diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.3233760459193985,
-  "global_step": 4000,
+  "epoch": 0.646752091838797,
+  "global_step": 8000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -4916,11 +4916,4822 @@
       "eval_samples_per_second": 0.092,
       "eval_steps_per_second": 0.046,
       "step": 4000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.461557118602959e-05,
+      "loss": 2.9516,
+      "step": 4005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.460883391273884e-05,
+      "loss": 3.2431,
+      "step": 4010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.460209663944808e-05,
+      "loss": 3.0873,
+      "step": 4015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.459535936615733e-05,
+      "loss": 2.8684,
+      "step": 4020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458862209286658e-05,
+      "loss": 3.1442,
+      "step": 4025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458188481957582e-05,
+      "loss": 3.1723,
+      "step": 4030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4575147546285065e-05,
+      "loss": 3.4121,
+      "step": 4035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4568410272994313e-05,
+      "loss": 2.8923,
+      "step": 4040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.456167299970356e-05,
+      "loss": 3.3188,
+      "step": 4045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455493572641281e-05,
+      "loss": 3.3509,
+      "step": 4050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454819845312205e-05,
+      "loss": 2.9415,
+      "step": 4055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45414611798313e-05,
+      "loss": 3.1391,
+      "step": 4060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453472390654055e-05,
+      "loss": 3.2693,
+      "step": 4065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.452798663324979e-05,
+      "loss": 3.1826,
+      "step": 4070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.452124935995904e-05,
+      "loss": 2.9731,
+      "step": 4075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4514512086668284e-05,
+      "loss": 3.0303,
+      "step": 4080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.450777481337753e-05,
+      "loss": 3.1995,
+      "step": 4085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4501037540086775e-05,
+      "loss": 3.0824,
+      "step": 4090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4494300266796024e-05,
+      "loss": 3.1416,
+      "step": 4095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448756299350527e-05,
+      "loss": 3.0063,
+      "step": 4100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448082572021452e-05,
+      "loss": 2.9765,
+      "step": 4105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4474088446923764e-05,
+      "loss": 3.1823,
+      "step": 4110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.446735117363301e-05,
+      "loss": 3.255,
+      "step": 4115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4460613900342255e-05,
+      "loss": 3.2149,
+      "step": 4120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.44538766270515e-05,
+      "loss": 3.0531,
+      "step": 4125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4447139353760746e-05,
+      "loss": 3.0287,
+      "step": 4130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4440402080469995e-05,
+      "loss": 2.8681,
+      "step": 4135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4433664807179243e-05,
+      "loss": 3.2937,
+      "step": 4140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4426927533888486e-05,
+      "loss": 3.1149,
+      "step": 4145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4420190260597734e-05,
+      "loss": 3.1983,
+      "step": 4150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441345298730698e-05,
+      "loss": 3.3325,
+      "step": 4155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4406715714016225e-05,
+      "loss": 3.1252,
+      "step": 4160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439997844072547e-05,
+      "loss": 3.2575,
+      "step": 4165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4393241167434716e-05,
+      "loss": 3.1463,
+      "step": 4170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4386503894143965e-05,
+      "loss": 3.002,
+      "step": 4175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437976662085321e-05,
+      "loss": 3.3059,
+      "step": 4180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4373029347562456e-05,
+      "loss": 3.2497,
+      "step": 4185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4366292074271705e-05,
+      "loss": 2.9685,
+      "step": 4190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4359554800980954e-05,
+      "loss": 2.9922,
+      "step": 4195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4352817527690196e-05,
+      "loss": 3.261,
+      "step": 4200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.434608025439944e-05,
+      "loss": 3.0169,
+      "step": 4205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.433934298110869e-05,
+      "loss": 3.0149,
+      "step": 4210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.433260570781793e-05,
+      "loss": 3.0762,
+      "step": 4215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.432586843452718e-05,
+      "loss": 3.1659,
+      "step": 4220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431913116123643e-05,
+      "loss": 3.1697,
+      "step": 4225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4312393887945676e-05,
+      "loss": 3.2132,
+      "step": 4230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.430565661465492e-05,
+      "loss": 2.9254,
+      "step": 4235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429891934136417e-05,
+      "loss": 3.1275,
+      "step": 4240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4292182068073416e-05,
+      "loss": 3.298,
+      "step": 4245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.428544479478266e-05,
+      "loss": 2.9652,
+      "step": 4250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.42787075214919e-05,
+      "loss": 2.8456,
+      "step": 4255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.427197024820115e-05,
+      "loss": 2.9501,
+      "step": 4260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.42652329749104e-05,
+      "loss": 2.9993,
+      "step": 4265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425849570161964e-05,
+      "loss": 3.1575,
+      "step": 4270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425175842832889e-05,
+      "loss": 2.9604,
+      "step": 4275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424502115503814e-05,
+      "loss": 3.1482,
+      "step": 4280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4238283881747386e-05,
+      "loss": 3.2198,
+      "step": 4285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423154660845663e-05,
+      "loss": 3.3901,
+      "step": 4290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422480933516587e-05,
+      "loss": 3.1776,
+      "step": 4295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421807206187512e-05,
+      "loss": 3.0075,
+      "step": 4300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421133478858437e-05,
+      "loss": 2.9798,
+      "step": 4305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.420459751529361e-05,
+      "loss": 2.9517,
+      "step": 4310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419786024200286e-05,
+      "loss": 3.1629,
+      "step": 4315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419112296871211e-05,
+      "loss": 3.1076,
+      "step": 4320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.418438569542135e-05,
+      "loss": 2.9667,
+      "step": 4325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41776484221306e-05,
+      "loss": 3.1091,
+      "step": 4330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.417091114883984e-05,
+      "loss": 2.9619,
+      "step": 4335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.416417387554909e-05,
+      "loss": 3.2804,
+      "step": 4340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.415743660225833e-05,
+      "loss": 2.9225,
+      "step": 4345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.415069932896758e-05,
+      "loss": 2.8438,
+      "step": 4350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414396205567683e-05,
+      "loss": 3.2374,
+      "step": 4355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413722478238608e-05,
+      "loss": 3.027,
+      "step": 4360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413048750909532e-05,
+      "loss": 3.1018,
+      "step": 4365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412375023580457e-05,
+      "loss": 3.0387,
+      "step": 4370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.411701296251382e-05,
+      "loss": 3.1002,
+      "step": 4375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.411027568922306e-05,
+      "loss": 3.0958,
+      "step": 4380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41035384159323e-05,
+      "loss": 3.0154,
+      "step": 4385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.409680114264155e-05,
+      "loss": 2.8887,
+      "step": 4390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40900638693508e-05,
+      "loss": 3.3108,
+      "step": 4395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408332659606004e-05,
+      "loss": 3.1558,
+      "step": 4400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407658932276929e-05,
+      "loss": 3.0019,
+      "step": 4405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406985204947854e-05,
+      "loss": 3.1111,
+      "step": 4410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406311477618779e-05,
+      "loss": 3.0919,
+      "step": 4415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405637750289703e-05,
+      "loss": 3.1809,
+      "step": 4420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404964022960627e-05,
+      "loss": 3.211,
+      "step": 4425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404290295631552e-05,
+      "loss": 3.0161,
+      "step": 4430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4036165683024764e-05,
+      "loss": 3.0918,
+      "step": 4435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402942840973401e-05,
+      "loss": 3.1814,
+      "step": 4440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402269113644326e-05,
+      "loss": 2.9785,
+      "step": 4445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401595386315251e-05,
+      "loss": 3.1153,
+      "step": 4450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400921658986175e-05,
+      "loss": 2.8357,
+      "step": 4455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4002479316571e-05,
+      "loss": 3.0285,
+      "step": 4460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3995742043280244e-05,
+      "loss": 3.2999,
+      "step": 4465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.398900476998949e-05,
+      "loss": 3.0133,
+      "step": 4470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3982267496698735e-05,
+      "loss": 3.3555,
+      "step": 4475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3975530223407984e-05,
+      "loss": 3.296,
+      "step": 4480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.396879295011723e-05,
+      "loss": 3.4159,
+      "step": 4485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3962055676826475e-05,
+      "loss": 3.1615,
