diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,115221 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.96,
+  "eval_steps": 500,
+  "global_step": 153600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1875e-06,
+      "loss": 3.7908,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 3.244,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.187499999999999e-06,
+      "loss": 2.6806,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.6875e-06,
+      "loss": 2.5457,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.21875e-05,
+      "loss": 2.5046,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4687500000000001e-05,
+      "loss": 2.4884,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.71875e-05,
+      "loss": 2.4491,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.96875e-05,
+      "loss": 2.459,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.21875e-05,
+      "loss": 2.4546,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4687500000000004e-05,
+      "loss": 2.4491,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.71875e-05,
+      "loss": 2.431,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.96875e-05,
+      "loss": 2.4526,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.21875e-05,
+      "loss": 2.4363,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.46875e-05,
+      "loss": 2.4337,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.71875e-05,
+      "loss": 2.437,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.96875e-05,
+      "loss": 2.43,
+      "step": 128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.21875e-05,
+      "loss": 2.4237,
+      "step": 136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.46875e-05,
+      "loss": 2.4397,
+      "step": 144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.71875e-05,
+      "loss": 2.4258,
+      "step": 152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.96875e-05,
+      "loss": 2.4191,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.21875e-05,
+      "loss": 2.4195,
+      "step": 168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.46875e-05,
+      "loss": 2.4165,
+      "step": 176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.718750000000001e-05,
+      "loss": 2.4126,
+      "step": 184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.968750000000001e-05,
+      "loss": 2.4364,
+      "step": 192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.21875e-05,
+      "loss": 2.4099,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.468750000000001e-05,
+      "loss": 2.4278,
+      "step": 208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.71875e-05,
+      "loss": 2.4437,
+      "step": 216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.96875e-05,
+      "loss": 2.4319,
+      "step": 224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.218750000000001e-05,
+      "loss": 2.4429,
+      "step": 232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.468749999999999e-05,
+      "loss": 2.4255,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.71875e-05,
+      "loss": 2.4345,
+      "step": 248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.96875e-05,
+      "loss": 2.4152,
+      "step": 256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.218750000000001e-05,
+      "loss": 2.4331,
+      "step": 264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.46875e-05,
+      "loss": 2.4294,
+      "step": 272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.71875e-05,
+      "loss": 2.4323,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.96875e-05,
+      "loss": 2.4397,
+      "step": 288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.21875e-05,
+      "loss": 2.4432,
+      "step": 296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.46875e-05,
+      "loss": 2.4435,
+      "step": 304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.718750000000001e-05,
+      "loss": 2.4259,
+      "step": 312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.96875e-05,
+      "loss": 2.433,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010218750000000001,
+      "loss": 2.4046,
+      "step": 328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001046875,
+      "loss": 2.4232,
+      "step": 336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001071875,
+      "loss": 2.4301,
+      "step": 344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001096875,
+      "loss": 2.4404,
+      "step": 352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001121875,
+      "loss": 2.4471,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011468750000000002,
+      "loss": 2.4418,
+      "step": 368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011718750000000001,
+      "loss": 2.4427,
+      "step": 376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001196875,
+      "loss": 2.4383,
+      "step": 384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001221875,
+      "loss": 2.4211,
+      "step": 392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001246875,
+      "loss": 2.4363,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012718750000000002,
+      "loss": 2.4359,
+      "step": 408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001296875,
+      "loss": 2.4287,
+      "step": 416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001321875,
+      "loss": 2.4371,
+      "step": 424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013468750000000001,
+      "loss": 2.4317,
+      "step": 432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001371875,
+      "loss": 2.4417,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013968750000000003,
+      "loss": 2.4406,
+      "step": 448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001421875,
+      "loss": 2.448,
+      "step": 456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001446875,
+      "loss": 2.4444,
+      "step": 464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014718750000000002,
+      "loss": 2.4265,
+      "step": 472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001496875,
+      "loss": 2.4423,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001521875,
+      "loss": 2.4551,
+      "step": 488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001546875,
+      "loss": 2.4311,
+      "step": 496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001571875,
+      "loss": 2.4289,
+      "step": 504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001596875,
+      "loss": 2.4416,
+      "step": 512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001621875,
+      "loss": 2.4463,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001646875,
+      "loss": 2.4513,
+      "step": 528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016718750000000002,
+      "loss": 2.4543,
+      "step": 536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001696875,
+      "loss": 2.4493,
+      "step": 544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001721875,
+      "loss": 2.4445,
+      "step": 552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001746875,
+      "loss": 2.4467,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017718750000000002,
+      "loss": 2.4626,
+      "step": 568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001796875,
+      "loss": 2.4415,
+      "step": 576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001821875,
+      "loss": 2.4567,
+      "step": 584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001846875,
+      "loss": 2.4502,
+      "step": 592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018718750000000002,
+      "loss": 2.4515,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018968750000000002,
+      "loss": 2.4469,
+      "step": 608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001921875,
+      "loss": 2.4481,
+      "step": 616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001946875,
+      "loss": 2.4434,
+      "step": 624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019718750000000002,
+      "loss": 2.4621,
+      "step": 632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001996875,
+      "loss": 2.4592,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999999548675498,
+      "loss": 2.4773,
+      "step": 648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999998194702038,
+      "loss": 2.4694,
+      "step": 656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999995938079756,
+      "loss": 2.4468,
+      "step": 664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999999277880888,
+      "loss": 2.4647,
+      "step": 672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999988716889723,
+      "loss": 2.4333,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999983752322698,
+      "loss": 2.4634,
+      "step": 688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999977885108293,
+      "loss": 2.462,
+      "step": 696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999971115247108,
+      "loss": 2.4678,
+      "step": 704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999963442739814,
+      "loss": 2.4698,
+      "step": 712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999954867587185,
+      "loss": 2.4364,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999945389790079,
+      "loss": 2.4617,
+      "step": 728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999993500934945,
+      "loss": 2.46,
+      "step": 736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999923726266329,
+      "loss": 2.4694,
+      "step": 744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999991154054186,
+      "loss": 2.4677,
+      "step": 752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999898452177257,
+      "loss": 2.4403,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999884461173836,
+      "loss": 2.4546,
+      "step": 768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999869567533,
+      "loss": 2.4709,
+      "step": 776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999853771256242,
+      "loss": 2.4434,
+      "step": 784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999837072345148,
+      "loss": 2.456,
+      "step": 792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999981947080139,
+      "loss": 2.4698,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999800966626736,
+      "loss": 2.4417,
+      "step": 808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999781559823044,
+      "loss": 2.4621,
+      "step": 816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999976125039225,
+      "loss": 2.4566,
+      "step": 824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199997400383364,
+      "loss": 2.4626,
+      "step": 832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999717923657623,
+      "loss": 2.4658,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999969490635813,
+      "loss": 2.452,
+      "step": 848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999670986440234,
+      "loss": 2.4648,
+      "step": 856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999646163906335,
+      "loss": 2.4578,
+      "step": 864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999620438758918,
+      "loss": 2.4739,
+      "step": 872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999959381100057,
+      "loss": 2.4797,
+      "step": 880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999566280633953,
+      "loss": 2.4588,
+      "step": 888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999537847661833,
+      "loss": 2.4708,
+      "step": 896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999508512087062,
+      "loss": 2.4584,
+      "step": 904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999947827391258,
+      "loss": 2.4612,
+      "step": 912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999447133141422,
+      "loss": 2.4454,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999415089776712,
+      "loss": 2.4417,
+      "step": 928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999382143821658,
+      "loss": 2.452,
+      "step": 936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999934829527957,
+      "loss": 2.4647,
+      "step": 944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999313544153844,
+      "loss": 2.4687,
+      "step": 952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999927789044796,
+      "loss": 2.4574,
+      "step": 960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999241334165495,
+      "loss": 2.4513,
+      "step": 968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999203875310117,
+      "loss": 2.4372,
+      "step": 976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999165513885582,
+      "loss": 2.4729,
+      "step": 984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999126249895742,
+      "loss": 2.4827,
+      "step": 992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999086083344526,
+      "loss": 2.4509,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999904501423597,
+      "loss": 2.4499,
+      "step": 1008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999900304257419,
+      "loss": 2.4495,
+      "step": 1016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998960168363396,
+      "loss": 2.4493,
+      "step": 1024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998916391607892,
+      "loss": 2.4783,
+      "step": 1032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999887171231206,
+      "loss": 2.4696,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998826130480387,
+      "loss": 2.458,
+      "step": 1048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998779646117443,
+      "loss": 2.4689,
+      "step": 1056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999873225922789,
+      "loss": 2.4648,
+      "step": 1064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998683969816482,
+      "loss": 2.4683,
+      "step": 1072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998634777888062,
+      "loss": 2.4603,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998584683447558,
+      "loss": 2.4625,
+      "step": 1088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998533686500005,
+      "loss": 2.4614,
+      "step": 1096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999848178705051,
+      "loss": 2.4478,
+      "step": 1104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998428985104277,
+      "loss": 2.4685,
+      "step": 1112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998375280666606,
+      "loss": 2.4549,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998320673742884,
+      "loss": 2.4505,
+      "step": 1128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998265164338584,
+      "loss": 2.4779,
+      "step": 1136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998208752459274,
+      "loss": 2.4484,
+      "step": 1144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998151438110616,
+      "loss": 2.4654,
+      "step": 1152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998093221298356,
+      "loss": 2.4586,
+      "step": 1160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998034102028327,
+      "loss": 2.4604,
+      "step": 1168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997974080306465,
+      "loss": 2.4624,
+      "step": 1176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999791315613879,
+      "loss": 2.4584,
+      "step": 1184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997851329531406,
+      "loss": 2.4749,
+      "step": 1192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997788600490524,
+      "loss": 2.4656,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997724969022428,
+      "loss": 2.47,
+      "step": 1208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199976604351335,
+      "loss": 2.4712,
+      "step": 1216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997594998830216,
+      "loss": 2.4586,
+      "step": 1224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997528660119134,
+      "loss": 2.4565,
+      "step": 1232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997461419006914,
+      "loss": 2.4474,
+      "step": 1240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997393275500292,
+      "loss": 2.4615,
+      "step": 1248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997324229606108,
+      "loss": 2.4584,
+      "step": 1256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997254281331283,
+      "loss": 2.4601,
+      "step": 1264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997183430682836,
+      "loss": 2.4461,
+      "step": 1272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997111677667875,
+      "loss": 2.4616,
+      "step": 1280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999703902229359,
+      "loss": 2.4537,
+      "step": 1288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019996965464567278,
+      "loss": 2.4584,
+      "step": 1296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199968910044963,
+      "loss": 2.4498,
+      "step": 1304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019996815642088137,
+      "loss": 2.4746,
+      "step": 1312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019996739377350345,
+      "loss": 2.4598,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999666221029057,
+      "loss": 2.4702,
+      "step": 1328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019996584140916557,
+      "loss": 2.4624,
+      "step": 1336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019996505169236125,
+      "loss": 2.447,
+      "step": 1344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999642529525721,
+      "loss": 2.472,
+      "step": 1352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999634451898781,
+      "loss": 2.4599,
+      "step": 1360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019996262840436032,
+      "loss": 2.4628,
+      "step": 1368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019996180259610063,
+      "loss": 2.4446,
+      "step": 1376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999609677651819,
+      "loss": 2.4503,
+      "step": 1384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999601239116879,
+      "loss": 2.4448,
+      "step": 1392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019995927103570317,
+      "loss": 2.4539,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019995840913731333,
+      "loss": 2.4493,
+      "step": 1408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019995753821660473,
+      "loss": 2.4523,
+      "step": 1416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019995665827366482,
+      "loss": 2.4513,
+      "step": 1424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999557693085818,
+      "loss": 2.4701,
+      "step": 1432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999548713214448,
+      "loss": 2.4596,
+      "step": 1440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019995396431234395,
+      "loss": 2.4566,
+      "step": 1448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999530482813702,
+      "loss": 2.4461,
+      "step": 1456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999521232286154,
+      "loss": 2.4557,
+      "step": 1464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019995118915417233,
+      "loss": 2.4603,
+      "step": 1472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019995024605813468,
+      "loss": 2.4615,
+      "step": 1480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019994929394059706,
+      "loss": 2.4428,
+      "step": 1488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999483328016549,
+      "loss": 2.4484,
+      "step": 1496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019994736264140466,
+      "loss": 2.4419,
+      "step": 1504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019994638345994362,
+      "loss": 2.4678,
+      "step": 1512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019994539525736998,
+      "loss": 2.4551,
+      "step": 1520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019994439803378282,
+      "loss": 2.4575,
+      "step": 1528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999433917892822,
+      "loss": 2.4426,
+      "step": 1536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019994237652396909,
+      "loss": 2.4511,
+      "step": 1544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999413522379452,
+      "loss": 2.4538,
+      "step": 1552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019994031893131334,
+      "loss": 2.4625,
+      "step": 1560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999392766041771,
+      "loss": 2.4645,
+      "step": 1568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019993822525664107,
+      "loss": 2.4668,
+      "step": 1576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019993716488881063,
+      "loss": 2.4399,
+      "step": 1584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019993609550079217,
+      "loss": 2.4444,
+      "step": 1592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019993501709269295,
+      "loss": 2.4504,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019993392966462112,
+      "loss": 2.4444,
+      "step": 1608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019993283321668572,
+      "loss": 2.457,
+      "step": 1616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019993172774899676,
+      "loss": 2.4689,
+      "step": 1624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999306132616651,
+      "loss": 2.4495,
+      "step": 1632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019992948975480244,
+      "loss": 2.4523,
+      "step": 1640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999283572285216,
+      "loss": 2.4741,
+      "step": 1648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019992721568293606,
+      "loss": 2.4476,
+      "step": 1656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999260651181603,
+      "loss": 2.4605,
+      "step": 1664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019992490553430981,
+      "loss": 2.435,
+      "step": 1672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019992373693150084,
+      "loss": 2.451,
+      "step": 1680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019992255930985058,
+      "loss": 2.4393,
+      "step": 1688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019992137266947716,
+      "loss": 2.4474,
+      "step": 1696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019992017701049957,
+      "loss": 2.4181,
+      "step": 1704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019991897233303773,
+      "loss": 2.4535,
+      "step": 1712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999177586372125,
+      "loss": 2.4633,
+      "step": 1720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019991653592314556,
+      "loss": 2.4442,
+      "step": 1728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019991530419095956,
+      "loss": 2.4517,
+      "step": 1736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999140634407781,
+      "loss": 2.4664,
+      "step": 1744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019991281367272548,
+      "loss": 2.4554,
+      "step": 1752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019991155488692714,
+      "loss": 2.428,
+      "step": 1760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999102870835093,
+      "loss": 2.4654,
+      "step": 1768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999090102625992,
+      "loss": 2.451,
+      "step": 1776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019990772442432474,
+      "loss": 2.4292,
+      "step": 1784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199906429568815,
+      "loss": 2.4508,
+      "step": 1792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999051256961998,
+      "loss": 2.4439,
+      "step": 1800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019990381280660991,
+      "loss": 2.4451,
+      "step": 1808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019990249090017707,
+      "loss": 2.4462,
+      "step": 1816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019990115997703379,
+      "loss": 2.4566,
+      "step": 1824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989982003731354,
+      "loss": 2.4673,
+      "step": 1832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989847108115076,
+      "loss": 2.4467,
+      "step": 1840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989711310868072,
+      "loss": 2.4426,
+      "step": 1848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989574612003962,
+      "loss": 2.4495,
+      "step": 1856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998943701153646,
+      "loss": 2.434,
+      "step": 1864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989298509479357,
+      "loss": 2.4699,
+      "step": 1872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989159105846555,
+      "loss": 2.4647,
+      "step": 1880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989018800652025,
+      "loss": 2.4523,
+      "step": 1888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988877593909844,
+      "loss": 2.4627,
+      "step": 1896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988735485634179,
+      "loss": 2.45,
+      "step": 1904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988592475839273,
+      "loss": 2.4496,
+      "step": 1912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988448564539475,
+      "loss": 2.4534,
+      "step": 1920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998830375174922,
+      "loss": 2.4373,
+      "step": 1928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988158037483026,
+      "loss": 2.4457,
+      "step": 1936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988011421755514,
+      "loss": 2.4543,
+      "step": 1944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998786390458138,
+      "loss": 2.447,
+      "step": 1952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998771548597543,
+      "loss": 2.4593,
+      "step": 1960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998756616595254,
+      "loss": 2.454,
+      "step": 1968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019987415944527692,
+      "loss": 2.4317,
+      "step": 1976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998726482171595,
+      "loss": 2.4334,
+      "step": 1984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998711279753247,
+      "loss": 2.4591,
+      "step": 1992
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019986959871992502,
+      "loss": 2.449,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001998680604511138,
+      "loss": 2.4296,
+      "step": 2008
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019986651316904536,
+      "loss": 2.4508,
+      "step": 2016
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001998649568738748,
+      "loss": 2.464,
+      "step": 2024
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019986339156575833,
+      "loss": 2.4447,
+      "step": 2032
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019986181724485286,
+      "loss": 2.4509,
+      "step": 2040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019986023391131632,
+      "loss": 2.4673,
+      "step": 2048
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019985864156530747,
+      "loss": 2.4549,
+      "step": 2056
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019985704020698605,
+      "loss": 2.4507,
+      "step": 2064
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019985542983651269,
+      "loss": 2.4632,
+      "step": 2072
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019985381045404881,
+      "loss": 2.4354,
+      "step": 2080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001998521820597569,
+      "loss": 2.4405,
+      "step": 2088
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019985054465380028,
+      "loss": 2.44,
+      "step": 2096
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001998488982363431,
+      "loss": 2.4376,
+      "step": 2104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001998472428075506,
+      "loss": 2.4441,
+      "step": 2112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019984557836758872,
+      "loss": 2.4635,
+      "step": 2120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019984390491662447,
+      "loss": 2.4606,
+      "step": 2128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019984222245482557,
+      "loss": 2.4428,
+      "step": 2136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019984053098236087,
+      "loss": 2.4565,
+      "step": 2144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019983883049939994,
+      "loss": 2.4409,
+      "step": 2152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001998371210061134,
+      "loss": 2.455,
+      "step": 2160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019983540250267267,
+      "loss": 2.4195,
+      "step": 2168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001998336749892501,
+      "loss": 2.4555,
+      "step": 2176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019983193846601896,
+      "loss": 2.4474,
+      "step": 2184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019983019293315338,
+      "loss": 2.4574,
+      "step": 2192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019982843839082848,
+      "loss": 2.4332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019982667483922018,
+      "loss": 2.4547,
+      "step": 2208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001998249022785054,
+      "loss": 2.4366,
+      "step": 2216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019982312070886192,
+      "loss": 2.4435,
+      "step": 2224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019982133013046836,
+      "loss": 2.4627,
+      "step": 2232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019981953054350434,
+      "loss": 2.4434,
+      "step": 2240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019981772194815038,
+      "loss": 2.4292,
+      "step": 2248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019981590434458785,
+      "loss": 2.452,
+      "step": 2256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199814077732999,
+      "loss": 2.4529,
+      "step": 2264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019981224211356707,
+      "loss": 2.4578,
+      "step": 2272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019981039748647621,
+      "loss": 2.4521,
+      "step": 2280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019980854385191134,
+      "loss": 2.4473,
+      "step": 2288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001998066812100584,
+      "loss": 2.4263,
+      "step": 2296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019980480956110423,
+      "loss": 2.446,
+      "step": 2304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019980292890523654,
+      "loss": 2.439,
+      "step": 2312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019980103924264388,
+      "loss": 2.4562,
+      "step": 2320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019979914057351584,
+      "loss": 2.4409,
+      "step": 2328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019979723289804285,
+      "loss": 2.4439,
+      "step": 2336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001997953162164162,
+      "loss": 2.4548,
+      "step": 2344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019979339052882814,
+      "loss": 2.4491,
+      "step": 2352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019979145583547182,
+      "loss": 2.4437,
+      "step": 2360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019978951213654129,
+      "loss": 2.4304,
+      "step": 2368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019978755943223143,
+      "loss": 2.4417,
+      "step": 2376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019978559772273813,
+      "loss": 2.4714,
+      "step": 2384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019978362700825815,
+      "loss": 2.4345,
+      "step": 2392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001997816472889891,
+      "loss": 2.4435,
+      "step": 2400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019977965856512954,
+      "loss": 2.4371,
+      "step": 2408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199777660836879,
+      "loss": 2.4583,
+      "step": 2416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019977565410443776,
