss": 0.7322,
+      "step": 21800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.919112549110239e-06,
+      "loss": 0.8475,
+      "step": 21810
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.914490409059396e-06,
+      "loss": 0.9903,
+      "step": 21820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.90986826900855e-06,
+      "loss": 0.6878,
+      "step": 21830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.905246128957709e-06,
+      "loss": 0.9359,
+      "step": 21840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.900623988906864e-06,
+      "loss": 0.8116,
+      "step": 21850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.896001848856022e-06,
+      "loss": 0.9351,
+      "step": 21860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.891379708805177e-06,
+      "loss": 0.8894,
+      "step": 21870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.886757568754334e-06,
+      "loss": 0.7676,
+      "step": 21880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.88213542870349e-06,
+      "loss": 0.8912,
+      "step": 21890
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.877513288652647e-06,
+      "loss": 1.0405,
+      "step": 21900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.872891148601804e-06,
+      "loss": 0.7991,
+      "step": 21910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.86826900855096e-06,
+      "loss": 0.8325,
+      "step": 21920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.863646868500117e-06,
+      "loss": 1.0111,
+      "step": 21930
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.859024728449272e-06,
+      "loss": 1.1903,
+      "step": 21940
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.854402588398429e-06,
+      "loss": 0.8289,
+      "step": 21950
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.849780448347585e-06,
+      "loss": 0.8619,
+      "step": 21960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.845158308296742e-06,
+      "loss": 0.792,
+      "step": 21970
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.840536168245899e-06,
+      "loss": 0.8578,
+      "step": 21980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.835914028195055e-06,
+      "loss": 0.983,
+      "step": 21990
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.831291888144212e-06,
+      "loss": 0.8386,
+      "step": 22000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.826669748093369e-06,
+      "loss": 0.6902,
+      "step": 22010
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.822047608042524e-06,
+      "loss": 1.0276,
+      "step": 22020
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.817425467991682e-06,
+      "loss": 0.7965,
+      "step": 22030
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.812803327940837e-06,
+      "loss": 0.8969,
+      "step": 22040
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.808181187889994e-06,
+      "loss": 0.9207,
+      "step": 22050
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.80355904783915e-06,
+      "loss": 0.7524,
+      "step": 22060
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.798936907788307e-06,
+      "loss": 0.7398,
+      "step": 22070
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.794314767737464e-06,
+      "loss": 0.9246,
+      "step": 22080
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.78969262768662e-06,
+      "loss": 0.8994,
+      "step": 22090
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.785070487635775e-06,
+      "loss": 0.9695,
+      "step": 22100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.780448347584932e-06,
+      "loss": 0.9925,
+      "step": 22110
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.775826207534089e-06,
+      "loss": 0.7844,
+      "step": 22120
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.771204067483245e-06,
+      "loss": 0.8611,
+      "step": 22130
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.766581927432402e-06,
+      "loss": 0.87,
+      "step": 22140
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.761959787381558e-06,
+      "loss": 0.7969,
+      "step": 22150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.757337647330715e-06,
+      "loss": 1.0234,
+      "step": 22160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.752715507279872e-06,
+      "loss": 1.1319,
+      "step": 22170
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.748093367229028e-06,
+      "loss": 0.6985,
+      "step": 22180
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.743471227178183e-06,
+      "loss": 1.0297,
+      "step": 22190
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.738849087127342e-06,
+      "loss": 1.0035,
+      "step": 22200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.734226947076497e-06,
+      "loss": 1.0273,
+      "step": 22210
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.729604807025653e-06,
+      "loss": 0.8961,
+      "step": 22220
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.72498266697481e-06,
+      "loss": 0.8001,
+      "step": 22230
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.720360526923967e-06,
+      "loss": 0.9669,
+      "step": 22240
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.715738386873123e-06,
+      "loss": 0.8454,
+      "step": 22250
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.71111624682228e-06,
+      "loss": 0.8133,
+      "step": 22260
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.706494106771435e-06,
+      "loss": 0.7349,
+      "step": 22270
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.701871966720593e-06,
+      "loss": 0.6697,
+      "step": 22280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.697249826669748e-06,
+      "loss": 0.8973,
+      "step": 22290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.692627686618905e-06,
+      "loss": 1.2562,
+      "step": 22300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.688005546568062e-06,
+      "loss": 0.871,
+      "step": 22310
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.683383406517218e-06,
+      "loss": 0.9271,
+      "step": 22320
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.678761266466375e-06,
+      "loss": 0.8655,
+      "step": 22330
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.674139126415532e-06,
+      "loss": 0.8413,
+      "step": 22340
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.669516986364688e-06,
+      "loss": 1.0043,
+      "step": 22350
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.664894846313843e-06,
+      "loss": 0.8739,
+      "step": 22360
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.660272706263002e-06,
+      "loss": 0.9009,
+      "step": 22370
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.655650566212157e-06,
+      "loss": 0.844,
+      "step": 22380
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.651028426161313e-06,
+      "loss": 1.0584,
+      "step": 22390
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.64640628611047e-06,
+      "loss": 0.8711,
+      "step": 22400
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.641784146059627e-06,
+      "loss": 0.8438,
+      "step": 22410
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.637162006008783e-06,
+      "loss": 1.0035,
+      "step": 22420
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.63253986595794e-06,
+      "loss": 0.8494,
+      "step": 22430
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.627917725907095e-06,
+      "loss": 1.0892,
+      "step": 22440
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.623295585856253e-06,
+      "loss": 0.7324,
+      "step": 22450
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.618673445805408e-06,
+      "loss": 0.9658,
+      "step": 22460
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.614051305754565e-06,
+      "loss": 0.9834,
+      "step": 22470
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.609429165703721e-06,
+      "loss": 0.8007,
+      "step": 22480
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.604807025652878e-06,
+      "loss": 1.031,
+      "step": 22490
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.600184885602035e-06,
+      "loss": 1.1524,
+      "step": 22500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.595562745551191e-06,
+      "loss": 0.737,
+      "step": 22510
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.590940605500348e-06,
+      "loss": 0.8428,
+      "step": 22520
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.586318465449503e-06,
+      "loss": 0.9316,
+      "step": 22530
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.581696325398661e-06,
+      "loss": 0.8354,
+      "step": 22540
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.577074185347816e-06,
+      "loss": 0.8014,
+      "step": 22550
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.572452045296973e-06,
+      "loss": 1.003,
+      "step": 22560
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.56782990524613e-06,
+      "loss": 1.2326,
+      "step": 22570
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.563207765195286e-06,
+      "loss": 1.0257,
+      "step": 22580
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.558585625144443e-06,
+      "loss": 0.8363,
+      "step": 22590
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.5539634850936e-06,
+      "loss": 0.9026,
+      "step": 22600
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.549341345042755e-06,
+      "loss": 0.8601,
+      "step": 22610
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.544719204991913e-06,
+      "loss": 0.8144,
+      "step": 22620
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.540097064941068e-06,
+      "loss": 0.7403,
+      "step": 22630
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.535474924890225e-06,
+      "loss": 0.7535,
+      "step": 22640
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.530852784839381e-06,
+      "loss": 0.9087,
+      "step": 22650
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.526230644788538e-06,
+      "loss": 0.9578,
+      "step": 22660
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.521608504737695e-06,
+      "loss": 0.8527,
+      "step": 22670
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.516986364686851e-06,
+      "loss": 0.7917,
+      "step": 22680
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.512364224636008e-06,
+      "loss": 0.9062,
+      "step": 22690
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.507742084585164e-06,
+      "loss": 0.8477,
+      "step": 22700
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.503119944534321e-06,
+      "loss": 0.8676,
+      "step": 22710
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.498497804483476e-06,
+      "loss": 0.8137,
+      "step": 22720
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.493875664432633e-06,
+      "loss": 0.9653,
+      "step": 22730
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.48925352438179e-06,
+      "loss": 0.7769,
+      "step": 22740
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.484631384330946e-06,
+      "loss": 0.9905,
+      "step": 22750
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.480009244280103e-06,
+      "loss": 0.9362,
+      "step": 22760
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.47538710422926e-06,
+      "loss": 1.0804,
+      "step": 22770
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.470764964178414e-06,
+      "loss": 0.7481,
+      "step": 22780
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.466142824127573e-06,
+      "loss": 0.8879,
+      "step": 22790
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.461520684076728e-06,
+      "loss": 0.8479,
+      "step": 22800
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.456898544025884e-06,
+      "loss": 1.0521,
+      "step": 22810
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.452276403975041e-06,
+      "loss": 1.0407,
+      "step": 22820
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.447654263924198e-06,
+      "loss": 0.9602,
+      "step": 22830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.443032123873354e-06,
+      "loss": 0.9742,
+      "step": 22840
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.438409983822511e-06,
+      "loss": 0.7545,
+      "step": 22850
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.433787843771668e-06,
+      "loss": 1.0003,
+      "step": 22860
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.429165703720824e-06,
+      "loss": 0.7833,
+      "step": 22870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.42454356366998e-06,
+      "loss": 0.7761,
+      "step": 22880
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.419921423619136e-06,
+      "loss": 0.8124,
+      "step": 22890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.415299283568293e-06,
+      "loss": 0.8137,
+      "step": 22900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.41067714351745e-06,
+      "loss": 0.9203,
+      "step": 22910
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.406055003466606e-06,
+      "loss": 0.7716,
+      "step": 22920
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.401432863415763e-06,
+      "loss": 0.9299,
+      "step": 22930
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.39681072336492e-06,
+      "loss": 0.9361,
+      "step": 22940
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.392188583314074e-06,
+      "loss": 0.8519,
+      "step": 22950
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.387566443263232e-06,
+      "loss": 0.8898,
+      "step": 22960
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.382944303212387e-06,
+      "loss": 0.7008,
+      "step": 22970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.378322163161546e-06,
+      "loss": 0.7612,
+      "step": 22980
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.3737000231107e-06,
+      "loss": 0.8519,
+      "step": 22990
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.369077883059857e-06,
+      "loss": 0.6735,
+      "step": 23000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.364455743009014e-06,
+      "loss": 1.2125,
+      "step": 23010
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.35983360295817e-06,
+      "loss": 1.0433,
+      "step": 23020
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.355211462907326e-06,
+      "loss": 0.8757,
+      "step": 23030
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.350589322856484e-06,
+      "loss": 0.9316,
+      "step": 23040
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.345967182805639e-06,
+      "loss": 0.7255,
+      "step": 23050
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.341345042754796e-06,
+      "loss": 0.8006,
+      "step": 23060
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.336722902703952e-06,
+      "loss": 0.7872,
+      "step": 23070
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.332100762653109e-06,
+      "loss": 0.9695,
+      "step": 23080
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.327478622602266e-06,
+      "loss": 0.8858,
+      "step": 23090
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.322856482551422e-06,
+      "loss": 0.5721,
+      "step": 23100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.318234342500579e-06,
+      "loss": 0.8212,
+      "step": 23110
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.313612202449736e-06,
+      "loss": 0.7938,
+      "step": 23120
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.308990062398892e-06,
+      "loss": 0.8981,
+      "step": 23130
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.304367922348047e-06,
+      "loss": 0.5163,
+      "step": 23140
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.299745782297206e-06,
+      "loss": 0.804,
+      "step": 23150
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.29512364224636e-06,
+      "loss": 0.9714,
+      "step": 23160
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.290501502195517e-06,
+      "loss": 0.8045,
+      "step": 23170
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.285879362144674e-06,
+      "loss": 0.9795,
+      "step": 23180
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.28125722209383e-06,
+      "loss": 1.0491,
+      "step": 23190
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.276635082042985e-06,
+      "loss": 1.0075,
+      "step": 23200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.272012941992144e-06,
+      "loss": 0.9494,
+      "step": 23210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.267390801941299e-06,
+      "loss": 0.9176,
+      "step": 23220
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.262768661890455e-06,
+      "loss": 1.0492,
+      "step": 23230
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.258146521839612e-06,
+      "loss": 0.6521,
+      "step": 23240
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.253524381788769e-06,
+      "loss": 1.0505,
+      "step": 23250
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.248902241737925e-06,
+      "loss": 0.9383,
+      "step": 23260
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.244280101687082e-06,
+      "loss": 0.7002,
+      "step": 23270
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.239657961636239e-06,
+      "loss": 0.9339,
+      "step": 23280
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.235035821585395e-06,
+      "loss": 0.653,
+      "step": 23290
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.230413681534552e-06,
+      "loss": 0.958,
+      "step": 23300
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.225791541483707e-06,
+      "loss": 0.941,
+      "step": 23310
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.221169401432865e-06,
+      "loss": 0.8605,
+      "step": 23320
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.21654726138202e-06,
+      "loss": 0.9308,
+      "step": 23330
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.211925121331177e-06,
+      "loss": 0.8214,
+      "step": 23340
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.207302981280334e-06,
+      "loss": 0.9443,
+      "step": 23350
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.20268084122949e-06,
+      "loss": 0.9344,
+      "step": 23360
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.198058701178645e-06,
+      "loss": 0.8099,
+      "step": 23370
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.193436561127804e-06,
+      "loss": 0.8545,
+      "step": 23380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.188814421076959e-06,
+      "loss": 0.8111,
+      "step": 23390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.184192281026117e-06,
+      "loss": 1.0546,
+      "step": 23400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.179570140975272e-06,
+      "loss": 0.8869,
+      "step": 23410
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.174948000924429e-06,
+      "loss": 1.0196,
+      "step": 23420
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.170325860873585e-06,
+      "loss": 0.8878,
+      "step": 23430
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.165703720822742e-06,
+      "loss": 0.9752,
+      "step": 23440
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.161081580771899e-06,
+      "loss": 0.7861,
+      "step": 23450
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.156459440721055e-06,
+      "loss": 0.7567,
+      "step": 23460
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.151837300670212e-06,
+      "loss": 1.0566,
+      "step": 23470
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.147215160619367e-06,
+      "loss": 0.7816,
+      "step": 23480
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.142593020568523e-06,
+      "loss": 0.7992,
+      "step": 23490
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.13797088051768e-06,
+      "loss": 0.8454,
+      "step": 23500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.133348740466837e-06,
+      "loss": 0.946,
+      "step": 23510
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.128726600415993e-06,
+      "loss": 0.8501,
+      "step": 23520
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.12410446036515e-06,
+      "loss": 0.9958,
+      "step": 23530
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.119482320314307e-06,
+      "loss": 0.7717,
+      "step": 23540
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.114860180263463e-06,
+      "loss": 0.9576,
+      "step": 23550
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.110238040212618e-06,
+      "loss": 0.9645,
+      "step": 23560
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.105615900161777e-06,
+      "loss": 0.9138,
+      "step": 23570
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.100993760110932e-06,
+      "loss": 0.9148,
+      "step": 23580
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.096371620060088e-06,
+      "loss": 0.8028,
+      "step": 23590
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.091749480009245e-06,
+      "loss": 0.7522,
+      "step": 23600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.087127339958402e-06,
+      "loss": 1.0922,
+      "step": 23610
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.082505199907558e-06,
+      "loss": 0.6681,
+      "step": 23620
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.077883059856715e-06,
+      "loss": 0.847,
+      "step": 23630
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.07326091980587e-06,
+      "loss": 0.8098,
+      "step": 23640
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.068638779755027e-06,
+      "loss": 0.8009,
+      "step": 23650
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.064016639704183e-06,
+      "loss": 0.9229,
+      "step": 23660
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.05939449965334e-06,
+      "loss": 0.8725,
+      "step": 23670
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.054772359602497e-06,
+      "loss": 0.7768,
+      "step": 23680
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.050150219551653e-06,
+      "loss": 0.9849,
+      "step": 23690
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.04552807950081e-06,
+      "loss": 0.6763,
+      "step": 23700
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.040905939449967e-06,
+      "loss": 0.9658,
+      "step": 23710
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.036283799399123e-06,
+      "loss": 0.8474,
+      "step": 23720
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.031661659348278e-06,
+      "loss": 0.8122,
+      "step": 23730
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.027039519297436e-06,
+      "loss": 0.9973,
+      "step": 23740
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.022417379246591e-06,
+      "loss": 0.8249,
+      "step": 23750
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.017795239195748e-06,
+      "loss": 0.9877,
+      "step": 23760
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.013173099144905e-06,
+      "loss": 0.7181,
+      "step": 23770
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.008550959094061e-06,
+      "loss": 0.857,
+      "step": 23780
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.003928819043218e-06,
+      "loss": 0.9475,
+      "step": 23790
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.999306678992375e-06,
+      "loss": 0.8696,
+      "step": 23800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.99468453894153e-06,
+      "loss": 0.8339,
+      "step": 23810
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.990062398890688e-06,
+      "loss": 1.0996,
+      "step": 23820
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.985440258839843e-06,
+      "loss": 1.0961,
+      "step": 23830
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.980818118789e-06,
+      "loss": 0.8156,
+      "step": 23840
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.976195978738156e-06,
+      "loss": 0.954,
+      "step": 23850
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.971573838687313e-06,
+      "loss": 0.8331,
+      "step": 23860
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.96695169863647e-06,
+      "loss": 0.6473,
+      "step": 23870
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.962329558585626e-06,
+      "loss": 0.6804,
+      "step": 23880
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.957707418534783e-06,
+      "loss": 0.9424,
+      "step": 23890
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.953085278483938e-06,
+      "loss": 0.8521,
+      "step": 23900
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.948463138433096e-06,
+      "loss": 0.8613,
+      "step": 23910
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.943840998382251e-06,
+      "loss": 0.838,
+      "step": 23920
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.939218858331408e-06,
+      "loss": 0.5908,
+      "step": 23930
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.934596718280565e-06,
+      "loss": 0.8873,
+      "step": 23940
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.929974578229721e-06,
+      "loss": 1.0882,
+      "step": 23950
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.925352438178878e-06,
+      "loss": 0.6851,
+      "step": 23960
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.920730298128035e-06,
+      "loss": 0.8564,
+      "step": 23970
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.91610815807719e-06,
+      "loss": 0.7541,
+      "step": 23980
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.911486018026348e-06,
+      "loss": 0.929,
+      "step": 23990
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.906863877975503e-06,
+      "loss": 1.0542,
+      "step": 24000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.90224173792466e-06,
+      "loss": 0.9233,
+      "step": 24010
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.897619597873816e-06,
+      "loss": 1.0838,
+      "step": 24020
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.892997457822973e-06,
+      "loss": 0.7336,
+      "step": 24030
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.88837531777213e-06,
+      "loss": 0.9847,
+      "step": 24040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.883753177721286e-06,
+      "loss": 0.9604,
+      "step": 24050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.879131037670443e-06,
+      "loss": 0.7175,
+      "step": 24060
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.874508897619598e-06,
+      "loss": 0.7646,
+      "step": 24070
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.869886757568756e-06,
+      "loss": 0.8674,
+      "step": 24080
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.865264617517911e-06,
+      "loss": 0.7047,
+      "step": 24090
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.860642477467068e-06,
+      "loss": 1.1347,
+      "step": 24100
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.856020337416224e-06,
+      "loss": 0.6839,
+      "step": 24110
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.851398197365381e-06,
+      "loss": 0.804,
+      "step": 24120
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.846776057314538e-06,
+      "loss": 0.8234,
+      "step": 24130
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.842153917263694e-06,
+      "loss": 0.8683,
+      "step": 24140
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.83753177721285e-06,
+      "loss": 0.9792,
+      "step": 24150
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.832909637162008e-06,
+      "loss": 1.0127,
+      "step": 24160
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.828287497111163e-06,
+      "loss": 1.1185,
+      "step": 24170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.82366535706032e-06,
+      "loss": 0.6551,
+      "step": 24180
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.819043217009476e-06,
+      "loss": 1.0182,
+      "step": 24190
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.814421076958633e-06,
+      "loss": 0.9794,
+      "step": 24200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.80979893690779e-06,
+      "loss": 0.8763,
+      "step": 24210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.805176796856946e-06,
+      "loss": 0.742,
+      "step": 24220
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.800554656806103e-06,
+      "loss": 0.9251,
+      "step": 24230
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.79593251675526e-06,
+      "loss": 0.8673,
+      "step": 24240
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.791310376704416e-06,
+      "loss": 1.0506,
+      "step": 24250
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.78668823665357e-06,
+      "loss": 0.838,
+      "step": 24260
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.782066096602727e-06,
+      "loss": 0.9414,
+      "step": 24270
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.777443956551884e-06,
+      "loss": 0.8974,
+      "step": 24280
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.77282181650104e-06,
+      "loss": 0.6754,
+      "step": 24290
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.768199676450197e-06,
+      "loss": 0.634,
+      "step": 24300
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.763577536399354e-06,
+      "loss": 0.8932,
+      "step": 24310
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.758955396348509e-06,
+      "loss": 0.6471,
+      "step": 24320
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.754333256297667e-06,
+      "loss": 0.9807,
+      "step": 24330
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.749711116246822e-06,
+      "loss": 1.0148,
+      "step": 24340
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.745088976195979e-06,
+      "loss": 0.8961,
+      "step": 24350
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.740466836145136e-06,
+      "loss": 0.884,
+      "step": 24360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.735844696094292e-06,
+      "loss": 0.8816,
+      "step": 24370
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.731222556043449e-06,
+      "loss": 1.0431,
+      "step": 24380
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.726600415992606e-06,
+      "loss": 0.8227,
+      "step": 24390
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.721978275941762e-06,
+      "loss": 0.7469,
+      "step": 24400
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.717356135890919e-06,
+      "loss": 0.7272,
+      "step": 24410
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.712733995840074e-06,
+      "loss": 1.0465,
+      "step": 24420
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.70811185578923e-06,
+      "loss": 0.5699,
+      "step": 24430
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.703489715738387e-06,
+      "loss": 0.7001,
+      "step": 24440
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.698867575687544e-06,
+      "loss": 0.804,
+      "step": 24450
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.6942454356367e-06,
+      "loss": 1.1156,
+      "step": 24460
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.689623295585857e-06,
+      "loss": 0.9704,
+      "step": 24470
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.685001155535014e-06,
+      "loss": 0.9821,
+      "step": 24480
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.680379015484169e-06,
+      "loss": 1.0867,
+      "step": 24490
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.675756875433327e-06,
+      "loss": 1.0561,