+      "step": 4490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3955318403535724e-05,
+      "loss": 3.1965,
+      "step": 4495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.394858113024497e-05,
+      "loss": 3.0436,
+      "step": 4500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3941843856954215e-05,
+      "loss": 3.1806,
+      "step": 4505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.393510658366346e-05,
+      "loss": 3.2201,
+      "step": 4510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3928369310372706e-05,
+      "loss": 3.0641,
+      "step": 4515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3921632037081954e-05,
+      "loss": 2.8647,
+      "step": 4520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3914894763791197e-05,
+      "loss": 2.8806,
+      "step": 4525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3908157490500445e-05,
+      "loss": 2.9292,
+      "step": 4530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3901420217209694e-05,
+      "loss": 3.0328,
+      "step": 4535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389468294391894e-05,
+      "loss": 3.1128,
+      "step": 4540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3887945670628185e-05,
+      "loss": 3.0104,
+      "step": 4545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3881208397337434e-05,
+      "loss": 2.7728,
+      "step": 4550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3874471124046676e-05,
+      "loss": 3.2391,
+      "step": 4555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3867733850755925e-05,
+      "loss": 2.8613,
+      "step": 4560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.386099657746517e-05,
+      "loss": 3.3067,
+      "step": 4565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3854259304174416e-05,
+      "loss": 2.9091,
+      "step": 4570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3847522030883665e-05,
+      "loss": 2.8817,
+      "step": 4575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.384078475759291e-05,
+      "loss": 2.9741,
+      "step": 4580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3834047484302156e-05,
+      "loss": 3.0892,
+      "step": 4585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3827310211011405e-05,
+      "loss": 3.2429,
+      "step": 4590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382057293772065e-05,
+      "loss": 2.9408,
+      "step": 4595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.381383566442989e-05,
+      "loss": 3.0757,
+      "step": 4600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.380709839113914e-05,
+      "loss": 2.8738,
+      "step": 4605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.380036111784839e-05,
+      "loss": 3.2346,
+      "step": 4610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3793623844557636e-05,
+      "loss": 2.9399,
+      "step": 4615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378688657126688e-05,
+      "loss": 3.195,
+      "step": 4620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3780149297976127e-05,
+      "loss": 2.7772,
+      "step": 4625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3773412024685375e-05,
+      "loss": 3.4355,
+      "step": 4630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.376667475139462e-05,
+      "loss": 3.2062,
+      "step": 4635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.375993747810386e-05,
+      "loss": 3.1914,
+      "step": 4640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.375320020481311e-05,
+      "loss": 3.1476,
+      "step": 4645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374646293152236e-05,
+      "loss": 3.0083,
+      "step": 4650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.37397256582316e-05,
+      "loss": 3.2261,
+      "step": 4655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373298838494085e-05,
+      "loss": 3.2371,
+      "step": 4660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.37262511116501e-05,
+      "loss": 2.9125,
+      "step": 4665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3719513838359346e-05,
+      "loss": 3.3789,
+      "step": 4670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371277656506859e-05,
+      "loss": 3.1501,
+      "step": 4675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370603929177784e-05,
+      "loss": 3.1306,
+      "step": 4680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369930201848708e-05,
+      "loss": 3.0257,
+      "step": 4685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369256474519632e-05,
+      "loss": 3.1348,
+      "step": 4690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368582747190557e-05,
+      "loss": 3.1772,
+      "step": 4695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367909019861482e-05,
+      "loss": 2.9573,
+      "step": 4700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367235292532407e-05,
+      "loss": 3.0109,
+      "step": 4705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.366561565203331e-05,
+      "loss": 3.1603,
+      "step": 4710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365887837874256e-05,
+      "loss": 3.0631,
+      "step": 4715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365214110545181e-05,
+      "loss": 3.2663,
+      "step": 4720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.364540383216105e-05,
+      "loss": 3.216,
+      "step": 4725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.363866655887029e-05,
+      "loss": 3.1248,
+      "step": 4730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.363192928557954e-05,
+      "loss": 3.0795,
+      "step": 4735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362519201228879e-05,
+      "loss": 3.2579,
+      "step": 4740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.361845473899803e-05,
+      "loss": 2.9879,
+      "step": 4745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.361171746570728e-05,
+      "loss": 3.0311,
+      "step": 4750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.360498019241653e-05,
+      "loss": 2.9413,
+      "step": 4755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.359824291912578e-05,
+      "loss": 2.9917,
+      "step": 4760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359150564583502e-05,
+      "loss": 2.9656,
+      "step": 4765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358476837254426e-05,
+      "loss": 2.833,
+      "step": 4770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357803109925351e-05,
+      "loss": 3.1525,
+      "step": 4775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357129382596276e-05,
+      "loss": 3.1786,
+      "step": 4780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3564556552672e-05,
+      "loss": 3.0506,
+      "step": 4785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355781927938125e-05,
+      "loss": 3.3234,
+      "step": 4790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35510820060905e-05,
+      "loss": 3.3562,
+      "step": 4795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.354434473279974e-05,
+      "loss": 3.1701,
+      "step": 4800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353760745950899e-05,
+      "loss": 2.8417,
+      "step": 4805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353087018621823e-05,
+      "loss": 3.1097,
+      "step": 4810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352413291292748e-05,
+      "loss": 3.3805,
+      "step": 4815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3517395639636724e-05,
+      "loss": 2.9212,
+      "step": 4820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.351065836634597e-05,
+      "loss": 2.9774,
+      "step": 4825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.350392109305522e-05,
+      "loss": 3.2141,
+      "step": 4830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3497183819764464e-05,
+      "loss": 3.2205,
+      "step": 4835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.349044654647371e-05,
+      "loss": 3.0818,
+      "step": 4840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.348370927318296e-05,
+      "loss": 3.0723,
+      "step": 4845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347697199989221e-05,
+      "loss": 3.3671,
+      "step": 4850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347023472660145e-05,
+      "loss": 3.2076,
+      "step": 4855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3463497453310695e-05,
+      "loss": 3.2495,
+      "step": 4860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3456760180019944e-05,
+      "loss": 2.7959,
+      "step": 4865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.345002290672919e-05,
+      "loss": 3.1347,
+      "step": 4870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3443285633438435e-05,
+      "loss": 2.9321,
+      "step": 4875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3436548360147684e-05,
+      "loss": 2.9697,
+      "step": 4880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.342981108685693e-05,
+      "loss": 3.0605,
+      "step": 4885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3423073813566175e-05,
+      "loss": 3.1212,
+      "step": 4890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3416336540275423e-05,
+      "loss": 2.9201,
+      "step": 4895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3409599266984666e-05,
+      "loss": 2.9829,
+      "step": 4900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3402861993693914e-05,
+      "loss": 3.3439,
+      "step": 4905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3396124720403157e-05,
+      "loss": 3.0339,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3389387447112405e-05,
+      "loss": 3.0685,
+      "step": 4915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3382650173821654e-05,
+      "loss": 3.0576,
+      "step": 4920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.33759129005309e-05,
+      "loss": 3.0614,
+      "step": 4925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3369175627240145e-05,
+      "loss": 3.1312,
+      "step": 4930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3362438353949394e-05,
+      "loss": 3.1098,
+      "step": 4935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3355701080658636e-05,
+      "loss": 3.2032,
+      "step": 4940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.334896380736788e-05,
+      "loss": 3.0584,
+      "step": 4945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.334222653407713e-05,
+      "loss": 2.87,
+      "step": 4950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3335489260786376e-05,
+      "loss": 2.872,
+      "step": 4955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3328751987495625e-05,
+      "loss": 3.0233,
+      "step": 4960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.332201471420487e-05,
+      "loss": 3.1497,
+      "step": 4965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3315277440914116e-05,
+      "loss": 3.3478,
+      "step": 4970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3308540167623365e-05,
+      "loss": 2.6513,
+      "step": 4975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3301802894332614e-05,
+      "loss": 3.1054,
+      "step": 4980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3295065621041856e-05,
+      "loss": 3.1679,
+      "step": 4985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.32883283477511e-05,
+      "loss": 2.9424,
+      "step": 4990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.328159107446035e-05,
+      "loss": 2.9678,