+      "loss": 2.4617,
+      "step": 2424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001997736383680071,
+      "loss": 2.4378,
+      "step": 2432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001997716136277892,
+      "loss": 2.4415,
+      "step": 2440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019976957988398713,
+      "loss": 2.4411,
+      "step": 2448
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019976753713680487,
+      "loss": 2.4488,
+      "step": 2456
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019976548538644729,
+      "loss": 2.4531,
+      "step": 2464
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019976342463312016,
+      "loss": 2.4361,
+      "step": 2472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019976135487703015,
+      "loss": 2.4234,
+      "step": 2480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019975927611838487,
+      "loss": 2.4434,
+      "step": 2488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001997571883573928,
+      "loss": 2.4423,
+      "step": 2496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019975509159426334,
+      "loss": 2.4479,
+      "step": 2504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019975298582920677,
+      "loss": 2.4344,
+      "step": 2512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019975087106243426,
+      "loss": 2.4431,
+      "step": 2520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199748747294158,
+      "loss": 2.4385,
+      "step": 2528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019974661452459088,
+      "loss": 2.4388,
+      "step": 2536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019974447275394687,
+      "loss": 2.4516,
+      "step": 2544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019974232198244078,
+      "loss": 2.4632,
+      "step": 2552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019974016221028826,
+      "loss": 2.4615,
+      "step": 2560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019973799343770603,
+      "loss": 2.4501,
+      "step": 2568
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001997358156649115,
+      "loss": 2.4451,
+      "step": 2576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019973362889212312,
+      "loss": 2.4358,
+      "step": 2584
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019973143311956024,
+      "loss": 2.4408,
+      "step": 2592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019972922834744308,
+      "loss": 2.4423,
+      "step": 2600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019972701457599272,
+      "loss": 2.4389,
+      "step": 2608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019972479180543123,
+      "loss": 2.4418,
+      "step": 2616
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019972256003598153,
+      "loss": 2.4507,
+      "step": 2624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019972031926786745,
+      "loss": 2.4392,
+      "step": 2632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019971806950131376,
+      "loss": 2.4504,
+      "step": 2640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019971581073654603,
+      "loss": 2.4225,
+      "step": 2648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019971354297379086,
+      "loss": 2.462,
+      "step": 2656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019971126621327564,
+      "loss": 2.4421,
+      "step": 2664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001997089804552288,
+      "loss": 2.4392,
+      "step": 2672
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019970668569987951,
+      "loss": 2.4402,
+      "step": 2680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019970438194745797,
+      "loss": 2.443,
+      "step": 2688
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019970206919819522,
+      "loss": 2.4299,
+      "step": 2696
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996997474523232,
+      "loss": 2.4463,
+      "step": 2704
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019969741671007477,
+      "loss": 2.4449,
+      "step": 2712
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996950769716837,
+      "loss": 2.4218,
+      "step": 2720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019969272823738466,
+      "loss": 2.4347,
+      "step": 2728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996903705074132,
+      "loss": 2.4463,
+      "step": 2736
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968800378200577,
+      "loss": 2.4388,
+      "step": 2744
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996856280613998,
+      "loss": 2.4269,
+      "step": 2752
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996832433458335,
+      "loss": 2.4495,
+      "step": 2760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968084963554604,
+      "loss": 2.4226,
+      "step": 2768
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967844693077758,
+      "loss": 2.4418,
+      "step": 2776
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967603523176898,
+      "loss": 2.4565,
+      "step": 2784
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996736145387622,
+      "loss": 2.4539,
+      "step": 2792
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199671184852,
+      "loss": 2.4393,
+      "step": 2800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001996687461717261,
+      "loss": 2.4553,
+      "step": 2808
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199666298498185,
+      "loss": 2.4382,
+      "step": 2816
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019966384183162225,
+      "loss": 2.427,
+      "step": 2824
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019966137617228422,
+      "loss": 2.4414,
+      "step": 2832
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019965890152041823,
+      "loss": 2.4332,
+      "step": 2840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019965641787627244,
+      "loss": 2.4351,
+      "step": 2848
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019965392524009599,
+      "loss": 2.4507,
+      "step": 2856
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001996514236121388,
+      "loss": 2.4253,
+      "step": 2864
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019964891299265188,
+      "loss": 2.45,
+      "step": 2872
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001996463933818869,
+      "loss": 2.4332,
+      "step": 2880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019964386478009665,
+      "loss": 2.444,
+      "step": 2888
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001996413271875347,
+      "loss": 2.44,
+      "step": 2896
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001996387806044556,
+      "loss": 2.4632,
+      "step": 2904
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019963622503111467,
+      "loss": 2.4233,
+      "step": 2912
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019963366046776834,
+      "loss": 2.4413,
+      "step": 2920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019963108691467373,
+      "loss": 2.4417,
+      "step": 2928
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019962850437208897,
+      "loss": 2.4445,
+      "step": 2936
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001996259128402731,
+      "loss": 2.4397,
+      "step": 2944
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019962331231948604,
+      "loss": 2.4274,
+      "step": 2952
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019962070280998858,
+      "loss": 2.4464,
+      "step": 2960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001996180843120424,
+      "loss": 2.4552,
+      "step": 2968
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001996154568259102,
+      "loss": 2.4477,
+      "step": 2976
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019961282035185546,
+      "loss": 2.4465,
+      "step": 2984
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019961017489014265,
+      "loss": 2.4439,
+      "step": 2992
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199607520441037,
+      "loss": 2.4297,
+      "step": 3000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019960485700480486,
+      "loss": 2.4379,
+      "step": 3008
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019960218458171323,
+      "loss": 2.4388,
+      "step": 3016
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019959950317203023,
+      "loss": 2.4465,
+      "step": 3024
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019959681277602475,
+      "loss": 2.4364,
+      "step": 3032
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019959411339396665,
+      "loss": 2.4373,
+      "step": 3040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019959140502612664,
+      "loss": 2.4383,
+      "step": 3048
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019958868767277634,
+      "loss": 2.4286,
+      "step": 3056
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019958596133418831,
+      "loss": 2.4224,
+      "step": 3064
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199583226010636,
+      "loss": 2.4136,
+      "step": 3072
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019958048170239372,
+      "loss": 2.4579,
+      "step": 3080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019957772840973668,
+      "loss": 2.4361,
+      "step": 3088
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019957496613294112,
+      "loss": 2.4405,
+      "step": 3096
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019957219487228397,
+      "loss": 2.452,
+      "step": 3104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019956941462804322,
+      "loss": 2.4343,
+      "step": 3112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019956662540049773,
+      "loss": 2.4397,
+      "step": 3120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019956382718992722,
+      "loss": 2.446,
+      "step": 3128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019956101999661234,
+      "loss": 2.4348,
+      "step": 3136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001995582038208346,
+      "loss": 2.435,
+      "step": 3144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019955537866287655,
+      "loss": 2.4519,
+      "step": 3152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019955254452302143,
+      "loss": 2.4423,
+      "step": 3160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019954970140155355,
+      "loss": 2.4262,
+      "step": 3168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199546849298758,
+      "loss": 2.4474,
+      "step": 3176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019954398821492086,
+      "loss": 2.4424,
+      "step": 3184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001995411181503291,
+      "loss": 2.4477,
+      "step": 3192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019953823910527058,
+      "loss": 2.4423,
+      "step": 3200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199535351080034,
+      "loss": 2.4388,
+      "step": 3208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019953245407490906,
+      "loss": 2.433,
+      "step": 3216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019952954809018627,
+      "loss": 2.4421,
+      "step": 3224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019952663312615716,
+      "loss": 2.4444,
+      "step": 3232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019952370918311402,
+      "loss": 2.4207,
+      "step": 3240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019952077626135005,
+      "loss": 2.4408,
+      "step": 3248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019951783436115953,
+      "loss": 2.4483,
+      "step": 3256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019951488348283745,
+      "loss": 2.4304,
+      "step": 3264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001995119236266798,
+      "loss": 2.4315,
+      "step": 3272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001995089547929834,
+      "loss": 2.4186,
+      "step": 3280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199505976982046,
+      "loss": 2.4568,
+      "step": 3288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001995029901941663,
+      "loss": 2.4397,
+      "step": 3296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019949999442964384,
+      "loss": 2.413,
+      "step": 3304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019949698968877902,
+      "loss": 2.4338,
+      "step": 3312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019949397597187333,
+      "loss": 2.4338,
+      "step": 3320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001994909532792289,
+      "loss": 2.4406,
+      "step": 3328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019948792161114895,
+      "loss": 2.4424,
+      "step": 3336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019948488096793753,
+      "loss": 2.4435,
+      "step": 3344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019948183134989962,
+      "loss": 2.4157,
+      "step": 3352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199478772757341,
+      "loss": 2.4086,
+      "step": 3360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019947570519056855,
+      "loss": 2.4379,
+      "step": 3368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019947262864988984,
+      "loss": 2.4355,
+      "step": 3376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019946954313561347,
+      "loss": 2.4351,
+      "step": 3384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019946644864804886,
+      "loss": 2.4495,
+      "step": 3392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001994633451875064,
+      "loss": 2.4345,
+      "step": 3400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019946023275429735,
+      "loss": 2.4333,
+      "step": 3408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001994571113487339,
+      "loss": 2.4329,
+      "step": 3416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019945398097112905,
+      "loss": 2.4431,
+      "step": 3424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019945084162179675,
+      "loss": 2.4366,
+      "step": 3432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001994476933010519,
+      "loss": 2.4247,
+      "step": 3440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019944453600921028,
+      "loss": 2.4589,
+      "step": 3448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019944136974658848,
+      "loss": 2.4422,
+      "step": 3456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019943819451350415,
+      "loss": 2.4318,
+      "step": 3464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019943501031027564,
+      "loss": 2.4404,
+      "step": 3472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001994318171372224,
+      "loss": 2.4272,
+      "step": 3480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019942861499466465,
+      "loss": 2.4378,
+      "step": 3488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019942540388292353,
+      "loss": 2.445,
+      "step": 3496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019942218380232112,
+      "loss": 2.4262,
+      "step": 3504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019941895475318039,
+      "loss": 2.4483,
+      "step": 3512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019941571673582516,
+      "loss": 2.4484,
+      "step": 3520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019941246975058017,
+      "loss": 2.3924,
+      "step": 3528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001994092137977711,
+      "loss": 2.4438,
+      "step": 3536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019940594887772454,
+      "loss": 2.4381,
+      "step": 3544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001994026749907679,
+      "loss": 2.426,
+      "step": 3552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019939939213722957,
+      "loss": 2.4433,
+      "step": 3560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019939610031743872,
+      "loss": 2.4425,
+      "step": 3568
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019939279953172558,
+      "loss": 2.4255,
+      "step": 3576
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019938948978042117,
+      "loss": 2.4279,
+      "step": 3584
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019938617106385746,
+      "loss": 2.4457,
+      "step": 3592
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019938284338236726,
+      "loss": 2.4399,
+      "step": 3600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019937950673628435,
+      "loss": 2.4434,
+      "step": 3608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001993761611259434,
+      "loss": 2.4532,
+      "step": 3616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019937280655167984,
+      "loss": 2.4479,
+      "step": 3624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019936944301383026,
+      "loss": 2.4249,
+      "step": 3632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019936607051273194,
+      "loss": 2.4197,
+      "step": 3640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001993626890487231,
+      "loss": 2.4326,
+      "step": 3648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001993592986221429,
+      "loss": 2.421,
+      "step": 3656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019935589923333142,
+      "loss": 2.4447,
+      "step": 3664
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019935249088262955,
+      "loss": 2.4518,
+      "step": 3672
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019934907357037912,
+      "loss": 2.4214,
+      "step": 3680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019934564729692293,
+      "loss": 2.4416,
+      "step": 3688
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019934221206260457,
+      "loss": 2.4305,
+      "step": 3696
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019933876786776855,
+      "loss": 2.4332,
+      "step": 3704
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019933531471276036,
+      "loss": 2.4378,
+      "step": 3712
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001993318525979263,
+      "loss": 2.4375,
+      "step": 3720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019932838152361362,
+      "loss": 2.4358,
+      "step": 3728
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001993249014901704,
+      "loss": 2.4444,
+      "step": 3736
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019932141249794574,
+      "loss": 2.4217,
+      "step": 3744
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019931791454728953,
+      "loss": 2.4397,
+      "step": 3752
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019931440763855255,
+      "loss": 2.4435,
+      "step": 3760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019931089177208658,
+      "loss": 2.441,
+      "step": 3768
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019930736694824425,
+      "loss": 2.4324,
+      "step": 3776
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019930383316737907,
+      "loss": 2.439,
+      "step": 3784
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019930029042984545,
+      "loss": 2.4413,
+      "step": 3792
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019929673873599865,
+      "loss": 2.4272,
+      "step": 3800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019929317808619498,
+      "loss": 2.4445,
+      "step": 3808
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001992896084807915,
+      "loss": 2.4309,
+      "step": 3816
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019928602992014624,
+      "loss": 2.4553,
+      "step": 3824
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019928244240461807,
+      "loss": 2.4391,
+      "step": 3832
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019927884593456686,
+      "loss": 2.4331,
+      "step": 3840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019927524051035328,
+      "loss": 2.4184,
+      "step": 3848
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019927162613233895,
+      "loss": 2.4233,
+      "step": 3856
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019926800280088634,
+      "loss": 2.4348,
+      "step": 3864
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019926437051635888,
+      "loss": 2.4239,
+      "step": 3872
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019926072927912086,
+      "loss": 2.4297,
+      "step": 3880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019925707908953746,
+      "loss": 2.4333,
+      "step": 3888
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001992534199479748,
+      "loss": 2.4244,
+      "step": 3896
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019924975185479986,
+      "loss": 2.4289,
+      "step": 3904
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001992460748103805,
+      "loss": 2.4404,
+      "step": 3912
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019924238881508554,
+      "loss": 2.4309,
+      "step": 3920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001992386938692847,
+      "loss": 2.4258,
+      "step": 3928
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019923498997334845,
+      "loss": 2.4052,
+      "step": 3936
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001992312771276484,
+      "loss": 2.4343,
+      "step": 3944
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001992275553325568,
+      "loss": 2.4511,
+      "step": 3952
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019922382458844703,
+      "loss": 2.4371,
+      "step": 3960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019922008489569327,
+      "loss": 2.4251,
+      "step": 3968
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019921633625467045,
+      "loss": 2.4434,
+      "step": 3976
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001992125786657547,
+      "loss": 2.4362,
+      "step": 3984
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001992088121293228,
+      "loss": 2.433,
+      "step": 3992
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019920503664575253,
+      "loss": 2.446,
+      "step": 4000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019920125221542254,
+      "loss": 2.4311,
+      "step": 4008
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019919745883871238,
+      "loss": 2.4436,
+      "step": 4016
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019919365651600257,
+      "loss": 2.4289,
+      "step": 4024
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019918984524767435,
+      "loss": 2.4192,
+      "step": 4032
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019918602503411005,
+      "loss": 2.4188,
+      "step": 4040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001991821958756928,
+      "loss": 2.4274,
+      "step": 4048
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019917835777280666,
+      "loss": 2.4346,
+      "step": 4056
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001991745107258365,
+      "loss": 2.4132,
+      "step": 4064
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019917065473516824,
+      "loss": 2.431,
+      "step": 4072
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019916678980118859,
+      "loss": 2.4237,
+      "step": 4080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019916291592428514,
+      "loss": 2.4356,
+      "step": 4088
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019915903310484644,
+      "loss": 2.4351,
+      "step": 4096
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019915514134326195,
+      "loss": 2.4252,
+      "step": 4104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019915124063992195,
+      "loss": 2.4243,
+      "step": 4112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001991473309952177,
+      "loss": 2.4173,
+      "step": 4120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019914341240954126,
+      "loss": 2.438,
+      "step": 4128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019913948488328566,
+      "loss": 2.4144,
+      "step": 4136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019913554841684484,
+      "loss": 2.4277,
+      "step": 4144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019913160301061357,
+      "loss": 2.4127,
+      "step": 4152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019912764866498756,
+      "loss": 2.446,
+      "step": 4160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019912368538036344,
+      "loss": 2.4476,
+      "step": 4168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019911971315713866,
+      "loss": 2.4323,
+      "step": 4176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019911573199571166,
+      "loss": 2.4243,
+      "step": 4184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019911174189648167,
+      "loss": 2.4372,
+      "step": 4192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001991077428598489,
+      "loss": 2.4262,
+      "step": 4200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019910373488621444,
+      "loss": 2.4549,
+      "step": 4208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019909971797598026,
+      "loss": 2.4264,
+      "step": 4216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019909569212954925,
+      "loss": 2.4239,
+      "step": 4224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019909165734732513,
+      "loss": 2.4095,
+      "step": 4232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019908761362971263,
+      "loss": 2.4324,
+      "step": 4240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019908356097711723,
+      "loss": 2.4469,
+      "step": 4248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001990794993899455,
+      "loss": 2.4226,
+      "step": 4256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019907542886860472,
+      "loss": 2.4307,
+      "step": 4264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019907134941350314,
+      "loss": 2.4383,
+      "step": 4272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019906726102504995,
+      "loss": 2.4283,
+      "step": 4280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001990631637036551,
+      "loss": 2.4128,
+      "step": 4288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019905905744972964,
+      "loss": 2.4227,
+      "step": 4296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019905494226368534,
+      "loss": 2.41,
+      "step": 4304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019905081814593494,
+      "loss": 2.4338,
+      "step": 4312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001990466850968921,
+      "loss": 2.4254,
+      "step": 4320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019904254311697126,
+      "loss": 2.4201,
+      "step": 4328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019903839220658791,
+      "loss": 2.423,
+      "step": 4336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019903423236615836,
+      "loss": 2.4335,
+      "step": 4344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019903006359609976,
+      "loss": 2.4316,
+      "step": 4352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001990258858968303,
+      "loss": 2.4102,
+      "step": 4360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019902169926876894,
+      "loss": 2.4341,
+      "step": 4368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001990175037123355,
+      "loss": 2.4377,
+      "step": 4376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019901329922795088,
+      "loss": 2.4223,
+      "step": 4384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019900908581603672,
+      "loss": 2.4407,
+      "step": 4392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001990048634770156,
+      "loss": 2.4118,
+      "step": 4400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019900063221131104,
+      "loss": 2.4226,
+      "step": 4408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019899639201934733,
+      "loss": 2.4124,
+      "step": 4416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001989921429015498,
+      "loss": 2.4221,
+      "step": 4424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001989878848583446,
+      "loss": 2.4428,
+      "step": 4432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019898361789015879,
+      "loss": 2.433,
+      "step": 4440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019897934199742027,
+      "loss": 2.4261,
+      "step": 4448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019897505718055798,
+      "loss": 2.443,
+      "step": 4456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001989707634400016,
+      "loss": 2.4349,
+      "step": 4464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019896646077618174,
+      "loss": 2.4319,
+      "step": 4472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019896214918953002,
+      "loss": 2.4292,
+      "step": 4480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019895782868047885,
+      "loss": 2.4261,
+      "step": 4488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019895349924946146,
+      "loss": 2.4266,
+      "step": 4496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001989491608969122,
+      "loss": 2.4458,
+      "step": 4504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019894481362326606,
+      "loss": 2.4249,
+      "step": 4512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019894045742895912,
+      "loss": 2.4271,
+      "step": 4520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001989360923144283,
+      "loss": 2.44,
+      "step": 4528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001989317182801113,
+      "loss": 2.4125,
+      "step": 4536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019892733532644692,
+      "loss": 2.4101,
+      "step": 4544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019892294345387468,
+      "loss": 2.4481,
+      "step": 4552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019891854266283507,
+      "loss": 2.4439,
+      "step": 4560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019891413295376948,
+      "loss": 2.4261,
+      "step": 4568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019890971432712018,
+      "loss": 2.4237,
+      "step": 4576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019890528678333033,
+      "loss": 2.4333,
+      "step": 4584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019890085032284394,
+      "loss": 2.4333,
+      "step": 4592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019889640494610603,
+      "loss": 2.4332,
+      "step": 4600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001988919506535624,
+      "loss": 2.4348,
+      "step": 4608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019888748744565984,
+      "loss": 2.3965,
+      "step": 4616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019888301532284592,
+      "loss": 2.441,
+      "step": 4624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019887853428556923,
+      "loss": 2.4317,
+      "step": 4632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019887404433427917,
+      "loss": 2.4223,
+      "step": 4640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000198869545469426,
+      "loss": 2.421,
+      "step": 4648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000198865037691461,
+      "loss": 2.4406,
+      "step": 4656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001988605210008363,
+      "loss": 2.449,
+      "step": 4664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001988559953980048,
+      "loss": 2.4437,
+      "step": 4672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001988514608834205,
+      "loss": 2.4063,
+      "step": 4680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019884691745753808,
+      "loss": 2.4317,
+      "step": 4688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019884236512081332,
+      "loss": 2.4154,
+      "step": 4696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019883780387370273,
+      "loss": 2.4344,
+      "step": 4704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001988332337166638,
+      "loss": 2.4386,
+      "step": 4712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019882865465015487,
+      "loss": 2.4452,
+      "step": 4720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019882406667463522,