+      "step": 24500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.671134735382482e-06,
+      "loss": 1.0991,
+      "step": 24510
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.66651259533164e-06,
+      "loss": 0.9112,
+      "step": 24520
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.661890455280795e-06,
+      "loss": 0.8459,
+      "step": 24530
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.657268315229952e-06,
+      "loss": 0.8564,
+      "step": 24540
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.652646175179109e-06,
+      "loss": 1.0557,
+      "step": 24550
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.648024035128265e-06,
+      "loss": 1.0189,
+      "step": 24560
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.64340189507742e-06,
+      "loss": 0.7821,
+      "step": 24570
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.638779755026579e-06,
+      "loss": 0.5513,
+      "step": 24580
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.634157614975734e-06,
+      "loss": 0.9305,
+      "step": 24590
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.62953547492489e-06,
+      "loss": 0.8948,
+      "step": 24600
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.624913334874047e-06,
+      "loss": 1.0453,
+      "step": 24610
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.620291194823204e-06,
+      "loss": 0.7534,
+      "step": 24620
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.61566905477236e-06,
+      "loss": 0.7355,
+      "step": 24630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.611046914721517e-06,
+      "loss": 0.8485,
+      "step": 24640
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.606424774670674e-06,
+      "loss": 0.9669,
+      "step": 24650
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.60180263461983e-06,
+      "loss": 0.9852,
+      "step": 24660
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.597180494568987e-06,
+      "loss": 0.87,
+      "step": 24670
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.592558354518142e-06,
+      "loss": 1.0162,
+      "step": 24680
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.5879362144673e-06,
+      "loss": 0.9247,
+      "step": 24690
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.583314074416455e-06,
+      "loss": 1.0942,
+      "step": 24700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.578691934365612e-06,
+      "loss": 0.7425,
+      "step": 24710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.574069794314769e-06,
+      "loss": 0.8693,
+      "step": 24720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.569447654263925e-06,
+      "loss": 0.8129,
+      "step": 24730
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.56482551421308e-06,
+      "loss": 0.9842,
+      "step": 24740
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.560203374162239e-06,
+      "loss": 0.9679,
+      "step": 24750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.555581234111394e-06,
+      "loss": 0.8833,
+      "step": 24760
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.55095909406055e-06,
+      "loss": 0.6988,
+      "step": 24770
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.546336954009707e-06,
+      "loss": 0.9362,
+      "step": 24780
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.541714813958863e-06,
+      "loss": 1.0149,
+      "step": 24790
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.53709267390802e-06,
+      "loss": 0.8466,
+      "step": 24800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.532470533857177e-06,
+      "loss": 0.9058,
+      "step": 24810
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.527848393806333e-06,
+      "loss": 0.8385,
+      "step": 24820
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.52322625375549e-06,
+      "loss": 0.7963,
+      "step": 24830
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.518604113704647e-06,
+      "loss": 0.9455,
+      "step": 24840
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.513981973653802e-06,
+      "loss": 0.6225,
+      "step": 24850
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.50935983360296e-06,
+      "loss": 0.8692,
+      "step": 24860
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.504737693552115e-06,
+      "loss": 0.9472,
+      "step": 24870
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.500115553501272e-06,
+      "loss": 0.9731,
+      "step": 24880
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.495493413450428e-06,
+      "loss": 0.8951,
+      "step": 24890
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.490871273399585e-06,
+      "loss": 0.8077,
+      "step": 24900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.48624913334874e-06,
+      "loss": 0.8705,
+      "step": 24910
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.481626993297898e-06,
+      "loss": 0.85,
+      "step": 24920
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.477004853247053e-06,
+      "loss": 0.864,
+      "step": 24930
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.472382713196212e-06,
+      "loss": 1.011,
+      "step": 24940
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.467760573145367e-06,
+      "loss": 0.7056,
+      "step": 24950
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.463138433094523e-06,
+      "loss": 0.7856,
+      "step": 24960
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.45851629304368e-06,
+      "loss": 1.0237,
+      "step": 24970
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.453894152992837e-06,
+      "loss": 0.6897,
+      "step": 24980
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.449272012941993e-06,
+      "loss": 0.6947,
+      "step": 24990
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.44464987289115e-06,
+      "loss": 1.02,
+      "step": 25000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.440027732840307e-06,
+      "loss": 0.9227,
+      "step": 25010
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.435405592789462e-06,
+      "loss": 0.7662,
+      "step": 25020
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.430783452738618e-06,
+      "loss": 0.825,
+      "step": 25030
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.426161312687775e-06,
+      "loss": 0.8166,
+      "step": 25040
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.421539172636931e-06,
+      "loss": 0.8658,
+      "step": 25050
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.416917032586088e-06,
+      "loss": 0.9965,
+      "step": 25060
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.412294892535245e-06,
+      "loss": 0.8424,
+      "step": 25070
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.407672752484401e-06,
+      "loss": 1.0854,
+      "step": 25080
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.403050612433558e-06,
+      "loss": 0.8403,
+      "step": 25090
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.398428472382713e-06,
+      "loss": 1.2058,
+      "step": 25100
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.393806332331871e-06,
+      "loss": 0.8759,
+      "step": 25110
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.389184192281026e-06,
+      "loss": 1.1957,
+      "step": 25120
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.384562052230183e-06,
+      "loss": 1.1047,
+      "step": 25130
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.37993991217934e-06,
+      "loss": 0.7602,
+      "step": 25140
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.375317772128496e-06,
+      "loss": 0.6938,
+      "step": 25150
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.370695632077653e-06,
+      "loss": 0.75,
+      "step": 25160
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.36607349202681e-06,
+      "loss": 0.6918,
+      "step": 25170
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.361451351975965e-06,
+      "loss": 0.8083,
+      "step": 25180
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.356829211925121e-06,
+      "loss": 1.026,
+      "step": 25190
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.352207071874278e-06,
+      "loss": 0.9123,
+      "step": 25200
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.347584931823435e-06,
+      "loss": 1.0341,
+      "step": 25210
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.342962791772591e-06,
+      "loss": 1.1153,
+      "step": 25220
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.338340651721748e-06,
+      "loss": 0.9839,
+      "step": 25230
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.333718511670905e-06,
+      "loss": 0.9697,
+      "step": 25240
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.329096371620061e-06,
+      "loss": 1.1724,
+      "step": 25250
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.324474231569218e-06,
+      "loss": 0.8453,
+      "step": 25260
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.319852091518373e-06,
+      "loss": 0.843,
+      "step": 25270
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.315229951467531e-06,
+      "loss": 0.9656,
+      "step": 25280
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.310607811416686e-06,
+      "loss": 1.0792,
+      "step": 25290
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.305985671365843e-06,
+      "loss": 0.7434,
+      "step": 25300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.301363531315e-06,
+      "loss": 0.8482,
+      "step": 25310
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.296741391264156e-06,
+      "loss": 0.9359,
+      "step": 25320
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.292119251213313e-06,
+      "loss": 0.927,
+      "step": 25330
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.28749711116247e-06,
+      "loss": 0.8104,
+      "step": 25340
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.282874971111624e-06,
+      "loss": 0.9688,
+      "step": 25350
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.278252831060783e-06,
+      "loss": 0.9255,
+      "step": 25360
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.273630691009938e-06,
+      "loss": 0.7265,
+      "step": 25370
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.269008550959094e-06,
+      "loss": 0.7843,
+      "step": 25380
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.264386410908251e-06,
+      "loss": 0.7088,
+      "step": 25390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.259764270857408e-06,
+      "loss": 0.6457,
+      "step": 25400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.255142130806564e-06,
+      "loss": 1.1505,
+      "step": 25410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.250519990755721e-06,
+      "loss": 1.0553,
+      "step": 25420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.245897850704878e-06,
+      "loss": 0.9515,
+      "step": 25430
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.241275710654033e-06,
+      "loss": 0.8825,
+      "step": 25440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.236653570603191e-06,
+      "loss": 0.8206,
+      "step": 25450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.232031430552346e-06,
+      "loss": 0.6315,
+      "step": 25460
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.227409290501503e-06,
+      "loss": 1.0637,
+      "step": 25470
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.22278715045066e-06,
+      "loss": 0.8162,
+      "step": 25480
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.218165010399816e-06,
+      "loss": 0.7747,
+      "step": 25490
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.213542870348973e-06,
+      "loss": 0.9516,
+      "step": 25500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.20892073029813e-06,
+      "loss": 0.8392,
+      "step": 25510
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.204298590247284e-06,
+      "loss": 0.8624,
+      "step": 25520
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.199676450196443e-06,
+      "loss": 1.1022,
+      "step": 25530
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.195054310145598e-06,
+      "loss": 0.7795,
+      "step": 25540
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.190432170094754e-06,
+      "loss": 0.5964,
+      "step": 25550
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.18581003004391e-06,
+      "loss": 1.1009,
+      "step": 25560
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.181187889993067e-06,
+      "loss": 0.9862,
+      "step": 25570
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.176565749942224e-06,
+      "loss": 0.9525,
+      "step": 25580
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.17194360989138e-06,
+      "loss": 0.9146,
+      "step": 25590
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.167321469840537e-06,
+      "loss": 0.7231,
+      "step": 25600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.162699329789692e-06,
+      "loss": 0.7637,
+      "step": 25610
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.15807718973885e-06,
+      "loss": 0.9597,
+      "step": 25620
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.153455049688006e-06,
+      "loss": 0.9176,
+      "step": 25630
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.148832909637162e-06,
+      "loss": 0.7302,
+      "step": 25640
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.144210769586319e-06,
+      "loss": 0.8081,
+      "step": 25650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.139588629535476e-06,
+      "loss": 0.9379,
+      "step": 25660
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.134966489484632e-06,
+      "loss": 0.8345,
+      "step": 25670
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.130344349433789e-06,
+      "loss": 0.8588,
+      "step": 25680
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.125722209382944e-06,
+      "loss": 0.8166,
+      "step": 25690
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.121100069332102e-06,
+      "loss": 0.8178,
+      "step": 25700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.116477929281257e-06,
+      "loss": 0.9118,
+      "step": 25710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.111855789230414e-06,
+      "loss": 0.7257,
+      "step": 25720
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.10723364917957e-06,
+      "loss": 1.1563,
+      "step": 25730
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.102611509128727e-06,
+      "loss": 1.0816,
+      "step": 25740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.097989369077884e-06,
+      "loss": 1.0412,
+      "step": 25750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.09336722902704e-06,
+      "loss": 0.7828,
+      "step": 25760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.088745088976197e-06,
+      "loss": 0.9031,
+      "step": 25770
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.084122948925354e-06,
+      "loss": 0.7639,
+      "step": 25780
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.07950080887451e-06,
+      "loss": 0.9254,
+      "step": 25790
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.074878668823666e-06,
+      "loss": 0.9509,
+      "step": 25800
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.070256528772822e-06,
+      "loss": 0.6472,
+      "step": 25810
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.065634388721979e-06,
+      "loss": 0.8004,
+      "step": 25820
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.061012248671135e-06,
+      "loss": 0.7271,
+      "step": 25830
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.056390108620292e-06,
+      "loss": 0.8606,
+      "step": 25840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.051767968569449e-06,
+      "loss": 0.761,
+      "step": 25850
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.047145828518604e-06,
+      "loss": 0.9006,
+      "step": 25860
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.042523688467762e-06,
+      "loss": 0.8759,
+      "step": 25870
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.037901548416917e-06,
+      "loss": 0.8467,
+      "step": 25880
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.033279408366074e-06,
+      "loss": 1.0313,
+      "step": 25890
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.02865726831523e-06,
+      "loss": 0.9655,
+      "step": 25900
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.024035128264387e-06,
+      "loss": 0.8988,
+      "step": 25910
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.019412988213544e-06,
+      "loss": 0.7485,
+      "step": 25920
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.0147908481627e-06,
+      "loss": 0.7995,
+      "step": 25930
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.010168708111857e-06,
+      "loss": 1.045,
+      "step": 25940
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.005546568061014e-06,
+      "loss": 0.8604,
+      "step": 25950
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.000924428010169e-06,
+      "loss": 1.0344,
+      "step": 25960
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.7023000736678399,
+      "eval_loss": 0.8277474045753479,
+      "eval_runtime": 370.3275,
+      "eval_samples_per_second": 32.99,
+      "eval_steps_per_second": 4.126,
+      "step": 25962
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.996302287959325e-06,
+      "loss": 1.0651,
+      "step": 25970
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.991680147908482e-06,
+      "loss": 1.0033,
+      "step": 25980
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.987058007857639e-06,
+      "loss": 0.8618,
+      "step": 25990
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.982435867806795e-06,
+      "loss": 0.9764,
+      "step": 26000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.977813727755952e-06,
+      "loss": 0.7547,
+      "step": 26010
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.973191587705109e-06,
+      "loss": 0.7769,
+      "step": 26020
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.968569447654264e-06,
+      "loss": 0.53,
+      "step": 26030
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.963947307603422e-06,
+      "loss": 0.6506,
+      "step": 26040
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.959325167552577e-06,
+      "loss": 0.8994,
+      "step": 26050
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.954703027501734e-06,
+      "loss": 0.7206,
+      "step": 26060
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.95008088745089e-06,
+      "loss": 0.6982,
+      "step": 26070
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.945458747400047e-06,
+      "loss": 1.0016,
+      "step": 26080
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.940836607349203e-06,
+      "loss": 0.9596,
+      "step": 26090
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.93621446729836e-06,
+      "loss": 0.8941,
+      "step": 26100
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.931592327247515e-06,
+      "loss": 1.0188,
+      "step": 26110
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.926970187196673e-06,
+      "loss": 0.894,
+      "step": 26120
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.922348047145828e-06,
+      "loss": 0.7879,
+      "step": 26130
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.917725907094985e-06,
+      "loss": 0.9023,
+      "step": 26140
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.913103767044142e-06,
+      "loss": 0.7074,
+      "step": 26150
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.908481626993298e-06,
+      "loss": 0.8804,
+      "step": 26160
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.903859486942455e-06,
+      "loss": 1.0399,
+      "step": 26170
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.899237346891612e-06,
+      "loss": 0.9455,
+      "step": 26180
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.894615206840768e-06,
+      "loss": 0.7583,
+      "step": 26190
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.889993066789923e-06,
+      "loss": 0.9301,
+      "step": 26200
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.885370926739082e-06,
+      "loss": 0.6127,
+      "step": 26210
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.880748786688237e-06,
+      "loss": 0.9057,
+      "step": 26220
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.876126646637395e-06,
+      "loss": 0.8209,
+      "step": 26230
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.87150450658655e-06,
+      "loss": 0.7997,
+      "step": 26240
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.866882366535707e-06,
+      "loss": 0.9574,
+      "step": 26250
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.862260226484863e-06,
+      "loss": 0.6283,
+      "step": 26260
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.85763808643402e-06,
+      "loss": 0.7524,
+      "step": 26270
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.853015946383175e-06,
+      "loss": 0.926,
+      "step": 26280
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.848393806332333e-06,
+      "loss": 0.6489,
+      "step": 26290
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.843771666281488e-06,
+      "loss": 1.043,
+      "step": 26300
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.839149526230645e-06,
+      "loss": 0.9733,
+      "step": 26310
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.834527386179802e-06,
+      "loss": 0.7837,
+      "step": 26320
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.829905246128958e-06,
+      "loss": 0.849,
+      "step": 26330
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.825283106078115e-06,
+      "loss": 0.7408,
+      "step": 26340
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.820660966027271e-06,
+      "loss": 1.0131,
+      "step": 26350
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.816038825976428e-06,
+      "loss": 0.7062,
+      "step": 26360
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.811416685925585e-06,
+      "loss": 1.0453,
+      "step": 26370
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.806794545874741e-06,
+      "loss": 0.759,
+      "step": 26380
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.802172405823896e-06,
+      "loss": 0.5866,
+      "step": 26390
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.797550265773055e-06,
+      "loss": 0.8697,
+      "step": 26400
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.79292812572221e-06,
+      "loss": 0.8353,
+      "step": 26410
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.788305985671366e-06,
+      "loss": 0.8428,
+      "step": 26420
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.783683845620523e-06,
+      "loss": 1.049,
+      "step": 26430
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.77906170556968e-06,
+      "loss": 0.8694,
+      "step": 26440
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.774439565518835e-06,
+      "loss": 0.8205,
+      "step": 26450
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.769817425467993e-06,
+      "loss": 1.1194,
+      "step": 26460
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.765195285417148e-06,
+      "loss": 0.7783,
+      "step": 26470
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.760573145366305e-06,
+      "loss": 0.8308,
+      "step": 26480
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.755951005315461e-06,
+      "loss": 0.8369,
+      "step": 26490
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.751328865264618e-06,
+      "loss": 0.873,
+      "step": 26500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.746706725213775e-06,
+      "loss": 0.9502,
+      "step": 26510
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.742084585162931e-06,
+      "loss": 1.0375,
+      "step": 26520
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.737462445112088e-06,
+      "loss": 0.9357,
+      "step": 26530
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.732840305061245e-06,
+      "loss": 0.9021,
+      "step": 26540
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.728218165010401e-06,
+      "loss": 0.8244,
+      "step": 26550
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.723596024959556e-06,
+      "loss": 0.8593,
+      "step": 26560
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.718973884908713e-06,
+      "loss": 0.7352,
+      "step": 26570
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.71435174485787e-06,
+      "loss": 0.983,
+      "step": 26580
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.709729604807026e-06,
+      "loss": 1.0018,
+      "step": 26590
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.705107464756183e-06,
+      "loss": 0.564,
+      "step": 26600
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.70048532470534e-06,
+      "loss": 0.9231,
+      "step": 26610
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.695863184654494e-06,
+      "loss": 0.6584,
+      "step": 26620
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.691241044603653e-06,
+      "loss": 0.8984,
+      "step": 26630
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.686618904552808e-06,
+      "loss": 0.6742,
+      "step": 26640
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.681996764501966e-06,
+      "loss": 0.6271,
+      "step": 26650
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.677374624451121e-06,
+      "loss": 0.8894,
+      "step": 26660
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.672752484400278e-06,
+      "loss": 0.9462,
+      "step": 26670
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.668130344349434e-06,
+      "loss": 0.8703,
+      "step": 26680
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.663508204298591e-06,
+      "loss": 0.8538,
+      "step": 26690
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.658886064247748e-06,
+      "loss": 0.8599,
+      "step": 26700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.654263924196904e-06,
+      "loss": 0.9368,
+      "step": 26710
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.64964178414606e-06,
+      "loss": 0.7219,
+      "step": 26720
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.645019644095216e-06,
+      "loss": 0.909,
+      "step": 26730
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.640397504044373e-06,
+      "loss": 0.6926,
+      "step": 26740
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.63577536399353e-06,
+      "loss": 0.8226,
+      "step": 26750
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.631153223942686e-06,
+      "loss": 0.914,
+      "step": 26760
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.626531083891843e-06,
+      "loss": 0.7198,
+      "step": 26770
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.6219089438409984e-06,
+      "loss": 0.965,
+      "step": 26780
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.617286803790156e-06,
+      "loss": 0.8865,
+      "step": 26790
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.612664663739312e-06,
+      "loss": 0.7788,
+      "step": 26800
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.608042523688468e-06,
+      "loss": 0.8349,
+      "step": 26810
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.603420383637625e-06,
+      "loss": 0.8148,
+      "step": 26820
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.598798243586782e-06,
+      "loss": 0.7252,
+      "step": 26830
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.5941761035359375e-06,
+      "loss": 0.8338,
+      "step": 26840
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.589553963485094e-06,
+      "loss": 0.9437,
+      "step": 26850
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.584931823434251e-06,
+      "loss": 0.7967,
+      "step": 26860
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.580309683383407e-06,
+      "loss": 0.9614,
+      "step": 26870
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.575687543332564e-06,
+      "loss": 0.8728,
+      "step": 26880
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.57106540328172e-06,
+      "loss": 0.786,
+      "step": 26890
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.566443263230876e-06,
+      "loss": 1.1031,
+      "step": 26900
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.561821123180033e-06,
+      "loss": 0.9725,
+      "step": 26910
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.557198983129189e-06,
+      "loss": 0.8606,
+      "step": 26920
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.552576843078347e-06,
+      "loss": 1.0195,
+      "step": 26930
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.547954703027502e-06,
+      "loss": 0.8931,
+      "step": 26940
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.543332562976658e-06,
+      "loss": 1.1264,
+      "step": 26950
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.538710422925816e-06,
+      "loss": 1.0112,
+      "step": 26960
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.5340882828749715e-06,
+      "loss": 0.6051,
+      "step": 26970
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.529466142824128e-06,
+      "loss": 0.9505,
+      "step": 26980
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.524844002773285e-06,
+      "loss": 0.8713,
+      "step": 26990
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.520221862722441e-06,
+      "loss": 0.9696,
+      "step": 27000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.515599722671597e-06,
+      "loss": 0.9612,
+      "step": 27010
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.510977582620754e-06,
+      "loss": 0.7082,
+      "step": 27020
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.506355442569911e-06,
+      "loss": 0.9683,
+      "step": 27030
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.5017333025190664e-06,
+      "loss": 0.7916,
+      "step": 27040
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.497111162468224e-06,
+      "loss": 0.8413,
+      "step": 27050
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.49248902241738e-06,
+      "loss": 0.953,
+      "step": 27060
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.487866882366536e-06,
+      "loss": 0.8584,
+      "step": 27070
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.483244742315693e-06,
+      "loss": 0.8541,
+      "step": 27080
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.478622602264849e-06,
+      "loss": 0.7962,
+      "step": 27090
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.474000462214006e-06,
+      "loss": 0.9031,
+      "step": 27100
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.469378322163162e-06,
+      "loss": 0.6497,
+      "step": 27110
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.464756182112318e-06,
+      "loss": 0.7564,
+      "step": 27120
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.4601340420614755e-06,
+      "loss": 0.7227,
+      "step": 27130
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.455511902010631e-06,
+      "loss": 0.9616,
+      "step": 27140
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.450889761959787e-06,
+      "loss": 0.8521,
+      "step": 27150
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.446267621908945e-06,
+      "loss": 0.9299,
+      "step": 27160
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.4416454818581004e-06,
+      "loss": 0.8233,
+      "step": 27170
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.437023341807257e-06,
+      "loss": 0.8294,