+      "step": 4995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.327485380116959e-05,
+      "loss": 2.8093,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.326811652787884e-05,
+      "loss": 3.1868,
+      "step": 5005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3261379254588087e-05,
+      "loss": 3.1751,
+      "step": 5010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3254641981297335e-05,
+      "loss": 3.1772,
+      "step": 5015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324790470800658e-05,
+      "loss": 2.9964,
+      "step": 5020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3241167434715826e-05,
+      "loss": 2.8733,
+      "step": 5025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323443016142507e-05,
+      "loss": 2.9992,
+      "step": 5030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.322769288813432e-05,
+      "loss": 3.0003,
+      "step": 5035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.322095561484356e-05,
+      "loss": 3.0774,
+      "step": 5040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321421834155281e-05,
+      "loss": 2.9182,
+      "step": 5045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320748106826206e-05,
+      "loss": 3.325,
+      "step": 5050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32007437949713e-05,
+      "loss": 2.7606,
+      "step": 5055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319400652168055e-05,
+      "loss": 2.9947,
+      "step": 5060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.31872692483898e-05,
+      "loss": 3.024,
+      "step": 5065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.318053197509904e-05,
+      "loss": 3.0752,
+      "step": 5070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.317379470180828e-05,
+      "loss": 3.1484,
+      "step": 5075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316705742851753e-05,
+      "loss": 3.0559,
+      "step": 5080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316032015522678e-05,
+      "loss": 3.1059,
+      "step": 5085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315358288193602e-05,
+      "loss": 3.062,
+      "step": 5090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.314684560864527e-05,
+      "loss": 2.8981,
+      "step": 5095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.314010833535452e-05,
+      "loss": 3.0228,
+      "step": 5100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313337106206377e-05,
+      "loss": 3.1474,
+      "step": 5105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.312663378877301e-05,
+      "loss": 3.053,
+      "step": 5110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.311989651548226e-05,
+      "loss": 2.8585,
+      "step": 5115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.31131592421915e-05,
+      "loss": 2.927,
+      "step": 5120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.310642196890075e-05,
+      "loss": 2.9696,
+      "step": 5125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.309968469560999e-05,
+      "loss": 3.0903,
+      "step": 5130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309294742231924e-05,
+      "loss": 3.1462,
+      "step": 5135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308621014902849e-05,
+      "loss": 3.0195,
+      "step": 5140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307947287573773e-05,
+      "loss": 2.931,
+      "step": 5145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307273560244698e-05,
+      "loss": 3.1945,
+      "step": 5150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306599832915623e-05,
+      "loss": 2.9829,
+      "step": 5155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305926105586547e-05,
+      "loss": 3.0799,
+      "step": 5160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3052523782574714e-05,
+      "loss": 2.9349,
+      "step": 5165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.304578650928396e-05,
+      "loss": 3.1081,
+      "step": 5170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303904923599321e-05,
+      "loss": 2.9171,
+      "step": 5175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303231196270246e-05,
+      "loss": 3.1681,
+      "step": 5180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30255746894117e-05,
+      "loss": 3.1256,
+      "step": 5185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.301883741612095e-05,
+      "loss": 3.0517,
+      "step": 5190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30121001428302e-05,
+      "loss": 3.1481,
+      "step": 5195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.300536286953944e-05,
+      "loss": 3.1047,
+      "step": 5200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2998625596248684e-05,
+      "loss": 3.0044,
+      "step": 5205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.299188832295793e-05,
+      "loss": 3.2102,
+      "step": 5210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.298515104966718e-05,
+      "loss": 3.383,
+      "step": 5215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2978413776376424e-05,
+      "loss": 3.2182,
+      "step": 5220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297167650308567e-05,
+      "loss": 3.0297,
+      "step": 5225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.296493922979492e-05,
+      "loss": 3.2297,
+      "step": 5230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.295820195650417e-05,
+      "loss": 3.2441,
+      "step": 5235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.295146468321341e-05,
+      "loss": 3.0492,
+      "step": 5240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2944727409922655e-05,
+      "loss": 2.9498,
+      "step": 5245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2937990136631904e-05,
+      "loss": 2.8979,
+      "step": 5250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2931252863341146e-05,
+      "loss": 3.0763,
+      "step": 5255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2924515590050395e-05,
+      "loss": 3.071,
+      "step": 5260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2917778316759644e-05,
+      "loss": 3.1906,
+      "step": 5265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291238849812704e-05,
+      "loss": 3.9365,
+      "step": 5270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290565122483629e-05,
+      "loss": 3.0687,
+      "step": 5275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2898913951545536e-05,
+      "loss": 2.8605,
+      "step": 5280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289217667825478e-05,
+      "loss": 3.0541,
+      "step": 5285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288543940496402e-05,
+      "loss": 3.1296,
+      "step": 5290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.287870213167327e-05,
+      "loss": 3.1929,
+      "step": 5295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.287196485838252e-05,
+      "loss": 3.0888,
+      "step": 5300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286522758509177e-05,
+      "loss": 3.1341,
+      "step": 5305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285849031180101e-05,
+      "loss": 3.2743,
+      "step": 5310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285175303851026e-05,
+      "loss": 2.9379,
+      "step": 5315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284501576521951e-05,
+      "loss": 2.93,
+      "step": 5320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283827849192875e-05,
+      "loss": 2.9956,
+      "step": 5325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283154121863799e-05,
+      "loss": 2.9685,
+      "step": 5330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282480394534724e-05,
+      "loss": 3.1619,
+      "step": 5335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281806667205649e-05,
+      "loss": 3.085,
+      "step": 5340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281132939876573e-05,
+      "loss": 2.9773,
+      "step": 5345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.280459212547498e-05,
+      "loss": 3.2456,
+      "step": 5350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.279785485218423e-05,
+      "loss": 2.9965,
+      "step": 5355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.279111757889348e-05,
+      "loss": 3.0508,
+      "step": 5360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278438030560272e-05,
+      "loss": 3.123,
+      "step": 5365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.277764303231196e-05,
+      "loss": 3.0023,
+      "step": 5370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.277090575902121e-05,
+      "loss": 3.3,
+      "step": 5375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.276416848573045e-05,
+      "loss": 3.139,
+      "step": 5380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27574312124397e-05,
+      "loss": 3.1068,
+      "step": 5385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275069393914895e-05,
+      "loss": 3.07,
+      "step": 5390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27439566658582e-05,
+      "loss": 3.1072,
+      "step": 5395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273721939256744e-05,
+      "loss": 3.0112,
+      "step": 5400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273048211927669e-05,
+      "loss": 3.1176,
+      "step": 5405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272374484598593e-05,
+      "loss": 2.7726,
+      "step": 5410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271700757269518e-05,
+      "loss": 2.8491,
+      "step": 5415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2710270299404423e-05,
+      "loss": 2.8852,
+      "step": 5420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270353302611367e-05,
+      "loss": 2.7856,
+      "step": 5425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269679575282292e-05,
+      "loss": 3.1395,
+      "step": 5430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269005847953216e-05,
+      "loss": 3.2047,
+      "step": 5435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268332120624141e-05,
+      "loss": 3.033,
+      "step": 5440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.267658393295066e-05,
+      "loss": 2.987,
+      "step": 5445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.266984665965991e-05,
+      "loss": 3.0016,
+      "step": 5450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.266310938636915e-05,
+      "loss": 3.1097,
+      "step": 5455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2656372113078394e-05,
+      "loss": 3.0633,
+      "step": 5460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.264963483978764e-05,
+      "loss": 2.8894,
+      "step": 5465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.264289756649689e-05,
+      "loss": 3.026,
+      "step": 5470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2636160293206134e-05,
+      "loss": 2.9712,
+      "step": 5475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.262942301991538e-05,
+      "loss": 3.2726,
+      "step": 5480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.262268574662463e-05,
+      "loss": 3.1535,
+      "step": 5485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2615948473333874e-05,
+      "loss": 3.0379,
+      "step": 5490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.260921120004312e-05,
+      "loss": 3.3582,
+      "step": 5495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2602473926752365e-05,
+      "loss": 2.9169,