+      "loss": 2.4318,
+      "step": 4728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000198819469790565,
+      "loss": 2.425,
+      "step": 4736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019881486399840522,
+      "loss": 2.4279,
+      "step": 4744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019881024929861786,
+      "loss": 2.4195,
+      "step": 4752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019880562569166567,
+      "loss": 2.4193,
+      "step": 4760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019880099317801247,
+      "loss": 2.4098,
+      "step": 4768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001987963517581228,
+      "loss": 2.4202,
+      "step": 4776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001987917014324622,
+      "loss": 2.4201,
+      "step": 4784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019878704220149706,
+      "loss": 2.4045,
+      "step": 4792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019878237406569473,
+      "loss": 2.4389,
+      "step": 4800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001987776970255233,
+      "loss": 2.4165,
+      "step": 4808
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019877301108145193,
+      "loss": 2.4148,
+      "step": 4816
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019876831623395053,
+      "loss": 2.4255,
+      "step": 4824
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019876361248349003,
+      "loss": 2.422,
+      "step": 4832
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019875889983054213,
+      "loss": 2.4315,
+      "step": 4840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019875417827557956,
+      "loss": 2.4249,
+      "step": 4848
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019874944781907578,
+      "loss": 2.4238,
+      "step": 4856
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001987447084615053,
+      "loss": 2.4289,
+      "step": 4864
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019873996020334337,
+      "loss": 2.4268,
+      "step": 4872
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019873520304506632,
+      "loss": 2.4279,
+      "step": 4880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019873043698715115,
+      "loss": 2.4372,
+      "step": 4888
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019872566203007596,
+      "loss": 2.4103,
+      "step": 4896
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019872087817431958,
+      "loss": 2.4103,
+      "step": 4904
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019871608542036188,
+      "loss": 2.4324,
+      "step": 4912
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019871128376868348,
+      "loss": 2.4135,
+      "step": 4920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000198706473219766,
+      "loss": 2.4204,
+      "step": 4928
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019870165377409188,
+      "loss": 2.4385,
+      "step": 4936
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001986968254321445,
+      "loss": 2.4291,
+      "step": 4944
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019869198819440814,
+      "loss": 2.4131,
+      "step": 4952
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019868714206136787,
+      "loss": 2.4162,
+      "step": 4960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019868228703350983,
+      "loss": 2.4106,
+      "step": 4968
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019867742311132088,
+      "loss": 2.4347,
+      "step": 4976
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019867255029528885,
+      "loss": 2.4151,
+      "step": 4984
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001986676685859025,
+      "loss": 2.437,
+      "step": 4992
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019866277798365138,
+      "loss": 2.4154,
+      "step": 5000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019865787848902605,
+      "loss": 2.4272,
+      "step": 5008
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019865297010251786,
+      "loss": 2.4273,
+      "step": 5016
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001986480528246191,
+      "loss": 2.4389,
+      "step": 5024
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019864312665582293,
+      "loss": 2.4238,
+      "step": 5032
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019863819159662347,
+      "loss": 2.443,
+      "step": 5040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019863324764751562,
+      "loss": 2.4488,
+      "step": 5048
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001986282948089953,
+      "loss": 2.4082,
+      "step": 5056
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019862333308155917,
+      "loss": 2.4275,
+      "step": 5064
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001986183624657049,
+      "loss": 2.4202,
+      "step": 5072
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019861338296193098,
+      "loss": 2.4266,
+      "step": 5080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001986083945707369,
+      "loss": 2.4172,
+      "step": 5088
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019860339729262295,
+      "loss": 2.417,
+      "step": 5096
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019859839112809025,
+      "loss": 2.4428,
+      "step": 5104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019859337607764098,
+      "loss": 2.4217,
+      "step": 5112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001985883521417781,
+      "loss": 2.4276,
+      "step": 5120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019858331932100543,
+      "loss": 2.4209,
+      "step": 5128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019857827761582782,
+      "loss": 2.4141,
+      "step": 5136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019857322702675082,
+      "loss": 2.4232,
+      "step": 5144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019856816755428106,
+      "loss": 2.4265,
+      "step": 5152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019856309919892597,
+      "loss": 2.4222,
+      "step": 5160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019855802196119384,
+      "loss": 2.4155,
+      "step": 5168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001985529358415939,
+      "loss": 2.3992,
+      "step": 5176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001985478408406363,
+      "loss": 2.4052,
+      "step": 5184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019854273695883192,
+      "loss": 2.4209,
+      "step": 5192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019853762419669283,
+      "loss": 2.4297,
+      "step": 5200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019853250255473165,
+      "loss": 2.4004,
+      "step": 5208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019852737203346213,
+      "loss": 2.4514,
+      "step": 5216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001985222326333988,
+      "loss": 2.4228,
+      "step": 5224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019851708435505717,
+      "loss": 2.4332,
+      "step": 5232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019851192719895356,
+      "loss": 2.4207,
+      "step": 5240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019850676116560517,
+      "loss": 2.4237,
+      "step": 5248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001985015862555301,
+      "loss": 2.4269,
+      "step": 5256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019849640246924743,
+      "loss": 2.4073,
+      "step": 5264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019849120980727707,
+      "loss": 2.4031,
+      "step": 5272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019848600827013975,
+      "loss": 2.431,
+      "step": 5280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019848079785835722,
+      "loss": 2.4159,
+      "step": 5288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000198475578572452,
+      "loss": 2.4308,
+      "step": 5296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001984703504129476,
+      "loss": 2.4225,
+      "step": 5304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019846511338036832,
+      "loss": 2.4148,
+      "step": 5312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019845986747523946,
+      "loss": 2.4143,
+      "step": 5320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019845461269808715,
+      "loss": 2.4298,
+      "step": 5328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001984493490494384,
+      "loss": 2.4081,
+      "step": 5336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001984440765298211,
+      "loss": 2.4295,
+      "step": 5344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001984387951397641,
+      "loss": 2.4363,
+      "step": 5352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019843350487979702,
+      "loss": 2.4087,
+      "step": 5360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001984282057504505,
+      "loss": 2.4297,
+      "step": 5368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019842289775225604,
+      "loss": 2.4083,
+      "step": 5376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019841758088574597,
+      "loss": 2.4181,
+      "step": 5384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001984122551514535,
+      "loss": 2.4235,
+      "step": 5392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019840692054991287,
+      "loss": 2.4441,
+      "step": 5400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019840157708165898,
+      "loss": 2.444,
+      "step": 5408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019839622474722784,
+      "loss": 2.41,
+      "step": 5416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019839086354715628,
+      "loss": 2.4071,
+      "step": 5424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019838549348198193,
+      "loss": 2.4191,
+      "step": 5432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019838011455224338,
+      "loss": 2.4208,
+      "step": 5440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019837472675848013,
+      "loss": 2.4122,
+      "step": 5448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001983693301012326,
+      "loss": 2.4252,
+      "step": 5456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019836392458104195,
+      "loss": 2.4186,
+      "step": 5464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019835851019845037,
+      "loss": 2.413,
+      "step": 5472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001983530869540009,
+      "loss": 2.4262,
+      "step": 5480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001983476548482374,
+      "loss": 2.4401,
+      "step": 5488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019834221388170478,
+      "loss": 2.417,
+      "step": 5496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019833676405494862,
+      "loss": 2.4109,
+      "step": 5504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001983313053685156,
+      "loss": 2.4255,
+      "step": 5512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019832583782295314,
+      "loss": 2.4205,
+      "step": 5520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001983203614188097,
+      "loss": 2.419,
+      "step": 5528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001983148761566344,
+      "loss": 2.4319,
+      "step": 5536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019830938203697747,
+      "loss": 2.4192,
+      "step": 5544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019830387906038987,
+      "loss": 2.4212,
+      "step": 5552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019829836722742362,
+      "loss": 2.4213,
+      "step": 5560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001982928465386314,
+      "loss": 2.4333,
+      "step": 5568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000198287316994567,
+      "loss": 2.4257,
+      "step": 5576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019828177859578495,
+      "loss": 2.4195,
+      "step": 5584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019827623134284074,
+      "loss": 2.4254,
+      "step": 5592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019827067523629075,
+      "loss": 2.4437,
+      "step": 5600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001982651102766922,
+      "loss": 2.4077,
+      "step": 5608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019825953646460324,
+      "loss": 2.4221,
+      "step": 5616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019825395380058283,
+      "loss": 2.4183,
+      "step": 5624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019824836228519096,
+      "loss": 2.3967,
+      "step": 5632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019824276191898838,
+      "loss": 2.4104,
+      "step": 5640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001982371527025368,
+      "loss": 2.4196,
+      "step": 5648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019823153463639878,
+      "loss": 2.4218,
+      "step": 5656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001982259077211378,
+      "loss": 2.4108,
+      "step": 5664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019822027195731817,
+      "loss": 2.4433,
+      "step": 5672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019821462734550516,
+      "loss": 2.4173,
+      "step": 5680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019820897388626488,
+      "loss": 2.4108,
+      "step": 5688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019820331158016435,
+      "loss": 2.4267,
+      "step": 5696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019819764042777142,
+      "loss": 2.4162,
+      "step": 5704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001981919604296549,
+      "loss": 2.4167,
+      "step": 5712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019818627158638453,
+      "loss": 2.405,
+      "step": 5720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019818057389853078,
+      "loss": 2.4068,
+      "step": 5728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019817486736666514,
+      "loss": 2.4205,
+      "step": 5736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001981691519913599,
+      "loss": 2.4149,
+      "step": 5744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019816342777318832,
+      "loss": 2.4238,
+      "step": 5752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019815769471272452,
+      "loss": 2.4134,
+      "step": 5760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019815195281054344,
+      "loss": 2.4193,
+      "step": 5768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000198146202067221,
+      "loss": 2.4233,
+      "step": 5776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019814044248333396,
+      "loss": 2.4386,
+      "step": 5784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019813467405945998,
+      "loss": 2.4293,
+      "step": 5792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019812889679617757,
+      "loss": 2.4222,
+      "step": 5800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001981231106940662,
+      "loss": 2.4344,
+      "step": 5808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019811731575370613,
+      "loss": 2.4337,
+      "step": 5816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019811151197567862,
+      "loss": 2.4287,
+      "step": 5824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019810569936056573,
+      "loss": 2.418,
+      "step": 5832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001980998779089504,
+      "loss": 2.4213,
+      "step": 5840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019809404762141653,
+      "loss": 2.4079,
+      "step": 5848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019808820849854882,
+      "loss": 2.4137,
+      "step": 5856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000198082360540933,
+      "loss": 2.416,
+      "step": 5864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019807650374915547,
+      "loss": 2.4408,
+      "step": 5872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019807063812380372,
+      "loss": 2.4035,
+      "step": 5880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019806476366546595,
+      "loss": 2.4086,
+      "step": 5888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019805888037473143,
+      "loss": 2.4334,
+      "step": 5896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019805298825219017,
+      "loss": 2.434,
+      "step": 5904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001980470872984331,
+      "loss": 2.4105,
+      "step": 5912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001980411775140521,
+      "loss": 2.4249,
+      "step": 5920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019803525889963984,
+      "loss": 2.4301,
+      "step": 5928
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019802933145579,
+      "loss": 2.4166,
+      "step": 5936
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000198023395183097,
+      "loss": 2.4214,
+      "step": 5944
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019801745008215618,
+      "loss": 2.4225,
+      "step": 5952
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001980114961535639,
+      "loss": 2.4261,
+      "step": 5960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019800553339791724,
+      "loss": 2.4148,
+      "step": 5968
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019799956181581425,
+      "loss": 2.4137,
+      "step": 5976
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019799358140785381,
+      "loss": 2.4184,
+      "step": 5984
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001979875921746358,
+      "loss": 2.4309,
+      "step": 5992
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019798159411676085,
+      "loss": 2.4377,
+      "step": 6000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019797558723483053,
+      "loss": 2.4096,
+      "step": 6008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019796957152944732,
+      "loss": 2.4495,
+      "step": 6016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019796354700121453,
+      "loss": 2.4204,
+      "step": 6024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019795751365073644,
+      "loss": 2.431,
+      "step": 6032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019795147147861809,
+      "loss": 2.4253,
+      "step": 6040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001979454204854655,
+      "loss": 2.4138,
+      "step": 6048
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001979393606718856,
+      "loss": 2.4097,
+      "step": 6056
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001979332920384861,
+      "loss": 2.4326,
+      "step": 6064
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019792721458587566,
+      "loss": 2.4241,
+      "step": 6072
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019792112831466382,
+      "loss": 2.4096,
+      "step": 6080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019791503322546102,
+      "loss": 2.4228,
+      "step": 6088
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019790892931887853,
+      "loss": 2.4267,
+      "step": 6096
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019790281659552855,
+      "loss": 2.4323,
+      "step": 6104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019789669505602416,
+      "loss": 2.4442,
+      "step": 6112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019789056470097927,
+      "loss": 2.4554,
+      "step": 6120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001978844255310088,
+      "loss": 2.4064,
+      "step": 6128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019787827754672838,
+      "loss": 2.4185,
+      "step": 6136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019787212074875472,
+      "loss": 2.4449,
+      "step": 6144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019786595513770523,
+      "loss": 2.4167,
+      "step": 6152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019785978071419833,
+      "loss": 2.3857,
+      "step": 6160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019785359747885324,
+      "loss": 2.4374,
+      "step": 6168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019784740543229014,
+      "loss": 2.4185,
+      "step": 6176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019784120457513003,
+      "loss": 2.3988,
+      "step": 6184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001978349949079949,
+      "loss": 2.4153,
+      "step": 6192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001978287764315074,
+      "loss": 2.4346,
+      "step": 6200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019782254914629138,
+      "loss": 2.4229,
+      "step": 6208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019781631305297123,
+      "loss": 2.429,
+      "step": 6216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019781006815217252,
+      "loss": 2.4171,
+      "step": 6224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001978038144445215,
+      "loss": 2.4402,
+      "step": 6232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019779755193064544,
+      "loss": 2.4371,
+      "step": 6240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019779128061117243,
+      "loss": 2.4127,
+      "step": 6248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001977850004867314,
+      "loss": 2.4162,
+      "step": 6256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019777871155795222,
+      "loss": 2.4168,
+      "step": 6264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019777241382546568,
+      "loss": 2.4037,
+      "step": 6272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019776610728990338,
+      "loss": 2.4039,
+      "step": 6280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019775979195189782,
+      "loss": 2.4355,
+      "step": 6288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001977534678120824,
+      "loss": 2.42,
+      "step": 6296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001977471348710914,
+      "loss": 2.4269,
+      "step": 6304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019774079312955995,
+      "loss": 2.4303,
+      "step": 6312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019773444258812414,
+      "loss": 2.4375,
+      "step": 6320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019772808324742088,
+      "loss": 2.42,
+      "step": 6328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019772171510808794,
+      "loss": 2.415,
+      "step": 6336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019771533817076407,
+      "loss": 2.4304,
+      "step": 6344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001977089524360888,
+      "loss": 2.4173,
+      "step": 6352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019770255790470254,
+      "loss": 2.4176,
+      "step": 6360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001976961545772467,
+      "loss": 2.4532,
+      "step": 6368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019768974245436348,
+      "loss": 2.3959,
+      "step": 6376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019768332153669596,
+      "loss": 2.4328,
+      "step": 6384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019767689182488812,
+      "loss": 2.4021,
+      "step": 6392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019767045331958486,
+      "loss": 2.3979,
+      "step": 6400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019766400602143187,
+      "loss": 2.4045,
+      "step": 6408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001976575499310758,
+      "loss": 2.422,
+      "step": 6416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001976510850491642,
+      "loss": 2.4299,
+      "step": 6424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019764461137634543,
+      "loss": 2.4217,
+      "step": 6432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019763812891326874,
+      "loss": 2.4129,
+      "step": 6440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019763163766058433,
+      "loss": 2.4204,
+      "step": 6448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001976251376189432,
+      "loss": 2.4248,
+      "step": 6456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019761862878899726,
+      "loss": 2.4222,
+      "step": 6464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019761211117139932,
+      "loss": 2.4271,
+      "step": 6472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001976055847668031,
+      "loss": 2.4303,
+      "step": 6480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019759904957586313,
+      "loss": 2.4209,
+      "step": 6488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001975925055992348,
+      "loss": 2.406,
+      "step": 6496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019758595283757453,
+      "loss": 2.4003,
+      "step": 6504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019757939129153948,
+      "loss": 2.4214,
+      "step": 6512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019757282096178773,
+      "loss": 2.4195,
+      "step": 6520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019756624184897825,
+      "loss": 2.4205,
+      "step": 6528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001975596539537709,
+      "loss": 2.4012,
+      "step": 6536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019755305727682637,
+      "loss": 2.4396,
+      "step": 6544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019754645181880632,
+      "loss": 2.4492,
+      "step": 6552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019753983758037322,
+      "loss": 2.425,
+      "step": 6560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019753321456219045,
+      "loss": 2.4277,
+      "step": 6568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001975265827649222,
+      "loss": 2.4428,
+      "step": 6576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019751994218923367,
+      "loss": 2.4237,
+      "step": 6584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001975132928357909,
+      "loss": 2.4228,
+      "step": 6592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001975066347052607,
+      "loss": 2.4164,
+      "step": 6600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001974999677983109,
+      "loss": 2.4236,
+      "step": 6608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019749329211561013,
+      "loss": 2.4256,
+      "step": 6616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001974866076578279,
+      "loss": 2.4231,
+      "step": 6624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019747991442563466,
+      "loss": 2.4295,
+      "step": 6632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019747321241970166,
+      "loss": 2.4168,
+      "step": 6640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019746650164070116,
+      "loss": 2.4107,
+      "step": 6648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019745978208930614,
+      "loss": 2.4104,
+      "step": 6656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019745305376619052,
+      "loss": 2.421,
+      "step": 6664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019744631667202917,
+      "loss": 2.4239,
+      "step": 6672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019743957080749778,
+      "loss": 2.4177,
+      "step": 6680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019743281617327286,
+      "loss": 2.4119,
+      "step": 6688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019742605277003192,
+      "loss": 2.4283,
+      "step": 6696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019741928059845325,
+      "loss": 2.4355,
+      "step": 6704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019741249965921607,
+      "loss": 2.4084,
+      "step": 6712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019740570995300052,
+      "loss": 2.4132,
+      "step": 6720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019739891148048752,
+      "loss": 2.4178,
+      "step": 6728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019739210424235892,
+      "loss": 2.4083,
+      "step": 6736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019738528823929746,
+      "loss": 2.4052,
+      "step": 6744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001973784634719867,
+      "loss": 2.4327,
+      "step": 6752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019737162994111124,
+      "loss": 2.4188,
+      "step": 6760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019736478764735636,
+      "loss": 2.4104,
+      "step": 6768
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019735793659140827,
+      "loss": 2.4135,
+      "step": 6776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019735107677395415,
+      "loss": 2.4231,
+      "step": 6784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000197344208195682,
+      "loss": 2.4132,
+      "step": 6792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019733733085728068,
+      "loss": 2.436,
+      "step": 6800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019733044475943998,
+      "loss": 2.4167,
+      "step": 6808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001973235499028505,
+      "loss": 2.4176,
+      "step": 6816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019731664628820376,
+      "loss": 2.4045,
+      "step": 6824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019730973391619215,
+      "loss": 2.4262,
+      "step": 6832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019730281278750898,
+      "loss": 2.4304,
+      "step": 6840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019729588290284835,
+      "loss": 2.4195,
+      "step": 6848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001972889442629053,
+      "loss": 2.4183,
+      "step": 6856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001972819968683758,
+      "loss": 2.4232,
+      "step": 6864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001972750407199565,
+      "loss": 2.4193,
+      "step": 6872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001972680758183452,
+      "loss": 2.4227,
+      "step": 6880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019726110216424037,
+      "loss": 2.401,
+      "step": 6888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019725411975834146,
+      "loss": 2.4279,
+      "step": 6896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019724712860134873,
+      "loss": 2.4032,
+      "step": 6904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019724012869396337,
+      "loss": 2.4079,
+      "step": 6912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001972331200368874,
+      "loss": 2.4222,
+      "step": 6920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019722610263082387,
+      "loss": 2.4291,
+      "step": 6928
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001972190764764764,
+      "loss": 2.403,
+      "step": 6936
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019721204157454981,
+      "loss": 2.4193,
+      "step": 6944
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019720499792574964,
+      "loss": 2.428,
+      "step": 6952
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001971979455307823,
+      "loss": 2.4367,
+      "step": 6960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019719088439035509,
+      "loss": 2.3963,
+      "step": 6968
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019718381450517625,
+      "loss": 2.4068,
+      "step": 6976
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001971767358759548,
+      "loss": 2.4293,
+      "step": 6984
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019716964850340074,
+      "loss": 2.4212,
+      "step": 6992
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019716255238822486,
+      "loss": 2.4169,
+      "step": 7000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019715544753113885,
+      "loss": 2.4429,
+      "step": 7008
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019714833393285532,
+      "loss": 2.4328,
+      "step": 7016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001971412115940877,
+      "loss": 2.4077,