+      "step": 27180
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.432401201756414e-06,
+      "loss": 0.6649,
+      "step": 27190
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.42777906170557e-06,
+      "loss": 0.8865,
+      "step": 27200
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.423156921654727e-06,
+      "loss": 0.7499,
+      "step": 27210
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.418534781603883e-06,
+      "loss": 0.8151,
+      "step": 27220
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.4139126415530395e-06,
+      "loss": 0.5837,
+      "step": 27230
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.409290501502196e-06,
+      "loss": 1.0249,
+      "step": 27240
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.404668361451353e-06,
+      "loss": 0.8906,
+      "step": 27250
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.400046221400509e-06,
+      "loss": 0.9419,
+      "step": 27260
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.395424081349666e-06,
+      "loss": 0.7645,
+      "step": 27270
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.390801941298822e-06,
+      "loss": 0.695,
+      "step": 27280
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.386179801247978e-06,
+      "loss": 0.7142,
+      "step": 27290
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.381557661197135e-06,
+      "loss": 0.6427,
+      "step": 27300
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.376935521146291e-06,
+      "loss": 1.062,
+      "step": 27310
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.372313381095447e-06,
+      "loss": 0.952,
+      "step": 27320
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.367691241044604e-06,
+      "loss": 0.812,
+      "step": 27330
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.36306910099376e-06,
+      "loss": 0.9711,
+      "step": 27340
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.358446960942918e-06,
+      "loss": 0.8381,
+      "step": 27350
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.3538248208920735e-06,
+      "loss": 0.7715,
+      "step": 27360
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.34920268084123e-06,
+      "loss": 1.0561,
+      "step": 27370
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.344580540790387e-06,
+      "loss": 0.9088,
+      "step": 27380
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.339958400739543e-06,
+      "loss": 0.9373,
+      "step": 27390
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.335336260688699e-06,
+      "loss": 0.868,
+      "step": 27400
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.330714120637856e-06,
+      "loss": 0.8987,
+      "step": 27410
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.326091980587013e-06,
+      "loss": 0.8139,
+      "step": 27420
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.3214698405361684e-06,
+      "loss": 0.8238,
+      "step": 27430
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.316847700485326e-06,
+      "loss": 0.8926,
+      "step": 27440
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.312225560434482e-06,
+      "loss": 0.8604,
+      "step": 27450
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.3076034203836376e-06,
+      "loss": 0.8719,
+      "step": 27460
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.302981280332795e-06,
+      "loss": 0.6954,
+      "step": 27470
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.298359140281951e-06,
+      "loss": 0.8915,
+      "step": 27480
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.293737000231108e-06,
+      "loss": 0.8223,
+      "step": 27490
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.289114860180264e-06,
+      "loss": 0.937,
+      "step": 27500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.28449272012942e-06,
+      "loss": 0.6879,
+      "step": 27510
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.2798705800785775e-06,
+      "loss": 0.8692,
+      "step": 27520
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.275248440027733e-06,
+      "loss": 0.8584,
+      "step": 27530
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.270626299976889e-06,
+      "loss": 0.7817,
+      "step": 27540
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.266004159926047e-06,
+      "loss": 0.8713,
+      "step": 27550
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.2613820198752025e-06,
+      "loss": 0.8519,
+      "step": 27560
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.256759879824359e-06,
+      "loss": 0.7981,
+      "step": 27570
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.252137739773516e-06,
+      "loss": 0.9577,
+      "step": 27580
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.247515599722672e-06,
+      "loss": 0.8811,
+      "step": 27590
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.242893459671828e-06,
+      "loss": 1.0366,
+      "step": 27600
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.238271319620985e-06,
+      "loss": 0.9348,
+      "step": 27610
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.2336491795701415e-06,
+      "loss": 0.7859,
+      "step": 27620
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.229027039519298e-06,
+      "loss": 0.8648,
+      "step": 27630
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.224404899468455e-06,
+      "loss": 0.7652,
+      "step": 27640
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.219782759417611e-06,
+      "loss": 0.8385,
+      "step": 27650
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.215160619366768e-06,
+      "loss": 0.8738,
+      "step": 27660
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.210538479315924e-06,
+      "loss": 0.8985,
+      "step": 27670
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.20591633926508e-06,
+      "loss": 0.6767,
+      "step": 27680
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.201294199214237e-06,
+      "loss": 0.9578,
+      "step": 27690
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.196672059163393e-06,
+      "loss": 0.8474,
+      "step": 27700
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.192049919112549e-06,
+      "loss": 0.7264,
+      "step": 27710
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.1874277790617064e-06,
+      "loss": 0.9314,
+      "step": 27720
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.182805639010862e-06,
+      "loss": 0.9336,
+      "step": 27730
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.178183498960019e-06,
+      "loss": 0.804,
+      "step": 27740
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.1735613589091755e-06,
+      "loss": 0.8373,
+      "step": 27750
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.168939218858331e-06,
+      "loss": 0.7973,
+      "step": 27760
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.164317078807489e-06,
+      "loss": 0.7196,
+      "step": 27770
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.159694938756645e-06,
+      "loss": 0.8309,
+      "step": 27780
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.155072798705801e-06,
+      "loss": 0.8427,
+      "step": 27790
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.150450658654958e-06,
+      "loss": 1.0412,
+      "step": 27800
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.145828518604115e-06,
+      "loss": 0.8376,
+      "step": 27810
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.1412063785532705e-06,
+      "loss": 0.9849,
+      "step": 27820
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.136584238502428e-06,
+      "loss": 0.8494,
+      "step": 27830
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.131962098451584e-06,
+      "loss": 1.0088,
+      "step": 27840
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.12733995840074e-06,
+      "loss": 1.0019,
+      "step": 27850
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.122717818349897e-06,
+      "loss": 0.8025,
+      "step": 27860
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.118095678299053e-06,
+      "loss": 0.9366,
+      "step": 27870
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.113473538248209e-06,
+      "loss": 0.8027,
+      "step": 27880
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.108851398197366e-06,
+      "loss": 0.8367,
+      "step": 27890
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.104229258146522e-06,
+      "loss": 0.7323,
+      "step": 27900
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.0996071180956795e-06,
+      "loss": 0.9788,
+      "step": 27910
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.094984978044835e-06,
+      "loss": 0.7633,
+      "step": 27920
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.090362837993991e-06,
+      "loss": 0.7821,
+      "step": 27930
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.085740697943149e-06,
+      "loss": 0.8761,
+      "step": 27940
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.0811185578923045e-06,
+      "loss": 0.8363,
+      "step": 27950
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.076496417841461e-06,
+      "loss": 0.7009,
+      "step": 27960
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.071874277790618e-06,
+      "loss": 0.9333,
+      "step": 27970
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.0672521377397744e-06,
+      "loss": 0.6261,
+      "step": 27980
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.06262999768893e-06,
+      "loss": 0.9721,
+      "step": 27990
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.058007857638087e-06,
+      "loss": 0.9991,
+      "step": 28000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.0533857175872436e-06,
+      "loss": 0.9079,
+      "step": 28010
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.048763577536399e-06,
+      "loss": 0.8932,
+      "step": 28020
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.044141437485557e-06,
+      "loss": 0.8673,
+      "step": 28030
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.039519297434713e-06,
+      "loss": 0.8408,
+      "step": 28040
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.03489715738387e-06,
+      "loss": 0.9871,
+      "step": 28050
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.030275017333026e-06,
+      "loss": 0.9033,
+      "step": 28060
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.025652877282182e-06,
+      "loss": 0.8382,
+      "step": 28070
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.021030737231339e-06,
+      "loss": 0.9183,
+      "step": 28080
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.016408597180495e-06,
+      "loss": 0.8098,
+      "step": 28090
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.011786457129651e-06,
+      "loss": 0.7502,
+      "step": 28100
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.0071643170788084e-06,
+      "loss": 0.99,
+      "step": 28110
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.002542177027964e-06,
+      "loss": 0.9041,
+      "step": 28120
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.997920036977121e-06,
+      "loss": 1.0339,
+      "step": 28130
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.9932978969262776e-06,
+      "loss": 0.7962,
+      "step": 28140
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.988675756875433e-06,
+      "loss": 0.9557,
+      "step": 28150
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.98405361682459e-06,
+      "loss": 0.7668,
+      "step": 28160
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.979431476773747e-06,
+      "loss": 0.8209,
+      "step": 28170
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.974809336722903e-06,
+      "loss": 0.9335,
+      "step": 28180
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.97018719667206e-06,
+      "loss": 0.6805,
+      "step": 28190
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.965565056621217e-06,
+      "loss": 0.7766,
+      "step": 28200
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.9609429165703725e-06,
+      "loss": 0.7428,
+      "step": 28210
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.956320776519529e-06,
+      "loss": 0.6647,
+      "step": 28220
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.951698636468686e-06,
+      "loss": 0.9651,
+      "step": 28230
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.947076496417842e-06,
+      "loss": 0.9835,
+      "step": 28240
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.942454356366999e-06,
+      "loss": 0.9615,
+      "step": 28250
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.937832216316155e-06,
+      "loss": 0.8178,
+      "step": 28260
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.933210076265311e-06,
+      "loss": 1.0031,
+      "step": 28270
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.928587936214468e-06,
+      "loss": 0.8079,
+      "step": 28280
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.923965796163624e-06,
+      "loss": 0.8104,
+      "step": 28290
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.91934365611278e-06,
+      "loss": 0.7849,
+      "step": 28300
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.914721516061937e-06,
+      "loss": 0.7279,
+      "step": 28310
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.910099376011093e-06,
+      "loss": 0.6253,
+      "step": 28320
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.905477235960251e-06,
+      "loss": 0.8961,
+      "step": 28330
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.9008550959094065e-06,
+      "loss": 0.751,
+      "step": 28340
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.896232955858563e-06,
+      "loss": 0.9986,
+      "step": 28350
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.89161081580772e-06,
+      "loss": 0.6315,
+      "step": 28360
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.8869886757568764e-06,
+      "loss": 0.8451,
+      "step": 28370
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.882366535706032e-06,
+      "loss": 0.6771,
+      "step": 28380
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.877744395655189e-06,
+      "loss": 1.0568,
+      "step": 28390
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.8731222556043456e-06,
+      "loss": 0.9744,
+      "step": 28400
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.868500115553501e-06,
+      "loss": 0.9114,
+      "step": 28410
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.863877975502659e-06,
+      "loss": 0.8228,
+      "step": 28420
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.859255835451815e-06,
+      "loss": 0.9338,
+      "step": 28430
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.8546336954009705e-06,
+      "loss": 0.9738,
+      "step": 28440
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.850011555350128e-06,
+      "loss": 0.8113,
+      "step": 28450
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.845389415299284e-06,
+      "loss": 0.6404,
+      "step": 28460
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.840767275248441e-06,
+      "loss": 0.9907,
+      "step": 28470
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.836145135197597e-06,
+      "loss": 0.7706,
+      "step": 28480
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.831522995146753e-06,
+      "loss": 0.6776,
+      "step": 28490
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.8269008550959104e-06,
+      "loss": 1.0682,
+      "step": 28500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.822278715045066e-06,
+      "loss": 0.839,
+      "step": 28510
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.817656574994223e-06,
+      "loss": 0.8554,
+      "step": 28520
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.8130344349433796e-06,
+      "loss": 0.988,
+      "step": 28530
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.808412294892535e-06,
+      "loss": 0.9763,
+      "step": 28540
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.803790154841692e-06,
+      "loss": 0.8827,
+      "step": 28550
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.799168014790849e-06,
+      "loss": 0.9516,
+      "step": 28560
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.794545874740005e-06,
+      "loss": 0.8742,
+      "step": 28570
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.789923734689161e-06,
+      "loss": 0.7259,
+      "step": 28580
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.785301594638319e-06,
+      "loss": 0.7773,
+      "step": 28590
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.7806794545874745e-06,
+      "loss": 0.9354,
+      "step": 28600
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.776057314536631e-06,
+      "loss": 1.0987,
+      "step": 28610
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.771435174485788e-06,
+      "loss": 0.8863,
+      "step": 28620
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.766813034434944e-06,
+      "loss": 0.877,
+      "step": 28630
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.762190894384101e-06,
+      "loss": 1.0547,
+      "step": 28640
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.757568754333257e-06,
+      "loss": 0.7584,
+      "step": 28650
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.752946614282413e-06,
+      "loss": 0.8784,
+      "step": 28660
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.74832447423157e-06,
+      "loss": 0.8218,
+      "step": 28670
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.743702334180726e-06,
+      "loss": 0.709,
+      "step": 28680
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.739080194129882e-06,
+      "loss": 0.9779,
+      "step": 28690
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.734458054079039e-06,
+      "loss": 0.8283,
+      "step": 28700
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.729835914028195e-06,
+      "loss": 0.8149,
+      "step": 28710
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.725213773977352e-06,
+      "loss": 0.7666,
+      "step": 28720
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.7205916339265085e-06,
+      "loss": 0.8385,
+      "step": 28730
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.715969493875665e-06,
+      "loss": 0.7309,
+      "step": 28740
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.711347353824822e-06,
+      "loss": 0.8958,
+      "step": 28750
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.706725213773978e-06,
+      "loss": 0.7202,
+      "step": 28760
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.702103073723134e-06,
+      "loss": 0.5692,
+      "step": 28770
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.697480933672291e-06,
+      "loss": 0.7282,
+      "step": 28780
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.6928587936214476e-06,
+      "loss": 0.8439,
+      "step": 28790
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.688236653570603e-06,
+      "loss": 0.7984,
+      "step": 28800
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.683614513519761e-06,
+      "loss": 0.6564,
+      "step": 28810
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.678992373468917e-06,
+      "loss": 0.8698,
+      "step": 28820
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.6743702334180725e-06,
+      "loss": 0.671,
+      "step": 28830
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.66974809336723e-06,
+      "loss": 0.8469,
+      "step": 28840
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.665125953316386e-06,
+      "loss": 0.9068,
+      "step": 28850
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.660503813265542e-06,
+      "loss": 0.658,
+      "step": 28860
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.655881673214699e-06,
+      "loss": 0.9194,
+      "step": 28870
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.651259533163855e-06,
+      "loss": 0.7324,
+      "step": 28880
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.6466373931130124e-06,
+      "loss": 1.0145,
+      "step": 28890
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.642015253062168e-06,
+      "loss": 0.7628,
+      "step": 28900
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.637393113011325e-06,
+      "loss": 0.7524,
+      "step": 28910
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.6327709729604816e-06,
+      "loss": 0.892,
+      "step": 28920
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.628148832909637e-06,
+      "loss": 0.8747,
+      "step": 28930
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.623526692858794e-06,
+      "loss": 0.8772,
+      "step": 28940
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.618904552807951e-06,
+      "loss": 0.9123,
+      "step": 28950
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.614282412757107e-06,
+      "loss": 1.0148,
+      "step": 28960
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.609660272706263e-06,
+      "loss": 0.8173,
+      "step": 28970
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.605038132655421e-06,
+      "loss": 0.8268,
+      "step": 28980
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.6004159926045765e-06,
+      "loss": 1.0235,
+      "step": 28990
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.595793852553732e-06,
+      "loss": 0.6544,
+      "step": 29000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.59117171250289e-06,
+      "loss": 0.8179,
+      "step": 29010
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.586549572452046e-06,
+      "loss": 0.8549,
+      "step": 29020
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.581927432401203e-06,
+      "loss": 0.8156,
+      "step": 29030
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.577305292350359e-06,
+      "loss": 0.9411,
+      "step": 29040
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.572683152299515e-06,
+      "loss": 0.6342,
+      "step": 29050
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.568061012248672e-06,
+      "loss": 0.779,
+      "step": 29060
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.563438872197828e-06,
+      "loss": 0.8512,
+      "step": 29070
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.558816732146984e-06,
+      "loss": 0.7971,
+      "step": 29080
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.554194592096141e-06,
+      "loss": 0.7516,
+      "step": 29090
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.549572452045297e-06,
+      "loss": 0.8885,
+      "step": 29100
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.544950311994454e-06,
+      "loss": 0.8636,
+      "step": 29110
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.5403281719436105e-06,
+      "loss": 0.8115,
+      "step": 29120
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.535706031892767e-06,
+      "loss": 0.7288,
+      "step": 29130
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.531083891841923e-06,
+      "loss": 1.0775,
+      "step": 29140
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.52646175179108e-06,
+      "loss": 0.8737,
+      "step": 29150
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.521839611740236e-06,
+      "loss": 1.0076,
+      "step": 29160
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.517217471689393e-06,
+      "loss": 1.0064,
+      "step": 29170
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.5125953316385496e-06,
+      "loss": 0.7731,
+      "step": 29180
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.507973191587705e-06,
+      "loss": 1.0141,
+      "step": 29190
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.503351051536863e-06,
+      "loss": 0.6796,
+      "step": 29200
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.498728911486019e-06,
+      "loss": 0.7433,
+      "step": 29210
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.4941067714351745e-06,
+      "loss": 0.8014,
+      "step": 29220
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.489484631384332e-06,
+      "loss": 0.7212,
+      "step": 29230
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.484862491333488e-06,
+      "loss": 0.8814,
+      "step": 29240
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.480240351282644e-06,
+      "loss": 1.2298,
+      "step": 29250
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.475618211231801e-06,
+      "loss": 0.8067,
+      "step": 29260
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.470996071180957e-06,
+      "loss": 0.8472,
+      "step": 29270
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.466373931130114e-06,
+      "loss": 0.6584,
+      "step": 29280
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.46175179107927e-06,
+      "loss": 0.7473,
+      "step": 29290
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.457129651028426e-06,
+      "loss": 0.6881,
+      "step": 29300
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.4525075109775836e-06,
+      "loss": 0.8737,
+      "step": 29310
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.447885370926739e-06,
+      "loss": 0.729,
+      "step": 29320
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.443263230875896e-06,
+      "loss": 0.8868,
+      "step": 29330
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.438641090825053e-06,
+      "loss": 0.7016,
+      "step": 29340
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.434018950774209e-06,
+      "loss": 0.6573,
+      "step": 29350
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.429396810723365e-06,
+      "loss": 0.822,
+      "step": 29360
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.424774670672523e-06,
+      "loss": 1.1005,
+      "step": 29370
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.4201525306216785e-06,
+      "loss": 0.8621,
+      "step": 29380
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.415530390570834e-06,
+      "loss": 1.0037,
+      "step": 29390
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.410908250519992e-06,
+      "loss": 0.8602,
+      "step": 29400
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.406286110469148e-06,
+      "loss": 0.7757,
+      "step": 29410
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.401663970418303e-06,
+      "loss": 0.7273,
+      "step": 29420
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.397041830367461e-06,
+      "loss": 0.8747,
+      "step": 29430
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.392419690316617e-06,
+      "loss": 0.8926,
+      "step": 29440
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.387797550265774e-06,
+      "loss": 0.634,
+      "step": 29450
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.38317541021493e-06,
+      "loss": 0.9905,
+      "step": 29460
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.378553270164086e-06,
+      "loss": 0.8824,
+      "step": 29470
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.373931130113243e-06,
+      "loss": 0.815,
+      "step": 29480
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.369308990062399e-06,
+      "loss": 1.0707,
+      "step": 29490
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.364686850011556e-06,
+      "loss": 0.9669,
+      "step": 29500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.3600647099607125e-06,
+      "loss": 0.9408,
+      "step": 29510
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.355442569909869e-06,
+      "loss": 0.895,
+      "step": 29520
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.350820429859025e-06,
+      "loss": 0.8228,
+      "step": 29530
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.346198289808182e-06,
+      "loss": 0.7528,
+      "step": 29540
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.341576149757338e-06,
+      "loss": 0.8785,
+      "step": 29550
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.336954009706494e-06,
+      "loss": 0.8125,
+      "step": 29560
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.332331869655652e-06,
+      "loss": 0.8091,
+      "step": 29570
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.327709729604807e-06,
+      "loss": 0.7012,
+      "step": 29580
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.323087589553965e-06,
+      "loss": 0.8517,
+      "step": 29590
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.318465449503121e-06,
+      "loss": 0.6918,
+      "step": 29600
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.3138433094522765e-06,
+      "loss": 0.732,
+      "step": 29610
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.309221169401434e-06,
+      "loss": 0.983,
+      "step": 29620
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.30459902935059e-06,
+      "loss": 0.9221,
+      "step": 29630
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.299976889299746e-06,
+      "loss": 0.6968,
+      "step": 29640
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.295354749248903e-06,
+      "loss": 0.9248,
+      "step": 29650
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.290732609198059e-06,
+      "loss": 0.849,
+      "step": 29660
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.286110469147216e-06,
+      "loss": 0.889,
+      "step": 29670
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.281488329096372e-06,
+      "loss": 1.0205,
+      "step": 29680
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.276866189045528e-06,
+      "loss": 0.8745,
+      "step": 29690
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.272244048994685e-06,
+      "loss": 0.8305,
+      "step": 29700
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.267621908943841e-06,
+      "loss": 0.7744,
+      "step": 29710
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.262999768892998e-06,
+      "loss": 0.9345,
+      "step": 29720
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.258377628842155e-06,
+      "loss": 0.7373,
+      "step": 29730
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.253755488791311e-06,
+      "loss": 0.8527,
+      "step": 29740
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.249133348740467e-06,
+      "loss": 0.7444,
+      "step": 29750
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.244511208689624e-06,
+      "loss": 0.9182,
+      "step": 29760
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.2398890686387805e-06,
+      "loss": 0.9149,
+      "step": 29770
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.235266928587936e-06,
+      "loss": 0.7782,
+      "step": 29780
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.230644788537094e-06,
+      "loss": 0.9068,
+      "step": 29790
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.22602264848625e-06,
+      "loss": 0.9879,
+      "step": 29800
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.221400508435405e-06,
+      "loss": 0.7616,
+      "step": 29810
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.216778368384563e-06,
+      "loss": 0.6894,
+      "step": 29820
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.212156228333719e-06,
+      "loss": 0.8776,
+      "step": 29830
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.2075340882828745e-06,
+      "loss": 0.7742,
+      "step": 29840
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.202911948232032e-06,
+      "loss": 1.078,
+      "step": 29850
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.198289808181188e-06,
+      "loss": 0.7034,
+      "step": 29860
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.193667668130345e-06,
+      "loss": 0.7158,
+      "step": 29870