+      "step": 5500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2595736653461614e-05,
+      "loss": 3.0502,
+      "step": 5505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2588999380170856e-05,
+      "loss": 3.2374,
+      "step": 5510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2582262106880105e-05,
+      "loss": 3.0465,
+      "step": 5515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2575524833589353e-05,
+      "loss": 3.4409,
+      "step": 5520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2568787560298596e-05,
+      "loss": 2.8746,
+      "step": 5525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2562050287007844e-05,
+      "loss": 2.9985,
+      "step": 5530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255531301371709e-05,
+      "loss": 3.2513,
+      "step": 5535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2548575740426335e-05,
+      "loss": 3.142,
+      "step": 5540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.254183846713558e-05,
+      "loss": 2.9062,
+      "step": 5545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2535101193844826e-05,
+      "loss": 2.9976,
+      "step": 5550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2528363920554075e-05,
+      "loss": 3.2293,
+      "step": 5555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2521626647263324e-05,
+      "loss": 2.7799,
+      "step": 5560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2514889373972566e-05,
+      "loss": 2.9451,
+      "step": 5565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2508152100681815e-05,
+      "loss": 2.9368,
+      "step": 5570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2501414827391064e-05,
+      "loss": 2.7972,
+      "step": 5575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2494677554100306e-05,
+      "loss": 2.9851,
+      "step": 5580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2487940280809555e-05,
+      "loss": 2.9721,
+      "step": 5585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.24812030075188e-05,
+      "loss": 3.0178,
+      "step": 5590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2474465734228046e-05,
+      "loss": 3.0758,
+      "step": 5595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.246772846093729e-05,
+      "loss": 3.1676,
+      "step": 5600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.246099118764654e-05,
+      "loss": 3.232,
+      "step": 5605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2454253914355786e-05,
+      "loss": 3.1686,
+      "step": 5610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2447516641065035e-05,
+      "loss": 3.1155,
+      "step": 5615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.244077936777428e-05,
+      "loss": 3.328,
+      "step": 5620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2434042094483526e-05,
+      "loss": 2.904,
+      "step": 5625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242730482119277e-05,
+      "loss": 2.9753,
+      "step": 5630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242056754790201e-05,
+      "loss": 2.9243,
+      "step": 5635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241383027461126e-05,
+      "loss": 3.1304,
+      "step": 5640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240709300132051e-05,
+      "loss": 3.0722,
+      "step": 5645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2400355728029756e-05,
+      "loss": 3.0274,
+      "step": 5650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2393618454739e-05,
+      "loss": 2.9222,
+      "step": 5655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.238688118144825e-05,
+      "loss": 3.0571,
+      "step": 5660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2380143908157496e-05,
+      "loss": 2.8809,
+      "step": 5665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.237340663486674e-05,
+      "loss": 3.025,
+      "step": 5670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236666936157598e-05,
+      "loss": 3.1147,
+      "step": 5675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235993208828523e-05,
+      "loss": 3.1962,
+      "step": 5680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235319481499448e-05,
+      "loss": 3.2954,
+      "step": 5685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.234645754170372e-05,
+      "loss": 3.0492,
+      "step": 5690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233972026841297e-05,
+      "loss": 2.9564,
+      "step": 5695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233298299512222e-05,
+      "loss": 2.8588,
+      "step": 5700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.232624572183147e-05,
+      "loss": 2.9818,
+      "step": 5705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231950844854071e-05,
+      "loss": 3.3102,
+      "step": 5710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231277117524995e-05,
+      "loss": 3.0969,
+      "step": 5715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.23060339019592e-05,
+      "loss": 3.1605,
+      "step": 5720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229929662866845e-05,
+      "loss": 3.1074,
+      "step": 5725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229255935537769e-05,
+      "loss": 3.5156,
+      "step": 5730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.228582208208694e-05,
+      "loss": 3.0901,
+      "step": 5735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.227908480879619e-05,
+      "loss": 3.1218,
+      "step": 5740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.227234753550543e-05,
+      "loss": 2.7888,
+      "step": 5745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.226561026221468e-05,
+      "loss": 2.8804,
+      "step": 5750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225887298892393e-05,
+      "loss": 3.1825,
+      "step": 5755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225213571563317e-05,
+      "loss": 3.0061,
+      "step": 5760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.224539844234241e-05,
+      "loss": 3.0918,
+      "step": 5765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223866116905166e-05,
+      "loss": 2.8022,
+      "step": 5770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223192389576091e-05,
+      "loss": 3.1484,
+      "step": 5775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222518662247016e-05,
+      "loss": 2.9748,
+      "step": 5780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22184493491794e-05,
+      "loss": 3.053,
+      "step": 5785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221171207588865e-05,
+      "loss": 3.1931,
+      "step": 5790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22049748025979e-05,
+      "loss": 3.5065,
+      "step": 5795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219823752930714e-05,
+      "loss": 2.9901,
+      "step": 5800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219150025601638e-05,
+      "loss": 3.3565,
+      "step": 5805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218476298272563e-05,
+      "loss": 2.9501,
+      "step": 5810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.217802570943488e-05,
+      "loss": 2.9877,
+      "step": 5815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.217128843614412e-05,
+      "loss": 3.5543,
+      "step": 5820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216455116285337e-05,
+      "loss": 3.1097,
+      "step": 5825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.215781388956262e-05,
+      "loss": 2.9987,
+      "step": 5830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.215107661627186e-05,
+      "loss": 3.0965,
+      "step": 5835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.214433934298111e-05,
+      "loss": 3.0131,
+      "step": 5840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2137602069690354e-05,
+      "loss": 3.4088,
+      "step": 5845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.21308647963996e-05,
+      "loss": 2.932,
+      "step": 5850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2124127523108845e-05,
+      "loss": 3.1906,
+      "step": 5855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2117390249818094e-05,
+      "loss": 3.3301,
+      "step": 5860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.211065297652734e-05,
+      "loss": 3.0456,
+      "step": 5865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.210391570323659e-05,
+      "loss": 2.8092,
+      "step": 5870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2097178429945834e-05,
+      "loss": 3.1395,
+      "step": 5875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.209044115665508e-05,
+      "loss": 2.8979,
+      "step": 5880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208370388336433e-05,
+      "loss": 3.1059,
+      "step": 5885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2076966610073574e-05,
+      "loss": 3.2633,
+      "step": 5890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2070229336782816e-05,
+      "loss": 2.9585,
+      "step": 5895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2063492063492065e-05,
+      "loss": 3.2317,
+      "step": 5900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2056754790201313e-05,
+      "loss": 2.9057,
+      "step": 5905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2050017516910556e-05,
+      "loss": 2.8944,
+      "step": 5910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2043280243619804e-05,
+      "loss": 3.0617,
+      "step": 5915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203654297032905e-05,
+      "loss": 3.1542,
+      "step": 5920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20298056970383e-05,
+      "loss": 3.0791,
+      "step": 5925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2023068423747544e-05,
+      "loss": 3.2551,
+      "step": 5930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2016331150456786e-05,
+      "loss": 3.2556,
+      "step": 5935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2009593877166035e-05,
+      "loss": 2.9848,
+      "step": 5940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.200285660387528e-05,
+      "loss": 3.2682,
+      "step": 5945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1996119330584526e-05,
+      "loss": 2.9622,
+      "step": 5950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1989382057293775e-05,
+      "loss": 3.4463,
+      "step": 5955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1982644784003024e-05,
+      "loss": 3.197,
+      "step": 5960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1975907510712266e-05,
+      "loss": 2.9737,
+      "step": 5965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1969170237421515e-05,
+      "loss": 3.1124,
+      "step": 5970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.196243296413076e-05,
+      "loss": 3.067,
+      "step": 5975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1955695690840006e-05,
+      "loss": 3.224,
+      "step": 5980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.194895841754925e-05,
+      "loss": 3.2847,
+      "step": 5985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.19422211442585e-05,
+      "loss": 3.0231,
+      "step": 5990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1935483870967746e-05,
+      "loss": 2.8844,
+      "step": 5995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192874659767699e-05,
+      "loss": 3.0196,
+      "step": 6000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192200932438624e-05,
+      "loss": 2.9583,