+      "step": 7024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019713408051555032,
+      "loss": 2.4,
+      "step": 7032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019712694069795838,
+      "loss": 2.4126,
+      "step": 7040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000197119792142028,
+      "loss": 2.4092,
+      "step": 7048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019711263484847614,
+      "loss": 2.4034,
+      "step": 7056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019710546881802056,
+      "loss": 2.4072,
+      "step": 7064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019709829405138005,
+      "loss": 2.4212,
+      "step": 7072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019709111054927415,
+      "loss": 2.4267,
+      "step": 7080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001970839183124234,
+      "loss": 2.3955,
+      "step": 7088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019707671734154906,
+      "loss": 2.4315,
+      "step": 7096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019706950763737337,
+      "loss": 2.4216,
+      "step": 7104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001970622892006194,
+      "loss": 2.4091,
+      "step": 7112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001970550620320112,
+      "loss": 2.4044,
+      "step": 7120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019704782613227354,
+      "loss": 2.4281,
+      "step": 7128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019704058150213214,
+      "loss": 2.426,
+      "step": 7136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001970333281423136,
+      "loss": 2.4268,
+      "step": 7144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019702606605354544,
+      "loss": 2.4068,
+      "step": 7152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019701879523655594,
+      "loss": 2.392,
+      "step": 7160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001970115156920744,
+      "loss": 2.4122,
+      "step": 7168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019700422742083084,
+      "loss": 2.4233,
+      "step": 7176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019699693042355627,
+      "loss": 2.411,
+      "step": 7184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019698962470098252,
+      "loss": 2.4073,
+      "step": 7192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019698231025384234,
+      "loss": 2.415,
+      "step": 7200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001969749870828693,
+      "loss": 2.4199,
+      "step": 7208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001969676551887979,
+      "loss": 2.417,
+      "step": 7216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019696031457236345,
+      "loss": 2.4365,
+      "step": 7224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001969529652343022,
+      "loss": 2.4072,
+      "step": 7232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019694560717535125,
+      "loss": 2.4145,
+      "step": 7240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019693824039624853,
+      "loss": 2.392,
+      "step": 7248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019693086489773297,
+      "loss": 2.4205,
+      "step": 7256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001969234806805442,
+      "loss": 2.4091,
+      "step": 7264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019691608774542282,
+      "loss": 2.4027,
+      "step": 7272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019690868609311036,
+      "loss": 2.4007,
+      "step": 7280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019690127572434917,
+      "loss": 2.4057,
+      "step": 7288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019689385663988242,
+      "loss": 2.4352,
+      "step": 7296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019688642884045423,
+      "loss": 2.4196,
+      "step": 7304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001968789923268095,
+      "loss": 2.4064,
+      "step": 7312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019687154709969415,
+      "loss": 2.4117,
+      "step": 7320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001968640931598549,
+      "loss": 2.4342,
+      "step": 7328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019685663050803925,
+      "loss": 2.4191,
+      "step": 7336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019684915914499576,
+      "loss": 2.4205,
+      "step": 7344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001968416790714737,
+      "loss": 2.4048,
+      "step": 7352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001968341902882233,
+      "loss": 2.4143,
+      "step": 7360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019682669279599566,
+      "loss": 2.3994,
+      "step": 7368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019681918659554272,
+      "loss": 2.3929,
+      "step": 7376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001968116716876173,
+      "loss": 2.4121,
+      "step": 7384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019680414807297312,
+      "loss": 2.4415,
+      "step": 7392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019679661575236474,
+      "loss": 2.4189,
+      "step": 7400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019678907472654762,
+      "loss": 2.4103,
+      "step": 7408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019678152499627807,
+      "loss": 2.4095,
+      "step": 7416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001967739665623133,
+      "loss": 2.4099,
+      "step": 7424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019676639942541139,
+      "loss": 2.4088,
+      "step": 7432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019675882358633124,
+      "loss": 2.3963,
+      "step": 7440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001967512390458327,
+      "loss": 2.433,
+      "step": 7448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019674364580467643,
+      "loss": 2.4292,
+      "step": 7456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000196736043863624,
+      "loss": 2.4088,
+      "step": 7464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019672843322343783,
+      "loss": 2.4036,
+      "step": 7472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019672081388488128,
+      "loss": 2.4336,
+      "step": 7480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019671318584871847,
+      "loss": 2.4031,
+      "step": 7488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019670554911571447,
+      "loss": 2.403,
+      "step": 7496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019669790368663517,
+      "loss": 2.4226,
+      "step": 7504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001966902495622474,
+      "loss": 2.4025,
+      "step": 7512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001966825867433188,
+      "loss": 2.4083,
+      "step": 7520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019667491523061794,
+      "loss": 2.4034,
+      "step": 7528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001966672350249142,
+      "loss": 2.397,
+      "step": 7536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019665954612697786,
+      "loss": 2.4118,
+      "step": 7544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019665184853758012,
+      "loss": 2.4071,
+      "step": 7552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019664414225749295,
+      "loss": 2.4014,
+      "step": 7560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019663642728748925,
+      "loss": 2.4183,
+      "step": 7568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001966287036283428,
+      "loss": 2.4104,
+      "step": 7576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019662097128082825,
+      "loss": 2.4273,
+      "step": 7584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019661323024572114,
+      "loss": 2.4006,
+      "step": 7592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019660548052379776,
+      "loss": 2.4077,
+      "step": 7600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019659772211583543,
+      "loss": 2.4195,
+      "step": 7608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001965899550226123,
+      "loss": 2.4095,
+      "step": 7616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019658217924490727,
+      "loss": 2.4084,
+      "step": 7624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019657439478350028,
+      "loss": 2.4152,
+      "step": 7632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019656660163917206,
+      "loss": 2.4218,
+      "step": 7640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001965587998127042,
+      "loss": 2.4112,
+      "step": 7648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001965509893048792,
+      "loss": 2.4048,
+      "step": 7656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001965431701164804,
+      "loss": 2.4148,
+      "step": 7664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019653534224829203,
+      "loss": 2.4049,
+      "step": 7672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019652750570109913,
+      "loss": 2.4004,
+      "step": 7680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019651966047568774,
+      "loss": 2.4197,
+      "step": 7688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001965118065728446,
+      "loss": 2.4023,
+      "step": 7696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019650394399335752,
+      "loss": 2.4011,
+      "step": 7704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000196496072738015,
+      "loss": 2.4232,
+      "step": 7712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019648819280760648,
+      "loss": 2.4206,
+      "step": 7720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019648030420292228,
+      "loss": 2.4185,
+      "step": 7728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001964724069247536,
+      "loss": 2.4076,
+      "step": 7736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001964645009738925,
+      "loss": 2.4251,
+      "step": 7744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019645658635113192,
+      "loss": 2.3953,
+      "step": 7752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019644866305726556,
+      "loss": 2.3957,
+      "step": 7760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019644073109308818,
+      "loss": 2.4033,
+      "step": 7768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019643279045939525,
+      "loss": 2.3995,
+      "step": 7776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019642484115698318,
+      "loss": 2.41,
+      "step": 7784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001964168831866493,
+      "loss": 2.4021,
+      "step": 7792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019640891654919168,
+      "loss": 2.4041,
+      "step": 7800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019640094124540934,
+      "loss": 2.4291,
+      "step": 7808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019639295727610215,
+      "loss": 2.4106,
+      "step": 7816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001963849646420709,
+      "loss": 2.3936,
+      "step": 7824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019637696334411718,
+      "loss": 2.4082,
+      "step": 7832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019636895338304345,
+      "loss": 2.4291,
+      "step": 7840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019636093475965312,
+      "loss": 2.4239,
+      "step": 7848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019635290747475038,
+      "loss": 2.3975,
+      "step": 7856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019634487152914032,
+      "loss": 2.4098,
+      "step": 7864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019633682692362888,
+      "loss": 2.4143,
+      "step": 7872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019632877365902295,
+      "loss": 2.4115,
+      "step": 7880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019632071173613015,
+      "loss": 2.4082,
+      "step": 7888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019631264115575913,
+      "loss": 2.4044,
+      "step": 7896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019630456191871922,
+      "loss": 2.4067,
+      "step": 7904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001962964740258208,
+      "loss": 2.4153,
+      "step": 7912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019628837747787502,
+      "loss": 2.4341,
+      "step": 7920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001962802722756939,
+      "loss": 2.4136,
+      "step": 7928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019627215842009038,
+      "loss": 2.4327,
+      "step": 7936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019626403591187824,
+      "loss": 2.4059,
+      "step": 7944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019625590475187205,
+      "loss": 2.4112,
+      "step": 7952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019624776494088739,
+      "loss": 2.4184,
+      "step": 7960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019623961647974062,
+      "loss": 2.4053,
+      "step": 7968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000196231459369249,
+      "loss": 2.4093,
+      "step": 7976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019622329361023056,
+      "loss": 2.4293,
+      "step": 7984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019621511920350434,
+      "loss": 2.3955,
+      "step": 7992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019620693614989023,
+      "loss": 2.4111,
+      "step": 8000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019619874445020888,
+      "loss": 2.395,
+      "step": 8008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001961905441052819,
+      "loss": 2.4094,
+      "step": 8016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001961823351159317,
+      "loss": 2.4233,
+      "step": 8024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019617411748298167,
+      "loss": 2.4006,
+      "step": 8032
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019616589120725593,
+      "loss": 2.4228,
+      "step": 8040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019615765628957954,
+      "loss": 2.4056,
+      "step": 8048
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001961494127307784,
+      "loss": 2.4062,
+      "step": 8056
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019614116053167936,
+      "loss": 2.4097,
+      "step": 8064
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019613289969310996,
+      "loss": 2.4107,
+      "step": 8072
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001961246302158988,
+      "loss": 2.4202,
+      "step": 8080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019611635210087523,
+      "loss": 2.417,
+      "step": 8088
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019610806534886953,
+      "loss": 2.3945,
+      "step": 8096
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001960997699607128,
+      "loss": 2.4208,
+      "step": 8104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000196091465937237,
+      "loss": 2.3983,
+      "step": 8112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019608315327927496,
+      "loss": 2.401,
+      "step": 8120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019607483198766045,
+      "loss": 2.4182,
+      "step": 8128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019606650206322802,
+      "loss": 2.412,
+      "step": 8136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001960581635068131,
+      "loss": 2.3973,
+      "step": 8144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000196049816319252,
+      "loss": 2.4032,
+      "step": 8152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019604146050138194,
+      "loss": 2.3854,
+      "step": 8160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001960330960540409,
+      "loss": 2.4162,
+      "step": 8168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019602472297806784,
+      "loss": 2.4235,
+      "step": 8176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001960163412743025,
+      "loss": 2.4081,
+      "step": 8184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019600795094358554,
+      "loss": 2.3797,
+      "step": 8192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019599955198675842,
+      "loss": 2.4345,
+      "step": 8200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019599114440466356,
+      "loss": 2.4186,
+      "step": 8208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019598272819814413,
+      "loss": 2.3856,
+      "step": 8216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001959743033680443,
+      "loss": 2.4097,
+      "step": 8224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019596586991520897,
+      "loss": 2.42,
+      "step": 8232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019595742784048404,
+      "loss": 2.4136,
+      "step": 8240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019594897714471615,
+      "loss": 2.4032,
+      "step": 8248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019594051782875286,
+      "loss": 2.4065,
+      "step": 8256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001959320498934426,
+      "loss": 2.4106,
+      "step": 8264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001959235733396346,
+      "loss": 2.4144,
+      "step": 8272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019591508816817915,
+      "loss": 2.3956,
+      "step": 8280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019590659437992714,
+      "loss": 2.395,
+      "step": 8288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019589809197573046,
+      "loss": 2.3663,
+      "step": 8296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019588958095644192,
+      "loss": 2.419,
+      "step": 8304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019588106132291505,
+      "loss": 2.3952,
+      "step": 8312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019587253307600438,
+      "loss": 2.4023,
+      "step": 8320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019586399621656525,
+      "loss": 2.3927,
+      "step": 8328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001958554507454538,
+      "loss": 2.4151,
+      "step": 8336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019584689666352713,
+      "loss": 2.4198,
+      "step": 8344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019583833397164316,
+      "loss": 2.4017,
+      "step": 8352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001958297626706607,
+      "loss": 2.3867,
+      "step": 8360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019582118276143936,
+      "loss": 2.4177,
+      "step": 8368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019581259424483968,
+      "loss": 2.4213,
+      "step": 8376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019580399712172304,
+      "loss": 2.4288,
+      "step": 8384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019579539139295168,
+      "loss": 2.4018,
+      "step": 8392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019578677705938872,
+      "loss": 2.3918,
+      "step": 8400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019577815412189807,
+      "loss": 2.4063,
+      "step": 8408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019576952258134464,
+      "loss": 2.3996,
+      "step": 8416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001957608824385941,
+      "loss": 2.4193,
+      "step": 8424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019575223369451298,
+      "loss": 2.3985,
+      "step": 8432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019574357634996874,
+      "loss": 2.4147,
+      "step": 8440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019573491040582964,
+      "loss": 2.3948,
+      "step": 8448
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019572623586296484,
+      "loss": 2.3941,
+      "step": 8456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001957175527222443,
+      "loss": 2.4034,
+      "step": 8464
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019570886098453893,
+      "loss": 2.4085,
+      "step": 8472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019570016065072044,
+      "loss": 2.4159,
+      "step": 8480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001956914517216615,
+      "loss": 2.4078,
+      "step": 8488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019568273419823547,
+      "loss": 2.4148,
+      "step": 8496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019567400808131671,
+      "loss": 2.414,
+      "step": 8504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019566527337178038,
+      "loss": 2.3968,
+      "step": 8512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019565653007050254,
+      "loss": 2.4054,
+      "step": 8520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019564777817836012,
+      "loss": 2.4014,
+      "step": 8528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019563901769623082,
+      "loss": 2.4193,
+      "step": 8536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019563024862499334,
+      "loss": 2.412,
+      "step": 8544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001956214709655271,
+      "loss": 2.4057,
+      "step": 8552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019561268471871248,
+      "loss": 2.4084,
+      "step": 8560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001956038898854307,
+      "loss": 2.4129,
+      "step": 8568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019559508646656385,
+      "loss": 2.4073,
+      "step": 8576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001955862744629948,
+      "loss": 2.4017,
+      "step": 8584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001955774538756074,
+      "loss": 2.4086,
+      "step": 8592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019556862470528625,
+      "loss": 2.4054,
+      "step": 8600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019555978695291694,
+      "loss": 2.413,
+      "step": 8608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019555094061938578,
+      "loss": 2.3965,
+      "step": 8616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019554208570558005,
+      "loss": 2.4004,
+      "step": 8624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001955332222123878,
+      "loss": 2.3984,
+      "step": 8632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019552435014069805,
+      "loss": 2.4077,
+      "step": 8640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001955154694914006,
+      "loss": 2.4086,
+      "step": 8648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019550658026538614,
+      "loss": 2.403,
+      "step": 8656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001954976824635461,
+      "loss": 2.4241,
+      "step": 8664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019548877608677304,
+      "loss": 2.4181,
+      "step": 8672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001954798611359601,
+      "loss": 2.4077,
+      "step": 8680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019547093761200146,
+      "loss": 2.3978,
+      "step": 8688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019546200551579212,
+      "loss": 2.406,
+      "step": 8696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019545306484822784,
+      "loss": 2.4211,
+      "step": 8704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019544411561020535,
+      "loss": 2.398,
+      "step": 8712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019543515780262225,
+      "loss": 2.3854,
+      "step": 8720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001954261914263769,
+      "loss": 2.3961,
+      "step": 8728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019541721648236863,
+      "loss": 2.4074,
+      "step": 8736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001954082329714975,
+      "loss": 2.4012,
+      "step": 8744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019539924089466459,
+      "loss": 2.396,
+      "step": 8752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001953902402527717,
+      "loss": 2.4133,
+      "step": 8760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019538123104672158,
+      "loss": 2.4083,
+      "step": 8768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019537221327741778,
+      "loss": 2.3939,
+      "step": 8776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019536318694576472,
+      "loss": 2.3961,
+      "step": 8784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001953541520526677,
+      "loss": 2.4101,
+      "step": 8792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001953451085990329,
+      "loss": 2.4035,
+      "step": 8800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019533605658576727,
+      "loss": 2.3905,
+      "step": 8808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001953269960137787,
+      "loss": 2.4051,
+      "step": 8816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019531792688397596,
+      "loss": 2.4191,
+      "step": 8824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019530884919726854,
+      "loss": 2.4101,
+      "step": 8832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019529976295456697,
+      "loss": 2.3873,
+      "step": 8840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019529066815678251,
+      "loss": 2.408,
+      "step": 8848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001952815648048273,
+      "loss": 2.4048,
+      "step": 8856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019527245289961437,
+      "loss": 2.3951,
+      "step": 8864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019526333244205762,
+      "loss": 2.416,
+      "step": 8872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019525420343307172,
+      "loss": 2.3938,
+      "step": 8880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001952450658735723,
+      "loss": 2.4105,
+      "step": 8888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019523591976447583,
+      "loss": 2.4079,
+      "step": 8896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019522676510669954,
+      "loss": 2.4108,
+      "step": 8904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019521760190116167,
+      "loss": 2.4022,
+      "step": 8912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001952084301487812,
+      "loss": 2.3883,
+      "step": 8920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019519924985047794,
+      "loss": 2.4126,
+      "step": 8928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019519006100717274,
+      "loss": 2.423,
+      "step": 8936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019518086361978713,
+      "loss": 2.4109,
+      "step": 8944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019517165768924354,
+      "loss": 2.4021,
+      "step": 8952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019516244321646532,
+      "loss": 2.4075,
+      "step": 8960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019515322020237662,
+      "loss": 2.385,
+      "step": 8968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001951439886479024,
+      "loss": 2.4168,
+      "step": 8976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019513474855396856,
+      "loss": 2.4148,
+      "step": 8984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019512549992150186,
+      "loss": 2.4234,
+      "step": 8992
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019511624275142986,
+      "loss": 2.411,
+      "step": 9000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019510697704468099,
+      "loss": 2.4034,
+      "step": 9008
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001950977028021846,
+      "loss": 2.4339,
+      "step": 9016
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001950884200248708,
+      "loss": 2.4157,
+      "step": 9024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019507912871367063,
+      "loss": 2.4059,
+      "step": 9032
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001950698288695159,
+      "loss": 2.406,
+      "step": 9040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019506052049333937,
+      "loss": 2.4054,
+      "step": 9048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019505120358607463,
+      "loss": 2.3838,
+      "step": 9056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019504187814865608,
+      "loss": 2.4067,
+      "step": 9064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019503254418201904,
+      "loss": 2.3902,
+      "step": 9072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019502320168709962,
+      "loss": 2.4173,
+      "step": 9080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019501385066483484,
+      "loss": 2.4129,
+      "step": 9088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019500449111616258,
+      "loss": 2.4028,
+      "step": 9096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001949951230420215,
+      "loss": 2.4211,
+      "step": 9104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019498574644335123,
+      "loss": 2.4257,
+      "step": 9112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019497636132109208,
+      "loss": 2.4118,
+      "step": 9120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019496696767618544,
+      "loss": 2.4061,
+      "step": 9128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019495756550957338,
+      "loss": 2.4011,
+      "step": 9136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019494815482219886,
+      "loss": 2.4206,
+      "step": 9144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001949387356150058,
+      "loss": 2.4149,
+      "step": 9152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001949293078889388,
+      "loss": 2.4166,
+      "step": 9160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019491987164494353,
+      "loss": 2.4275,
+      "step": 9168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001949104268839663,
+      "loss": 2.3892,
+      "step": 9176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019490097360695433,
+      "loss": 2.4166,
+      "step": 9184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001948915118148558,
+      "loss": 2.408,
+      "step": 9192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019488204150861968,
+      "loss": 2.4122,
+      "step": 9200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019487256268919573,
+      "loss": 2.4185,
+      "step": 9208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001948630753575347,
+      "loss": 2.3912,
+      "step": 9216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019485357951458808,
+      "loss": 2.4062,
+      "step": 9224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019484407516130817,
+      "loss": 2.3918,
+      "step": 9232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001948345622986483,
+      "loss": 2.3841,
+      "step": 9240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001948250409275626,
+      "loss": 2.4323,
+      "step": 9248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019481551104900583,
+      "loss": 2.3951,
+      "step": 9256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019480597266393398,
+      "loss": 2.3887,
+      "step": 9264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001947964257733036,
+      "loss": 2.4072,
+      "step": 9272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019478687037807215,
+      "loss": 2.4259,
+      "step": 9280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019477730647919807,
+      "loss": 2.4183,
+      "step": 9288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019476773407764049,
+      "loss": 2.4153,
+      "step": 9296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019475815317435952,
+      "loss": 2.4182,
+      "step": 9304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001947485637703161,
+      "loss": 2.4176,
+      "step": 9312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019473896586647186,
+      "loss": 2.4123,
+      "step": 9320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019472935946378956,
+      "loss": 2.3762,