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.189045528079501e-06,
+      "loss": 0.7022,
+      "step": 29880
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.184423388028658e-06,
+      "loss": 1.044,
+      "step": 29890
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.1798012479778145e-06,
+      "loss": 0.7976,
+      "step": 29900
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.175179107926971e-06,
+      "loss": 1.044,
+      "step": 29910
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.170556967876127e-06,
+      "loss": 0.6926,
+      "step": 29920
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.165934827825284e-06,
+      "loss": 0.945,
+      "step": 29930
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.16131268777444e-06,
+      "loss": 0.9021,
+      "step": 29940
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.156690547723596e-06,
+      "loss": 0.9632,
+      "step": 29950
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.152068407672754e-06,
+      "loss": 0.8855,
+      "step": 29960
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.147446267621909e-06,
+      "loss": 0.7423,
+      "step": 29970
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.142824127571065e-06,
+      "loss": 0.8642,
+      "step": 29980
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.138201987520223e-06,
+      "loss": 0.7475,
+      "step": 29990
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.1335798474693785e-06,
+      "loss": 0.9552,
+      "step": 30000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.128957707418536e-06,
+      "loss": 0.7389,
+      "step": 30010
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.124335567367692e-06,
+      "loss": 0.8187,
+      "step": 30020
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.119713427316848e-06,
+      "loss": 0.8994,
+      "step": 30030
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.115091287266005e-06,
+      "loss": 0.8383,
+      "step": 30040
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.110469147215161e-06,
+      "loss": 0.7675,
+      "step": 30050
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.105847007164318e-06,
+      "loss": 0.8097,
+      "step": 30060
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.101224867113474e-06,
+      "loss": 0.9229,
+      "step": 30070
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.09660272706263e-06,
+      "loss": 0.7139,
+      "step": 30080
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.091980587011787e-06,
+      "loss": 1.0467,
+      "step": 30090
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.087358446960943e-06,
+      "loss": 0.7595,
+      "step": 30100
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.0827363069101e-06,
+      "loss": 1.0367,
+      "step": 30110
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.078114166859256e-06,
+      "loss": 0.8861,
+      "step": 30120
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.073492026808413e-06,
+      "loss": 0.8908,
+      "step": 30130
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.068869886757569e-06,
+      "loss": 0.6384,
+      "step": 30140
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.064247746706726e-06,
+      "loss": 0.98,
+      "step": 30150
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.0596256066558825e-06,
+      "loss": 0.6826,
+      "step": 30160
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.055003466605038e-06,
+      "loss": 0.794,
+      "step": 30170
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.050381326554196e-06,
+      "loss": 0.9946,
+      "step": 30180
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.045759186503352e-06,
+      "loss": 0.8076,
+      "step": 30190
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.0411370464525074e-06,
+      "loss": 0.923,
+      "step": 30200
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.036514906401665e-06,
+      "loss": 0.845,
+      "step": 30210
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.031892766350821e-06,
+      "loss": 0.9595,
+      "step": 30220
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.0272706262999766e-06,
+      "loss": 0.8638,
+      "step": 30230
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.022648486249134e-06,
+      "loss": 0.9146,
+      "step": 30240
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.01802634619829e-06,
+      "loss": 0.9238,
+      "step": 30250
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.0134042061474465e-06,
+      "loss": 0.9476,
+      "step": 30260
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.008782066096603e-06,
+      "loss": 0.8826,
+      "step": 30270
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.00415992604576e-06,
+      "loss": 0.9541,
+      "step": 30280
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.9995377859949165e-06,
+      "loss": 1.0633,
+      "step": 30290
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.994915645944072e-06,
+      "loss": 0.9129,
+      "step": 30300
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.990293505893229e-06,
+      "loss": 0.8856,
+      "step": 30310
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.985671365842386e-06,
+      "loss": 0.9816,
+      "step": 30320
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.981049225791542e-06,
+      "loss": 0.8002,
+      "step": 30330
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.976427085740698e-06,
+      "loss": 0.6676,
+      "step": 30340
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.971804945689856e-06,
+      "loss": 0.9051,
+      "step": 30350
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.967182805639011e-06,
+      "loss": 0.8283,
+      "step": 30360
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.962560665588167e-06,
+      "loss": 0.7252,
+      "step": 30370
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.957938525537325e-06,
+      "loss": 0.7683,
+      "step": 30380
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9533163854864805e-06,
+      "loss": 0.9603,
+      "step": 30390
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.948694245435636e-06,
+      "loss": 0.9123,
+      "step": 30400
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.944072105384794e-06,
+      "loss": 0.8642,
+      "step": 30410
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.93944996533395e-06,
+      "loss": 0.7925,
+      "step": 30420
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.934827825283107e-06,
+      "loss": 0.8801,
+      "step": 30430
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.930205685232263e-06,
+      "loss": 0.8615,
+      "step": 30440
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.92558354518142e-06,
+      "loss": 0.9567,
+      "step": 30450
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.920961405130576e-06,
+      "loss": 0.7802,
+      "step": 30460
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.916339265079732e-06,
+      "loss": 1.1759,
+      "step": 30470
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.911717125028889e-06,
+      "loss": 0.8243,
+      "step": 30480
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.907094984978045e-06,
+      "loss": 0.7658,
+      "step": 30490
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.902472844927202e-06,
+      "loss": 0.779,
+      "step": 30500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.897850704876358e-06,
+      "loss": 0.8353,
+      "step": 30510
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.893228564825515e-06,
+      "loss": 0.9356,
+      "step": 30520
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.888606424774671e-06,
+      "loss": 0.7757,
+      "step": 30530
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.883984284723827e-06,
+      "loss": 0.8806,
+      "step": 30540
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8793621446729845e-06,
+      "loss": 0.8379,
+      "step": 30550
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.87474000462214e-06,
+      "loss": 0.9109,
+      "step": 30560
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.870117864571298e-06,
+      "loss": 1.0452,
+      "step": 30570
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.865495724520454e-06,
+      "loss": 0.8265,
+      "step": 30580
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8608735844696094e-06,
+      "loss": 0.8656,
+      "step": 30590
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.856251444418767e-06,
+      "loss": 0.8771,
+      "step": 30600
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.851629304367923e-06,
+      "loss": 0.798,
+      "step": 30610
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.8470071643170786e-06,
+      "loss": 0.9063,
+      "step": 30620
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.842385024266236e-06,
+      "loss": 1.0025,
+      "step": 30630
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.837762884215392e-06,
+      "loss": 0.6315,
+      "step": 30640
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.8331407441645485e-06,
+      "loss": 0.9338,
+      "step": 30650
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.828518604113705e-06,
+      "loss": 0.8425,
+      "step": 30660
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.823896464062862e-06,
+      "loss": 0.8984,
+      "step": 30670
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.819274324012018e-06,
+      "loss": 0.9077,
+      "step": 30680
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.814652183961174e-06,
+      "loss": 0.9365,
+      "step": 30690
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.810030043910331e-06,
+      "loss": 0.8355,
+      "step": 30700
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.805407903859488e-06,
+      "loss": 0.6667,
+      "step": 30710
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.800785763808644e-06,
+      "loss": 0.7739,
+      "step": 30720
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.7961636237578e-06,
+      "loss": 0.8172,
+      "step": 30730
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.791541483706958e-06,
+      "loss": 1.0454,
+      "step": 30740
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.786919343656113e-06,
+      "loss": 0.8318,
+      "step": 30750
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.782297203605269e-06,
+      "loss": 0.9377,
+      "step": 30760
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.777675063554427e-06,
+      "loss": 0.6052,
+      "step": 30770
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.7730529235035825e-06,
+      "loss": 1.0585,
+      "step": 30780
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.768430783452738e-06,
+      "loss": 0.8736,
+      "step": 30790
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.763808643401896e-06,
+      "loss": 1.0315,
+      "step": 30800
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.759186503351052e-06,
+      "loss": 0.7712,
+      "step": 30810
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.754564363300208e-06,
+      "loss": 0.6027,
+      "step": 30820
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.749942223249365e-06,
+      "loss": 0.7691,
+      "step": 30830
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.745320083198521e-06,
+      "loss": 0.6528,
+      "step": 30840
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.740697943147678e-06,
+      "loss": 0.6408,
+      "step": 30850
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.736075803096834e-06,
+      "loss": 0.9642,
+      "step": 30860
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.731453663045991e-06,
+      "loss": 0.9317,
+      "step": 30870
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.726831522995147e-06,
+      "loss": 0.9001,
+      "step": 30880
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.722209382944304e-06,
+      "loss": 0.9841,
+      "step": 30890
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.71758724289346e-06,
+      "loss": 0.7443,
+      "step": 30900
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.712965102842617e-06,
+      "loss": 0.8652,
+      "step": 30910
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.708342962791773e-06,
+      "loss": 0.8468,
+      "step": 30920
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.703720822740929e-06,
+      "loss": 0.7214,
+      "step": 30930
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6990986826900865e-06,
+      "loss": 0.7537,
+      "step": 30940
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.694476542639242e-06,
+      "loss": 0.8986,
+      "step": 30950
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.689854402588398e-06,
+      "loss": 0.7873,
+      "step": 30960
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.685232262537556e-06,
+      "loss": 0.803,
+      "step": 30970
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6806101224867114e-06,
+      "loss": 0.7862,
+      "step": 30980
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.675987982435869e-06,
+      "loss": 1.0793,
+      "step": 30990
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.671365842385025e-06,
+      "loss": 0.7539,
+      "step": 31000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6667437023341806e-06,
+      "loss": 0.8763,
+      "step": 31010
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.662121562283338e-06,
+      "loss": 0.6781,
+      "step": 31020
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.657499422232494e-06,
+      "loss": 0.9579,
+      "step": 31030
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6528772821816505e-06,
+      "loss": 0.9341,
+      "step": 31040
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.648255142130807e-06,
+      "loss": 1.1077,
+      "step": 31050
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.643633002079964e-06,
+      "loss": 0.9129,
+      "step": 31060
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.63901086202912e-06,
+      "loss": 0.856,
+      "step": 31070
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.634388721978276e-06,
+      "loss": 0.8224,
+      "step": 31080
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.629766581927433e-06,
+      "loss": 0.7852,
+      "step": 31090
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.625144441876589e-06,
+      "loss": 0.7073,
+      "step": 31100
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.620522301825746e-06,
+      "loss": 0.8302,
+      "step": 31110
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.615900161774902e-06,
+      "loss": 0.8213,
+      "step": 31120
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.61127802172406e-06,
+      "loss": 0.8312,
+      "step": 31130
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.606655881673215e-06,
+      "loss": 1.1522,
+      "step": 31140
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.602033741622371e-06,
+      "loss": 0.7461,
+      "step": 31150
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.597411601571529e-06,
+      "loss": 0.6285,
+      "step": 31160
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5927894615206845e-06,
+      "loss": 0.8482,
+      "step": 31170
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.58816732146984e-06,
+      "loss": 0.8016,
+      "step": 31180
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.583545181418998e-06,
+      "loss": 0.8694,
+      "step": 31190
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.578923041368154e-06,
+      "loss": 0.748,
+      "step": 31200
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.57430090131731e-06,
+      "loss": 0.6325,
+      "step": 31210
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.569678761266467e-06,
+      "loss": 0.9593,
+      "step": 31220
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.565056621215623e-06,
+      "loss": 0.8622,
+      "step": 31230
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.5604344811647794e-06,
+      "loss": 0.7673,
+      "step": 31240
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.555812341113936e-06,
+      "loss": 0.743,
+      "step": 31250
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.551190201063093e-06,
+      "loss": 0.7737,
+      "step": 31260
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.546568061012249e-06,
+      "loss": 1.0571,
+      "step": 31270
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.541945920961406e-06,
+      "loss": 0.8764,
+      "step": 31280
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.537323780910562e-06,
+      "loss": 0.8038,
+      "step": 31290
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5327016408597185e-06,
+      "loss": 0.8628,
+      "step": 31300
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.528079500808875e-06,
+      "loss": 0.8848,
+      "step": 31310
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.523457360758031e-06,
+      "loss": 0.9369,
+      "step": 31320
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5188352207071885e-06,
+      "loss": 0.882,
+      "step": 31330
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.514213080656344e-06,
+      "loss": 0.9396,
+      "step": 31340
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5095909406055e-06,
+      "loss": 0.8998,
+      "step": 31350
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.504968800554658e-06,
+      "loss": 1.3042,
+      "step": 31360
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5003466605038134e-06,
+      "loss": 0.9423,
+      "step": 31370
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.495724520452969e-06,
+      "loss": 0.7841,
+      "step": 31380
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.491102380402127e-06,
+      "loss": 0.9359,
+      "step": 31390
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.4864802403512826e-06,
+      "loss": 0.762,
+      "step": 31400
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.48185810030044e-06,
+      "loss": 0.706,
+      "step": 31410
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.477235960249596e-06,
+      "loss": 0.7816,
+      "step": 31420
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.4726138201987525e-06,
+      "loss": 0.966,
+      "step": 31430
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.467991680147909e-06,
+      "loss": 0.8295,
+      "step": 31440
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.463369540097066e-06,
+      "loss": 0.6668,
+      "step": 31450
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.458747400046222e-06,
+      "loss": 0.8141,
+      "step": 31460
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.454125259995378e-06,
+      "loss": 0.9401,
+      "step": 31470
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.449503119944535e-06,
+      "loss": 0.7878,
+      "step": 31480
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.444880979893691e-06,
+      "loss": 0.8414,
+      "step": 31490
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.440258839842848e-06,
+      "loss": 0.8171,
+      "step": 31500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.435636699792004e-06,
+      "loss": 0.6889,
+      "step": 31510
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.43101455974116e-06,
+      "loss": 1.1446,
+      "step": 31520
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.426392419690317e-06,
+      "loss": 0.9803,
+      "step": 31530
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.421770279639473e-06,
+      "loss": 0.7923,
+      "step": 31540
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.417148139588631e-06,
+      "loss": 0.9544,
+      "step": 31550
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.4125259995377865e-06,
+      "loss": 0.9628,
+      "step": 31560
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.407903859486942e-06,
+      "loss": 0.9337,
+      "step": 31570
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.4032817194361e-06,
+      "loss": 0.6862,
+      "step": 31580
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.398659579385256e-06,
+      "loss": 0.9313,
+      "step": 31590
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.394037439334412e-06,
+      "loss": 1.0059,
+      "step": 31600
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.389415299283569e-06,
+      "loss": 0.7732,
+      "step": 31610
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.384793159232725e-06,
+      "loss": 0.9112,
+      "step": 31620
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.3801710191818815e-06,
+      "loss": 0.6584,
+      "step": 31630
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.375548879131038e-06,
+      "loss": 1.0028,
+      "step": 31640
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.370926739080195e-06,
+      "loss": 0.5654,
+      "step": 31650
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.3663045990293506e-06,
+      "loss": 1.01,
+      "step": 31660
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.361682458978508e-06,
+      "loss": 1.0378,
+      "step": 31670
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.357060318927664e-06,
+      "loss": 0.8025,
+      "step": 31680
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.3524381788768205e-06,
+      "loss": 0.7708,
+      "step": 31690
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.347816038825977e-06,
+      "loss": 0.7094,
+      "step": 31700
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.343193898775133e-06,
+      "loss": 0.8695,
+      "step": 31710
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.3385717587242905e-06,
+      "loss": 0.9695,
+      "step": 31720
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.333949618673446e-06,
+      "loss": 1.1341,
+      "step": 31730
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.329327478622602e-06,
+      "loss": 0.8134,
+      "step": 31740
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.32470533857176e-06,
+      "loss": 0.7116,
+      "step": 31750
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.3200831985209155e-06,
+      "loss": 0.8082,
+      "step": 31760
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.315461058470071e-06,
+      "loss": 0.9589,
+      "step": 31770
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.310838918419229e-06,
+      "loss": 0.9333,
+      "step": 31780
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.306216778368385e-06,
+      "loss": 1.074,
+      "step": 31790
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.301594638317541e-06,
+      "loss": 0.9684,
+      "step": 31800
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.296972498266698e-06,
+      "loss": 0.8997,
+      "step": 31810
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.2923503582158545e-06,
+      "loss": 0.8261,
+      "step": 31820
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.287728218165011e-06,
+      "loss": 0.896,
+      "step": 31830
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.283106078114167e-06,
+      "loss": 0.8647,
+      "step": 31840
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.278483938063324e-06,
+      "loss": 1.0863,
+      "step": 31850
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.27386179801248e-06,
+      "loss": 0.8158,
+      "step": 31860
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.269239657961637e-06,
+      "loss": 1.0486,
+      "step": 31870
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.264617517910793e-06,
+      "loss": 0.7684,
+      "step": 31880
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.25999537785995e-06,
+      "loss": 0.9442,
+      "step": 31890
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.255373237809106e-06,
+      "loss": 0.7393,
+      "step": 31900
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.250751097758262e-06,
+      "loss": 0.9354,
+      "step": 31910
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.2461289577074194e-06,
+      "loss": 1.1204,
+      "step": 31920
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.241506817656575e-06,
+      "loss": 0.9393,
+      "step": 31930
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.236884677605731e-06,
+      "loss": 0.7269,
+      "step": 31940
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.2322625375548886e-06,
+      "loss": 0.902,
+      "step": 31950
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.227640397504044e-06,
+      "loss": 1.013,
+      "step": 31960
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.223018257453202e-06,
+      "loss": 0.8245,
+      "step": 31970
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.218396117402358e-06,
+      "loss": 0.9236,
+      "step": 31980
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.213773977351514e-06,
+      "loss": 0.7506,
+      "step": 31990
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.209151837300671e-06,
+      "loss": 0.7866,
+      "step": 32000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.204529697249827e-06,
+      "loss": 0.9782,
+      "step": 32010
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.1999075571989835e-06,
+      "loss": 0.9821,
+      "step": 32020
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.19528541714814e-06,
+      "loss": 0.7737,
+      "step": 32030
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.190663277097297e-06,
+      "loss": 1.1422,
+      "step": 32040
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.186041137046453e-06,
+      "loss": 0.8644,
+      "step": 32050
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.18141899699561e-06,
+      "loss": 0.7484,
+      "step": 32060
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.176796856944766e-06,
+      "loss": 1.0387,
+      "step": 32070
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.172174716893922e-06,
+      "loss": 0.8395,
+      "step": 32080
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.167552576843079e-06,
+      "loss": 0.9372,
+      "step": 32090
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.162930436792235e-06,
+      "loss": 0.8797,
+      "step": 32100
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.1583082967413925e-06,
+      "loss": 0.7911,
+      "step": 32110
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.153686156690548e-06,
+      "loss": 0.7839,
+      "step": 32120
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.149064016639704e-06,
+      "loss": 0.824,
+      "step": 32130
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.144441876588862e-06,
+      "loss": 0.9054,
+      "step": 32140
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.1398197365380175e-06,
+      "loss": 0.7858,
+      "step": 32150
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.135197596487173e-06,
+      "loss": 0.9093,
+      "step": 32160
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.130575456436331e-06,
+      "loss": 0.7226,
+      "step": 32170
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.125953316385487e-06,
+      "loss": 0.8678,
+      "step": 32180
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.121331176334643e-06,
+      "loss": 0.8112,
+      "step": 32190
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.1167090362838e-06,
+      "loss": 0.9278,
+      "step": 32200
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.1120868962329566e-06,
+      "loss": 0.7571,
+      "step": 32210
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.107464756182112e-06,
+      "loss": 0.7609,
+      "step": 32220
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.102842616131269e-06,
+      "loss": 1.0585,
+      "step": 32230
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.098220476080426e-06,
+      "loss": 0.8875,
+      "step": 32240
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.093598336029582e-06,
+      "loss": 0.7976,
+      "step": 32250
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.088976195978739e-06,
+      "loss": 0.9127,
+      "step": 32260
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.084354055927895e-06,
+      "loss": 0.7682,
+      "step": 32270
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.079731915877052e-06,
+      "loss": 0.6877,
+      "step": 32280
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.075109775826208e-06,
+      "loss": 0.6954,
+      "step": 32290
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.070487635775364e-06,
+      "loss": 0.971,
+      "step": 32300
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.0658654957245214e-06,
+      "loss": 0.9932,
+      "step": 32310
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.061243355673677e-06,
+      "loss": 0.9673,
+      "step": 32320
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.056621215622833e-06,
+      "loss": 0.8421,
+      "step": 32330
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.0519990755719906e-06,
+      "loss": 0.9211,
+      "step": 32340
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.047376935521146e-06,
+      "loss": 0.726,
+      "step": 32350
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.042754795470303e-06,
+      "loss": 0.8816,
+      "step": 32360
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.03813265541946e-06,
+      "loss": 0.8957,
+      "step": 32370
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.0335105153686155e-06,
+      "loss": 1.0791,
+      "step": 32380
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.028888375317773e-06,
+      "loss": 0.8557,
+      "step": 32390
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.024266235266929e-06,
+      "loss": 0.7816,
+      "step": 32400
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.0196440952160855e-06,
+      "loss": 0.8009,
+      "step": 32410
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.015021955165242e-06,
+      "loss": 0.7598,
+      "step": 32420
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.010399815114399e-06,
+      "loss": 0.6727,
+      "step": 32430
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.005777675063555e-06,
+      "loss": 0.9336,
+      "step": 32440
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.001155535012712e-06,
+      "loss": 0.6683,
+      "step": 32450
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.996533394961868e-06,
+      "loss": 0.771,
+      "step": 32460
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.9919112549110246e-06,
+      "loss": 1.0298,
+      "step": 32470
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.98728911486018e-06,
+      "loss": 0.9055,
+      "step": 32480
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.982666974809337e-06,
+      "loss": 0.9042,
+      "step": 32490
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.978044834758494e-06,
+      "loss": 0.9432,
+      "step": 32500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.97342269470765e-06,
+      "loss": 0.8511,
+      "step": 32510
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.968800554656806e-06,
+      "loss": 0.8693,
+      "step": 32520
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.964178414605963e-06,
+      "loss": 0.8486,
+      "step": 32530
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.9595562745551195e-06,
+      "loss": 0.8052,
+      "step": 32540
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.954934134504275e-06,
+      "loss": 0.9511,
+      "step": 32550
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.950311994453432e-06,
+      "loss": 0.849,
+      "step": 32560
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.945689854402589e-06,