+      "step": 6005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1915272051095486e-05,
+      "loss": 3.0375,
+      "step": 6010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1908534777804734e-05,
+      "loss": 2.9743,
+      "step": 6015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1901797504513977e-05,
+      "loss": 2.8241,
+      "step": 6020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189506023122322e-05,
+      "loss": 2.836,
+      "step": 6025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.188832295793247e-05,
+      "loss": 3.1118,
+      "step": 6030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1881585684641716e-05,
+      "loss": 2.8837,
+      "step": 6035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187484841135096e-05,
+      "loss": 3.093,
+      "step": 6040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186811113806021e-05,
+      "loss": 2.9789,
+      "step": 6045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1861373864769456e-05,
+      "loss": 2.9944,
+      "step": 6050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18546365914787e-05,
+      "loss": 3.1287,
+      "step": 6055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184789931818795e-05,
+      "loss": 3.1213,
+      "step": 6060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184116204489719e-05,
+      "loss": 3.2943,
+      "step": 6065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.183442477160644e-05,
+      "loss": 3.1028,
+      "step": 6070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.182768749831568e-05,
+      "loss": 3.243,
+      "step": 6075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.182095022502493e-05,
+      "loss": 2.8142,
+      "step": 6080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.181421295173418e-05,
+      "loss": 3.1794,
+      "step": 6085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.180747567844342e-05,
+      "loss": 3.3118,
+      "step": 6090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.180073840515267e-05,
+      "loss": 3.1506,
+      "step": 6095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.179400113186192e-05,
+      "loss": 3.177,
+      "step": 6100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.178726385857116e-05,
+      "loss": 3.0311,
+      "step": 6105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.17805265852804e-05,
+      "loss": 3.0866,
+      "step": 6110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.177378931198965e-05,
+      "loss": 3.5669,
+      "step": 6115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.17670520386989e-05,
+      "loss": 2.942,
+      "step": 6120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176031476540815e-05,
+      "loss": 3.1499,
+      "step": 6125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175357749211739e-05,
+      "loss": 3.0345,
+      "step": 6130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174684021882664e-05,
+      "loss": 2.9467,
+      "step": 6135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174010294553589e-05,
+      "loss": 3.0161,
+      "step": 6140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173336567224513e-05,
+      "loss": 3.2259,
+      "step": 6145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172662839895437e-05,
+      "loss": 2.9918,
+      "step": 6150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171989112566362e-05,
+      "loss": 2.7739,
+      "step": 6155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171315385237287e-05,
+      "loss": 3.0321,
+      "step": 6160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.170641657908211e-05,
+      "loss": 3.378,
+      "step": 6165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169967930579136e-05,
+      "loss": 3.0588,
+      "step": 6170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169294203250061e-05,
+      "loss": 2.8973,
+      "step": 6175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168620475920986e-05,
+      "loss": 3.0668,
+      "step": 6180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16794674859191e-05,
+      "loss": 3.1018,
+      "step": 6185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.167273021262835e-05,
+      "loss": 2.729,
+      "step": 6190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.166599293933759e-05,
+      "loss": 3.2127,
+      "step": 6195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1659255666046834e-05,
+      "loss": 3.0852,
+      "step": 6200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.165251839275608e-05,
+      "loss": 2.9025,
+      "step": 6205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.164578111946533e-05,
+      "loss": 3.0034,
+      "step": 6210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.163904384617458e-05,
+      "loss": 2.9844,
+      "step": 6215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.163230657288382e-05,
+      "loss": 2.8532,
+      "step": 6220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.162556929959307e-05,
+      "loss": 3.0747,
+      "step": 6225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.161883202630232e-05,
+      "loss": 2.9875,
+      "step": 6230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.161209475301156e-05,
+      "loss": 3.0558,
+      "step": 6235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1605357479720805e-05,
+      "loss": 2.968,
+      "step": 6240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1598620206430054e-05,
+      "loss": 2.9673,
+      "step": 6245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15918829331393e-05,
+      "loss": 2.948,
+      "step": 6250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1585145659848545e-05,
+      "loss": 3.173,
+      "step": 6255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1578408386557794e-05,
+      "loss": 3.1657,
+      "step": 6260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157167111326704e-05,
+      "loss": 3.011,
+      "step": 6265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156493383997629e-05,
+      "loss": 3.2625,
+      "step": 6270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1558196566685534e-05,
+      "loss": 3.1201,
+      "step": 6275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1551459293394776e-05,
+      "loss": 3.258,
+      "step": 6280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1544722020104024e-05,
+      "loss": 3.428,
+      "step": 6285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.153798474681327e-05,
+      "loss": 3.257,
+      "step": 6290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1531247473522515e-05,
+      "loss": 3.2086,
+      "step": 6295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1524510200231764e-05,
+      "loss": 3.2408,
+      "step": 6300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151777292694101e-05,
+      "loss": 3.0592,
+      "step": 6305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1511035653650255e-05,
+      "loss": 3.2017,
+      "step": 6310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1504298380359504e-05,
+      "loss": 2.9562,
+      "step": 6315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.149756110706875e-05,
+      "loss": 3.2331,
+      "step": 6320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1490823833777995e-05,
+      "loss": 3.1174,
+      "step": 6325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.148408656048724e-05,
+      "loss": 3.0513,
+      "step": 6330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1477349287196486e-05,
+      "loss": 3.0752,
+      "step": 6335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1470612013905735e-05,
+      "loss": 2.9025,
+      "step": 6340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1463874740614984e-05,
+      "loss": 3.0994,
+      "step": 6345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1457137467324226e-05,
+      "loss": 3.1386,
+      "step": 6350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1450400194033475e-05,
+      "loss": 3.0328,
+      "step": 6355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1443662920742724e-05,
+      "loss": 2.7842,
+      "step": 6360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1436925647451966e-05,
+      "loss": 2.8813,
+      "step": 6365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.143018837416121e-05,
+      "loss": 3.3991,
+      "step": 6370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142345110087046e-05,
+      "loss": 3.0849,
+      "step": 6375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1416713827579706e-05,
+      "loss": 3.5411,
+      "step": 6380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140997655428895e-05,
+      "loss": 2.8933,
+      "step": 6385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1403239280998197e-05,
+      "loss": 2.9984,
+      "step": 6390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1396502007707445e-05,
+      "loss": 3.1318,
+      "step": 6395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138976473441669e-05,
+      "loss": 3.2986,
+      "step": 6400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1383027461125936e-05,
+      "loss": 2.9299,
+      "step": 6405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137629018783518e-05,
+      "loss": 3.0908,
+      "step": 6410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.136955291454443e-05,
+      "loss": 3.0864,
+      "step": 6415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.136281564125367e-05,
+      "loss": 2.9794,
+      "step": 6420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.135607836796292e-05,
+      "loss": 3.1128,
+      "step": 6425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134934109467217e-05,
+      "loss": 2.9951,
+      "step": 6430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1342603821381416e-05,
+      "loss": 3.187,
+      "step": 6435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.133586654809066e-05,
+      "loss": 3.2286,
+      "step": 6440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132912927479991e-05,
+      "loss": 2.9073,
+      "step": 6445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132239200150915e-05,
+      "loss": 2.9606,
+      "step": 6450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.13156547282184e-05,
+      "loss": 3.3927,
+      "step": 6455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.130891745492764e-05,
+      "loss": 2.9814,
+      "step": 6460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.130218018163689e-05,
+      "loss": 3.134,
+      "step": 6465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129544290834614e-05,
+      "loss": 3.1801,
+      "step": 6470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.128870563505538e-05,
+      "loss": 3.3085,
+      "step": 6475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.128196836176463e-05,
+      "loss": 3.1814,
+      "step": 6480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.127523108847388e-05,
+      "loss": 3.2744,
+      "step": 6485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.126849381518313e-05,
+      "loss": 2.9528,
+      "step": 6490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126175654189237e-05,
+      "loss": 3.1222,
+      "step": 6495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125501926860161e-05,
+      "loss": 2.8767,
+      "step": 6500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124828199531086e-05,
+      "loss": 2.9187,
+      "step": 6505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12415447220201e-05,