+      "step": 9328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001947197445632326,
+      "loss": 2.4025,
+      "step": 9336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001947101211657653,
+      "loss": 2.4291,
+      "step": 9344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019470048927235285,
+      "loss": 2.3911,
+      "step": 9352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019469084888396122,
+      "loss": 2.4181,
+      "step": 9360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001946812000015574,
+      "loss": 2.4105,
+      "step": 9368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000194671542626109,
+      "loss": 2.4023,
+      "step": 9376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001946618767585847,
+      "loss": 2.4063,
+      "step": 9384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019465220239995383,
+      "loss": 2.4072,
+      "step": 9392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019464251955118677,
+      "loss": 2.4167,
+      "step": 9400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019463282821325457,
+      "loss": 2.4303,
+      "step": 9408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001946231283871293,
+      "loss": 2.3933,
+      "step": 9416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001946134200737837,
+      "loss": 2.4019,
+      "step": 9424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019460370327419155,
+      "loss": 2.4053,
+      "step": 9432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019459397798932734,
+      "loss": 2.4121,
+      "step": 9440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019458424422016648,
+      "loss": 2.4365,
+      "step": 9448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019457450196768518,
+      "loss": 2.4131,
+      "step": 9456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019456475123286058,
+      "loss": 2.386,
+      "step": 9464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001945549920166706,
+      "loss": 2.3913,
+      "step": 9472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019454522432009404,
+      "loss": 2.4037,
+      "step": 9480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019453544814411055,
+      "loss": 2.3996,
+      "step": 9488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019452566348970057,
+      "loss": 2.411,
+      "step": 9496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019451587035784556,
+      "loss": 2.4048,
+      "step": 9504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019450606874952758,
+      "loss": 2.4188,
+      "step": 9512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019449625866572982,
+      "loss": 2.3966,
+      "step": 9520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019448644010743604,
+      "loss": 2.4092,
+      "step": 9528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019447661307563102,
+      "loss": 2.4065,
+      "step": 9536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019446677757130042,
+      "loss": 2.3949,
+      "step": 9544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019445693359543064,
+      "loss": 2.3914,
+      "step": 9552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000194447081149009,
+      "loss": 2.4077,
+      "step": 9560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001944372202330236,
+      "loss": 2.4113,
+      "step": 9568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019442735084846343,
+      "loss": 2.4031,
+      "step": 9576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019441747299631841,
+      "loss": 2.4068,
+      "step": 9584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019440758667757914,
+      "loss": 2.4021,
+      "step": 9592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019439769189323727,
+      "loss": 2.3959,
+      "step": 9600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019438778864428507,
+      "loss": 2.3896,
+      "step": 9608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019437787693171588,
+      "loss": 2.4095,
+      "step": 9616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019436795675652367,
+      "loss": 2.4162,
+      "step": 9624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001943580281197035,
+      "loss": 2.3949,
+      "step": 9632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019434809102225113,
+      "loss": 2.3892,
+      "step": 9640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019433814546516314,
+      "loss": 2.4045,
+      "step": 9648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019432819144943707,
+      "loss": 2.3994,
+      "step": 9656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019431822897607122,
+      "loss": 2.3975,
+      "step": 9664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001943082580460648,
+      "loss": 2.4102,
+      "step": 9672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019429827866041777,
+      "loss": 2.387,
+      "step": 9680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019428829082013106,
+      "loss": 2.3991,
+      "step": 9688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019427829452620641,
+      "loss": 2.4186,
+      "step": 9696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019426828977964634,
+      "loss": 2.4008,
+      "step": 9704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019425827658145432,
+      "loss": 2.4016,
+      "step": 9712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001942482549326346,
+      "loss": 2.3905,
+      "step": 9720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019423822483419226,
+      "loss": 2.423,
+      "step": 9728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019422818628713327,
+      "loss": 2.3927,
+      "step": 9736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019421813929246451,
+      "loss": 2.403,
+      "step": 9744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019420808385119353,
+      "loss": 2.4096,
+      "step": 9752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019419801996432894,
+      "loss": 2.4053,
+      "step": 9760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019418794763288,
+      "loss": 2.4083,
+      "step": 9768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000194177866857857,
+      "loss": 2.4248,
+      "step": 9776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001941677776402709,
+      "loss": 2.3963,
+      "step": 9784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001941576799811336,
+      "loss": 2.4204,
+      "step": 9792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019414757388145788,
+      "loss": 2.3917,
+      "step": 9800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001941374593422573,
+      "loss": 2.4247,
+      "step": 9808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019412733636454632,
+      "loss": 2.4044,
+      "step": 9816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019411720494934019,
+      "loss": 2.4155,
+      "step": 9824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019410706509765503,
+      "loss": 2.413,
+      "step": 9832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019409691681050782,
+      "loss": 2.4246,
+      "step": 9840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019408676008891637,
+      "loss": 2.4097,
+      "step": 9848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019407659493389935,
+      "loss": 2.418,
+      "step": 9856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019406642134647625,
+      "loss": 2.4174,
+      "step": 9864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019405623932766743,
+      "loss": 2.3935,
+      "step": 9872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019404604887849413,
+      "loss": 2.4197,
+      "step": 9880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019403584999997832,
+      "loss": 2.4144,
+      "step": 9888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019402564269314294,
+      "loss": 2.4046,
+      "step": 9896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001940154269590117,
+      "loss": 2.4087,
+      "step": 9904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019400520279860926,
+      "loss": 2.3906,
+      "step": 9912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001939949702129609,
+      "loss": 2.4114,
+      "step": 9920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000193984729203093,
+      "loss": 2.4047,
+      "step": 9928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019397447977003264,
+      "loss": 2.4079,
+      "step": 9936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019396422191480782,
+      "loss": 2.4078,
+      "step": 9944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019395395563844725,
+      "loss": 2.4187,
+      "step": 9952
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019394368094198067,
+      "loss": 2.4035,
+      "step": 9960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019393339782643854,
+      "loss": 2.3984,
+      "step": 9968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019392310629285219,
+      "loss": 2.394,
+      "step": 9976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019391280634225387,
+      "loss": 2.3946,
+      "step": 9984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019390249797567648,
+      "loss": 2.404,
+      "step": 9992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019389218119415398,
+      "loss": 2.3919,
+      "step": 10000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001938818559987211,
+      "loss": 2.3999,
+      "step": 10008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019387152239041332,
+      "loss": 2.4205,
+      "step": 10016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019386118037026707,
+      "loss": 2.3882,
+      "step": 10024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019385082993931966,
+      "loss": 2.3965,
+      "step": 10032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019384047109860913,
+      "loss": 2.4062,
+      "step": 10040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019383010384917438,
+      "loss": 2.4062,
+      "step": 10048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019381972819205526,
+      "loss": 2.418,
+      "step": 10056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019380934412829232,
+      "loss": 2.4023,
+      "step": 10064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001937989516589271,
+      "loss": 2.4137,
+      "step": 10072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001937885507850018,
+      "loss": 2.4112,
+      "step": 10080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019377814150755963,
+      "loss": 2.3932,
+      "step": 10088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019376772382764465,
+      "loss": 2.387,
+      "step": 10096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019375729774630157,
+      "loss": 2.379,
+      "step": 10104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019374686326457613,
+      "loss": 2.4042,
+      "step": 10112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019373642038351483,
+      "loss": 2.4076,
+      "step": 10120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019372596910416506,
+      "loss": 2.4092,
+      "step": 10128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000193715509427575,
+      "loss": 2.4144,
+      "step": 10136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001937050413547937,
+      "loss": 2.3994,
+      "step": 10144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019369456488687106,
+      "loss": 2.4128,
+      "step": 10152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001936840800248578,
+      "loss": 2.3965,
+      "step": 10160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019367358676980546,
+      "loss": 2.3984,
+      "step": 10168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019366308512276653,
+      "loss": 2.4089,
+      "step": 10176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001936525750847942,
+      "loss": 2.4114,
+      "step": 10184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019364205665694262,
+      "loss": 2.4011,
+      "step": 10192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019363152984026668,
+      "loss": 2.3928,
+      "step": 10200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019362099463582217,
+      "loss": 2.3791,
+      "step": 10208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019361045104466574,
+      "loss": 2.3928,
+      "step": 10216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001935998990678548,
+      "loss": 2.4021,
+      "step": 10224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019358933870644774,
+      "loss": 2.3935,
+      "step": 10232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019357876996150362,
+      "loss": 2.4221,
+      "step": 10240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019356819283408242,
+      "loss": 2.4058,
+      "step": 10248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019355760732524507,
+      "loss": 2.3928,
+      "step": 10256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019354701343605313,
+      "loss": 2.3875,
+      "step": 10264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019353641116756913,
+      "loss": 2.4003,
+      "step": 10272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019352580052085647,
+      "loss": 2.3862,
+      "step": 10280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001935151814969793,
+      "loss": 2.4032,
+      "step": 10288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001935045540970026,
+      "loss": 2.4051,
+      "step": 10296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019349391832199235,
+      "loss": 2.4073,
+      "step": 10304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019348327417301517,
+      "loss": 2.4072,
+      "step": 10312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019347262165113864,
+      "loss": 2.3915,
+      "step": 10320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019346196075743113,
+      "loss": 2.3974,
+      "step": 10328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019345129149296188,
+      "loss": 2.4104,
+      "step": 10336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019344061385880094,
+      "loss": 2.391,
+      "step": 10344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019342992785601927,
+      "loss": 2.3983,
+      "step": 10352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019341923348568853,
+      "loss": 2.3955,
+      "step": 10360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001934085307488814,
+      "loss": 2.3923,
+      "step": 10368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019339781964667122,
+      "loss": 2.3957,
+      "step": 10376
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019338710018013228,
+      "loss": 2.4006,
+      "step": 10384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001933763723503397,
+      "loss": 2.398,
+      "step": 10392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001933656361583694,
+      "loss": 2.4,
+      "step": 10400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019335489160529818,
+      "loss": 2.405,
+      "step": 10408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019334413869220363,
+      "loss": 2.4081,
+      "step": 10416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019333337742016425,
+      "loss": 2.4103,
+      "step": 10424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019332260779025931,
+      "loss": 2.4046,
+      "step": 10432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019331182980356892,
+      "loss": 2.3788,
+      "step": 10440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001933010434611741,
+      "loss": 2.3966,
+      "step": 10448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019329024876415661,
+      "loss": 2.3871,
+      "step": 10456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019327944571359912,
+      "loss": 2.3946,
+      "step": 10464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019326863431058512,
+      "loss": 2.4228,
+      "step": 10472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019325781455619892,
+      "loss": 2.3953,
+      "step": 10480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001932469864515257,
+      "loss": 2.3816,
+      "step": 10488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001932361499976514,
+      "loss": 2.396,
+      "step": 10496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019322530519566293,
+      "loss": 2.3901,
+      "step": 10504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019321445204664793,
+      "loss": 2.3853,
+      "step": 10512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019320359055169487,
+      "loss": 2.4011,
+      "step": 10520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019319272071189318,
+      "loss": 2.4012,
+      "step": 10528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019318184252833297,
+      "loss": 2.3994,
+      "step": 10536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001931709560021053,
+      "loss": 2.3926,
+      "step": 10544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019316006113430202,
+      "loss": 2.395,
+      "step": 10552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019314915792601581,
+      "loss": 2.4044,
+      "step": 10560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019313824637834023,
+      "loss": 2.3973,
+      "step": 10568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001931273264923696,
+      "loss": 2.3845,
+      "step": 10576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019311639826919918,
+      "loss": 2.396,
+      "step": 10584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019310546170992493,
+      "loss": 2.4032,
+      "step": 10592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019309451681564382,
+      "loss": 2.4047,
+      "step": 10600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001930835635874535,
+      "loss": 2.3954,
+      "step": 10608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001930726020264525,
+      "loss": 2.3902,
+      "step": 10616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019306163213374025,
+      "loss": 2.3962,
+      "step": 10624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019305065391041695,
+      "loss": 2.3868,
+      "step": 10632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019303966735758366,
+      "loss": 2.3881,
+      "step": 10640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019302867247634225,
+      "loss": 2.4076,
+      "step": 10648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001930176692677955,
+      "loss": 2.3905,
+      "step": 10656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019300665773304686,
+      "loss": 2.3978,
+      "step": 10664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019299563787320082,
+      "loss": 2.392,
+      "step": 10672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001929846096893626,
+      "loss": 2.3802,
+      "step": 10680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019297357318263822,
+      "loss": 2.3949,
+      "step": 10688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001929625283541346,
+      "loss": 2.4079,
+      "step": 10696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001929514752049595,
+      "loss": 2.4035,
+      "step": 10704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019294041373622148,
+      "loss": 2.4049,
+      "step": 10712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019292934394902987,
+      "loss": 2.3937,
+      "step": 10720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019291826584449503,
+      "loss": 2.3969,
+      "step": 10728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019290717942372794,
+      "loss": 2.3707,
+      "step": 10736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019289608468784053,
+      "loss": 2.3787,
+      "step": 10744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019288498163794554,
+      "loss": 2.4003,
+      "step": 10752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019287387027515653,
+      "loss": 2.3972,
+      "step": 10760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019286275060058795,
+      "loss": 2.4054,
+      "step": 10768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019285162261535496,
+      "loss": 2.3946,
+      "step": 10776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019284048632057374,
+      "loss": 2.3795,
+      "step": 10784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001928293417173611,
+      "loss": 2.4216,
+      "step": 10792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019281818880683487,
+      "loss": 2.3951,
+      "step": 10800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019280702759011354,
+      "loss": 2.3861,
+      "step": 10808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019279585806831657,
+      "loss": 2.3896,
+      "step": 10816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019278468024256417,
+      "loss": 2.4081,
+      "step": 10824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019277349411397744,
+      "loss": 2.3815,
+      "step": 10832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019276229968367826,
+      "loss": 2.4099,
+      "step": 10840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001927510969527894,
+      "loss": 2.403,
+      "step": 10848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019273988592243438,
+      "loss": 2.4118,
+      "step": 10856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019272866659373765,
+      "loss": 2.4056,
+      "step": 10864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019271743896782443,
+      "loss": 2.4091,
+      "step": 10872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019270620304582078,
+      "loss": 2.3838,
+      "step": 10880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001926949588288536,
+      "loss": 2.4042,
+      "step": 10888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001926837063180506,
+      "loss": 2.3995,
+      "step": 10896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001926724455145404,
+      "loss": 2.4141,
+      "step": 10904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019266117641945235,
+      "loss": 2.3887,
+      "step": 10912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001926498990339167,
+      "loss": 2.4013,
+      "step": 10920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001926386133590645,
+      "loss": 2.4007,
+      "step": 10928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001926273193960276,
+      "loss": 2.3902,
+      "step": 10936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001926160171459388,
+      "loss": 2.4156,
+      "step": 10944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001926047066099316,
+      "loss": 2.4051,
+      "step": 10952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019259338778914036,
+      "loss": 2.3947,
+      "step": 10960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019258206068470035,
+      "loss": 2.3846,
+      "step": 10968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001925707252977476,
+      "loss": 2.3745,
+      "step": 10976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019255938162941898,
+      "loss": 2.3938,
+      "step": 10984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019254802968085216,
+      "loss": 2.3954,
+      "step": 10992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019253666945318572,
+      "loss": 2.3911,
+      "step": 11000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019252530094755898,
+      "loss": 2.4118,
+      "step": 11008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019251392416511225,
+      "loss": 2.3904,
+      "step": 11016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001925025391069864,
+      "loss": 2.3712,
+      "step": 11024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001924911457743234,
+      "loss": 2.4212,
+      "step": 11032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019247974416826587,
+      "loss": 2.3729,
+      "step": 11040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001924683342899574,
+      "loss": 2.3915,
+      "step": 11048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019245691614054226,
+      "loss": 2.3949,
+      "step": 11056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019244548972116566,
+      "loss": 2.3909,
+      "step": 11064
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019243405503297357,
+      "loss": 2.4039,
+      "step": 11072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001924226120771129,
+      "loss": 2.378,
+      "step": 11080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019241116085473128,
+      "loss": 2.3805,
+      "step": 11088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019239970136697716,
+      "loss": 2.3944,
+      "step": 11096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019238823361499995,
+      "loss": 2.4051,
+      "step": 11104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001923767575999497,
+      "loss": 2.4014,
+      "step": 11112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019236527332297746,
+      "loss": 2.4039,
+      "step": 11120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019235378078523498,
+      "loss": 2.3894,
+      "step": 11128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019234227998787492,
+      "loss": 2.3844,
+      "step": 11136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001923307709320508,
+      "loss": 2.4063,
+      "step": 11144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019231925361891682,
+      "loss": 2.4008,
+      "step": 11152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001923077280496282,
+      "loss": 2.394,
+      "step": 11160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001922961942253408,
+      "loss": 2.4016,
+      "step": 11168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019228465214721145,
+      "loss": 2.4011,
+      "step": 11176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019227310181639775,
+      "loss": 2.4036,
+      "step": 11184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019226154323405813,
+      "loss": 2.4048,
+      "step": 11192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001922499764013518,
+      "loss": 2.3917,
+      "step": 11200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001922384013194389,
+      "loss": 2.4117,
+      "step": 11208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001922268179894804,
+      "loss": 2.3976,
+      "step": 11216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001922152264126379,
+      "loss": 2.3852,
+      "step": 11224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001922036265900741,
+      "loss": 2.3877,
+      "step": 11232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001921920185229523,
+      "loss": 2.4108,
+      "step": 11240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001921804022124368,
+      "loss": 2.4057,
+      "step": 11248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001921687776596926,
+      "loss": 2.4048,
+      "step": 11256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001921571448658856,
+      "loss": 2.402,
+      "step": 11264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019214550383218253,
+      "loss": 2.3869,
+      "step": 11272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019213385455975086,
+      "loss": 2.3903,
+      "step": 11280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019212219704975894,
+      "loss": 2.405,
+      "step": 11288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019211053130337605,
+      "loss": 2.4182,
+      "step": 11296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001920988573217721,
+      "loss": 2.3914,
+      "step": 11304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019208717510611797,
+      "loss": 2.3852,
+      "step": 11312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001920754846575853,
+      "loss": 2.3982,
+      "step": 11320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019206378597734657,
+      "loss": 2.3879,
+      "step": 11328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019205207906657513,
+      "loss": 2.3958,
+      "step": 11336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001920403639264451,
+      "loss": 2.4021,
+      "step": 11344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001920286405581314,
+      "loss": 2.4259,
+      "step": 11352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019201690896280987,
+      "loss": 2.3875,
+      "step": 11360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019200516914165712,
+      "loss": 2.4076,
+      "step": 11368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019199342109585055,
+      "loss": 2.3825,
+      "step": 11376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019198166482656842,
+      "loss": 2.3666,
+      "step": 11384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019196990033498986,
+      "loss": 2.3811,
+      "step": 11392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001919581276222948,
+      "loss": 2.3987,
+      "step": 11400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019194634668966392,
+      "loss": 2.3926,
+      "step": 11408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019193455753827878,
+      "loss": 2.3905,
+      "step": 11416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019192276016932183,
+      "loss": 2.3905,
+      "step": 11424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019191095458397618,
+      "loss": 2.3871,
+      "step": 11432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019189914078342595,
+      "loss": 2.4105,
+      "step": 11440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019188731876885597,
+      "loss": 2.4054,
+      "step": 11448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001918754885414519,
+      "loss": 2.411,
+      "step": 11456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019186365010240026,
+      "loss": 2.3928,
+      "step": 11464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001918518034528884,
+      "loss": 2.3947,
+      "step": 11472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019183994859410446,
+      "loss": 2.3808,
+      "step": 11480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019182808552723737,
+      "loss": 2.4019,
+      "step": 11488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000191816214253477,
+      "loss": 2.3943,
+      "step": 11496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019180433477401388,
+      "loss": 2.3967,
+      "step": 11504
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019179244709003956,
+      "loss": 2.3844,
+      "step": 11512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019178055120274624,
+      "loss": 2.407,
+      "step": 11520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019176864711332705,
+      "loss": 2.394,
+      "step": 11528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019175673482297586,
+      "loss": 2.3861,
+      "step": 11536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019174481433288746,
+      "loss": 2.4107,
+      "step": 11544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019173288564425733,
+      "loss": 2.3898,
+      "step": 11552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001917209487582819,
+      "loss": 2.3807,
+      "step": 11560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001917090036761584,
+      "loss": 2.4037,
+      "step": 11568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019169705039908479,
+      "loss": 2.3935,
+      "step": 11576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019168508892825994,
+      "loss": 2.4099,
+      "step": 11584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019167311926488354,
+      "loss": 2.3978,
+      "step": 11592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019166114141015608,
+      "loss": 2.4021,
+      "step": 11600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001916491553652788,
+      "loss": 2.3752,
+      "step": 11608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019163716113145395,
+      "loss": 2.4018,