+      "loss": 0.6478,
+      "step": 32570
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.941067714351745e-06,
+      "loss": 0.7908,
+      "step": 32580
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.936445574300902e-06,
+      "loss": 0.7656,
+      "step": 32590
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.9318234342500586e-06,
+      "loss": 0.5802,
+      "step": 32600
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.927201294199214e-06,
+      "loss": 0.6914,
+      "step": 32610
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.922579154148371e-06,
+      "loss": 0.7908,
+      "step": 32620
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.917957014097528e-06,
+      "loss": 0.8673,
+      "step": 32630
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.913334874046684e-06,
+      "loss": 0.7313,
+      "step": 32640
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.908712733995841e-06,
+      "loss": 0.7159,
+      "step": 32650
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.904090593944997e-06,
+      "loss": 0.6851,
+      "step": 32660
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.8994684538941535e-06,
+      "loss": 0.6576,
+      "step": 32670
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.89484631384331e-06,
+      "loss": 0.9487,
+      "step": 32680
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.890224173792466e-06,
+      "loss": 0.7275,
+      "step": 32690
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.885602033741623e-06,
+      "loss": 0.942,
+      "step": 32700
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.880979893690779e-06,
+      "loss": 0.5889,
+      "step": 32710
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.876357753639936e-06,
+      "loss": 0.8014,
+      "step": 32720
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.871735613589092e-06,
+      "loss": 0.6518,
+      "step": 32730
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.867113473538248e-06,
+      "loss": 0.7904,
+      "step": 32740
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.862491333487405e-06,
+      "loss": 0.7874,
+      "step": 32750
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.857869193436562e-06,
+      "loss": 0.8205,
+      "step": 32760
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8532470533857175e-06,
+      "loss": 0.8301,
+      "step": 32770
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.848624913334874e-06,
+      "loss": 0.9569,
+      "step": 32780
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.844002773284031e-06,
+      "loss": 0.8445,
+      "step": 32790
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8393806332331875e-06,
+      "loss": 0.8738,
+      "step": 32800
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.834758493182344e-06,
+      "loss": 0.8151,
+      "step": 32810
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.830136353131501e-06,
+      "loss": 0.7504,
+      "step": 32820
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.825514213080657e-06,
+      "loss": 0.7596,
+      "step": 32830
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.820892073029813e-06,
+      "loss": 0.9487,
+      "step": 32840
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.81626993297897e-06,
+      "loss": 0.9567,
+      "step": 32850
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8116477929281266e-06,
+      "loss": 0.789,
+      "step": 32860
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.807025652877282e-06,
+      "loss": 0.8866,
+      "step": 32870
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.802403512826439e-06,
+      "loss": 0.7828,
+      "step": 32880
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.797781372775596e-06,
+      "loss": 0.9807,
+      "step": 32890
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.7931592327247515e-06,
+      "loss": 0.8544,
+      "step": 32900
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.788537092673908e-06,
+      "loss": 0.6157,
+      "step": 32910
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.783914952623065e-06,
+      "loss": 1.1017,
+      "step": 32920
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.7792928125722215e-06,
+      "loss": 0.8704,
+      "step": 32930
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.774670672521377e-06,
+      "loss": 0.8481,
+      "step": 32940
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.770048532470534e-06,
+      "loss": 0.7485,
+      "step": 32950
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.765426392419691e-06,
+      "loss": 0.8658,
+      "step": 32960
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.760804252368847e-06,
+      "loss": 0.7358,
+      "step": 32970
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.756182112318004e-06,
+      "loss": 0.6847,
+      "step": 32980
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.7515599722671606e-06,
+      "loss": 0.8797,
+      "step": 32990
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.746937832216316e-06,
+      "loss": 0.796,
+      "step": 33000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.742315692165473e-06,
+      "loss": 0.8004,
+      "step": 33010
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.73769355211463e-06,
+      "loss": 0.8445,
+      "step": 33020
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.733071412063786e-06,
+      "loss": 0.9541,
+      "step": 33030
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.728449272012942e-06,
+      "loss": 0.7739,
+      "step": 33040
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.723827131962099e-06,
+      "loss": 0.7691,
+      "step": 33050
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.7192049919112555e-06,
+      "loss": 0.8311,
+      "step": 33060
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.714582851860412e-06,
+      "loss": 0.8351,
+      "step": 33070
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.709960711809568e-06,
+      "loss": 0.7763,
+      "step": 33080
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.705338571758725e-06,
+      "loss": 0.8614,
+      "step": 33090
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.700716431707881e-06,
+      "loss": 0.8305,
+      "step": 33100
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.696094291657037e-06,
+      "loss": 0.7194,
+      "step": 33110
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.691472151606194e-06,
+      "loss": 0.7676,
+      "step": 33120
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.68685001155535e-06,
+      "loss": 0.9108,
+      "step": 33130
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.682227871504507e-06,
+      "loss": 1.0234,
+      "step": 33140
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.677605731453663e-06,
+      "loss": 0.9036,
+      "step": 33150
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.6729835914028195e-06,
+      "loss": 0.8526,
+      "step": 33160
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.668361451351976e-06,
+      "loss": 0.6643,
+      "step": 33170
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.663739311301133e-06,
+      "loss": 0.9805,
+      "step": 33180
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.6591171712502895e-06,
+      "loss": 0.917,
+      "step": 33190
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.654495031199446e-06,
+      "loss": 0.9361,
+      "step": 33200
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.649872891148603e-06,
+      "loss": 1.0172,
+      "step": 33210
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.645250751097759e-06,
+      "loss": 0.93,
+      "step": 33220
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.640628611046915e-06,
+      "loss": 0.7856,
+      "step": 33230
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.636006470996072e-06,
+      "loss": 0.7703,
+      "step": 33240
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.631384330945228e-06,
+      "loss": 0.6893,
+      "step": 33250
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.626762190894384e-06,
+      "loss": 0.8037,
+      "step": 33260
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.622140050843541e-06,
+      "loss": 0.8606,
+      "step": 33270
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.617517910792698e-06,
+      "loss": 0.7846,
+      "step": 33280
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.6128957707418535e-06,
+      "loss": 0.5493,
+      "step": 33290
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.60827363069101e-06,
+      "loss": 0.874,
+      "step": 33300
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.603651490640167e-06,
+      "loss": 0.6667,
+      "step": 33310
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.599029350589323e-06,
+      "loss": 0.8831,
+      "step": 33320
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.594407210538479e-06,
+      "loss": 0.8478,
+      "step": 33330
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.589785070487636e-06,
+      "loss": 0.7208,
+      "step": 33340
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.585162930436793e-06,
+      "loss": 0.8044,
+      "step": 33350
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.580540790385949e-06,
+      "loss": 0.8109,
+      "step": 33360
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.575918650335106e-06,
+      "loss": 0.8922,
+      "step": 33370
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.571296510284262e-06,
+      "loss": 0.8493,
+      "step": 33380
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.566674370233418e-06,
+      "loss": 0.7479,
+      "step": 33390
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.562052230182575e-06,
+      "loss": 0.8697,
+      "step": 33400
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.557430090131732e-06,
+      "loss": 0.8479,
+      "step": 33410
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.552807950080888e-06,
+      "loss": 1.0081,
+      "step": 33420
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.548185810030044e-06,
+      "loss": 0.7775,
+      "step": 33430
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.543563669979201e-06,
+      "loss": 0.8812,
+      "step": 33440
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.5389415299283575e-06,
+      "loss": 0.6718,
+      "step": 33450
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.534319389877513e-06,
+      "loss": 0.9486,
+      "step": 33460
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.52969724982667e-06,
+      "loss": 0.9903,
+      "step": 33470
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.525075109775827e-06,
+      "loss": 0.8399,
+      "step": 33480
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.520452969724983e-06,
+      "loss": 1.0769,
+      "step": 33490
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.515830829674139e-06,
+      "loss": 0.5462,
+      "step": 33500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.511208689623296e-06,
+      "loss": 0.9348,
+      "step": 33510
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.506586549572452e-06,
+      "loss": 1.0336,
+      "step": 33520
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.501964409521609e-06,
+      "loss": 0.8455,
+      "step": 33530
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.497342269470765e-06,
+      "loss": 0.6238,
+      "step": 33540
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4927201294199215e-06,
+      "loss": 0.6453,
+      "step": 33550
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.488097989369078e-06,
+      "loss": 0.9314,
+      "step": 33560
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.483475849318235e-06,
+      "loss": 1.0848,
+      "step": 33570
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4788537092673915e-06,
+      "loss": 0.5453,
+      "step": 33580
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.474231569216548e-06,
+      "loss": 0.7674,
+      "step": 33590
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.469609429165704e-06,
+      "loss": 0.7785,
+      "step": 33600
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.464987289114861e-06,
+      "loss": 0.6392,
+      "step": 33610
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.460365149064017e-06,
+      "loss": 0.7415,
+      "step": 33620
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.455743009013174e-06,
+      "loss": 0.8186,
+      "step": 33630
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.45112086896233e-06,
+      "loss": 0.7765,
+      "step": 33640
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.446498728911486e-06,
+      "loss": 1.0137,
+      "step": 33650
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.441876588860643e-06,
+      "loss": 0.8581,
+      "step": 33660
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.437254448809799e-06,
+      "loss": 0.7715,
+      "step": 33670
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.4326323087589555e-06,
+      "loss": 0.9221,
+      "step": 33680
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.428010168708112e-06,
+      "loss": 1.0978,
+      "step": 33690
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.423388028657269e-06,
+      "loss": 0.9445,
+      "step": 33700
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.418765888606425e-06,
+      "loss": 0.8869,
+      "step": 33710
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.414143748555581e-06,
+      "loss": 0.9934,
+      "step": 33720
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.409521608504738e-06,
+      "loss": 0.849,
+      "step": 33730
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.404899468453895e-06,
+      "loss": 0.6828,
+      "step": 33740
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.400277328403051e-06,
+      "loss": 0.7154,
+      "step": 33750
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.395655188352208e-06,
+      "loss": 0.9535,
+      "step": 33760
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.391033048301364e-06,
+      "loss": 0.8457,
+      "step": 33770
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.38641090825052e-06,
+      "loss": 0.77,
+      "step": 33780
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.381788768199677e-06,
+      "loss": 0.8939,
+      "step": 33790
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.377166628148834e-06,
+      "loss": 0.6355,
+      "step": 33800
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3725444880979895e-06,
+      "loss": 0.812,
+      "step": 33810
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.367922348047146e-06,
+      "loss": 0.8799,
+      "step": 33820
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.363300207996303e-06,
+      "loss": 0.6812,
+      "step": 33830
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3586780679454595e-06,
+      "loss": 0.9337,
+      "step": 33840
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.354055927894615e-06,
+      "loss": 0.7518,
+      "step": 33850
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.349433787843772e-06,
+      "loss": 0.945,
+      "step": 33860
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.344811647792929e-06,
+      "loss": 1.0115,
+      "step": 33870
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.340189507742084e-06,
+      "loss": 0.7932,
+      "step": 33880
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.335567367691241e-06,
+      "loss": 0.9372,
+      "step": 33890
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.330945227640398e-06,
+      "loss": 1.0376,
+      "step": 33900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.326323087589554e-06,
+      "loss": 0.8187,
+      "step": 33910
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.32170094753871e-06,
+      "loss": 0.9372,
+      "step": 33920
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.317078807487867e-06,
+      "loss": 0.6733,
+      "step": 33930
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3124566674370235e-06,
+      "loss": 0.7882,
+      "step": 33940
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.30783452738618e-06,
+      "loss": 0.7345,
+      "step": 33950
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.303212387335337e-06,
+      "loss": 0.8703,
+      "step": 33960
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.2985902472844935e-06,
+      "loss": 0.782,
+      "step": 33970
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.29396810723365e-06,
+      "loss": 0.904,
+      "step": 33980
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.289345967182806e-06,
+      "loss": 0.8714,
+      "step": 33990
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.284723827131963e-06,
+      "loss": 0.7232,
+      "step": 34000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.280101687081119e-06,
+      "loss": 0.7414,
+      "step": 34010
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.275479547030275e-06,
+      "loss": 0.7351,
+      "step": 34020
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.270857406979432e-06,
+      "loss": 0.892,
+      "step": 34030
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.266235266928588e-06,
+      "loss": 0.7479,
+      "step": 34040
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.261613126877745e-06,
+      "loss": 0.7142,
+      "step": 34050
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.256990986826901e-06,
+      "loss": 0.8768,
+      "step": 34060
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.2523688467760575e-06,
+      "loss": 0.8008,
+      "step": 34070
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.247746706725214e-06,
+      "loss": 0.7866,
+      "step": 34080
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.24312456667437e-06,
+      "loss": 0.8818,
+      "step": 34090
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.238502426623527e-06,
+      "loss": 0.793,
+      "step": 34100
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.233880286572683e-06,
+      "loss": 0.8238,
+      "step": 34110
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.22925814652184e-06,
+      "loss": 0.8727,
+      "step": 34120
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.224636006470997e-06,
+      "loss": 0.8885,
+      "step": 34130
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.220013866420153e-06,
+      "loss": 1.0344,
+      "step": 34140
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.215391726369309e-06,
+      "loss": 0.8353,
+      "step": 34150
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.210769586318466e-06,
+      "loss": 0.83,
+      "step": 34160
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.206147446267622e-06,
+      "loss": 0.9764,
+      "step": 34170
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.201525306216779e-06,
+      "loss": 0.9879,
+      "step": 34180
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.196903166165936e-06,
+      "loss": 0.7826,
+      "step": 34190
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.1922810261150915e-06,
+      "loss": 0.8169,
+      "step": 34200
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.187658886064248e-06,
+      "loss": 0.6759,
+      "step": 34210
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.183036746013405e-06,
+      "loss": 0.8889,
+      "step": 34220
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.178414605962561e-06,
+      "loss": 0.8383,
+      "step": 34230
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.173792465911717e-06,
+      "loss": 0.9674,
+      "step": 34240
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.169170325860874e-06,
+      "loss": 0.6895,
+      "step": 34250
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.164548185810031e-06,
+      "loss": 0.8246,
+      "step": 34260
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.1599260457591864e-06,
+      "loss": 0.7988,
+      "step": 34270
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.155303905708343e-06,
+      "loss": 0.8485,
+      "step": 34280
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.1506817656575e-06,
+      "loss": 0.7348,
+      "step": 34290
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.146059625606656e-06,
+      "loss": 0.6433,
+      "step": 34300
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.141437485555812e-06,
+      "loss": 0.843,
+      "step": 34310
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.136815345504969e-06,
+      "loss": 0.7484,
+      "step": 34320
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1321932054541255e-06,
+      "loss": 1.0112,
+      "step": 34330
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.127571065403282e-06,
+      "loss": 0.9275,
+      "step": 34340
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.122948925352439e-06,
+      "loss": 0.7421,
+      "step": 34350
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1183267853015955e-06,
+      "loss": 0.8663,
+      "step": 34360
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.113704645250751e-06,
+      "loss": 0.8148,
+      "step": 34370
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.109082505199908e-06,
+      "loss": 1.0629,
+      "step": 34380
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.104460365149065e-06,
+      "loss": 0.8802,
+      "step": 34390
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.099838225098221e-06,
+      "loss": 0.7113,
+      "step": 34400
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.095216085047377e-06,
+      "loss": 0.7695,
+      "step": 34410
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.090593944996534e-06,
+      "loss": 0.8435,
+      "step": 34420
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.08597180494569e-06,
+      "loss": 0.958,
+      "step": 34430
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.081349664894846e-06,
+      "loss": 0.9479,
+      "step": 34440
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.076727524844003e-06,
+      "loss": 0.8277,
+      "step": 34450
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.0721053847931595e-06,
+      "loss": 1.0545,
+      "step": 34460
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.067483244742316e-06,
+      "loss": 0.9099,
+      "step": 34470
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.062861104691472e-06,
+      "loss": 0.9452,
+      "step": 34480
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.058238964640629e-06,
+      "loss": 0.8064,
+      "step": 34490
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.053616824589785e-06,
+      "loss": 0.7501,
+      "step": 34500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.048994684538942e-06,
+      "loss": 0.7351,
+      "step": 34510
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.044372544488099e-06,
+      "loss": 0.647,
+      "step": 34520
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.039750404437255e-06,
+      "loss": 0.9834,
+      "step": 34530
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.035128264386411e-06,
+      "loss": 0.8692,
+      "step": 34540
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.030506124335568e-06,
+      "loss": 0.7937,
+      "step": 34550
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.025883984284724e-06,
+      "loss": 0.8545,
+      "step": 34560
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.021261844233881e-06,
+      "loss": 0.8305,
+      "step": 34570
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.016639704183037e-06,
+      "loss": 0.7454,
+      "step": 34580
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0120175641321935e-06,
+      "loss": 0.9621,
+      "step": 34590
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.00739542408135e-06,
+      "loss": 1.4021,
+      "step": 34600
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.002773284030507e-06,
+      "loss": 0.8853,
+      "step": 34610
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.7099942702791192,
+      "eval_loss": 0.8126096725463867,
+      "eval_runtime": 367.0628,
+      "eval_samples_per_second": 33.283,
+      "eval_steps_per_second": 4.163,
+      "step": 34616
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.998151143979663e-06,
+      "loss": 0.7069,
+      "step": 34620
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.993529003928819e-06,
+      "loss": 0.8001,
+      "step": 34630
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.988906863877976e-06,
+      "loss": 0.8772,
+      "step": 34640
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.984284723827132e-06,
+      "loss": 0.8171,
+      "step": 34650
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9796625837762884e-06,
+      "loss": 0.7894,
+      "step": 34660
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.975040443725445e-06,
+      "loss": 0.9996,
+      "step": 34670
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.970418303674602e-06,
+      "loss": 0.956,
+      "step": 34680
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9657961636237576e-06,
+      "loss": 0.8379,
+      "step": 34690
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.961174023572914e-06,
+      "loss": 0.7121,
+      "step": 34700
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.956551883522071e-06,
+      "loss": 0.8737,
+      "step": 34710
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9519297434712275e-06,
+      "loss": 0.7063,
+      "step": 34720
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.947307603420384e-06,
+      "loss": 0.9014,
+      "step": 34730
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.942685463369541e-06,
+      "loss": 0.6742,
+      "step": 34740
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9380633233186975e-06,
+      "loss": 0.7062,
+      "step": 34750
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.933441183267853e-06,
+      "loss": 0.8461,
+      "step": 34760
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.92881904321701e-06,
+      "loss": 0.8971,
+      "step": 34770
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.924196903166167e-06,
+      "loss": 0.8831,
+      "step": 34780
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9195747631153224e-06,
+      "loss": 0.9111,
+      "step": 34790
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.914952623064479e-06,
+      "loss": 0.9948,
+      "step": 34800
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.910330483013636e-06,
+      "loss": 0.9234,
+      "step": 34810
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.905708342962792e-06,
+      "loss": 0.853,
+      "step": 34820
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.901086202911948e-06,
+      "loss": 1.0511,
+      "step": 34830
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.896464062861105e-06,
+      "loss": 0.901,
+      "step": 34840
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8918419228102615e-06,
+      "loss": 1.006,
+      "step": 34850
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.887219782759417e-06,
+      "loss": 0.7138,
+      "step": 34860
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.882597642708574e-06,
+      "loss": 1.0092,
+      "step": 34870
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.877975502657731e-06,
+      "loss": 0.8362,
+      "step": 34880
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.873353362606887e-06,
+      "loss": 1.0254,
+      "step": 34890
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.868731222556044e-06,
+      "loss": 0.8378,
+      "step": 34900
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.864109082505201e-06,
+      "loss": 1.035,
+      "step": 34910
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8594869424543564e-06,
+      "loss": 0.8403,
+      "step": 34920
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.854864802403513e-06,
+      "loss": 0.5308,
+      "step": 34930
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.85024266235267e-06,
+      "loss": 0.7483,
+      "step": 34940
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.845620522301826e-06,
+      "loss": 0.8705,
+      "step": 34950
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.840998382250983e-06,
+      "loss": 0.7976,
+      "step": 34960
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.836376242200139e-06,
+      "loss": 0.8729,
+      "step": 34970
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8317541021492955e-06,
+      "loss": 0.528,
+      "step": 34980
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.827131962098452e-06,
+      "loss": 0.817,
+      "step": 34990
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.822509822047608e-06,
+      "loss": 0.8395,
+      "step": 35000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.817887681996765e-06,
+      "loss": 0.6564,
+      "step": 35010
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8132655419459213e-06,
+      "loss": 0.8115,
+      "step": 35020
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.808643401895078e-06,
+      "loss": 0.7753,
+      "step": 35030
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.804021261844234e-06,
+      "loss": 0.9006,
+      "step": 35040
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.799399121793391e-06,
+      "loss": 0.8409,
+      "step": 35050
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.794776981742547e-06,
+      "loss": 0.8017,
+      "step": 35060
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7901548416917033e-06,
+      "loss": 0.756,
+      "step": 35070
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.78553270164086e-06,
+      "loss": 0.7219,
+      "step": 35080
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7809105615900166e-06,
+      "loss": 0.8108,
+      "step": 35090
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7762884215391733e-06,
+      "loss": 0.8035,
+      "step": 35100
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.771666281488329e-06,
+      "loss": 0.8371,
+      "step": 35110
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7670441414374858e-06,
+      "loss": 0.8685,
+      "step": 35120
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7624220013866424e-06,
+      "loss": 0.9147,
+      "step": 35130
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7577998613357987e-06,
+      "loss": 0.7649,
+      "step": 35140
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7531777212849553e-06,
+      "loss": 0.9778,
+      "step": 35150
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.748555581234112e-06,
+      "loss": 0.8469,
+      "step": 35160
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.743933441183268e-06,
+      "loss": 0.8757,
+      "step": 35170
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7393113011324244e-06,
+      "loss": 0.9069,
+      "step": 35180
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.734689161081581e-06,
+      "loss": 1.0769,
+      "step": 35190
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7300670210307378e-06,
+      "loss": 0.6899,
+      "step": 35200
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7254448809798936e-06,
+      "loss": 0.972,
+      "step": 35210
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7208227409290502e-06,
+      "loss": 0.9369,
+      "step": 35220
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.716200600878207e-06,
+      "loss": 0.7243,
+      "step": 35230
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7115784608273635e-06,
+      "loss": 0.7076,
+      "step": 35240