+      "loss": 3.1005,
+      "step": 6510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123480744872935e-05,
+      "loss": 3.0222,
+      "step": 6515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12280701754386e-05,
+      "loss": 3.0137,
+      "step": 6520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122133290214785e-05,
+      "loss": 3.1305,
+      "step": 6525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.121459562885709e-05,
+      "loss": 2.9975,
+      "step": 6530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120785835556634e-05,
+      "loss": 2.8521,
+      "step": 6535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120112108227558e-05,
+      "loss": 2.8654,
+      "step": 6540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119438380898483e-05,
+      "loss": 3.0614,
+      "step": 6545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118764653569407e-05,
+      "loss": 2.9413,
+      "step": 6550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118090926240332e-05,
+      "loss": 2.9269,
+      "step": 6555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.117417198911257e-05,
+      "loss": 3.1327,
+      "step": 6560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.116743471582181e-05,
+      "loss": 3.1815,
+      "step": 6565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.116069744253106e-05,
+      "loss": 3.3745,
+      "step": 6570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115396016924031e-05,
+      "loss": 3.0903,
+      "step": 6575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.114722289594955e-05,
+      "loss": 3.0028,
+      "step": 6580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1140485622658794e-05,
+      "loss": 3.1658,
+      "step": 6585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.113374834936804e-05,
+      "loss": 3.0588,
+      "step": 6590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.112701107607729e-05,
+      "loss": 2.9141,
+      "step": 6595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.112027380278654e-05,
+      "loss": 3.0221,
+      "step": 6600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.111353652949578e-05,
+      "loss": 3.0241,
+      "step": 6605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.110679925620503e-05,
+      "loss": 3.1218,
+      "step": 6610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.110006198291428e-05,
+      "loss": 3.1217,
+      "step": 6615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109332470962352e-05,
+      "loss": 2.8752,
+      "step": 6620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108658743633277e-05,
+      "loss": 3.3187,
+      "step": 6625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1079850163042014e-05,
+      "loss": 3.1058,
+      "step": 6630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107311288975126e-05,
+      "loss": 3.142,
+      "step": 6635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1066375616460505e-05,
+      "loss": 3.048,
+      "step": 6640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1059638343169754e-05,
+      "loss": 2.75,
+      "step": 6645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1052901069879e-05,
+      "loss": 3.0339,
+      "step": 6650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104616379658825e-05,
+      "loss": 3.113,
+      "step": 6655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1039426523297493e-05,
+      "loss": 2.8537,
+      "step": 6660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.103268925000674e-05,
+      "loss": 3.1454,
+      "step": 6665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1025951976715984e-05,
+      "loss": 2.9773,
+      "step": 6670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1019214703425227e-05,
+      "loss": 2.959,
+      "step": 6675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1012477430134475e-05,
+      "loss": 3.2781,
+      "step": 6680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1005740156843724e-05,
+      "loss": 3.3036,
+      "step": 6685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.099900288355297e-05,
+      "loss": 3.0932,
+      "step": 6690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0992265610262215e-05,
+      "loss": 2.9254,
+      "step": 6695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0985528336971464e-05,
+      "loss": 2.9804,
+      "step": 6700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.097879106368071e-05,
+      "loss": 2.8821,
+      "step": 6705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0972053790389955e-05,
+      "loss": 3.245,
+      "step": 6710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.09653165170992e-05,
+      "loss": 3.04,
+      "step": 6715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0958579243808446e-05,
+      "loss": 3.2381,
+      "step": 6720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0951841970517695e-05,
+      "loss": 3.176,
+      "step": 6725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.094510469722694e-05,
+      "loss": 3.1522,
+      "step": 6730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0938367423936186e-05,
+      "loss": 3.071,
+      "step": 6735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0931630150645435e-05,
+      "loss": 3.195,
+      "step": 6740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0924892877354684e-05,
+      "loss": 3.317,
+      "step": 6745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0918155604063926e-05,
+      "loss": 3.0664,
+      "step": 6750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0911418330773175e-05,
+      "loss": 3.04,
+      "step": 6755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090468105748242e-05,
+      "loss": 2.9613,
+      "step": 6760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.089794378419166e-05,
+      "loss": 3.2902,
+      "step": 6765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.089120651090091e-05,
+      "loss": 2.9059,
+      "step": 6770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0884469237610157e-05,
+      "loss": 3.0708,
+      "step": 6775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0877731964319405e-05,
+      "loss": 3.2298,
+      "step": 6780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08723421456868e-05,
+      "loss": 3.3366,
+      "step": 6785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086560487239605e-05,
+      "loss": 2.985,
+      "step": 6790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085886759910529e-05,
+      "loss": 2.9827,
+      "step": 6795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0852130325814534e-05,
+      "loss": 3.1584,
+      "step": 6800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084539305252378e-05,
+      "loss": 2.8316,
+      "step": 6805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083865577923303e-05,
+      "loss": 3.1549,
+      "step": 6810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083191850594228e-05,
+      "loss": 3.2837,
+      "step": 6815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.082518123265152e-05,
+      "loss": 2.9758,
+      "step": 6820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081844395936077e-05,
+      "loss": 3.2667,
+      "step": 6825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081170668607002e-05,
+      "loss": 3.1173,
+      "step": 6830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.080496941277926e-05,
+      "loss": 3.0736,
+      "step": 6835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0798232139488504e-05,
+      "loss": 2.9801,
+      "step": 6840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.079149486619775e-05,
+      "loss": 2.9754,
+      "step": 6845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0784757592907e-05,
+      "loss": 2.8494,
+      "step": 6850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0778020319616244e-05,
+      "loss": 3.2473,
+      "step": 6855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.077128304632549e-05,
+      "loss": 2.8755,
+      "step": 6860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.076454577303474e-05,
+      "loss": 3.0215,
+      "step": 6865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075780849974399e-05,
+      "loss": 2.9058,
+      "step": 6870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075107122645323e-05,
+      "loss": 3.2889,
+      "step": 6875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0744333953162475e-05,
+      "loss": 2.8474,
+      "step": 6880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0737596679871724e-05,
+      "loss": 2.9836,
+      "step": 6885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0730859406580966e-05,
+      "loss": 2.9131,
+      "step": 6890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0724122133290215e-05,
+      "loss": 2.9475,
+      "step": 6895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0717384859999464e-05,
+      "loss": 3.168,
+      "step": 6900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071064758670871e-05,
+      "loss": 3.2896,
+      "step": 6905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0703910313417955e-05,
+      "loss": 2.8471,
+      "step": 6910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.06971730401272e-05,
+      "loss": 2.9173,
+      "step": 6915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069043576683645e-05,
+      "loss": 3.1674,
+      "step": 6920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0683698493545694e-05,
+      "loss": 3.1038,
+      "step": 6925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0676961220254936e-05,
+      "loss": 3.0604,
+      "step": 6930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0670223946964185e-05,
+      "loss": 2.9545,
+      "step": 6935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0663486673673434e-05,
+      "loss": 3.1475,
+      "step": 6940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0656749400382676e-05,
+      "loss": 2.9257,
+      "step": 6945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0650012127091925e-05,
+      "loss": 2.6838,
+      "step": 6950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0643274853801174e-05,
+      "loss": 3.1874,
+      "step": 6955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063653758051042e-05,
+      "loss": 3.2852,
+      "step": 6960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0629800307219665e-05,
+      "loss": 2.8721,
+      "step": 6965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062306303392891e-05,
+      "loss": 3.1328,
+      "step": 6970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0616325760638156e-05,
+      "loss": 3.207,
+      "step": 6975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0609588487347405e-05,
+      "loss": 3.2351,
+      "step": 6980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.060285121405665e-05,
+      "loss": 3.0004,
+      "step": 6985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0596113940765896e-05,
+      "loss": 3.0114,
+      "step": 6990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0589376667475145e-05,
+      "loss": 3.3165,
+      "step": 6995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058263939418439e-05,
+      "loss": 3.212,
+      "step": 7000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0575902120893636e-05,
+      "loss": 2.9537,
+      "step": 7005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056916484760288e-05,
+      "loss": 3.0058,
+      "step": 7010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056242757431213e-05,