+      "step": 11616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019162515870988438,
+      "loss": 2.3985,
+      "step": 11624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001916131481017739,
+      "loss": 2.3943,
+      "step": 11632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019160112930832713,
+      "loss": 2.3845,
+      "step": 11640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019158910233074946,
+      "loss": 2.4047,
+      "step": 11648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019157706717024716,
+      "loss": 2.3881,
+      "step": 11656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019156502382802727,
+      "loss": 2.3916,
+      "step": 11664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001915529723052976,
+      "loss": 2.3867,
+      "step": 11672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019154091260326696,
+      "loss": 2.3928,
+      "step": 11680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001915288447231448,
+      "loss": 2.3753,
+      "step": 11688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019151676866614153,
+      "loss": 2.3932,
+      "step": 11696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001915046844334682,
+      "loss": 2.3908,
+      "step": 11704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019149259202633686,
+      "loss": 2.4105,
+      "step": 11712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019148049144596037,
+      "loss": 2.3955,
+      "step": 11720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001914683826935522,
+      "loss": 2.395,
+      "step": 11728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001914562657703269,
+      "loss": 2.3903,
+      "step": 11736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001914441406774997,
+      "loss": 2.3949,
+      "step": 11744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019143200741628672,
+      "loss": 2.3807,
+      "step": 11752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019141986598790475,
+      "loss": 2.3761,
+      "step": 11760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019140771639357163,
+      "loss": 2.3888,
+      "step": 11768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019139555863450578,
+      "loss": 2.394,
+      "step": 11776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019138339271192664,
+      "loss": 2.3867,
+      "step": 11784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019137121862705436,
+      "loss": 2.3905,
+      "step": 11792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019135903638110993,
+      "loss": 2.3857,
+      "step": 11800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019134684597531514,
+      "loss": 2.3907,
+      "step": 11808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019133464741089262,
+      "loss": 2.3862,
+      "step": 11816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019132244068906587,
+      "loss": 2.3944,
+      "step": 11824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001913102258110591,
+      "loss": 2.4102,
+      "step": 11832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001912980027780974,
+      "loss": 2.3836,
+      "step": 11840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001912857715914067,
+      "loss": 2.3724,
+      "step": 11848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019127353225221368,
+      "loss": 2.403,
+      "step": 11856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019126128476174592,
+      "loss": 2.3876,
+      "step": 11864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019124902912123174,
+      "loss": 2.3896,
+      "step": 11872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019123676533190032,
+      "loss": 2.3701,
+      "step": 11880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019122449339498162,
+      "loss": 2.3926,
+      "step": 11888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019121221331170653,
+      "loss": 2.363,
+      "step": 11896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019119992508330659,
+      "loss": 2.3798,
+      "step": 11904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019118762871101426,
+      "loss": 2.4065,
+      "step": 11912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019117532419606284,
+      "loss": 2.4163,
+      "step": 11920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019116301153968633,
+      "loss": 2.3922,
+      "step": 11928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001911506907431197,
+      "loss": 2.4043,
+      "step": 11936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019113836180759857,
+      "loss": 2.4003,
+      "step": 11944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001911260247343595,
+      "loss": 2.3917,
+      "step": 11952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019111367952463987,
+      "loss": 2.395,
+      "step": 11960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001911013261796778,
+      "loss": 2.4013,
+      "step": 11968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019108896470071226,
+      "loss": 2.3944,
+      "step": 11976
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019107659508898299,
+      "loss": 2.4123,
+      "step": 11984
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019106421734573067,
+      "loss": 2.412,
+      "step": 11992
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019105183147219673,
+      "loss": 2.4034,
+      "step": 12000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019103943746962327,
+      "loss": 2.3749,
+      "step": 12008
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019102703533925347,
+      "loss": 2.3911,
+      "step": 12016
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019101462508233115,
+      "loss": 2.3802,
+      "step": 12024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000191002206700101,
+      "loss": 2.386,
+      "step": 12032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019098978019380848,
+      "loss": 2.3939,
+      "step": 12040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019097734556469995,
+      "loss": 2.3811,
+      "step": 12048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019096490281402247,
+      "loss": 2.3995,
+      "step": 12056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019095245194302403,
+      "loss": 2.4037,
+      "step": 12064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019093999295295336,
+      "loss": 2.3969,
+      "step": 12072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019092752584506,
+      "loss": 2.391,
+      "step": 12080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001909150506205944,
+      "loss": 2.3966,
+      "step": 12088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019090256728080768,
+      "loss": 2.4,
+      "step": 12096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001908900758269519,
+      "loss": 2.3954,
+      "step": 12104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019087757626027986,
+      "loss": 2.3941,
+      "step": 12112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001908650685820452,
+      "loss": 2.38,
+      "step": 12120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019085255279350236,
+      "loss": 2.3901,
+      "step": 12128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019084002889590662,
+      "loss": 2.3823,
+      "step": 12136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019082749689051405,
+      "loss": 2.3969,
+      "step": 12144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019081495677858153,
+      "loss": 2.3891,
+      "step": 12152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019080240856136677,
+      "loss": 2.3703,
+      "step": 12160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019078985224012826,
+      "loss": 2.3761,
+      "step": 12168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001907772878161254,
+      "loss": 2.3934,
+      "step": 12176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019076471529061825,
+      "loss": 2.3807,
+      "step": 12184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019075213466486778,
+      "loss": 2.3938,
+      "step": 12192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019073954594013578,
+      "loss": 2.3723,
+      "step": 12200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001907269491176848,
+      "loss": 2.4025,
+      "step": 12208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001907143441987783,
+      "loss": 2.3859,
+      "step": 12216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001907017311846804,
+      "loss": 2.3931,
+      "step": 12224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019068911007665614,
+      "loss": 2.4062,
+      "step": 12232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019067648087597137,
+      "loss": 2.3993,
+      "step": 12240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019066384358389267,
+      "loss": 2.369,
+      "step": 12248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019065119820168753,
+      "loss": 2.3985,
+      "step": 12256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019063854473062424,
+      "loss": 2.3921,
+      "step": 12264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019062588317197182,
+      "loss": 2.4,
+      "step": 12272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019061321352700016,
+      "loss": 2.3945,
+      "step": 12280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019060053579697997,
+      "loss": 2.4166,
+      "step": 12288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019058784998318273,
+      "loss": 2.3814,
+      "step": 12296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001905751560868808,
+      "loss": 2.3883,
+      "step": 12304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019056245410934726,
+      "loss": 2.3767,
+      "step": 12312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019054974405185605,
+      "loss": 2.3678,
+      "step": 12320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019053702591568196,
+      "loss": 2.4128,
+      "step": 12328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019052429970210053,
+      "loss": 2.3963,
+      "step": 12336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001905115654123881,
+      "loss": 2.3806,
+      "step": 12344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019049882304782186,
+      "loss": 2.3816,
+      "step": 12352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001904860726096798,
+      "loss": 2.3895,
+      "step": 12360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001904733140992407,
+      "loss": 2.4059,
+      "step": 12368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019046054751778422,
+      "loss": 2.3718,
+      "step": 12376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019044777286659071,
+      "loss": 2.3923,
+      "step": 12384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019043499014694142,
+      "loss": 2.3875,
+      "step": 12392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001904221993601184,
+      "loss": 2.3916,
+      "step": 12400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019040940050740448,
+      "loss": 2.3685,
+      "step": 12408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019039659359008332,
+      "loss": 2.3798,
+      "step": 12416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019038377860943937,
+      "loss": 2.3655,
+      "step": 12424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001903709555667579,
+      "loss": 2.3882,
+      "step": 12432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000190358124463325,
+      "loss": 2.3843,
+      "step": 12440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019034528530042754,
+      "loss": 2.3885,
+      "step": 12448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019033243807935326,
+      "loss": 2.3696,
+      "step": 12456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001903195828013906,
+      "loss": 2.4002,
+      "step": 12464
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001903067194678289,
+      "loss": 2.3927,
+      "step": 12472
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001902938480799583,
+      "loss": 2.3861,
+      "step": 12480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019028096863906967,
+      "loss": 2.3856,
+      "step": 12488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019026808114645482,
+      "loss": 2.4021,
+      "step": 12496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019025518560340625,
+      "loss": 2.3694,
+      "step": 12504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001902422820112173,
+      "loss": 2.3779,
+      "step": 12512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019022937037118217,
+      "loss": 2.399,
+      "step": 12520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019021645068459576,
+      "loss": 2.3916,
+      "step": 12528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019020352295275386,
+      "loss": 2.3976,
+      "step": 12536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001901905871769531,
+      "loss": 2.3913,
+      "step": 12544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019017764335849084,
+      "loss": 2.38,
+      "step": 12552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019016469149866526,
+      "loss": 2.3972,
+      "step": 12560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019015173159877535,
+      "loss": 2.3888,
+      "step": 12568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019013876366012094,
+      "loss": 2.3768,
+      "step": 12576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001901257876840026,
+      "loss": 2.4004,
+      "step": 12584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001901128036717218,
+      "loss": 2.3968,
+      "step": 12592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019009981162458072,
+      "loss": 2.3689,
+      "step": 12600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001900868115438824,
+      "loss": 2.3897,
+      "step": 12608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001900738034309307,
+      "loss": 2.3933,
+      "step": 12616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001900607872870302,
+      "loss": 2.3817,
+      "step": 12624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019004776311348646,
+      "loss": 2.367,
+      "step": 12632
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019003473091160558,
+      "loss": 2.3762,
+      "step": 12640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001900216906826947,
+      "loss": 2.3858,
+      "step": 12648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019000864242806168,
+      "loss": 2.3918,
+      "step": 12656
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018999558614901522,
+      "loss": 2.3958,
+      "step": 12664
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001899825218468647,
+      "loss": 2.3904,
+      "step": 12672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018996944952292047,
+      "loss": 2.3878,
+      "step": 12680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018995636917849357,
+      "loss": 2.3922,
+      "step": 12688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001899432808148959,
+      "loss": 2.4021,
+      "step": 12696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018993018443344017,
+      "loss": 2.3734,
+      "step": 12704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018991708003543981,
+      "loss": 2.3859,
+      "step": 12712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018990396762220922,
+      "loss": 2.3797,
+      "step": 12720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018989084719506338,
+      "loss": 2.3693,
+      "step": 12728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018987771875531826,
+      "loss": 2.3839,
+      "step": 12736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018986458230429057,
+      "loss": 2.3936,
+      "step": 12744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018985143784329785,
+      "loss": 2.3934,
+      "step": 12752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018983828537365833,
+      "loss": 2.3911,
+      "step": 12760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018982512489669118,
+      "loss": 2.4017,
+      "step": 12768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001898119564137163,
+      "loss": 2.3777,
+      "step": 12776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001897987799260545,
+      "loss": 2.3865,
+      "step": 12784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001897855954350272,
+      "loss": 2.3924,
+      "step": 12792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018977240294195676,
+      "loss": 2.3995,
+      "step": 12800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018975920244816631,
+      "loss": 2.382,
+      "step": 12808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001897459939549798,
+      "loss": 2.3879,
+      "step": 12816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018973277746372197,
+      "loss": 2.3702,
+      "step": 12824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018971955297571836,
+      "loss": 2.3681,
+      "step": 12832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018970632049229532,
+      "loss": 2.3798,
+      "step": 12840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018969308001477993,
+      "loss": 2.3962,
+      "step": 12848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001896798315445002,
+      "loss": 2.3764,
+      "step": 12856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018966657508278487,
+      "loss": 2.3801,
+      "step": 12864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018965331063096346,
+      "loss": 2.3941,
+      "step": 12872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018964003819036632,
+      "loss": 2.3866,
+      "step": 12880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018962675776232463,
+      "loss": 2.3785,
+      "step": 12888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018961346934817034,
+      "loss": 2.3872,
+      "step": 12896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001896001729492362,
+      "loss": 2.3892,
+      "step": 12904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001895868685668557,
+      "loss": 2.3971,
+      "step": 12912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001895735562023633,
+      "loss": 2.3821,
+      "step": 12920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018956023585709405,
+      "loss": 2.4062,
+      "step": 12928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000189546907532384,
+      "loss": 2.3856,
+      "step": 12936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018953357122956986,
+      "loss": 2.3817,
+      "step": 12944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001895202269499892,
+      "loss": 2.3994,
+      "step": 12952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001895068746949803,
+      "loss": 2.3995,
+      "step": 12960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001894935144658824,
+      "loss": 2.3993,
+      "step": 12968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018948014626403548,
+      "loss": 2.3906,
+      "step": 12976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001894667700907802,
+      "loss": 2.3824,
+      "step": 12984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001894533859474582,
+      "loss": 2.3823,
+      "step": 12992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018943999383541177,
+      "loss": 2.4173,
+      "step": 13000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001894265937559841,
+      "loss": 2.3766,
+      "step": 13008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018941318571051916,
+      "loss": 2.4138,
+      "step": 13016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018939976970036164,
+      "loss": 2.378,
+      "step": 13024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018938634572685713,
+      "loss": 2.4069,
+      "step": 13032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018937291379135196,
+      "loss": 2.3895,
+      "step": 13040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001893594738951933,
+      "loss": 2.3897,
+      "step": 13048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001893460260397291,
+      "loss": 2.3892,
+      "step": 13056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018933257022630805,
+      "loss": 2.3839,
+      "step": 13064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018931910645627977,
+      "loss": 2.3975,
+      "step": 13072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018930563473099453,
+      "loss": 2.3792,
+      "step": 13080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018929215505180349,
+      "loss": 2.4199,
+      "step": 13088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018927866742005862,
+      "loss": 2.3644,
+      "step": 13096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001892651718371126,
+      "loss": 2.385,
+      "step": 13104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000189251668304319,
+      "loss": 2.4003,
+      "step": 13112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018923815682303215,
+      "loss": 2.388,
+      "step": 13120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018922463739460714,
+      "loss": 2.3739,
+      "step": 13128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018921111002039994,
+      "loss": 2.3848,
+      "step": 13136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001891975747017672,
+      "loss": 2.3927,
+      "step": 13144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001891840314400665,
+      "loss": 2.3622,
+      "step": 13152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018917048023665609,
+      "loss": 2.3949,
+      "step": 13160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018915692109289516,
+      "loss": 2.4043,
+      "step": 13168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018914335401014353,
+      "loss": 2.3888,
+      "step": 13176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018912977898976198,
+      "loss": 2.3936,
+      "step": 13184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018911619603311197,
+      "loss": 2.3957,
+      "step": 13192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018910260514155578,
+      "loss": 2.39,
+      "step": 13200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018908900631645653,
+      "loss": 2.3914,
+      "step": 13208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001890753995591781,
+      "loss": 2.3779,
+      "step": 13216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018906178487108511,
+      "loss": 2.3808,
+      "step": 13224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001890481622535431,
+      "loss": 2.3676,
+      "step": 13232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018903453170791837,
+      "loss": 2.3929,
+      "step": 13240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018902089323557793,
+      "loss": 2.4018,
+      "step": 13248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001890072468378896,
+      "loss": 2.3905,
+      "step": 13256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018899359251622216,
+      "loss": 2.3968,
+      "step": 13264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000188979930271945,
+      "loss": 2.4097,
+      "step": 13272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001889662601064283,
+      "loss": 2.371,
+      "step": 13280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018895258202104317,
+      "loss": 2.406,
+      "step": 13288
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018893889601716147,
+      "loss": 2.3976,
+      "step": 13296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018892520209615575,
+      "loss": 2.3921,
+      "step": 13304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018891150025939952,
+      "loss": 2.3893,
+      "step": 13312
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018889779050826694,
+      "loss": 2.3908,
+      "step": 13320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018888407284413302,
+      "loss": 2.3696,
+      "step": 13328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001888703472683736,
+      "loss": 2.3959,
+      "step": 13336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018885661378236518,
+      "loss": 2.3824,
+      "step": 13344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018884287238748527,
+      "loss": 2.4161,
+      "step": 13352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000188829123085112,
+      "loss": 2.4001,
+      "step": 13360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018881536587662438,
+      "loss": 2.3359,
+      "step": 13368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018880160076340213,
+      "loss": 2.3931,
+      "step": 13376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018878782774682586,
+      "loss": 2.3966,
+      "step": 13384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018877404682827688,
+      "loss": 2.4013,
+      "step": 13392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018876025800913736,
+      "loss": 2.3765,
+      "step": 13400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018874646129079029,
+      "loss": 2.3761,
+      "step": 13408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001887326566746193,
+      "loss": 2.4023,
+      "step": 13416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000188718844162009,
+      "loss": 2.3807,
+      "step": 13424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018870502375434463,
+      "loss": 2.3871,
+      "step": 13432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001886911954530124,
+      "loss": 2.396,
+      "step": 13440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018867735925939914,
+      "loss": 2.3976,
+      "step": 13448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018866351517489254,
+      "loss": 2.3788,
+      "step": 13456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001886496632008811,
+      "loss": 2.3786,
+      "step": 13464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018863580333875412,
+      "loss": 2.3943,
+      "step": 13472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018862193558990165,
+      "loss": 2.3964,
+      "step": 13480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018860805995571446,
+      "loss": 2.4179,
+      "step": 13488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018859417643758437,
+      "loss": 2.37,
+      "step": 13496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018858028503690365,
+      "loss": 2.3721,
+      "step": 13504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018856638575506565,
+      "loss": 2.3841,
+      "step": 13512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001885524785934643,
+      "loss": 2.4048,
+      "step": 13520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018853856355349448,
+      "loss": 2.3759,
+      "step": 13528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018852464063655176,
+      "loss": 2.3836,
+      "step": 13536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001885107098440325,
+      "loss": 2.3986,
+      "step": 13544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018849677117733392,
+      "loss": 2.3937,
+      "step": 13552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000188482824637854,
+      "loss": 2.3821,
+      "step": 13560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018846887022699147,
+      "loss": 2.3863,
+      "step": 13568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018845490794614587,
+      "loss": 2.3616,
+      "step": 13576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001884409377967176,
+      "loss": 2.378,
+      "step": 13584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018842695978010766,
+      "loss": 2.4052,
+      "step": 13592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001884129738977181,
+      "loss": 2.4167,
+      "step": 13600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018839898015095155,
+      "loss": 2.3874,
+      "step": 13608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018838497854121154,
+      "loss": 2.3905,
+      "step": 13616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018837096906990233,
+      "loss": 2.4128,
+      "step": 13624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000188356951738429,
+      "loss": 2.4001,
+      "step": 13632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018834292654819739,
+      "loss": 2.4033,
+      "step": 13640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018832889350061418,
+      "loss": 2.4065,
+      "step": 13648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018831485259708676,
+      "loss": 2.3896,
+      "step": 13656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001883008038390234,
+      "loss": 2.38,
+      "step": 13664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001882867472278331,
+      "loss": 2.3854,
+      "step": 13672
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001882726827649256,
+      "loss": 2.4,
+      "step": 13680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018825861045171163,
+      "loss": 2.4093,
+      "step": 13688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001882445302896024,
+      "loss": 2.3892,
+      "step": 13696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018823044228001017,
+      "loss": 2.4066,
+      "step": 13704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018821634642434784,
+      "loss": 2.3769,
+      "step": 13712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001882022427240292,
+      "loss": 2.387,
+      "step": 13720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018818813118046866,
+      "loss": 2.4067,
+      "step": 13728
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018817401179508167,
+      "loss": 2.3972,
+      "step": 13736
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018815988456928422,
+      "loss": 2.3924,
+      "step": 13744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018814574950449324,
+      "loss": 2.3757,
+      "step": 13752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018813160660212638,
+      "loss": 2.397,
+      "step": 13760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001881174558636021,
+      "loss": 2.3774,
+      "step": 13768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001881032972903396,
+      "loss": 2.4052,
+      "step": 13776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000188089130883759,
+      "loss": 2.3775,
+      "step": 13784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000188074956645281,
+      "loss": 2.3826,
+      "step": 13792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018806077457632724,
+      "loss": 2.4138,
+      "step": 13800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018804658467832012,
+      "loss": 2.4132,
+      "step": 13808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018803238695268282,
+      "loss": 2.3717,
+      "step": 13816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018801818140083926,
+      "loss": 2.3741,
+      "step": 13824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018800396802421413,
+      "loss": 2.3942,
+      "step": 13832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018798974682423305,
+      "loss": 2.4012,
+      "step": 13840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001879755178023223,
+      "loss": 2.4156,
+      "step": 13848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001879612809599089,
+      "loss": 2.4123,
+      "step": 13856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018794703629842075,
+      "loss": 2.3687,
+      "step": 13864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001879327838192866,
+      "loss": 2.3836,
+      "step": 13872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001879185235239358,
+      "loss": 2.3718,
+      "step": 13880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018790425541379859,
+      "loss": 2.3938,
+      "step": 13888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018788997949030598,
+      "loss": 2.4019,
+      "step": 13896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018787569575488982,