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7069563207765198e-06,
+      "loss": 0.6653,
+      "step": 35250
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7023341807256764e-06,
+      "loss": 0.6735,
+      "step": 35260
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.697712040674833e-06,
+      "loss": 0.6894,
+      "step": 35270
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.693089900623989e-06,
+      "loss": 0.9681,
+      "step": 35280
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6884677605731456e-06,
+      "loss": 0.7619,
+      "step": 35290
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.683845620522302e-06,
+      "loss": 0.8986,
+      "step": 35300
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.679223480471459e-06,
+      "loss": 0.8525,
+      "step": 35310
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.674601340420615e-06,
+      "loss": 0.7147,
+      "step": 35320
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6699792003697713e-06,
+      "loss": 0.8369,
+      "step": 35330
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.665357060318928e-06,
+      "loss": 0.7731,
+      "step": 35340
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6607349202680842e-06,
+      "loss": 0.7981,
+      "step": 35350
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.656112780217241e-06,
+      "loss": 0.8217,
+      "step": 35360
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6514906401663975e-06,
+      "loss": 0.7144,
+      "step": 35370
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.646868500115554e-06,
+      "loss": 0.5527,
+      "step": 35380
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.64224636006471e-06,
+      "loss": 0.5499,
+      "step": 35390
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6376242200138667e-06,
+      "loss": 0.7583,
+      "step": 35400
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6330020799630233e-06,
+      "loss": 0.9539,
+      "step": 35410
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6283799399121796e-06,
+      "loss": 0.9497,
+      "step": 35420
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.623757799861336e-06,
+      "loss": 0.8312,
+      "step": 35430
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6191356598104924e-06,
+      "loss": 0.7448,
+      "step": 35440
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.614513519759649e-06,
+      "loss": 0.6801,
+      "step": 35450
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6098913797088053e-06,
+      "loss": 0.9351,
+      "step": 35460
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.605269239657962e-06,
+      "loss": 0.725,
+      "step": 35470
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6006470996071187e-06,
+      "loss": 0.7147,
+      "step": 35480
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.5960249595562745e-06,
+      "loss": 0.7567,
+      "step": 35490
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.591402819505431e-06,
+      "loss": 0.7854,
+      "step": 35500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.5867806794545878e-06,
+      "loss": 0.7827,
+      "step": 35510
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.5821585394037444e-06,
+      "loss": 0.7397,
+      "step": 35520
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5775363993529007e-06,
+      "loss": 0.8328,
+      "step": 35530
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5729142593020573e-06,
+      "loss": 0.9609,
+      "step": 35540
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.568292119251214e-06,
+      "loss": 0.703,
+      "step": 35550
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.56366997920037e-06,
+      "loss": 0.7651,
+      "step": 35560
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5590478391495264e-06,
+      "loss": 0.7464,
+      "step": 35570
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.554425699098683e-06,
+      "loss": 0.8806,
+      "step": 35580
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5498035590478398e-06,
+      "loss": 0.7511,
+      "step": 35590
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5451814189969956e-06,
+      "loss": 0.8027,
+      "step": 35600
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5405592789461522e-06,
+      "loss": 1.0548,
+      "step": 35610
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.535937138895309e-06,
+      "loss": 0.7947,
+      "step": 35620
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.531314998844465e-06,
+      "loss": 0.8239,
+      "step": 35630
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5266928587936218e-06,
+      "loss": 0.7181,
+      "step": 35640
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5220707187427784e-06,
+      "loss": 0.8644,
+      "step": 35650
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.517448578691935e-06,
+      "loss": 0.8889,
+      "step": 35660
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.512826438641091e-06,
+      "loss": 0.8374,
+      "step": 35670
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5082042985902476e-06,
+      "loss": 1.1417,
+      "step": 35680
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5035821585394042e-06,
+      "loss": 0.8656,
+      "step": 35690
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4989600184885604e-06,
+      "loss": 0.7859,
+      "step": 35700
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4943378784377167e-06,
+      "loss": 0.8274,
+      "step": 35710
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4897157383868733e-06,
+      "loss": 0.9646,
+      "step": 35720
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.48509359833603e-06,
+      "loss": 0.7969,
+      "step": 35730
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4804714582851862e-06,
+      "loss": 0.8105,
+      "step": 35740
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.475849318234343e-06,
+      "loss": 0.6962,
+      "step": 35750
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4712271781834995e-06,
+      "loss": 1.0439,
+      "step": 35760
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4666050381326554e-06,
+      "loss": 0.9131,
+      "step": 35770
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.461982898081812e-06,
+      "loss": 0.9365,
+      "step": 35780
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4573607580309687e-06,
+      "loss": 0.7393,
+      "step": 35790
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4527386179801253e-06,
+      "loss": 0.7054,
+      "step": 35800
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4481164779292816e-06,
+      "loss": 0.7896,
+      "step": 35810
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4434943378784382e-06,
+      "loss": 0.9841,
+      "step": 35820
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4388721978275945e-06,
+      "loss": 0.7137,
+      "step": 35830
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4342500577767507e-06,
+      "loss": 0.723,
+      "step": 35840
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4296279177259073e-06,
+      "loss": 0.8539,
+      "step": 35850
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.425005777675064e-06,
+      "loss": 0.94,
+      "step": 35860
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4203836376242207e-06,
+      "loss": 1.2955,
+      "step": 35870
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.4157614975733765e-06,
+      "loss": 0.9427,
+      "step": 35880
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.411139357522533e-06,
+      "loss": 0.6974,
+      "step": 35890
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.4065172174716898e-06,
+      "loss": 0.7849,
+      "step": 35900
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.401895077420846e-06,
+      "loss": 0.741,
+      "step": 35910
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.3972729373700027e-06,
+      "loss": 0.8569,
+      "step": 35920
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.3926507973191593e-06,
+      "loss": 0.9231,
+      "step": 35930
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.3880286572683156e-06,
+      "loss": 0.7245,
+      "step": 35940
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.383406517217472e-06,
+      "loss": 0.8532,
+      "step": 35950
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3787843771666285e-06,
+      "loss": 0.665,
+      "step": 35960
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.374162237115785e-06,
+      "loss": 0.7541,
+      "step": 35970
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.369540097064941e-06,
+      "loss": 0.7781,
+      "step": 35980
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3649179570140976e-06,
+      "loss": 0.7973,
+      "step": 35990
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3602958169632542e-06,
+      "loss": 0.7882,
+      "step": 36000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.355673676912411e-06,
+      "loss": 0.9376,
+      "step": 36010
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.351051536861567e-06,
+      "loss": 0.6816,
+      "step": 36020
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3464293968107238e-06,
+      "loss": 0.6851,
+      "step": 36030
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3418072567598804e-06,
+      "loss": 0.8546,
+      "step": 36040
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3371851167090363e-06,
+      "loss": 0.8972,
+      "step": 36050
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.332562976658193e-06,
+      "loss": 0.691,
+      "step": 36060
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3279408366073496e-06,
+      "loss": 0.6403,
+      "step": 36070
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3233186965565062e-06,
+      "loss": 0.9738,
+      "step": 36080
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3186965565056625e-06,
+      "loss": 0.7173,
+      "step": 36090
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3140744164548187e-06,
+      "loss": 0.8744,
+      "step": 36100
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3094522764039753e-06,
+      "loss": 0.7739,
+      "step": 36110
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3048301363531316e-06,
+      "loss": 0.6332,
+      "step": 36120
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3002079963022882e-06,
+      "loss": 0.895,
+      "step": 36130
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.295585856251445e-06,
+      "loss": 0.7222,
+      "step": 36140
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2909637162006016e-06,
+      "loss": 0.8747,
+      "step": 36150
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2863415761497574e-06,
+      "loss": 0.8193,
+      "step": 36160
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.281719436098914e-06,
+      "loss": 0.9051,
+      "step": 36170
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2770972960480707e-06,
+      "loss": 0.755,
+      "step": 36180
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.272475155997227e-06,
+      "loss": 0.7635,
+      "step": 36190
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2678530159463836e-06,
+      "loss": 0.6589,
+      "step": 36200
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.26323087589554e-06,
+      "loss": 0.6896,
+      "step": 36210
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2586087358446965e-06,
+      "loss": 0.8495,
+      "step": 36220
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2539865957938527e-06,
+      "loss": 0.9678,
+      "step": 36230
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2493644557430093e-06,
+      "loss": 0.8095,
+      "step": 36240
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.244742315692166e-06,
+      "loss": 0.756,
+      "step": 36250
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.240120175641322e-06,
+      "loss": 0.6845,
+      "step": 36260
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2354980355904785e-06,
+      "loss": 0.8443,
+      "step": 36270
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.230875895539635e-06,
+      "loss": 0.7264,
+      "step": 36280
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2262537554887918e-06,
+      "loss": 1.0056,
+      "step": 36290
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.221631615437948e-06,
+      "loss": 0.5267,
+      "step": 36300
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2170094753871047e-06,
+      "loss": 0.9955,
+      "step": 36310
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2123873353362613e-06,
+      "loss": 0.7008,
+      "step": 36320
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.207765195285417e-06,
+      "loss": 0.8374,
+      "step": 36330
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.203143055234574e-06,
+      "loss": 0.696,
+      "step": 36340
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.1985209151837305e-06,
+      "loss": 0.8003,
+      "step": 36350
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.193898775132887e-06,
+      "loss": 0.6519,
+      "step": 36360
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.189276635082043e-06,
+      "loss": 0.8804,
+      "step": 36370
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.1846544950311996e-06,
+      "loss": 0.9999,
+      "step": 36380
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.1800323549803562e-06,
+      "loss": 0.5891,
+      "step": 36390
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1754102149295125e-06,
+      "loss": 0.7401,
+      "step": 36400
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.170788074878669e-06,
+      "loss": 0.9722,
+      "step": 36410
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.166165934827826e-06,
+      "loss": 0.7862,
+      "step": 36420
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1615437947769824e-06,
+      "loss": 0.8041,
+      "step": 36430
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1569216547261383e-06,
+      "loss": 0.9561,
+      "step": 36440
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.152299514675295e-06,
+      "loss": 1.0503,
+      "step": 36450
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1476773746244516e-06,
+      "loss": 0.8244,
+      "step": 36460
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.143055234573608e-06,
+      "loss": 0.9081,
+      "step": 36470
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.138433094522764e-06,
+      "loss": 0.9342,
+      "step": 36480
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1338109544719207e-06,
+      "loss": 0.7321,
+      "step": 36490
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1291888144210774e-06,
+      "loss": 0.8728,
+      "step": 36500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1245666743702336e-06,
+      "loss": 0.8298,
+      "step": 36510
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1199445343193902e-06,
+      "loss": 0.7774,
+      "step": 36520
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.115322394268547e-06,
+      "loss": 0.8077,
+      "step": 36530
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1107002542177027e-06,
+      "loss": 0.7509,
+      "step": 36540
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1060781141668594e-06,
+      "loss": 0.8812,
+      "step": 36550
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.101455974116016e-06,
+      "loss": 0.8235,
+      "step": 36560
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0968338340651727e-06,
+      "loss": 0.7962,
+      "step": 36570
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.092211694014329e-06,
+      "loss": 0.7675,
+      "step": 36580
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0875895539634856e-06,
+      "loss": 0.7777,
+      "step": 36590
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.082967413912642e-06,
+      "loss": 0.7933,
+      "step": 36600
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.078345273861798e-06,
+      "loss": 0.9538,
+      "step": 36610
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0737231338109547e-06,
+      "loss": 0.9384,
+      "step": 36620
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0691009937601114e-06,
+      "loss": 0.7537,
+      "step": 36630
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.064478853709268e-06,
+      "loss": 0.7733,
+      "step": 36640
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.059856713658424e-06,
+      "loss": 0.8655,
+      "step": 36650
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0552345736075805e-06,
+      "loss": 0.5404,
+      "step": 36660
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.050612433556737e-06,
+      "loss": 0.7121,
+      "step": 36670
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0459902935058934e-06,
+      "loss": 0.749,
+      "step": 36680
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.04136815345505e-06,
+      "loss": 0.8609,
+      "step": 36690
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0367460134042067e-06,
+      "loss": 1.0,
+      "step": 36700
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.032123873353363e-06,
+      "loss": 0.7724,
+      "step": 36710
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.027501733302519e-06,
+      "loss": 0.5046,
+      "step": 36720
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.022879593251676e-06,
+      "loss": 0.8254,
+      "step": 36730
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.0182574532008325e-06,
+      "loss": 0.8006,
+      "step": 36740
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.0136353131499883e-06,
+      "loss": 0.8566,
+      "step": 36750
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.009013173099145e-06,
+      "loss": 0.8376,
+      "step": 36760
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.0043910330483016e-06,
+      "loss": 0.8486,
+      "step": 36770
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9997688929974582e-06,
+      "loss": 0.8148,
+      "step": 36780
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9951467529466145e-06,
+      "loss": 0.7262,
+      "step": 36790
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.990524612895771e-06,
+      "loss": 0.6713,
+      "step": 36800
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.985902472844928e-06,
+      "loss": 0.9347,
+      "step": 36810
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9812803327940836e-06,
+      "loss": 0.695,
+      "step": 36820
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9766581927432403e-06,
+      "loss": 0.8646,
+      "step": 36830
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.972036052692397e-06,
+      "loss": 0.7767,
+      "step": 36840
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9674139126415536e-06,
+      "loss": 1.033,
+      "step": 36850
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.96279177259071e-06,
+      "loss": 0.8918,
+      "step": 36860
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.958169632539866e-06,
+      "loss": 0.6394,
+      "step": 36870
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9535474924890227e-06,
+      "loss": 0.7291,
+      "step": 36880
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.948925352438179e-06,
+      "loss": 1.011,
+      "step": 36890
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9443032123873356e-06,
+      "loss": 0.879,
+      "step": 36900
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9396810723364922e-06,
+      "loss": 0.8506,
+      "step": 36910
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.935058932285649e-06,
+      "loss": 0.8317,
+      "step": 36920
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9304367922348047e-06,
+      "loss": 0.8928,
+      "step": 36930
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9258146521839614e-06,
+      "loss": 0.822,
+      "step": 36940
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.921192512133118e-06,
+      "loss": 0.7867,
+      "step": 36950
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9165703720822743e-06,
+      "loss": 0.7249,
+      "step": 36960
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.911948232031431e-06,
+      "loss": 0.9639,
+      "step": 36970
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.907326091980587e-06,
+      "loss": 0.6177,
+      "step": 36980
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.902703951929744e-06,
+      "loss": 1.04,
+      "step": 36990
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8980818118789e-06,
+      "loss": 0.8999,
+      "step": 37000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8934596718280567e-06,
+      "loss": 0.7201,
+      "step": 37010
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8888375317772134e-06,
+      "loss": 0.8562,
+      "step": 37020
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.884215391726369e-06,
+      "loss": 0.7591,
+      "step": 37030
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.879593251675526e-06,
+      "loss": 0.6808,
+      "step": 37040
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8749711116246825e-06,
+      "loss": 0.7226,
+      "step": 37050
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.870348971573839e-06,
+      "loss": 0.7607,
+      "step": 37060
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8657268315229954e-06,
+      "loss": 0.6859,
+      "step": 37070
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.861104691472152e-06,
+      "loss": 0.807,
+      "step": 37080
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8564825514213087e-06,
+      "loss": 0.541,
+      "step": 37090
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8518604113704645e-06,
+      "loss": 0.9842,
+      "step": 37100
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.847238271319621e-06,
+      "loss": 0.6039,
+      "step": 37110
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.842616131268778e-06,
+      "loss": 0.8673,
+      "step": 37120
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8379939912179345e-06,
+      "loss": 0.7599,
+      "step": 37130
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8333718511670903e-06,
+      "loss": 0.6206,
+      "step": 37140
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.828749711116247e-06,
+      "loss": 0.7236,
+      "step": 37150
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8241275710654036e-06,
+      "loss": 0.6601,
+      "step": 37160
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.81950543101456e-06,
+      "loss": 0.7511,
+      "step": 37170
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8148832909637165e-06,
+      "loss": 0.7755,
+      "step": 37180
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.810261150912873e-06,
+      "loss": 0.6762,
+      "step": 37190
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.80563901086203e-06,
+      "loss": 0.8101,
+      "step": 37200
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8010168708111856e-06,
+      "loss": 0.9237,
+      "step": 37210
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.7963947307603423e-06,
+      "loss": 0.9949,
+      "step": 37220
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.791772590709499e-06,
+      "loss": 0.8406,
+      "step": 37230
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.787150450658655e-06,
+      "loss": 0.7871,
+      "step": 37240
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.7825283106078114e-06,
+      "loss": 0.8524,
+      "step": 37250
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.777906170556968e-06,
+      "loss": 0.9479,
+      "step": 37260
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7732840305061247e-06,
+      "loss": 0.6727,
+      "step": 37270
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.768661890455281e-06,
+      "loss": 0.7475,
+      "step": 37280
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7640397504044376e-06,
+      "loss": 0.8576,
+      "step": 37290
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7594176103535943e-06,
+      "loss": 0.7843,
+      "step": 37300
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.75479547030275e-06,
+      "loss": 0.7232,
+      "step": 37310
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7501733302519067e-06,
+      "loss": 0.7849,
+      "step": 37320
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7455511902010634e-06,
+      "loss": 0.8123,
+      "step": 37330
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.74092905015022e-06,
+      "loss": 0.8127,
+      "step": 37340
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7363069100993763e-06,
+      "loss": 0.8041,
+      "step": 37350
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.731684770048533e-06,
+      "loss": 0.8431,
+      "step": 37360
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.727062629997689e-06,
+      "loss": 0.804,
+      "step": 37370
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7224404899468454e-06,
+      "loss": 0.7621,
+      "step": 37380
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.717818349896002e-06,
+      "loss": 0.7351,
+      "step": 37390
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7131962098451587e-06,
+      "loss": 0.7024,
+      "step": 37400
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7085740697943154e-06,
+      "loss": 0.5587,
+      "step": 37410
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.703951929743471e-06,
+      "loss": 0.9377,
+      "step": 37420
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.699329789692628e-06,
+      "loss": 1.1282,
+      "step": 37430
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6947076496417845e-06,
+      "loss": 0.8122,
+      "step": 37440
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6900855095909407e-06,
+      "loss": 1.0603,
+      "step": 37450
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6854633695400974e-06,
+      "loss": 0.7829,
+      "step": 37460
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.680841229489254e-06,
+      "loss": 0.9305,
+      "step": 37470
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6762190894384103e-06,
+      "loss": 0.9139,
+      "step": 37480
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6715969493875665e-06,
+      "loss": 0.6857,
+      "step": 37490
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.666974809336723e-06,
+      "loss": 0.7759,
+      "step": 37500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.66235266928588e-06,
+      "loss": 0.8132,
+      "step": 37510
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6577305292350356e-06,
+      "loss": 0.856,
+      "step": 37520
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6531083891841923e-06,
+      "loss": 0.5726,
+      "step": 37530
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.648486249133349e-06,
+      "loss": 0.7664,
+      "step": 37540
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6438641090825056e-06,
+      "loss": 0.9584,
+      "step": 37550
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.639241969031662e-06,
+      "loss": 0.6453,
+      "step": 37560
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6346198289808185e-06,
+      "loss": 1.0094,
+      "step": 37570
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.629997688929975e-06,
+      "loss": 0.6951,
+      "step": 37580
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.625375548879131e-06,
+      "loss": 0.731,
+      "step": 37590
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6207534088282876e-06,
+      "loss": 0.779,
+      "step": 37600
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6161312687774443e-06,
+      "loss": 0.8964,
+      "step": 37610
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.611509128726601e-06,
+      "loss": 0.8582,
+      "step": 37620
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.606886988675757e-06,
+      "loss": 0.6743,
+      "step": 37630
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6022648486249134e-06,
+      "loss": 0.704,
+      "step": 37640
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.59764270857407e-06,
+      "loss": 0.8917,
+      "step": 37650
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.5930205685232263e-06,
+      "loss": 0.8824,
+      "step": 37660
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.588398428472383e-06,
+      "loss": 0.7287,
+      "step": 37670
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.5837762884215396e-06,
+      "loss": 0.5654,
+      "step": 37680
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5791541483706963e-06,
+      "loss": 0.9316,
+      "step": 37690
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.574532008319852e-06,
+      "loss": 0.6635,
+      "step": 37700
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5699098682690087e-06,
+      "loss": 1.0371,
+      "step": 37710
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5652877282181654e-06,
+      "loss": 0.8448,
+      "step": 37720
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5606655881673216e-06,
+      "loss": 1.0049,
+      "step": 37730
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5560434481164783e-06,
+      "loss": 0.7377,
+      "step": 37740
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5514213080656345e-06,
+      "loss": 0.8419,
+      "step": 37750
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.546799168014791e-06,
+      "loss": 0.7814,
+      "step": 37760
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5421770279639474e-06,
+      "loss": 0.7054,
+      "step": 37770
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.537554887913104e-06,
+      "loss": 0.9375,
+      "step": 37780
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5329327478622607e-06,
+      "loss": 0.9926,
+      "step": 37790
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5283106078114165e-06,
+      "loss": 0.7379,
+      "step": 37800
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.523688467760573e-06,
+      "loss": 0.8499,
+      "step": 37810
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.51906632770973e-06,
+      "loss": 0.8487,
+      "step": 37820
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5144441876588865e-06,
+      "loss": 0.8133,
+      "step": 37830
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5098220476080427e-06,
+      "loss": 0.8657,
+      "step": 37840
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5051999075571994e-06,
+      "loss": 0.8853,
+      "step": 37850
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.500577767506356e-06,
+      "loss": 0.7408,
+      "step": 37860
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4959556274555123e-06,
+      "loss": 0.7534,
+      "step": 37870
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4913334874046685e-06,
+      "loss": 0.6876,
+      "step": 37880
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.486711347353825e-06,
+      "loss": 0.7416,
+      "step": 37890
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4820892073029814e-06,
+      "loss": 0.8899,
+      "step": 37900
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4774670672521376e-06,
+      "loss": 0.8597,
+      "step": 37910
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4728449272012943e-06,
+      "loss": 0.8152,
+      "step": 37920
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.468222787150451e-06,
+      "loss": 0.8279,
+      "step": 37930
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.463600647099607e-06,