+      "loss": 2.9632,
+      "step": 7015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055569030102137e-05,
+      "loss": 3.0076,
+      "step": 7020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054895302773062e-05,
+      "loss": 3.118,
+      "step": 7025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0542215754439866e-05,
+      "loss": 3.0387,
+      "step": 7030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0535478481149115e-05,
+      "loss": 3.0755,
+      "step": 7035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053008866251651e-05,
+      "loss": 3.5368,
+      "step": 7040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052335138922575e-05,
+      "loss": 2.8257,
+      "step": 7045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0516614115935e-05,
+      "loss": 3.0709,
+      "step": 7050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0509876842644243e-05,
+      "loss": 2.8406,
+      "step": 7055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050313956935349e-05,
+      "loss": 3.1349,
+      "step": 7060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049640229606274e-05,
+      "loss": 2.9948,
+      "step": 7065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048966502277198e-05,
+      "loss": 2.9537,
+      "step": 7070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048292774948123e-05,
+      "loss": 2.7327,
+      "step": 7075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047619047619048e-05,
+      "loss": 2.9098,
+      "step": 7080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046945320289972e-05,
+      "loss": 3.1846,
+      "step": 7085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046271592960897e-05,
+      "loss": 3.046,
+      "step": 7090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0455978656318214e-05,
+      "loss": 3.0464,
+      "step": 7095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044924138302746e-05,
+      "loss": 3.3597,
+      "step": 7100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044250410973671e-05,
+      "loss": 3.1111,
+      "step": 7105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0435766836445954e-05,
+      "loss": 3.187,
+      "step": 7110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.04290295631552e-05,
+      "loss": 2.8657,
+      "step": 7115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042229228986445e-05,
+      "loss": 3.3056,
+      "step": 7120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0415555016573694e-05,
+      "loss": 2.8097,
+      "step": 7125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040881774328294e-05,
+      "loss": 3.078,
+      "step": 7130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0402080469992185e-05,
+      "loss": 3.1158,
+      "step": 7135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0395343196701434e-05,
+      "loss": 2.904,
+      "step": 7140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0388605923410676e-05,
+      "loss": 3.0502,
+      "step": 7145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0381868650119925e-05,
+      "loss": 3.0258,
+      "step": 7150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0375131376829173e-05,
+      "loss": 3.079,
+      "step": 7155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.036839410353842e-05,
+      "loss": 3.3828,
+      "step": 7160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0361656830247664e-05,
+      "loss": 3.3249,
+      "step": 7165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035491955695691e-05,
+      "loss": 3.1637,
+      "step": 7170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0348182283666155e-05,
+      "loss": 3.3418,
+      "step": 7175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03414450103754e-05,
+      "loss": 3.0859,
+      "step": 7180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0334707737084646e-05,
+      "loss": 3.1996,
+      "step": 7185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0327970463793895e-05,
+      "loss": 3.1428,
+      "step": 7190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0321233190503144e-05,
+      "loss": 3.2682,
+      "step": 7195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0314495917212386e-05,
+      "loss": 2.856,
+      "step": 7200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0307758643921635e-05,
+      "loss": 3.2575,
+      "step": 7205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0301021370630884e-05,
+      "loss": 3.1416,
+      "step": 7210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0294284097340126e-05,
+      "loss": 2.8951,
+      "step": 7215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028754682404937e-05,
+      "loss": 3.0688,
+      "step": 7220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028080955075862e-05,
+      "loss": 2.9216,
+      "step": 7225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0274072277467866e-05,
+      "loss": 3.0731,
+      "step": 7230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.026733500417711e-05,
+      "loss": 2.8017,
+      "step": 7235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026059773088636e-05,
+      "loss": 3.1149,
+      "step": 7240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0253860457595606e-05,
+      "loss": 2.9981,
+      "step": 7245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0247123184304855e-05,
+      "loss": 3.094,
+      "step": 7250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02403859110141e-05,
+      "loss": 3.0985,
+      "step": 7255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0233648637723346e-05,
+      "loss": 3.1172,
+      "step": 7260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022691136443259e-05,
+      "loss": 3.1349,
+      "step": 7265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022017409114183e-05,
+      "loss": 2.9795,
+      "step": 7270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021343681785108e-05,
+      "loss": 3.1452,
+      "step": 7275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020669954456033e-05,
+      "loss": 3.0341,
+      "step": 7280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0199962271269576e-05,
+      "loss": 3.3302,
+      "step": 7285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.019322499797882e-05,
+      "loss": 3.282,
+      "step": 7290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018648772468807e-05,
+      "loss": 3.1143,
+      "step": 7295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0179750451397316e-05,
+      "loss": 3.047,
+      "step": 7300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.017301317810656e-05,
+      "loss": 2.9338,
+      "step": 7305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01662759048158e-05,
+      "loss": 3.0096,
+      "step": 7310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015953863152505e-05,
+      "loss": 2.8297,
+      "step": 7315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01528013582343e-05,
+      "loss": 3.3948,
+      "step": 7320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.014606408494354e-05,
+      "loss": 3.1253,
+      "step": 7325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013932681165279e-05,
+      "loss": 3.2223,
+      "step": 7330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013258953836204e-05,
+      "loss": 2.956,
+      "step": 7335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012585226507129e-05,
+      "loss": 3.335,
+      "step": 7340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011911499178053e-05,
+      "loss": 3.1048,
+      "step": 7345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011237771848977e-05,
+      "loss": 3.0714,
+      "step": 7350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.010564044519902e-05,
+      "loss": 3.0881,
+      "step": 7355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009890317190827e-05,
+      "loss": 2.8853,
+      "step": 7360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009216589861751e-05,
+      "loss": 2.93,
+      "step": 7365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008542862532676e-05,
+      "loss": 3.0098,
+      "step": 7370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007869135203601e-05,
+      "loss": 2.8795,
+      "step": 7375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007195407874525e-05,
+      "loss": 3.1682,
+      "step": 7380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00652168054545e-05,
+      "loss": 2.9576,
+      "step": 7385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005847953216375e-05,
+      "loss": 3.1748,
+      "step": 7390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005174225887299e-05,
+      "loss": 2.9084,
+      "step": 7395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004500498558223e-05,
+      "loss": 3.3471,
+      "step": 7400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003826771229148e-05,
+      "loss": 3.0671,
+      "step": 7405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003153043900073e-05,
+      "loss": 2.7733,
+      "step": 7410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.002479316570998e-05,
+      "loss": 2.9595,
+      "step": 7415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001805589241922e-05,
+      "loss": 3.3316,
+      "step": 7420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001131861912847e-05,
+      "loss": 2.9916,
+      "step": 7425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.000458134583772e-05,
+      "loss": 3.11,
+      "step": 7430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999784407254696e-05,
+      "loss": 2.9916,
+      "step": 7435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.99911067992562e-05,
+      "loss": 2.9348,
+      "step": 7440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.998436952596545e-05,
+      "loss": 3.1977,
+      "step": 7445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.99776322526747e-05,
+      "loss": 3.1294,
+      "step": 7450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.997089497938394e-05,
+      "loss": 2.9127,
+      "step": 7455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996415770609319e-05,
+      "loss": 2.996,
+      "step": 7460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.995742043280244e-05,
+      "loss": 3.0576,
+      "step": 7465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.995068315951169e-05,
+      "loss": 3.1013,
+      "step": 7470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.994394588622093e-05,
+      "loss": 2.8199,
+      "step": 7475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9937208612930174e-05,
+      "loss": 3.0425,
+      "step": 7480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993047133963942e-05,
+      "loss": 2.7993,
+      "step": 7485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9923734066348665e-05,
+      "loss": 3.192,
+      "step": 7490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9916996793057914e-05,
+      "loss": 2.9459,
+      "step": 7495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991025951976716e-05,
+      "loss": 3.0381,
+      "step": 7500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990352224647641e-05,
+      "loss": 3.3137,
+      "step": 7505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9896784973185654e-05,
+      "loss": 2.9695,
+      "step": 7510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98900476998949e-05,
+      "loss": 2.8945,
+      "step": 7515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9883310426604145e-05,