+      "loss": 2.3951,
+      "step": 13904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001878614042089826,
+      "loss": 2.3883,
+      "step": 13912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018784710485401774,
+      "loss": 2.3959,
+      "step": 13920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018783279769142938,
+      "loss": 2.3872,
+      "step": 13928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018781848272265243,
+      "loss": 2.4153,
+      "step": 13936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001878041599491226,
+      "loss": 2.3818,
+      "step": 13944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001877898293722764,
+      "loss": 2.3965,
+      "step": 13952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018777549099355107,
+      "loss": 2.3899,
+      "step": 13960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018776114481438473,
+      "loss": 2.3974,
+      "step": 13968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018774679083621615,
+      "loss": 2.3958,
+      "step": 13976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000187732429060485,
+      "loss": 2.3926,
+      "step": 13984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018771805948863168,
+      "loss": 2.3912,
+      "step": 13992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018770368212209736,
+      "loss": 2.3841,
+      "step": 14000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018768929696232403,
+      "loss": 2.3808,
+      "step": 14008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018767490401075442,
+      "loss": 2.3908,
+      "step": 14016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018766050326883207,
+      "loss": 2.4197,
+      "step": 14024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018764609473800127,
+      "loss": 2.3863,
+      "step": 14032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018763167841970714,
+      "loss": 2.3756,
+      "step": 14040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018761725431539557,
+      "loss": 2.4033,
+      "step": 14048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018760282242651316,
+      "loss": 2.3984,
+      "step": 14056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018758838275450743,
+      "loss": 2.4083,
+      "step": 14064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018757393530082652,
+      "loss": 2.4045,
+      "step": 14072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018755948006691947,
+      "loss": 2.3828,
+      "step": 14080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018754501705423607,
+      "loss": 2.3839,
+      "step": 14088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001875305462642268,
+      "loss": 2.3793,
+      "step": 14096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001875160676983431,
+      "loss": 2.3988,
+      "step": 14104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018750158135803698,
+      "loss": 2.3787,
+      "step": 14112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018748708724476148,
+      "loss": 2.3917,
+      "step": 14120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018747258535997013,
+      "loss": 2.3884,
+      "step": 14128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018745807570511746,
+      "loss": 2.3827,
+      "step": 14136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001874435582816587,
+      "loss": 2.3758,
+      "step": 14144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001874290330910499,
+      "loss": 2.3704,
+      "step": 14152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001874145001347478,
+      "loss": 2.3942,
+      "step": 14160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018739995941420996,
+      "loss": 2.3794,
+      "step": 14168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018738541093089476,
+      "loss": 2.403,
+      "step": 14176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001873708546862614,
+      "loss": 2.3708,
+      "step": 14184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018735629068176966,
+      "loss": 2.4114,
+      "step": 14192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018734171891888034,
+      "loss": 2.4045,
+      "step": 14200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001873271393990548,
+      "loss": 2.3996,
+      "step": 14208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018731255212375538,
+      "loss": 2.3881,
+      "step": 14216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018729795709444502,
+      "loss": 2.3621,
+      "step": 14224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001872833543125876,
+      "loss": 2.3904,
+      "step": 14232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018726874377964764,
+      "loss": 2.3896,
+      "step": 14240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018725412549709048,
+      "loss": 2.4033,
+      "step": 14248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001872394994663823,
+      "loss": 2.375,
+      "step": 14256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018722486568899002,
+      "loss": 2.3939,
+      "step": 14264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018721022416638125,
+      "loss": 2.3886,
+      "step": 14272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018719557490002448,
+      "loss": 2.3943,
+      "step": 14280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018718091789138902,
+      "loss": 2.3837,
+      "step": 14288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018716625314194475,
+      "loss": 2.3855,
+      "step": 14296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001871515806531626,
+      "loss": 2.3876,
+      "step": 14304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018713690042651405,
+      "loss": 2.385,
+      "step": 14312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018712221246347147,
+      "loss": 2.3824,
+      "step": 14320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018710751676550799,
+      "loss": 2.3919,
+      "step": 14328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001870928133340975,
+      "loss": 2.4023,
+      "step": 14336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018707810217071465,
+      "loss": 2.4068,
+      "step": 14344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001870633832768349,
+      "loss": 2.3966,
+      "step": 14352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018704865665393448,
+      "loss": 2.3727,
+      "step": 14360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018703392230349036,
+      "loss": 2.3779,
+      "step": 14368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018701918022698036,
+      "loss": 2.3807,
+      "step": 14376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018700443042588298,
+      "loss": 2.3991,
+      "step": 14384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018698967290167758,
+      "loss": 2.3679,
+      "step": 14392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001869749076558442,
+      "loss": 2.3803,
+      "step": 14400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018696013468986374,
+      "loss": 2.3814,
+      "step": 14408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001869453540052179,
+      "loss": 2.394,
+      "step": 14416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018693056560338902,
+      "loss": 2.3942,
+      "step": 14424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018691576948586035,
+      "loss": 2.3973,
+      "step": 14432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001869009656541158,
+      "loss": 2.3868,
+      "step": 14440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018688615410964013,
+      "loss": 2.3882,
+      "step": 14448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018687133485391886,
+      "loss": 2.3745,
+      "step": 14456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001868565078884383,
+      "loss": 2.406,
+      "step": 14464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001868416732146855,
+      "loss": 2.3812,
+      "step": 14472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018682683083414827,
+      "loss": 2.3764,
+      "step": 14480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018681198074831524,
+      "loss": 2.3861,
+      "step": 14488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001867971229586758,
+      "loss": 2.3799,
+      "step": 14496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018678225746672006,
+      "loss": 2.3895,
+      "step": 14504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018676738427393899,
+      "loss": 2.3791,
+      "step": 14512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018675250338182425,
+      "loss": 2.3839,
+      "step": 14520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018673761479186832,
+      "loss": 2.3915,
+      "step": 14528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018672271850556443,
+      "loss": 2.4077,
+      "step": 14536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018670781452440665,
+      "loss": 2.3912,
+      "step": 14544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001866929028498897,
+      "loss": 2.3687,
+      "step": 14552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018667798348350915,
+      "loss": 2.3852,
+      "step": 14560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018666305642676135,
+      "loss": 2.407,
+      "step": 14568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018664812168114338,
+      "loss": 2.3976,
+      "step": 14576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018663317924815315,
+      "loss": 2.3687,
+      "step": 14584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001866182291292892,
+      "loss": 2.383,
+      "step": 14592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001866032713260511,
+      "loss": 2.3783,
+      "step": 14600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001865883058399389,
+      "loss": 2.3977,
+      "step": 14608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001865733326724536,
+      "loss": 2.3764,
+      "step": 14616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018655835182509689,
+      "loss": 2.382,
+      "step": 14624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018654336329937133,
+      "loss": 2.3731,
+      "step": 14632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018652836709678015,
+      "loss": 2.3948,
+      "step": 14640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018651336321882735,
+      "loss": 2.3867,
+      "step": 14648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001864983516670178,
+      "loss": 2.3776,
+      "step": 14656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018648333244285705,
+      "loss": 2.3812,
+      "step": 14664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018646830554785142,
+      "loss": 2.3921,
+      "step": 14672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018645327098350804,
+      "loss": 2.3925,
+      "step": 14680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018643822875133478,
+      "loss": 2.3856,
+      "step": 14688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001864231788528403,
+      "loss": 2.3744,
+      "step": 14696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018640812128953402,
+      "loss": 2.3906,
+      "step": 14704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018639305606292614,
+      "loss": 2.3844,
+      "step": 14712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001863779831745276,
+      "loss": 2.3816,
+      "step": 14720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018636290262585012,
+      "loss": 2.3888,
+      "step": 14728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001863478144184062,
+      "loss": 2.3826,
+      "step": 14736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018633271855370912,
+      "loss": 2.3709,
+      "step": 14744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001863176150332729,
+      "loss": 2.3821,
+      "step": 14752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001863025038586123,
+      "loss": 2.3784,
+      "step": 14760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018628738503124293,
+      "loss": 2.368,
+      "step": 14768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001862722585526811,
+      "loss": 2.3957,
+      "step": 14776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018625712442444395,
+      "loss": 2.3962,
+      "step": 14784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001862419826480493,
+      "loss": 2.3944,
+      "step": 14792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001862268332250158,
+      "loss": 2.3789,
+      "step": 14800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018621167615686282,
+      "loss": 2.364,
+      "step": 14808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018619651144511062,
+      "loss": 2.3748,
+      "step": 14816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018618133909128006,
+      "loss": 2.3811,
+      "step": 14824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018616615909689284,
+      "loss": 2.3867,
+      "step": 14832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018615097146347146,
+      "loss": 2.3912,
+      "step": 14840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001861357761925391,
+      "loss": 2.3685,
+      "step": 14848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018612057328561984,
+      "loss": 2.4118,
+      "step": 14856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018610536274423837,
+      "loss": 2.3947,
+      "step": 14864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001860901445699203,
+      "loss": 2.3841,
+      "step": 14872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018607491876419184,
+      "loss": 2.3682,
+      "step": 14880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001860596853285801,
+      "loss": 2.386,
+      "step": 14888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001860444442646129,
+      "loss": 2.3677,
+      "step": 14896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018602919557381885,
+      "loss": 2.3731,
+      "step": 14904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018601393925772724,
+      "loss": 2.3954,
+      "step": 14912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018599867531786827,
+      "loss": 2.3908,
+      "step": 14920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018598340375577282,
+      "loss": 2.3884,
+      "step": 14928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018596812457297247,
+      "loss": 2.3663,
+      "step": 14936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018595283777099973,
+      "loss": 2.3876,
+      "step": 14944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018593754335138773,
+      "loss": 2.3872,
+      "step": 14952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001859222413156704,
+      "loss": 2.4022,
+      "step": 14960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001859069316653825,
+      "loss": 2.3823,
+      "step": 14968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018589161440205943,
+      "loss": 2.3755,
+      "step": 14976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018587628952723744,
+      "loss": 2.3939,
+      "step": 14984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001858609570424536,
+      "loss": 2.3759,
+      "step": 14992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001858456169492456,
+      "loss": 2.3976,
+      "step": 15000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018583026924915197,
+      "loss": 2.3807,
+      "step": 15008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018581491394371202,
+      "loss": 2.3762,
+      "step": 15016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001857995510344658,
+      "loss": 2.3826,
+      "step": 15024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001857841805229541,
+      "loss": 2.407,
+      "step": 15032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018576880241071852,
+      "loss": 2.3696,
+      "step": 15040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001857534166993014,
+      "loss": 2.3774,
+      "step": 15048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018573802339024584,
+      "loss": 2.368,
+      "step": 15056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018572262248509568,
+      "loss": 2.3773,
+      "step": 15064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018570721398539553,
+      "loss": 2.3665,
+      "step": 15072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001856917978926908,
+      "loss": 2.3755,
+      "step": 15080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018567637420852762,
+      "loss": 2.41,
+      "step": 15088
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018566094293445297,
+      "loss": 2.4029,
+      "step": 15096
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018564550407201442,
+      "loss": 2.3754,
+      "step": 15104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018563005762276045,
+      "loss": 2.3788,
+      "step": 15112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018561460358824022,
+      "loss": 2.3852,
+      "step": 15120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018559914197000376,
+      "loss": 2.3849,
+      "step": 15128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018558367276960168,
+      "loss": 2.4004,
+      "step": 15136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018556819598858555,
+      "loss": 2.3679,
+      "step": 15144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018555271162850752,
+      "loss": 2.3715,
+      "step": 15152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018553721969092062,
+      "loss": 2.3428,
+      "step": 15160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001855217201773786,
+      "loss": 2.3644,
+      "step": 15168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018550621308943598,
+      "loss": 2.4002,
+      "step": 15176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018549069842864805,
+      "loss": 2.3878,
+      "step": 15184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001854751761965708,
+      "loss": 2.3729,
+      "step": 15192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018545964639476105,
+      "loss": 2.3946,
+      "step": 15200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018544410902477634,
+      "loss": 2.3885,
+      "step": 15208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000185428564088175,
+      "loss": 2.383,
+      "step": 15216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018541301158651608,
+      "loss": 2.3915,
+      "step": 15224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001853974515213594,
+      "loss": 2.3809,
+      "step": 15232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018538188389426554,
+      "loss": 2.3801,
+      "step": 15240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001853663087067959,
+      "loss": 2.3656,
+      "step": 15248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018535072596051255,
+      "loss": 2.3878,
+      "step": 15256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001853351356569783,
+      "loss": 2.4025,
+      "step": 15264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018531953779775683,
+      "loss": 2.3873,
+      "step": 15272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018530393238441251,
+      "loss": 2.3677,
+      "step": 15280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018528831941851047,
+      "loss": 2.3819,
+      "step": 15288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018527269890161662,
+      "loss": 2.3778,
+      "step": 15296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018525707083529757,
+      "loss": 2.3795,
+      "step": 15304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018524143522112075,
+      "loss": 2.381,
+      "step": 15312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001852257920606543,
+      "loss": 2.3834,
+      "step": 15320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001852101413554672,
+      "loss": 2.3797,
+      "step": 15328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018519448310712905,
+      "loss": 2.3915,
+      "step": 15336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018517881731721034,
+      "loss": 2.3796,
+      "step": 15344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018516314398728224,
+      "loss": 2.3695,
+      "step": 15352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018514746311891668,
+      "loss": 2.3761,
+      "step": 15360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018513177471368638,
+      "loss": 2.398,
+      "step": 15368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001851160787731648,
+      "loss": 2.3725,
+      "step": 15376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018510037529892613,
+      "loss": 2.3802,
+      "step": 15384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018508466429254542,
+      "loss": 2.3788,
+      "step": 15392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018506894575559825,
+      "loss": 2.3796,
+      "step": 15400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018505321968966128,
+      "loss": 2.3792,
+      "step": 15408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001850374860963116,
+      "loss": 2.3848,
+      "step": 15416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018502174497712724,
+      "loss": 2.3791,
+      "step": 15424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018500599633368693,
+      "loss": 2.3767,
+      "step": 15432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018499024016757027,
+      "loss": 2.3721,
+      "step": 15440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018497447648035736,
+      "loss": 2.3896,
+      "step": 15448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018495870527362936,
+      "loss": 2.4002,
+      "step": 15456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018494292654896793,
+      "loss": 2.3835,
+      "step": 15464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001849271403079556,
+      "loss": 2.3942,
+      "step": 15472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001849113465521757,
+      "loss": 2.4037,
+      "step": 15480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018489554528321216,
+      "loss": 2.3669,
+      "step": 15488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018487973650264985,
+      "loss": 2.4058,
+      "step": 15496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018486392021207422,
+      "loss": 2.3779,
+      "step": 15504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018484809641307163,
+      "loss": 2.3867,
+      "step": 15512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001848322651072291,
+      "loss": 2.3882,
+      "step": 15520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018481642629613435,
+      "loss": 2.3751,
+      "step": 15528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000184800579981376,
+      "loss": 2.3875,
+      "step": 15536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018478472616454333,
+      "loss": 2.3588,
+      "step": 15544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018476886484722636,
+      "loss": 2.3829,
+      "step": 15552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018475299603101593,
+      "loss": 2.3881,
+      "step": 15560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018473711971750355,
+      "loss": 2.4098,
+      "step": 15568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001847212359082816,
+      "loss": 2.3789,
+      "step": 15576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000184705344604943,
+      "loss": 2.3727,
+      "step": 15584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018468944580908171,
+      "loss": 2.3799,
+      "step": 15592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846735395222922,
+      "loss": 2.3727,
+      "step": 15600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846576257461698,
+      "loss": 2.3747,
+      "step": 15608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018464170448231055,
+      "loss": 2.3403,
+      "step": 15616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846257757323113,
+      "loss": 2.383,
+      "step": 15624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018460983949776958,
+      "loss": 2.3975,
+      "step": 15632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018459389578028373,
+      "loss": 2.3895,
+      "step": 15640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018457794458145283,
+      "loss": 2.3725,
+      "step": 15648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018456198590287664,
+      "loss": 2.3772,
+      "step": 15656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018454601974615575,
+      "loss": 2.3967,
+      "step": 15664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001845300461128915,
+      "loss": 2.3875,
+      "step": 15672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018451406500468595,
+      "loss": 2.3831,
+      "step": 15680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018449807642314188,
+      "loss": 2.3767,
+      "step": 15688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018448208036986286,
+      "loss": 2.3836,
+      "step": 15696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018446607684645324,
+      "loss": 2.3718,
+      "step": 15704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018445006585451806,
+      "loss": 2.3818,
+      "step": 15712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001844340473956631,
+      "loss": 2.394,
+      "step": 15720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000184418021471495,
+      "loss": 2.3908,
+      "step": 15728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018440198808362097,
+      "loss": 2.3763,
+      "step": 15736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018438594723364917,
+      "loss": 2.3853,
+      "step": 15744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001843698989231883,
+      "loss": 2.3745,
+      "step": 15752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018435384315384804,
+      "loss": 2.3803,
+      "step": 15760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001843377799272386,
+      "loss": 2.3671,
+      "step": 15768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018432170924497106,
+      "loss": 2.3745,
+      "step": 15776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001843056311086572,
+      "loss": 2.3819,
+      "step": 15784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018428954551990957,
+      "loss": 2.3738,
+      "step": 15792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018427345248034146,
+      "loss": 2.3784,
+      "step": 15800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018425735199156698,
+      "loss": 2.391,
+      "step": 15808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001842412440552008,
+      "loss": 2.3849,
+      "step": 15816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018422512867285855,
+      "loss": 2.3773,
+      "step": 15824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018420900584615647,
+      "loss": 2.3881,
+      "step": 15832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001841928755767116,
+      "loss": 2.392,
+      "step": 15840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001841767378661417,
+      "loss": 2.393,
+      "step": 15848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018416059271606527,
+      "loss": 2.3561,
+      "step": 15856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018414444012810163,
+      "loss": 2.3845,
+      "step": 15864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018412828010387077,
+      "loss": 2.3522,
+      "step": 15872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018411211264499346,
+      "loss": 2.3897,
+      "step": 15880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001840959377530912,
+      "loss": 2.3724,
+      "step": 15888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018407975542978624,
+      "loss": 2.4019,
+      "step": 15896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018406356567670154,
+      "loss": 2.3802,
+      "step": 15904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001840473684954609,
+      "loss": 2.3986,
+      "step": 15912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018403116388768877,
+      "loss": 2.3751,
+      "step": 15920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018401495185501037,
+      "loss": 2.3934,
+      "step": 15928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001839987323990517,
+      "loss": 2.3873,
+      "step": 15936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018398250552143945,
+      "loss": 2.3821,
+      "step": 15944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018396627122380116,
+      "loss": 2.3839,
+      "step": 15952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001839500295077649,
+      "loss": 2.3619,
+      "step": 15960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018393378037495975,
+      "loss": 2.3718,
+      "step": 15968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018391752382701537,
+      "loss": 2.3725,
+      "step": 15976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001839012598655622,
+      "loss": 2.3796,
+      "step": 15984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018388498849223136,
+      "loss": 2.3857,
+      "step": 15992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001838687097086549,
+      "loss": 2.3749,
+      "step": 16000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018385242351646533,
+      "loss": 2.3798,
+      "step": 16008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018383612991729623,
+      "loss": 2.3691,
+      "step": 16016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018381982891278164,
+      "loss": 2.3696,
+      "step": 16024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018380352050455652,
+      "loss": 2.3618,
+      "step": 16032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018378720469425648,
+      "loss": 2.3924,
+      "step": 16040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001837708814835179,
+      "loss": 2.3676,
+      "step": 16048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018375455087397792,
+      "loss": 2.3559,
+      "step": 16056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001837382128672744,
+      "loss": 2.384,
+      "step": 16064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018372186746504597,
+      "loss": 2.3827,
+      "step": 16072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018370551466893195,
+      "loss": 2.3815,
+      "step": 16080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018368915448057244,
+      "loss": 2.3795,
+      "step": 16088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001836727869016083,
+      "loss": 2.3768,
+      "step": 16096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001836564119336811,
+      "loss": 2.3837,
+      "step": 16104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018364002957843312,
+      "loss": 2.3766,
+      "step": 16112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018362363983750744,
+      "loss": 2.3723,
+      "step": 16120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018360724271254783,
+      "loss": 2.3822,
+      "step": 16128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001835908382051989,
+      "loss": 2.3849,
+      "step": 16136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018357442631710587,
+      "loss": 2.3693,
+      "step": 16144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018355800704991478,
+      "loss": 2.387,
+      "step": 16152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001835415804052724,
+      "loss": 2.3625,
+      "step": 16160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018352514638482622,
+      "loss": 2.3718,
+      "step": 16168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018350870499022443,
+      "loss": 2.3725,
+      "step": 16176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001834922562231161,
+      "loss": 2.3827,