+      "loss": 0.7891,
+      "step": 37940
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.458978507048764e-06,
+      "loss": 0.7798,
+      "step": 37950
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4543563669979205e-06,
+      "loss": 0.8365,
+      "step": 37960
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4497342269470767e-06,
+      "loss": 0.8375,
+      "step": 37970
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.445112086896233e-06,
+      "loss": 0.6968,
+      "step": 37980
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4404899468453896e-06,
+      "loss": 0.6553,
+      "step": 37990
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.435867806794546e-06,
+      "loss": 0.6821,
+      "step": 38000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4312456667437025e-06,
+      "loss": 0.7226,
+      "step": 38010
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4266235266928587e-06,
+      "loss": 0.9781,
+      "step": 38020
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4220013866420154e-06,
+      "loss": 0.8679,
+      "step": 38030
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.417379246591172e-06,
+      "loss": 0.8135,
+      "step": 38040
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.4127571065403283e-06,
+      "loss": 0.8498,
+      "step": 38050
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.408134966489485e-06,
+      "loss": 0.7852,
+      "step": 38060
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.403512826438641e-06,
+      "loss": 0.8999,
+      "step": 38070
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.398890686387798e-06,
+      "loss": 0.9457,
+      "step": 38080
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.394268546336954e-06,
+      "loss": 0.7311,
+      "step": 38090
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.3896464062861107e-06,
+      "loss": 1.0359,
+      "step": 38100
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.385024266235267e-06,
+      "loss": 0.5918,
+      "step": 38110
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.3804021261844236e-06,
+      "loss": 0.9023,
+      "step": 38120
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3757799861335803e-06,
+      "loss": 0.7401,
+      "step": 38130
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3711578460827365e-06,
+      "loss": 0.7755,
+      "step": 38140
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.366535706031893e-06,
+      "loss": 1.0756,
+      "step": 38150
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3619135659810494e-06,
+      "loss": 1.111,
+      "step": 38160
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.357291425930206e-06,
+      "loss": 0.8878,
+      "step": 38170
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3526692858793623e-06,
+      "loss": 1.015,
+      "step": 38180
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3480471458285185e-06,
+      "loss": 0.7015,
+      "step": 38190
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.343425005777675e-06,
+      "loss": 0.8411,
+      "step": 38200
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3388028657268314e-06,
+      "loss": 0.9048,
+      "step": 38210
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.334180725675988e-06,
+      "loss": 0.8259,
+      "step": 38220
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3295585856251447e-06,
+      "loss": 0.7908,
+      "step": 38230
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3249364455743014e-06,
+      "loss": 0.9188,
+      "step": 38240
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3203143055234576e-06,
+      "loss": 0.9833,
+      "step": 38250
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.315692165472614e-06,
+      "loss": 0.766,
+      "step": 38260
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3110700254217705e-06,
+      "loss": 0.7103,
+      "step": 38270
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3064478853709268e-06,
+      "loss": 0.8448,
+      "step": 38280
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3018257453200834e-06,
+      "loss": 0.766,
+      "step": 38290
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2972036052692396e-06,
+      "loss": 0.7764,
+      "step": 38300
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2925814652183963e-06,
+      "loss": 0.8038,
+      "step": 38310
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.287959325167553e-06,
+      "loss": 0.7558,
+      "step": 38320
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.283337185116709e-06,
+      "loss": 0.7038,
+      "step": 38330
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.278715045065866e-06,
+      "loss": 0.8888,
+      "step": 38340
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.274092905015022e-06,
+      "loss": 0.7891,
+      "step": 38350
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2694707649641787e-06,
+      "loss": 0.8708,
+      "step": 38360
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.264848624913335e-06,
+      "loss": 0.9766,
+      "step": 38370
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2602264848624916e-06,
+      "loss": 0.7565,
+      "step": 38380
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.255604344811648e-06,
+      "loss": 0.9294,
+      "step": 38390
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2509822047608045e-06,
+      "loss": 0.7993,
+      "step": 38400
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2463600647099608e-06,
+      "loss": 0.6835,
+      "step": 38410
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2417379246591174e-06,
+      "loss": 1.0095,
+      "step": 38420
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.237115784608274e-06,
+      "loss": 0.8261,
+      "step": 38430
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2324936445574303e-06,
+      "loss": 0.8136,
+      "step": 38440
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.227871504506587e-06,
+      "loss": 0.9287,
+      "step": 38450
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.223249364455743e-06,
+      "loss": 0.6646,
+      "step": 38460
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.2186272244048994e-06,
+      "loss": 0.8341,
+      "step": 38470
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.214005084354056e-06,
+      "loss": 0.8221,
+      "step": 38480
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.2093829443032123e-06,
+      "loss": 0.8717,
+      "step": 38490
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.204760804252369e-06,
+      "loss": 0.7171,
+      "step": 38500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.2001386642015256e-06,
+      "loss": 0.8499,
+      "step": 38510
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.195516524150682e-06,
+      "loss": 0.6939,
+      "step": 38520
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.1908943840998385e-06,
+      "loss": 0.9046,
+      "step": 38530
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.1862722440489948e-06,
+      "loss": 0.8706,
+      "step": 38540
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.1816501039981514e-06,
+      "loss": 0.7388,
+      "step": 38550
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1770279639473076e-06,
+      "loss": 0.8387,
+      "step": 38560
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1724058238964643e-06,
+      "loss": 0.9778,
+      "step": 38570
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1677836838456205e-06,
+      "loss": 0.7673,
+      "step": 38580
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.163161543794777e-06,
+      "loss": 0.8174,
+      "step": 38590
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1585394037439334e-06,
+      "loss": 0.9256,
+      "step": 38600
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.15391726369309e-06,
+      "loss": 0.5892,
+      "step": 38610
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1492951236422467e-06,
+      "loss": 0.6775,
+      "step": 38620
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.144672983591403e-06,
+      "loss": 0.9301,
+      "step": 38630
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1400508435405596e-06,
+      "loss": 0.6767,
+      "step": 38640
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.135428703489716e-06,
+      "loss": 0.8614,
+      "step": 38650
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1308065634388725e-06,
+      "loss": 0.7664,
+      "step": 38660
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1261844233880288e-06,
+      "loss": 0.7859,
+      "step": 38670
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.121562283337185e-06,
+      "loss": 1.1051,
+      "step": 38680
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1169401432863417e-06,
+      "loss": 0.6886,
+      "step": 38690
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1123180032354983e-06,
+      "loss": 0.73,
+      "step": 38700
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1076958631846545e-06,
+      "loss": 0.5472,
+      "step": 38710
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.103073723133811e-06,
+      "loss": 0.6179,
+      "step": 38720
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.098451583082968e-06,
+      "loss": 0.9279,
+      "step": 38730
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.093829443032124e-06,
+      "loss": 0.8562,
+      "step": 38740
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0892073029812803e-06,
+      "loss": 0.7964,
+      "step": 38750
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.084585162930437e-06,
+      "loss": 1.0026,
+      "step": 38760
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0799630228795932e-06,
+      "loss": 0.8679,
+      "step": 38770
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.07534088282875e-06,
+      "loss": 0.8446,
+      "step": 38780
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.070718742777906e-06,
+      "loss": 0.9137,
+      "step": 38790
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0660966027270628e-06,
+      "loss": 0.7973,
+      "step": 38800
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0614744626762194e-06,
+      "loss": 0.8514,
+      "step": 38810
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0568523226253757e-06,
+      "loss": 0.8002,
+      "step": 38820
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0522301825745323e-06,
+      "loss": 0.8189,
+      "step": 38830
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0476080425236885e-06,
+      "loss": 0.702,
+      "step": 38840
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.042985902472845e-06,
+      "loss": 0.7931,
+      "step": 38850
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0383637624220014e-06,
+      "loss": 0.8525,
+      "step": 38860
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.033741622371158e-06,
+      "loss": 0.7489,
+      "step": 38870
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0291194823203143e-06,
+      "loss": 0.7543,
+      "step": 38880
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.024497342269471e-06,
+      "loss": 0.7692,
+      "step": 38890
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.0198752022186276e-06,
+      "loss": 0.838,
+      "step": 38900
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.015253062167784e-06,
+      "loss": 0.7529,
+      "step": 38910
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.0106309221169405e-06,
+      "loss": 0.7866,
+      "step": 38920
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.0060087820660968e-06,
+      "loss": 0.853,
+      "step": 38930
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.0013866420152534e-06,
+      "loss": 0.8385,
+      "step": 38940
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.9967645019644097e-06,
+      "loss": 1.0043,
+      "step": 38950
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.992142361913566e-06,
+      "loss": 0.7267,
+      "step": 38960
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.9875202218627225e-06,
+      "loss": 0.6928,
+      "step": 38970
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.9828980818118788e-06,
+      "loss": 1.0287,
+      "step": 38980
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9782759417610354e-06,
+      "loss": 0.8203,
+      "step": 38990
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.973653801710192e-06,
+      "loss": 0.9839,
+      "step": 39000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9690316616593488e-06,
+      "loss": 0.962,
+      "step": 39010
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.964409521608505e-06,
+      "loss": 0.827,
+      "step": 39020
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9597873815576612e-06,
+      "loss": 0.8636,
+      "step": 39030
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.955165241506818e-06,
+      "loss": 0.728,
+      "step": 39040
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.950543101455974e-06,
+      "loss": 0.9913,
+      "step": 39050
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9459209614051308e-06,
+      "loss": 0.9632,
+      "step": 39060
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.941298821354287e-06,
+      "loss": 0.6621,
+      "step": 39070
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9366766813034437e-06,
+      "loss": 0.605,
+      "step": 39080
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9320545412526003e-06,
+      "loss": 1.1318,
+      "step": 39090
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9274324012017565e-06,
+      "loss": 0.7884,
+      "step": 39100
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.922810261150913e-06,
+      "loss": 0.8642,
+      "step": 39110
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9181881211000694e-06,
+      "loss": 0.6713,
+      "step": 39120
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.913565981049226e-06,
+      "loss": 0.8832,
+      "step": 39130
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9089438409983823e-06,
+      "loss": 0.7471,
+      "step": 39140
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.904321700947539e-06,
+      "loss": 0.8892,
+      "step": 39150
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8996995608966954e-06,
+      "loss": 0.8031,
+      "step": 39160
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8950774208458517e-06,
+      "loss": 0.9307,
+      "step": 39170
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8904552807950083e-06,
+      "loss": 0.7306,
+      "step": 39180
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8858331407441646e-06,
+      "loss": 0.9192,
+      "step": 39190
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8812110006933212e-06,
+      "loss": 0.841,
+      "step": 39200
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8765888606424777e-06,
+      "loss": 0.6903,
+      "step": 39210
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.871966720591634e-06,
+      "loss": 0.7733,
+      "step": 39220
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8673445805407905e-06,
+      "loss": 0.9847,
+      "step": 39230
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8627224404899468e-06,
+      "loss": 0.9649,
+      "step": 39240
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8581003004391034e-06,
+      "loss": 0.7399,
+      "step": 39250
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8534781603882599e-06,
+      "loss": 0.7935,
+      "step": 39260
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8488560203374165e-06,
+      "loss": 0.927,
+      "step": 39270
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8442338802865728e-06,
+      "loss": 1.0558,
+      "step": 39280
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8396117402357294e-06,
+      "loss": 0.9971,
+      "step": 39290
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8349896001848857e-06,
+      "loss": 0.6513,
+      "step": 39300
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8303674601340421e-06,
+      "loss": 1.125,
+      "step": 39310
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8257453200831988e-06,
+      "loss": 0.7183,
+      "step": 39320
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.821123180032355e-06,
+      "loss": 0.7917,
+      "step": 39330
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.8165010399815117e-06,
+      "loss": 1.0274,
+      "step": 39340
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.811878899930668e-06,
+      "loss": 0.821,
+      "step": 39350
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.8072567598798246e-06,
+      "loss": 0.8154,
+      "step": 39360
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.802634619828981e-06,
+      "loss": 0.681,
+      "step": 39370
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7980124797781372e-06,
+      "loss": 0.7506,
+      "step": 39380
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7933903397272939e-06,
+      "loss": 0.6765,
+      "step": 39390
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7887681996764503e-06,
+      "loss": 0.8094,
+      "step": 39400
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.784146059625607e-06,
+      "loss": 0.8857,
+      "step": 39410
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7795239195747632e-06,
+      "loss": 0.7577,
+      "step": 39420
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7749017795239199e-06,
+      "loss": 0.865,
+      "step": 39430
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7702796394730761e-06,
+      "loss": 0.9283,
+      "step": 39440
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7656574994222326e-06,
+      "loss": 0.8391,
+      "step": 39450
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7610353593713892e-06,
+      "loss": 0.8601,
+      "step": 39460
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7564132193205455e-06,
+      "loss": 0.7577,
+      "step": 39470
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7517910792697021e-06,
+      "loss": 0.6711,
+      "step": 39480
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7471689392188583e-06,
+      "loss": 0.6647,
+      "step": 39490
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.742546799168015e-06,
+      "loss": 0.7228,
+      "step": 39500
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7379246591171714e-06,
+      "loss": 0.6484,
+      "step": 39510
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7333025190663277e-06,
+      "loss": 0.868,
+      "step": 39520
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7286803790154843e-06,
+      "loss": 1.0486,
+      "step": 39530
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7240582389646408e-06,
+      "loss": 0.6293,
+      "step": 39540
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7194360989137972e-06,
+      "loss": 0.6916,
+      "step": 39550
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7148139588629537e-06,
+      "loss": 0.8302,
+      "step": 39560
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7101918188121103e-06,
+      "loss": 0.683,
+      "step": 39570
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7055696787612666e-06,
+      "loss": 0.7885,
+      "step": 39580
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.700947538710423e-06,
+      "loss": 0.7836,
+      "step": 39590
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6963253986595797e-06,
+      "loss": 0.9216,
+      "step": 39600
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.691703258608736e-06,
+      "loss": 0.7465,
+      "step": 39610
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6870811185578926e-06,
+      "loss": 0.8102,
+      "step": 39620
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6824589785070488e-06,
+      "loss": 0.8372,
+      "step": 39630
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6778368384562054e-06,
+      "loss": 0.797,
+      "step": 39640
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6732146984053619e-06,
+      "loss": 0.9152,
+      "step": 39650
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6685925583545181e-06,
+      "loss": 0.8541,
+      "step": 39660
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6639704183036748e-06,
+      "loss": 0.6513,
+      "step": 39670
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6593482782528312e-06,
+      "loss": 0.7553,
+      "step": 39680
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6547261382019877e-06,
+      "loss": 1.128,
+      "step": 39690
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6501039981511441e-06,
+      "loss": 0.7128,
+      "step": 39700
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6454818581003008e-06,
+      "loss": 0.8618,
+      "step": 39710
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.640859718049457e-06,
+      "loss": 0.8839,
+      "step": 39720
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6362375779986135e-06,
+      "loss": 0.9761,
+      "step": 39730
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.63161543794777e-06,
+      "loss": 0.6661,
+      "step": 39740
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6269932978969263e-06,
+      "loss": 1.0321,
+      "step": 39750
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.622371157846083e-06,
+      "loss": 0.8469,
+      "step": 39760
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6177490177952392e-06,
+      "loss": 1.1144,
+      "step": 39770
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6131268777443959e-06,
+      "loss": 0.9211,
+      "step": 39780
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6085047376935523e-06,
+      "loss": 0.7554,
+      "step": 39790
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6038825976427086e-06,
+      "loss": 0.8313,
+      "step": 39800
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5992604575918652e-06,
+      "loss": 0.8605,
+      "step": 39810
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5946383175410215e-06,
+      "loss": 1.0696,
+      "step": 39820
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5900161774901781e-06,
+      "loss": 0.7738,
+      "step": 39830
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5853940374393346e-06,
+      "loss": 0.8035,
+      "step": 39840
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5807718973884912e-06,
+      "loss": 0.7949,
+      "step": 39850
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5761497573376475e-06,
+      "loss": 0.6713,
+      "step": 39860
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.571527617286804e-06,
+      "loss": 0.8782,
+      "step": 39870
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5669054772359603e-06,
+      "loss": 0.8094,
+      "step": 39880
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5622833371851168e-06,
+      "loss": 1.0555,
+      "step": 39890
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5576611971342735e-06,
+      "loss": 0.6384,
+      "step": 39900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5530390570834297e-06,
+      "loss": 0.8541,
+      "step": 39910
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5484169170325863e-06,
+      "loss": 0.8051,
+      "step": 39920
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5437947769817428e-06,
+      "loss": 1.0088,
+      "step": 39930
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.539172636930899e-06,
+      "loss": 0.7512,
+      "step": 39940
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5345504968800557e-06,
+      "loss": 0.683,
+      "step": 39950
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.529928356829212e-06,
+      "loss": 0.6487,
+      "step": 39960
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5253062167783686e-06,
+      "loss": 0.9845,
+      "step": 39970
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.520684076727525e-06,
+      "loss": 0.8149,
+      "step": 39980
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5160619366766815e-06,
+      "loss": 0.646,
+      "step": 39990
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.511439796625838e-06,
+      "loss": 1.0718,
+      "step": 40000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5068176565749941e-06,
+      "loss": 0.9203,
+      "step": 40010
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5021955165241508e-06,
+      "loss": 0.8804,
+      "step": 40020
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4975733764733072e-06,
+      "loss": 1.0434,
+      "step": 40030
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.492951236422464e-06,
+      "loss": 0.8672,
+      "step": 40040
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4883290963716201e-06,
+      "loss": 0.9391,
+      "step": 40050
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4837069563207768e-06,
+      "loss": 0.7916,
+      "step": 40060
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.479084816269933e-06,
+      "loss": 0.8457,
+      "step": 40070
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4744626762190895e-06,
+      "loss": 0.765,
+      "step": 40080
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4698405361682461e-06,
+      "loss": 0.8014,
+      "step": 40090
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4652183961174024e-06,
+      "loss": 0.9513,
+      "step": 40100
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.460596256066559e-06,
+      "loss": 0.6928,
+      "step": 40110
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4559741160157155e-06,
+      "loss": 1.0236,
+      "step": 40120
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.451351975964872e-06,
+      "loss": 0.7326,
+      "step": 40130
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4467298359140284e-06,
+      "loss": 0.8985,
+      "step": 40140
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4421076958631846e-06,
+      "loss": 0.7906,
+      "step": 40150
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4374855558123412e-06,
+      "loss": 1.0059,
+      "step": 40160
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4328634157614977e-06,
+      "loss": 0.9045,
+      "step": 40170
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4282412757106543e-06,
+      "loss": 0.7234,
+      "step": 40180
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4236191356598106e-06,
+      "loss": 0.8198,
+      "step": 40190
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4189969956089672e-06,
+      "loss": 0.8259,
+      "step": 40200
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4143748555581235e-06,
+      "loss": 0.8541,
+      "step": 40210
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.40975271550728e-06,
+      "loss": 0.8396,
+      "step": 40220
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4051305754564366e-06,
+      "loss": 0.8265,
+      "step": 40230
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4005084354055928e-06,
+      "loss": 0.8588,
+      "step": 40240
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.3958862953547495e-06,
+      "loss": 0.7324,
+      "step": 40250
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.3912641553039057e-06,
+      "loss": 0.9739,
+      "step": 40260
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.3866420152530624e-06,
+      "loss": 0.6921,
+      "step": 40270
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.3820198752022188e-06,
+      "loss": 1.0302,
+      "step": 40280
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.377397735151375e-06,
+      "loss": 0.682,
+      "step": 40290
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3727755951005317e-06,
+      "loss": 0.6932,
+      "step": 40300
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3681534550496881e-06,
+      "loss": 0.6749,
+      "step": 40310
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3635313149988446e-06,
+      "loss": 1.0174,
+      "step": 40320
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.358909174948001e-06,
+      "loss": 0.9619,
+      "step": 40330
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3542870348971577e-06,
+      "loss": 0.8209,
+      "step": 40340
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.349664894846314e-06,
+      "loss": 0.9947,
+      "step": 40350
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3450427547954704e-06,
+      "loss": 0.8035,
+      "step": 40360
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.340420614744627e-06,
+      "loss": 0.7037,
+      "step": 40370
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3357984746937833e-06,
+      "loss": 0.7243,
+      "step": 40380
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.33117633464294e-06,
+      "loss": 0.7731,
+      "step": 40390
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3265541945920961e-06,
+      "loss": 0.7311,
+      "step": 40400
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3219320545412528e-06,
+      "loss": 0.8399,
+      "step": 40410
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3173099144904092e-06,
+      "loss": 0.7077,
+      "step": 40420
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3126877744395655e-06,
+      "loss": 0.809,
+      "step": 40430
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3080656343887221e-06,
+      "loss": 0.8577,
+      "step": 40440
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3034434943378786e-06,
+      "loss": 0.6869,
+      "step": 40450
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.298821354287035e-06,
+      "loss": 0.9247,
+      "step": 40460
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2941992142361915e-06,
+      "loss": 0.7559,
+      "step": 40470
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2895770741853481e-06,
+      "loss": 0.8214,
+      "step": 40480
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2849549341345044e-06,
+      "loss": 0.7947,
+      "step": 40490
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2803327940836608e-06,
+      "loss": 0.9728,
+      "step": 40500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2757106540328173e-06,
+      "loss": 0.8738,
+      "step": 40510
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2710885139819737e-06,
+      "loss": 0.6264,
+      "step": 40520
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2664663739311304e-06,
+      "loss": 1.0125,
+      "step": 40530
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2618442338802866e-06,
+      "loss": 0.7527,
+      "step": 40540
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2572220938294432e-06,
+      "loss": 0.86,
+      "step": 40550
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2525999537785997e-06,
+      "loss": 0.5716,
+      "step": 40560
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2479778137277561e-06,
+      "loss": 0.6269,
+      "step": 40570
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2433556736769126e-06,
+      "loss": 0.8941,
+      "step": 40580
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2387335336260688e-06,
+      "loss": 0.8534,
+      "step": 40590
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2341113935752255e-06,
+      "loss": 0.8418,
+      "step": 40600
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.229489253524382e-06,
+      "loss": 0.8226,
+      "step": 40610
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2248671134735384e-06,
+      "loss": 0.9161,
+      "step": 40620
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2202449734226948e-06,
+      "loss": 0.9764,
+      "step": 40630