+      "loss": 3.0771,
+      "step": 7520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987657315331339e-05,
+      "loss": 2.8429,
+      "step": 7525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9869835880022636e-05,
+      "loss": 3.35,
+      "step": 7530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986444606139004e-05,
+      "loss": 3.2072,
+      "step": 7535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9857708788099286e-05,
+      "loss": 3.4036,
+      "step": 7540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985097151480853e-05,
+      "loss": 2.8693,
+      "step": 7545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.984423424151778e-05,
+      "loss": 3.1092,
+      "step": 7550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.983749696822702e-05,
+      "loss": 2.8871,
+      "step": 7555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.983075969493627e-05,
+      "loss": 2.8921,
+      "step": 7560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.982402242164551e-05,
+      "loss": 3.216,
+      "step": 7565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981728514835476e-05,
+      "loss": 3.0464,
+      "step": 7570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981054787506401e-05,
+      "loss": 2.952,
+      "step": 7575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.980381060177325e-05,
+      "loss": 2.8887,
+      "step": 7580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.97970733284825e-05,
+      "loss": 3.2435,
+      "step": 7585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979033605519175e-05,
+      "loss": 3.0364,
+      "step": 7590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9783598781901e-05,
+      "loss": 3.008,
+      "step": 7595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.977686150861024e-05,
+      "loss": 2.9321,
+      "step": 7600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.977012423531948e-05,
+      "loss": 3.0257,
+      "step": 7605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976338696202873e-05,
+      "loss": 3.3022,
+      "step": 7610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.975664968873797e-05,
+      "loss": 3.286,
+      "step": 7615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974991241544722e-05,
+      "loss": 3.1355,
+      "step": 7620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974317514215647e-05,
+      "loss": 3.1771,
+      "step": 7625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973643786886572e-05,
+      "loss": 3.1195,
+      "step": 7630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972970059557496e-05,
+      "loss": 2.8496,
+      "step": 7635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972296332228421e-05,
+      "loss": 3.3008,
+      "step": 7640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971622604899345e-05,
+      "loss": 3.0312,
+      "step": 7645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9709488775702694e-05,
+      "loss": 3.4856,
+      "step": 7650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.970275150241194e-05,
+      "loss": 3.1483,
+      "step": 7655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969601422912119e-05,
+      "loss": 2.8638,
+      "step": 7660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968927695583044e-05,
+      "loss": 3.1239,
+      "step": 7665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968253968253968e-05,
+      "loss": 3.0082,
+      "step": 7670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.967580240924893e-05,
+      "loss": 2.9481,
+      "step": 7675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.966906513595818e-05,
+      "loss": 2.9585,
+      "step": 7680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.966232786266742e-05,
+      "loss": 2.8524,
+      "step": 7685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9655590589376664e-05,
+      "loss": 3.0689,
+      "step": 7690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964885331608591e-05,
+      "loss": 3.0235,
+      "step": 7695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964211604279516e-05,
+      "loss": 3.1008,
+      "step": 7700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9635378769504404e-05,
+      "loss": 3.1541,
+      "step": 7705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.962864149621365e-05,
+      "loss": 3.2348,
+      "step": 7710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.96219042229229e-05,
+      "loss": 3.0871,
+      "step": 7715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.961516694963215e-05,
+      "loss": 2.8276,
+      "step": 7720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.960842967634139e-05,
+      "loss": 2.9168,
+      "step": 7725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.960169240305064e-05,
+      "loss": 3.037,
+      "step": 7730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9594955129759884e-05,
+      "loss": 3.1009,
+      "step": 7735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958821785646913e-05,
+      "loss": 2.8545,
+      "step": 7740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9581480583178375e-05,
+      "loss": 3.2073,
+      "step": 7745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9574743309887624e-05,
+      "loss": 3.1771,
+      "step": 7750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956800603659687e-05,
+      "loss": 3.0321,
+      "step": 7755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9561268763306115e-05,
+      "loss": 2.9031,
+      "step": 7760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9554531490015364e-05,
+      "loss": 2.9989,
+      "step": 7765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954779421672461e-05,
+      "loss": 2.9791,
+      "step": 7770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9541056943433855e-05,
+      "loss": 3.1055,
+      "step": 7775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95343196701431e-05,
+      "loss": 3.2512,
+      "step": 7780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9527582396852346e-05,
+      "loss": 2.939,
+      "step": 7785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9520845123561594e-05,
+      "loss": 3.0405,
+      "step": 7790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.951410785027084e-05,
+      "loss": 3.4549,
+      "step": 7795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9507370576980085e-05,
+      "loss": 3.0582,
+      "step": 7800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9500633303689334e-05,
+      "loss": 2.8642,
+      "step": 7805
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.949389603039858e-05,
+      "loss": 2.9274,
+      "step": 7810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9487158757107825e-05,
+      "loss": 3.3588,
+      "step": 7815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.948042148381707e-05,
+      "loss": 3.177,
+      "step": 7820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9473684210526316e-05,
+      "loss": 2.9691,
+      "step": 7825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9466946937235565e-05,
+      "loss": 3.4174,
+      "step": 7830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.946020966394481e-05,
+      "loss": 2.9425,
+      "step": 7835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9453472390654056e-05,
+      "loss": 2.8462,
+      "step": 7840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9446735117363305e-05,
+      "loss": 3.307,
+      "step": 7845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9439997844072554e-05,
+      "loss": 2.9172,
+      "step": 7850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9433260570781796e-05,
+      "loss": 2.9545,
+      "step": 7855
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9426523297491045e-05,
+      "loss": 3.1286,
+      "step": 7860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941978602420029e-05,
+      "loss": 2.924,
+      "step": 7865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941304875090953e-05,
+      "loss": 2.8875,
+      "step": 7870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940631147761878e-05,
+      "loss": 3.1005,
+      "step": 7875
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939957420432803e-05,
+      "loss": 2.9367,
+      "step": 7880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9392836931037276e-05,
+      "loss": 2.9924,
+      "step": 7885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.938609965774652e-05,
+      "loss": 3.1627,
+      "step": 7890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9379362384455767e-05,
+      "loss": 2.8434,
+      "step": 7895
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9372625111165015e-05,
+      "loss": 3.1066,
+      "step": 7900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936588783787426e-05,
+      "loss": 3.3607,
+      "step": 7905
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93591505645835e-05,
+      "loss": 2.9483,
+      "step": 7910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935241329129275e-05,
+      "loss": 3.1395,
+      "step": 7915
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9345676018002e-05,
+      "loss": 2.8039,
+      "step": 7920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.933893874471124e-05,
+      "loss": 2.9289,
+      "step": 7925
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.933220147142049e-05,
+      "loss": 3.1207,
+      "step": 7930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.932546419812974e-05,
+      "loss": 2.956,
+      "step": 7935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9318726924838986e-05,
+      "loss": 3.1368,
+      "step": 7940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.931198965154823e-05,
+      "loss": 3.0236,
+      "step": 7945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.930525237825747e-05,
+      "loss": 3.1166,
+      "step": 7950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.929851510496672e-05,
+      "loss": 2.8586,
+      "step": 7955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.929177783167596e-05,
+      "loss": 3.1254,
+      "step": 7960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.928504055838521e-05,
+      "loss": 2.9588,
+      "step": 7965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.927830328509446e-05,
+      "loss": 2.9815,
+      "step": 7970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.927156601180371e-05,
+      "loss": 3.0071,
+      "step": 7975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926482873851295e-05,
+      "loss": 3.0211,
+      "step": 7980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.92580914652222e-05,
+      "loss": 2.8963,
+      "step": 7985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925135419193144e-05,
+      "loss": 3.0395,
+      "step": 7990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924461691864069e-05,
+      "loss": 2.8795,
+      "step": 7995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923787964534993e-05,
+      "loss": 3.0102,
+      "step": 8000
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 3.0175957679748535,
+      "eval_rouge2_fmeasure": 0.0061,
+      "eval_rouge2_precision": 0.0093,
+      "eval_rouge2_recall": 0.0066,
+      "eval_runtime": 2672.3133,
+      "eval_samples_per_second": 0.103,
+      "eval_steps_per_second": 0.052,
+      "step": 8000
     }
   ],
   "max_steps": 37107,
   "num_train_epochs": 3,
-  "total_flos": 1.56090117390336e+17,
+  "total_flos": 3.12180234780672e+17,
   "trial_name": null,
   "trial_params": null
 }