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2156228333718513e-06,
+      "loss": 0.7506,
+      "step": 40640
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2110006933210077e-06,
+      "loss": 0.8354,
+      "step": 40650
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2063785532701641e-06,
+      "loss": 0.825,
+      "step": 40660
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2017564132193206e-06,
+      "loss": 1.0832,
+      "step": 40670
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.197134273168477e-06,
+      "loss": 0.7905,
+      "step": 40680
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1925121331176335e-06,
+      "loss": 0.54,
+      "step": 40690
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1878899930667901e-06,
+      "loss": 0.9861,
+      "step": 40700
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1832678530159466e-06,
+      "loss": 0.8331,
+      "step": 40710
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.178645712965103e-06,
+      "loss": 0.8353,
+      "step": 40720
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1740235729142593e-06,
+      "loss": 0.7574,
+      "step": 40730
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1694014328634157e-06,
+      "loss": 1.0287,
+      "step": 40740
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1647792928125724e-06,
+      "loss": 1.0113,
+      "step": 40750
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1601571527617288e-06,
+      "loss": 0.7358,
+      "step": 40760
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1555350127108853e-06,
+      "loss": 0.8273,
+      "step": 40770
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1509128726600417e-06,
+      "loss": 0.7437,
+      "step": 40780
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1462907326091982e-06,
+      "loss": 0.8638,
+      "step": 40790
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1416685925583546e-06,
+      "loss": 1.1248,
+      "step": 40800
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.137046452507511e-06,
+      "loss": 0.7548,
+      "step": 40810
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1324243124566675e-06,
+      "loss": 1.0012,
+      "step": 40820
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.127802172405824e-06,
+      "loss": 1.0023,
+      "step": 40830
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1231800323549804e-06,
+      "loss": 0.9223,
+      "step": 40840
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.118557892304137e-06,
+      "loss": 0.7782,
+      "step": 40850
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1139357522532935e-06,
+      "loss": 1.0198,
+      "step": 40860
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1093136122024497e-06,
+      "loss": 0.8788,
+      "step": 40870
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1046914721516062e-06,
+      "loss": 0.8287,
+      "step": 40880
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1000693321007628e-06,
+      "loss": 0.719,
+      "step": 40890
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0954471920499193e-06,
+      "loss": 0.9037,
+      "step": 40900
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0908250519990757e-06,
+      "loss": 0.9093,
+      "step": 40910
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0862029119482322e-06,
+      "loss": 0.8286,
+      "step": 40920
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0815807718973886e-06,
+      "loss": 0.7253,
+      "step": 40930
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.076958631846545e-06,
+      "loss": 0.659,
+      "step": 40940
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0723364917957015e-06,
+      "loss": 0.9496,
+      "step": 40950
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.067714351744858e-06,
+      "loss": 1.036,
+      "step": 40960
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0630922116940144e-06,
+      "loss": 0.9976,
+      "step": 40970
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0584700716431708e-06,
+      "loss": 0.6983,
+      "step": 40980
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0538479315923273e-06,
+      "loss": 0.8223,
+      "step": 40990
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.049225791541484e-06,
+      "loss": 0.9578,
+      "step": 41000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0446036514906402e-06,
+      "loss": 0.8982,
+      "step": 41010
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0399815114397966e-06,
+      "loss": 0.9461,
+      "step": 41020
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.035359371388953e-06,
+      "loss": 0.7597,
+      "step": 41030
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0307372313381097e-06,
+      "loss": 0.8229,
+      "step": 41040
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0261150912872662e-06,
+      "loss": 0.7196,
+      "step": 41050
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0214929512364226e-06,
+      "loss": 0.6362,
+      "step": 41060
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.016870811185579e-06,
+      "loss": 0.8454,
+      "step": 41070
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0122486711347355e-06,
+      "loss": 0.8263,
+      "step": 41080
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.007626531083892e-06,
+      "loss": 0.7458,
+      "step": 41090
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0030043910330484e-06,
+      "loss": 0.9175,
+      "step": 41100
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.983822509822048e-07,
+      "loss": 0.8472,
+      "step": 41110
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.937601109313613e-07,
+      "loss": 0.8031,
+      "step": 41120
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.891379708805177e-07,
+      "loss": 0.8077,
+      "step": 41130
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.845158308296744e-07,
+      "loss": 0.9741,
+      "step": 41140
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.798936907788306e-07,
+      "loss": 0.8479,
+      "step": 41150
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.75271550727987e-07,
+      "loss": 0.7406,
+      "step": 41160
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.706494106771435e-07,
+      "loss": 0.6997,
+      "step": 41170
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.660272706263002e-07,
+      "loss": 0.9546,
+      "step": 41180
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.614051305754566e-07,
+      "loss": 0.8618,
+      "step": 41190
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.56782990524613e-07,
+      "loss": 0.8183,
+      "step": 41200
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.521608504737695e-07,
+      "loss": 0.7402,
+      "step": 41210
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.475387104229258e-07,
+      "loss": 0.9857,
+      "step": 41220
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.429165703720823e-07,
+      "loss": 0.8468,
+      "step": 41230
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.382944303212388e-07,
+      "loss": 0.7985,
+      "step": 41240
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.336722902703953e-07,
+      "loss": 0.6779,
+      "step": 41250
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.290501502195517e-07,
+      "loss": 0.8421,
+      "step": 41260
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.244280101687083e-07,
+      "loss": 0.7356,
+      "step": 41270
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.198058701178647e-07,
+      "loss": 0.7185,
+      "step": 41280
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.151837300670211e-07,
+      "loss": 0.8265,
+      "step": 41290
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.105615900161775e-07,
+      "loss": 0.7589,
+      "step": 41300
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.05939449965334e-07,
+      "loss": 0.7489,
+      "step": 41310
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.013173099144905e-07,
+      "loss": 0.8537,
+      "step": 41320
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.966951698636469e-07,
+      "loss": 0.9229,
+      "step": 41330
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.920730298128035e-07,
+      "loss": 1.038,
+      "step": 41340
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.874508897619599e-07,
+      "loss": 0.7859,
+      "step": 41350
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.828287497111163e-07,
+      "loss": 0.6604,
+      "step": 41360
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.782066096602727e-07,
+      "loss": 0.6185,
+      "step": 41370
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.735844696094292e-07,
+      "loss": 1.1035,
+      "step": 41380
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.689623295585857e-07,
+      "loss": 0.8818,
+      "step": 41390
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.643401895077422e-07,
+      "loss": 0.635,
+      "step": 41400
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.597180494568986e-07,
+      "loss": 1.0335,
+      "step": 41410
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.550959094060552e-07,
+      "loss": 0.7906,
+      "step": 41420
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.504737693552115e-07,
+      "loss": 0.7634,
+      "step": 41430
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.45851629304368e-07,
+      "loss": 0.6991,
+      "step": 41440
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.412294892535244e-07,
+      "loss": 0.7547,
+      "step": 41450
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.366073492026809e-07,
+      "loss": 0.8949,
+      "step": 41460
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.319852091518374e-07,
+      "loss": 0.7047,
+      "step": 41470
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.273630691009938e-07,
+      "loss": 0.8287,
+      "step": 41480
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.227409290501504e-07,
+      "loss": 0.8802,
+      "step": 41490
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.181187889993067e-07,
+      "loss": 0.8211,
+      "step": 41500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.134966489484632e-07,
+      "loss": 0.7476,
+      "step": 41510
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.088745088976196e-07,
+      "loss": 0.5697,
+      "step": 41520
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.042523688467762e-07,
+      "loss": 0.6267,
+      "step": 41530
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.996302287959326e-07,
+      "loss": 0.6687,
+      "step": 41540
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.950080887450891e-07,
+      "loss": 0.6749,
+      "step": 41550
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.903859486942456e-07,
+      "loss": 0.9505,
+      "step": 41560
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.85763808643402e-07,
+      "loss": 1.1076,
+      "step": 41570
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.811416685925584e-07,
+      "loss": 0.8154,
+      "step": 41580
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.765195285417148e-07,
+      "loss": 0.7719,
+      "step": 41590
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.718973884908714e-07,
+      "loss": 0.6438,
+      "step": 41600
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.672752484400278e-07,
+      "loss": 0.7093,
+      "step": 41610
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.626531083891843e-07,
+      "loss": 0.7584,
+      "step": 41620
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.580309683383407e-07,
+      "loss": 0.8208,
+      "step": 41630
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.534088282874971e-07,
+      "loss": 0.7233,
+      "step": 41640
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.487866882366536e-07,
+      "loss": 0.7593,
+      "step": 41650
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.441645481858101e-07,
+      "loss": 0.6811,
+      "step": 41660
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.395424081349665e-07,
+      "loss": 0.8348,
+      "step": 41670
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.349202680841231e-07,
+      "loss": 0.8877,
+      "step": 41680
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.302981280332795e-07,
+      "loss": 0.6969,
+      "step": 41690
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.25675987982436e-07,
+      "loss": 0.6676,
+      "step": 41700
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.210538479315923e-07,
+      "loss": 0.7548,
+      "step": 41710
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.164317078807488e-07,
+      "loss": 0.9627,
+      "step": 41720
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.118095678299053e-07,
+      "loss": 0.8813,
+      "step": 41730
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.071874277790617e-07,
+      "loss": 0.9023,
+      "step": 41740
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.025652877282183e-07,
+      "loss": 0.8035,
+      "step": 41750
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.979431476773747e-07,
+      "loss": 0.882,
+      "step": 41760
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.933210076265312e-07,
+      "loss": 0.7534,
+      "step": 41770
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.886988675756875e-07,
+      "loss": 0.7041,
+      "step": 41780
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.840767275248441e-07,
+      "loss": 0.7689,
+      "step": 41790
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.794545874740005e-07,
+      "loss": 0.6684,
+      "step": 41800
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.74832447423157e-07,
+      "loss": 0.824,
+      "step": 41810
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.702103073723135e-07,
+      "loss": 0.9621,
+      "step": 41820
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.6558816732147e-07,
+      "loss": 0.7783,
+      "step": 41830
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.609660272706264e-07,
+      "loss": 1.1654,
+      "step": 41840
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.563438872197827e-07,
+      "loss": 0.7988,
+      "step": 41850
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.517217471689393e-07,
+      "loss": 0.9677,
+      "step": 41860
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.470996071180957e-07,
+      "loss": 0.7021,
+      "step": 41870
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.424774670672522e-07,
+      "loss": 0.6715,
+      "step": 41880
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.378553270164086e-07,
+      "loss": 0.7793,
+      "step": 41890
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.332331869655652e-07,
+      "loss": 0.616,
+      "step": 41900
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.286110469147216e-07,
+      "loss": 0.8431,
+      "step": 41910
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.239889068638781e-07,
+      "loss": 0.8945,
+      "step": 41920
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.193667668130344e-07,
+      "loss": 0.8388,
+      "step": 41930
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.14744626762191e-07,
+      "loss": 0.7603,
+      "step": 41940
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.101224867113474e-07,
+      "loss": 0.683,
+      "step": 41950
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.055003466605039e-07,
+      "loss": 0.7064,
+      "step": 41960
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.008782066096603e-07,
+      "loss": 0.8207,
+      "step": 41970
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.962560665588167e-07,
+      "loss": 0.8594,
+      "step": 41980
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.916339265079733e-07,
+      "loss": 0.8947,
+      "step": 41990
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.870117864571296e-07,
+      "loss": 0.8772,
+      "step": 42000
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.823896464062862e-07,
+      "loss": 0.8328,
+      "step": 42010
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.777675063554426e-07,
+      "loss": 0.7813,
+      "step": 42020
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.731453663045991e-07,
+      "loss": 0.7217,
+      "step": 42030
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.685232262537555e-07,
+      "loss": 0.8375,
+      "step": 42040
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.63901086202912e-07,
+      "loss": 0.8941,
+      "step": 42050
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.592789461520685e-07,
+      "loss": 0.918,
+      "step": 42060
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.546568061012249e-07,
+      "loss": 0.8044,
+      "step": 42070
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.500346660503814e-07,
+      "loss": 0.7943,
+      "step": 42080
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.454125259995379e-07,
+      "loss": 0.7427,
+      "step": 42090
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.407903859486943e-07,
+      "loss": 0.8255,
+      "step": 42100
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.361682458978507e-07,
+      "loss": 0.8548,
+      "step": 42110
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.315461058470072e-07,
+      "loss": 0.6372,
+      "step": 42120
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.269239657961636e-07,
+      "loss": 0.7794,
+      "step": 42130
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.223018257453201e-07,
+      "loss": 0.9804,
+      "step": 42140
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.176796856944765e-07,
+      "loss": 0.8948,
+      "step": 42150
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.130575456436331e-07,
+      "loss": 0.7771,
+      "step": 42160
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.084354055927895e-07,
+      "loss": 0.8642,
+      "step": 42170
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.03813265541946e-07,
+      "loss": 0.7998,
+      "step": 42180
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.991911254911024e-07,
+      "loss": 0.6526,
+      "step": 42190
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.945689854402589e-07,
+      "loss": 0.6169,
+      "step": 42200
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.899468453894153e-07,
+      "loss": 0.7735,
+      "step": 42210
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.853247053385717e-07,
+      "loss": 0.9115,
+      "step": 42220
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.807025652877283e-07,
+      "loss": 0.917,
+      "step": 42230
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.7608042523688475e-07,
+      "loss": 0.9388,
+      "step": 42240
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.7145828518604114e-07,
+      "loss": 0.8131,
+      "step": 42250
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.6683614513519764e-07,
+      "loss": 1.0282,
+      "step": 42260
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.6221400508435414e-07,
+      "loss": 0.9411,
+      "step": 42270
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.5759186503351053e-07,
+      "loss": 0.938,
+      "step": 42280
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.52969724982667e-07,
+      "loss": 0.9582,
+      "step": 42290
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.4834758493182347e-07,
+      "loss": 0.741,
+      "step": 42300
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.4372544488097997e-07,
+      "loss": 0.9183,
+      "step": 42310
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.3910330483013636e-07,
+      "loss": 0.7874,
+      "step": 42320
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.3448116477929286e-07,
+      "loss": 0.8563,
+      "step": 42330
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.298590247284493e-07,
+      "loss": 0.7968,
+      "step": 42340
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.2523688467760575e-07,
+      "loss": 0.9793,
+      "step": 42350
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.206147446267622e-07,
+      "loss": 0.9982,
+      "step": 42360
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.159926045759187e-07,
+      "loss": 0.7095,
+      "step": 42370
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.113704645250752e-07,
+      "loss": 0.9736,
+      "step": 42380
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.067483244742316e-07,
+      "loss": 0.8531,
+      "step": 42390
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.021261844233881e-07,
+      "loss": 0.8678,
+      "step": 42400
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.9750404437254453e-07,
+      "loss": 0.7943,
+      "step": 42410
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.92881904321701e-07,
+      "loss": 0.7993,
+      "step": 42420
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.882597642708574e-07,
+      "loss": 0.9618,
+      "step": 42430
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.836376242200139e-07,
+      "loss": 0.7184,
+      "step": 42440
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.7901548416917036e-07,
+      "loss": 0.7651,
+      "step": 42450
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.743933441183268e-07,
+      "loss": 0.7157,
+      "step": 42460
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.6977120406748326e-07,
+      "loss": 0.7871,
+      "step": 42470
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.6514906401663975e-07,
+      "loss": 0.7309,
+      "step": 42480
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.6052692396579615e-07,
+      "loss": 0.9873,
+      "step": 42490
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.5590478391495264e-07,
+      "loss": 0.6802,
+      "step": 42500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.5128264386410914e-07,
+      "loss": 0.8875,
+      "step": 42510
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.466605038132656e-07,
+      "loss": 0.9746,
+      "step": 42520
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.4203836376242203e-07,
+      "loss": 0.798,
+      "step": 42530
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.374162237115785e-07,
+      "loss": 0.879,
+      "step": 42540
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.32794083660735e-07,
+      "loss": 0.8701,
+      "step": 42550
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.2817194360989137e-07,
+      "loss": 0.6338,
+      "step": 42560
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.2354980355904787e-07,
+      "loss": 0.7981,
+      "step": 42570
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.189276635082043e-07,
+      "loss": 0.789,
+      "step": 42580
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.143055234573608e-07,
+      "loss": 0.8245,
+      "step": 42590
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.096833834065172e-07,
+      "loss": 0.8833,
+      "step": 42600
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.050612433556737e-07,
+      "loss": 0.6714,
+      "step": 42610
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.0043910330483015e-07,
+      "loss": 0.9324,
+      "step": 42620
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.9581696325398665e-07,
+      "loss": 0.9003,
+      "step": 42630
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.911948232031431e-07,
+      "loss": 0.8213,
+      "step": 42640
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.8657268315229954e-07,
+      "loss": 1.2601,
+      "step": 42650
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.81950543101456e-07,
+      "loss": 0.9313,
+      "step": 42660
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.7732840305061243e-07,
+      "loss": 0.82,
+      "step": 42670
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.7270626299976893e-07,
+      "loss": 0.8705,
+      "step": 42680
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.6808412294892537e-07,
+      "loss": 1.1823,
+      "step": 42690
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.634619828980818e-07,
+      "loss": 0.7676,
+      "step": 42700
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5883984284723826e-07,
+      "loss": 0.9706,
+      "step": 42710
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5421770279639476e-07,
+      "loss": 0.9443,
+      "step": 42720
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.495955627455512e-07,
+      "loss": 0.706,
+      "step": 42730
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.4497342269470765e-07,
+      "loss": 0.8233,
+      "step": 42740
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.4035128264386415e-07,
+      "loss": 0.737,
+      "step": 42750
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.3572914259302057e-07,
+      "loss": 0.8844,
+      "step": 42760
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.3110700254217707e-07,
+      "loss": 0.8567,
+      "step": 42770
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.264848624913335e-07,
+      "loss": 0.6255,
+      "step": 42780
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2186272244048999e-07,
+      "loss": 0.8046,
+      "step": 42790
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.1724058238964643e-07,
+      "loss": 1.0321,
+      "step": 42800
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.1261844233880288e-07,
+      "loss": 0.9522,
+      "step": 42810
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.0799630228795935e-07,
+      "loss": 0.7474,
+      "step": 42820
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.033741622371158e-07,
+      "loss": 0.9196,
+      "step": 42830
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.9875202218627227e-07,
+      "loss": 0.6466,
+      "step": 42840
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.941298821354287e-07,
+      "loss": 0.7211,
+      "step": 42850
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.8950774208458518e-07,
+      "loss": 0.8246,
+      "step": 42860
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.8488560203374163e-07,
+      "loss": 0.9331,
+      "step": 42870
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.8026346198289807e-07,
+      "loss": 0.8622,
+      "step": 42880
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.7564132193205457e-07,
+      "loss": 0.7557,
+      "step": 42890
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.7101918188121102e-07,
+      "loss": 0.903,
+      "step": 42900
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.663970418303675e-07,
+      "loss": 0.7853,
+      "step": 42910
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6177490177952393e-07,
+      "loss": 0.9762,
+      "step": 42920
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.571527617286804e-07,
+      "loss": 0.8818,
+      "step": 42930
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5253062167783685e-07,
+      "loss": 0.8462,
+      "step": 42940
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.4790848162699332e-07,
+      "loss": 0.8265,
+      "step": 42950
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.4328634157614977e-07,
+      "loss": 0.833,
+      "step": 42960
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3866420152530621e-07,
+      "loss": 0.7779,
+      "step": 42970
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3404206147446269e-07,
+      "loss": 0.6876,
+      "step": 42980
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2941992142361913e-07,
+      "loss": 0.9143,
+      "step": 42990
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.247977813727756e-07,
+      "loss": 0.8746,
+      "step": 43000
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2017564132193208e-07,
+      "loss": 0.7074,
+      "step": 43010
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.1555350127108853e-07,
+      "loss": 1.0225,
+      "step": 43020
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.1093136122024499e-07,
+      "loss": 0.7904,
+      "step": 43030
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0630922116940144e-07,
+      "loss": 0.7089,
+      "step": 43040
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.016870811185579e-07,
+      "loss": 0.6949,
+      "step": 43050
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.706494106771436e-08,
+      "loss": 0.9937,
+      "step": 43060
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.244280101687081e-08,
+      "loss": 1.0286,
+      "step": 43070
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.782066096602729e-08,
+      "loss": 0.7941,
+      "step": 43080
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.319852091518374e-08,
+      "loss": 0.6805,
+      "step": 43090
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.85763808643402e-08,
+      "loss": 0.7183,
+      "step": 43100
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.395424081349666e-08,
+      "loss": 0.7769,
+      "step": 43110
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.933210076265311e-08,
+      "loss": 0.8894,
+      "step": 43120
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.470996071180957e-08,
+      "loss": 1.0142,
+      "step": 43130
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.008782066096604e-08,
+      "loss": 0.8407,
+      "step": 43140
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.5465680610122496e-08,
+      "loss": 0.7511,
+      "step": 43150
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.084354055927895e-08,
+      "loss": 0.7445,
+      "step": 43160
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.622140050843541e-08,
+      "loss": 0.9687,
+      "step": 43170
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.159926045759187e-08,
+      "loss": 0.8512,
+      "step": 43180
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.697712040674833e-08,
+      "loss": 0.9349,
+      "step": 43190
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.235498035590478e-08,
+      "loss": 0.7318,
+      "step": 43200
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.7732840305061248e-08,
+      "loss": 0.9102,
+      "step": 43210
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.3110700254217703e-08,
+      "loss": 0.9605,
+      "step": 43220
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.8488560203374165e-08,
+      "loss": 0.7449,
+      "step": 43230
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.3866420152530624e-08,
+      "loss": 0.8021,
+      "step": 43240
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.244280101687083e-09,
+      "loss": 0.6975,
+      "step": 43250
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.622140050843541e-09,
+      "loss": 0.9352,
+      "step": 43260
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.9557,
+      "step": 43270
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.7140050748956372,
+      "eval_loss": 0.8022026419639587,
+      "eval_runtime": 374.0744,
+      "eval_samples_per_second": 32.659,
+      "eval_steps_per_second": 4.085,
+      "step": 43270
+    },
+    {
+      "epoch": 5.0,
+      "step": 43270,
+      "total_flos": 8.699718760261724e+18,
+      "train_loss": 0.9448149496615829,
+      "train_runtime": 12156.2703,
+      "train_samples_per_second": 28.474,
+      "train_steps_per_second": 3.559
+    }
+  ],
+  "max_steps": 43270,
+  "num_train_epochs": 5,
+  "total_flos": 8.699718760261724e+18,
+  "trial_name": null,
+  "trial_params": null
+}