+      "step": 16184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018347580008515086,
+      "loss": 2.3833,
+      "step": 16192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018345933657797924,
+      "loss": 2.3621,
+      "step": 16200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018344286570325238,
+      "loss": 2.3729,
+      "step": 16208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018342638746262225,
+      "loss": 2.3715,
+      "step": 16216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018340990185774153,
+      "loss": 2.367,
+      "step": 16224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001833934088902636,
+      "loss": 2.3655,
+      "step": 16232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001833769085618426,
+      "loss": 2.367,
+      "step": 16240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001833604008741335,
+      "loss": 2.3616,
+      "step": 16248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018334388582879187,
+      "loss": 2.3609,
+      "step": 16256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018332736342747404,
+      "loss": 2.3617,
+      "step": 16264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018331083367183722,
+      "loss": 2.3633,
+      "step": 16272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001832942965635391,
+      "loss": 2.3614,
+      "step": 16280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001832777521042384,
+      "loss": 2.3657,
+      "step": 16288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018326120029559437,
+      "loss": 2.3888,
+      "step": 16296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018324464113926708,
+      "loss": 2.3891,
+      "step": 16304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001832280746369173,
+      "loss": 2.3752,
+      "step": 16312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018321150079020655,
+      "loss": 2.3781,
+      "step": 16320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018319491960079714,
+      "loss": 2.3646,
+      "step": 16328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000183178331070352,
+      "loss": 2.3754,
+      "step": 16336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018316173520053494,
+      "loss": 2.3706,
+      "step": 16344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001831451319930104,
+      "loss": 2.3674,
+      "step": 16352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018312852144944356,
+      "loss": 2.3666,
+      "step": 16360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001831119035715004,
+      "loss": 2.3627,
+      "step": 16368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001830952783608476,
+      "loss": 2.3828,
+      "step": 16376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018307864581915253,
+      "loss": 2.3803,
+      "step": 16384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001830620059480834,
+      "loss": 2.3684,
+      "step": 16392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018304535874930905,
+      "loss": 2.3804,
+      "step": 16400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018302870422449912,
+      "loss": 2.3843,
+      "step": 16408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018301204237532396,
+      "loss": 2.3739,
+      "step": 16416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018299537320345466,
+      "loss": 2.398,
+      "step": 16424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018297869671056303,
+      "loss": 2.3739,
+      "step": 16432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018296201289832163,
+      "loss": 2.3642,
+      "step": 16440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001829453217684038,
+      "loss": 2.3537,
+      "step": 16448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001829286233224835,
+      "loss": 2.3654,
+      "step": 16456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001829119175622355,
+      "loss": 2.3842,
+      "step": 16464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018289520448933538,
+      "loss": 2.372,
+      "step": 16472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018287848410545923,
+      "loss": 2.385,
+      "step": 16480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828617564122841,
+      "loss": 2.4041,
+      "step": 16488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018284502141148766,
+      "loss": 2.3851,
+      "step": 16496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018282827910474832,
+      "loss": 2.3728,
+      "step": 16504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018281152949374527,
+      "loss": 2.3858,
+      "step": 16512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001827947725801584,
+      "loss": 2.3586,
+      "step": 16520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001827780083656683,
+      "loss": 2.3819,
+      "step": 16528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018276123685195632,
+      "loss": 2.3959,
+      "step": 16536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018274445804070458,
+      "loss": 2.3927,
+      "step": 16544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018272767193359594,
+      "loss": 2.3581,
+      "step": 16552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018271087853231386,
+      "loss": 2.3796,
+      "step": 16560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018269407783854267,
+      "loss": 2.3681,
+      "step": 16568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001826772698539674,
+      "loss": 2.3837,
+      "step": 16576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018266045458027375,
+      "loss": 2.3791,
+      "step": 16584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018264363201914822,
+      "loss": 2.3623,
+      "step": 16592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018262680217227806,
+      "loss": 2.3698,
+      "step": 16600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018260996504135114,
+      "loss": 2.3674,
+      "step": 16608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018259312062805615,
+      "loss": 2.3728,
+      "step": 16616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001825762689340825,
+      "loss": 2.3722,
+      "step": 16624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001825594099611203,
+      "loss": 2.3819,
+      "step": 16632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001825425437108605,
+      "loss": 2.3665,
+      "step": 16640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018252567018499453,
+      "loss": 2.3813,
+      "step": 16648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001825087893852148,
+      "loss": 2.361,
+      "step": 16656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001824919013132144,
+      "loss": 2.3604,
+      "step": 16664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018247500597068703,
+      "loss": 2.3613,
+      "step": 16672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018245810335932723,
+      "loss": 2.3853,
+      "step": 16680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018244119348083024,
+      "loss": 2.3532,
+      "step": 16688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018242427633689202,
+      "loss": 2.3426,
+      "step": 16696
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018240735192920923,
+      "loss": 2.3635,
+      "step": 16704
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018239042025947936,
+      "loss": 2.3453,
+      "step": 16712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018237348132940054,
+      "loss": 2.3877,
+      "step": 16720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001823565351406716,
+      "loss": 2.3681,
+      "step": 16728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001823395816949922,
+      "loss": 2.371,
+      "step": 16736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001823226209940626,
+      "loss": 2.3846,
+      "step": 16744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000182305653039584,
+      "loss": 2.3735,
+      "step": 16752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018228867783325804,
+      "loss": 2.3651,
+      "step": 16760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018227169537678734,
+      "loss": 2.3645,
+      "step": 16768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001822547056718751,
+      "loss": 2.3768,
+      "step": 16776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001822377087202253,
+      "loss": 2.3717,
+      "step": 16784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018222070452354264,
+      "loss": 2.3736,
+      "step": 16792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018220369308353255,
+      "loss": 2.3747,
+      "step": 16800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018218667440190114,
+      "loss": 2.3809,
+      "step": 16808
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018216964848035533,
+      "loss": 2.368,
+      "step": 16816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018215261532060273,
+      "loss": 2.3652,
+      "step": 16824
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018213557492435166,
+      "loss": 2.3799,
+      "step": 16832
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018211852729331114,
+      "loss": 2.3597,
+      "step": 16840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018210147242919098,
+      "loss": 2.378,
+      "step": 16848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018208441033370167,
+      "loss": 2.365,
+      "step": 16856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018206734100855451,
+      "loss": 2.3826,
+      "step": 16864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018205026445546136,
+      "loss": 2.3758,
+      "step": 16872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018203318067613495,
+      "loss": 2.3749,
+      "step": 16880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001820160896722887,
+      "loss": 2.3445,
+      "step": 16888
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001819989914456367,
+      "loss": 2.3681,
+      "step": 16896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018198188599789384,
+      "loss": 2.3507,
+      "step": 16904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018196477333077567,
+      "loss": 2.3637,
+      "step": 16912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018194765344599852,
+      "loss": 2.3783,
+      "step": 16920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018193052634527942,
+      "loss": 2.3551,
+      "step": 16928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001819133920303361,
+      "loss": 2.3797,
+      "step": 16936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018189625050288706,
+      "loss": 2.3521,
+      "step": 16944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018187910176465147,
+      "loss": 2.3728,
+      "step": 16952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018186194581734923,
+      "loss": 2.3808,
+      "step": 16960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018184478266270108,
+      "loss": 2.365,
+      "step": 16968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018182761230242826,
+      "loss": 2.353,
+      "step": 16976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018181043473825293,
+      "loss": 2.3863,
+      "step": 16984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018179324997189792,
+      "loss": 2.3677,
+      "step": 16992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018177605800508674,
+      "loss": 2.3686,
+      "step": 17000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018175885883954364,
+      "loss": 2.3806,
+      "step": 17008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018174165247699362,
+      "loss": 2.3568,
+      "step": 17016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018172443891916235,
+      "loss": 2.3748,
+      "step": 17024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018170721816777627,
+      "loss": 2.3875,
+      "step": 17032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018168999022456255,
+      "loss": 2.3762,
+      "step": 17040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000181672755091249,
+      "loss": 2.3734,
+      "step": 17048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018165551276956424,
+      "loss": 2.3755,
+      "step": 17056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001816382632612376,
+      "loss": 2.3695,
+      "step": 17064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018162100656799905,
+      "loss": 2.3571,
+      "step": 17072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018160374269157937,
+      "loss": 2.361,
+      "step": 17080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018158647163371002,
+      "loss": 2.3679,
+      "step": 17088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001815691933961232,
+      "loss": 2.3724,
+      "step": 17096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018155190798055184,
+      "loss": 2.3684,
+      "step": 17104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018153461538872952,
+      "loss": 2.3734,
+      "step": 17112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001815173156223906,
+      "loss": 2.3753,
+      "step": 17120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018150000868327022,
+      "loss": 2.364,
+      "step": 17128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018148269457310406,
+      "loss": 2.3685,
+      "step": 17136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018146537329362869,
+      "loss": 2.356,
+      "step": 17144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018144804484658132,
+      "loss": 2.3608,
+      "step": 17152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001814307092336999,
+      "loss": 2.3891,
+      "step": 17160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001814133664567231,
+      "loss": 2.3568,
+      "step": 17168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001813960165173903,
+      "loss": 2.3695,
+      "step": 17176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001813786594174416,
+      "loss": 2.3599,
+      "step": 17184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001813612951586178,
+      "loss": 2.3915,
+      "step": 17192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018134392374266045,
+      "loss": 2.366,
+      "step": 17200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001813265451713118,
+      "loss": 2.3685,
+      "step": 17208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018130915944631487,
+      "loss": 2.3817,
+      "step": 17216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018129176656941323,
+      "loss": 2.373,
+      "step": 17224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018127436654235142,
+      "loss": 2.3615,
+      "step": 17232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001812569593668745,
+      "loss": 2.3635,
+      "step": 17240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018123954504472832,
+      "loss": 2.3688,
+      "step": 17248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018122212357765945,
+      "loss": 2.3604,
+      "step": 17256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001812046949674151,
+      "loss": 2.3814,
+      "step": 17264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018118725921574338,
+      "loss": 2.3913,
+      "step": 17272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018116981632439286,
+      "loss": 2.3708,
+      "step": 17280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001811523662951131,
+      "loss": 2.3765,
+      "step": 17288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018113490912965416,
+      "loss": 2.3437,
+      "step": 17296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018111744482976686,
+      "loss": 2.3527,
+      "step": 17304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018109997339720284,
+      "loss": 2.3859,
+      "step": 17312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018108249483371437,
+      "loss": 2.376,
+      "step": 17320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018106500914105445,
+      "loss": 2.3711,
+      "step": 17328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810475163209768,
+      "loss": 2.3689,
+      "step": 17336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018103001637523585,
+      "loss": 2.351,
+      "step": 17344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810125093055867,
+      "loss": 2.3794,
+      "step": 17352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018099499511378526,
+      "loss": 2.343,
+      "step": 17360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018097747380158808,
+      "loss": 2.3409,
+      "step": 17368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018095994537075245,
+      "loss": 2.3678,
+      "step": 17376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018094240982303642,
+      "loss": 2.3556,
+      "step": 17384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018092486716019863,
+      "loss": 2.3367,
+      "step": 17392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018090731738399857,
+      "loss": 2.3833,
+      "step": 17400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001808897604961963,
+      "loss": 2.3603,
+      "step": 17408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001808721964985528,
+      "loss": 2.3486,
+      "step": 17416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018085462539282955,
+      "loss": 2.3676,
+      "step": 17424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001808370471807889,
+      "loss": 2.3633,
+      "step": 17432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018081946186419373,
+      "loss": 2.374,
+      "step": 17440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001808018694448079,
+      "loss": 2.3513,
+      "step": 17448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001807842699243957,
+      "loss": 2.3595,
+      "step": 17456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018076666330472233,
+      "loss": 2.3797,
+      "step": 17464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018074904958755356,
+      "loss": 2.3481,
+      "step": 17472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018073142877465608,
+      "loss": 2.3589,
+      "step": 17480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018071380086779704,
+      "loss": 2.3441,
+      "step": 17488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018069616586874446,
+      "loss": 2.368,
+      "step": 17496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018067852377926704,
+      "loss": 2.3802,
+      "step": 17504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018066087460113419,
+      "loss": 2.3592,
+      "step": 17512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018064321833611598,
+      "loss": 2.3723,
+      "step": 17520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018062555498598326,
+      "loss": 2.3701,
+      "step": 17528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018060788455250758,
+      "loss": 2.3631,
+      "step": 17536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018059020703746115,
+      "loss": 2.36,
+      "step": 17544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001805725224426169,
+      "loss": 2.3722,
+      "step": 17552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001805548307697486,
+      "loss": 2.355,
+      "step": 17560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018053713202063055,
+      "loss": 2.375,
+      "step": 17568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018051942619703784,
+      "loss": 2.3581,
+      "step": 17576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018050171330074623,
+      "loss": 2.3616,
+      "step": 17584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001804839933335323,
+      "loss": 2.345,
+      "step": 17592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001804662662971732,
+      "loss": 2.367,
+      "step": 17600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018044853219344688,
+      "loss": 2.3672,
+      "step": 17608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018043079102413197,
+      "loss": 2.3736,
+      "step": 17616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001804130427910078,
+      "loss": 2.3985,
+      "step": 17624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018039528749585443,
+      "loss": 2.3715,
+      "step": 17632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001803775251404526,
+      "loss": 2.3789,
+      "step": 17640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018035975572658377,
+      "loss": 2.3686,
+      "step": 17648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018034197925603015,
+      "loss": 2.367,
+      "step": 17656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018032419573057456,
+      "loss": 2.3625,
+      "step": 17664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001803064051520006,
+      "loss": 2.3787,
+      "step": 17672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018028860752209263,
+      "loss": 2.3771,
+      "step": 17680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018027080284263556,
+      "loss": 2.3664,
+      "step": 17688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018025299111541516,
+      "loss": 2.3747,
+      "step": 17696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018023517234221783,
+      "loss": 2.3469,
+      "step": 17704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018021734652483068,
+      "loss": 2.3606,
+      "step": 17712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018019951366504157,
+      "loss": 2.3712,
+      "step": 17720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018018167376463899,
+      "loss": 2.368,
+      "step": 17728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001801638268254122,
+      "loss": 2.3606,
+      "step": 17736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018014597284915116,
+      "loss": 2.3681,
+      "step": 17744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001801281118376465,
+      "loss": 2.3725,
+      "step": 17752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001801102437926896,
+      "loss": 2.354,
+      "step": 17760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001800923687160725,
+      "loss": 2.3515,
+      "step": 17768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018007448660958803,
+      "loss": 2.3513,
+      "step": 17776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018005659747502957,
+      "loss": 2.3711,
+      "step": 17784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018003870131419136,
+      "loss": 2.3651,
+      "step": 17792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001800207981288683,
+      "loss": 2.3764,
+      "step": 17800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001800028879208559,
+      "loss": 2.3589,
+      "step": 17808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017998497069195052,
+      "loss": 2.3654,
+      "step": 17816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017996704644394912,
+      "loss": 2.3619,
+      "step": 17824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017994911517864946,
+      "loss": 2.3661,
+      "step": 17832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017993117689784984,
+      "loss": 2.3719,
+      "step": 17840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017991323160334945,
+      "loss": 2.3806,
+      "step": 17848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001798952792969481,
+      "loss": 2.3697,
+      "step": 17856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017987731998044627,
+      "loss": 2.3605,
+      "step": 17864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001798593536556452,
+      "loss": 2.3587,
+      "step": 17872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001798413803243468,
+      "loss": 2.3695,
+      "step": 17880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017982339998835368,
+      "loss": 2.3715,
+      "step": 17888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001798054126494692,
+      "loss": 2.3741,
+      "step": 17896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017978741830949738,
+      "loss": 2.3634,
+      "step": 17904
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017976941697024295,
+      "loss": 2.337,
+      "step": 17912
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001797514086335113,
+      "loss": 2.3751,
+      "step": 17920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017973339330110862,
+      "loss": 2.3771,
+      "step": 17928
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001797153709748417,
+      "loss": 2.3653,
+      "step": 17936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017969734165651814,
+      "loss": 2.3601,
+      "step": 17944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001796793053479461,
+      "loss": 2.3713,
+      "step": 17952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017966126205093457,
+      "loss": 2.3614,
+      "step": 17960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017964321176729318,
+      "loss": 2.3672,
+      "step": 17968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017962515449883225,
+      "loss": 2.3607,
+      "step": 17976
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001796070902473629,
+      "loss": 2.3666,
+      "step": 17984
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017958901901469676,
+      "loss": 2.3635,
+      "step": 17992
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017957094080264634,
+      "loss": 2.3672,
+      "step": 18000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017955285561302477,
+      "loss": 2.3717,
+      "step": 18008
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001795347634476459,
+      "loss": 2.368,
+      "step": 18016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017951666430832423,
+      "loss": 2.3663,
+      "step": 18024
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017949855819687506,
+      "loss": 2.367,
+      "step": 18032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001794804451151143,
+      "loss": 2.3878,
+      "step": 18040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017946232506485857,
+      "loss": 2.3693,
+      "step": 18048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017944419804792527,
+      "loss": 2.347,
+      "step": 18056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017942606406613242,
+      "loss": 2.346,
+      "step": 18064
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001794079231212987,
+      "loss": 2.3581,
+      "step": 18072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017938977521524355,
+      "loss": 2.3576,
+      "step": 18080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017937162034978714,
+      "loss": 2.3717,
+      "step": 18088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001793534585267503,
+      "loss": 2.3828,
+      "step": 18096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017933528974795456,
+      "loss": 2.3749,
+      "step": 18104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001793171140152221,
+      "loss": 2.3684,
+      "step": 18112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017929893133037593,
+      "loss": 2.3748,
+      "step": 18120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017928074169523957,
+      "loss": 2.3783,
+      "step": 18128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001792625451116374,
+      "loss": 2.355,
+      "step": 18136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017924434158139442,
+      "loss": 2.3709,
+      "step": 18144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017922613110633634,
+      "loss": 2.3436,
+      "step": 18152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001792079136882896,
+      "loss": 2.3698,
+      "step": 18160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001791896893290812,
+      "loss": 2.3607,
+      "step": 18168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017917145803053906,
+      "loss": 2.353,
+      "step": 18176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017915321979449162,
+      "loss": 2.3529,
+      "step": 18184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017913497462276808,
+      "loss": 2.3774,
+      "step": 18192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001791167225171983,
+      "loss": 2.363,
+      "step": 18200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017909846347961294,
+      "loss": 2.3738,
+      "step": 18208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017908019751184317,
+      "loss": 2.362,
+      "step": 18216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000179061924615721,
+      "loss": 2.374,
+      "step": 18224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001790436447930792,
+      "loss": 2.3554,
+      "step": 18232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017902535804575104,
+      "loss": 2.3519,
+      "step": 18240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017900706437557054,
+      "loss": 2.3493,
+      "step": 18248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017898876378437251,
+      "loss": 2.3605,
+      "step": 18256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017897045627399241,
+      "loss": 2.355,
+      "step": 18264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017895214184626633,
+      "loss": 2.3568,
+      "step": 18272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017893382050303112,
+      "loss": 2.3796,
+      "step": 18280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017891549224612435,
+      "loss": 2.3619,
+      "step": 18288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017889715707738416,
+      "loss": 2.3586,
+      "step": 18296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017887881499864953,
+      "loss": 2.3753,
+      "step": 18304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017886046601176006,
+      "loss": 2.3635,
+      "step": 18312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000178842110118556,
+      "loss": 2.3602,
+      "step": 18320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001788237473208784,
+      "loss": 2.3755,
+      "step": 18328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017880537762056894,
+      "loss": 2.3766,
+      "step": 18336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017878700101946993,
+      "loss": 2.3753,
+      "step": 18344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001787686175194245,
+      "loss": 2.3663,
+      "step": 18352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017875022712227642,
+      "loss": 2.3524,
+      "step": 18360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017873182982987008,
+      "loss": 2.335,
+      "step": 18368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001787134256440507,
+      "loss": 2.3788,
+      "step": 18376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001786950145666641,
+      "loss": 2.3763,
+      "step": 18384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017867659659955674,
+      "loss": 2.3478,
+      "step": 18392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001786581717445759,
+      "loss": 2.3608,
+      "step": 18400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017863974000356946,
+      "loss": 2.3671,
+      "step": 18408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017862130137838604,
+      "loss": 2.3752,
+      "step": 18416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017860285587087494,
+      "loss": 2.3711,
+      "step": 18424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001785844034828861,
+      "loss": 2.3499,
+      "step": 18432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017856594421627022,
+      "loss": 2.3592,
+      "step": 18440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017854747807287866,
+      "loss": 2.365,
+      "step": 18448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017852900505456346,
+      "loss": 2.3702,
+      "step": 18456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017851052516317735,
+      "loss": 2.3503,
+      "step": 18464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017849203840057375,
+      "loss": 2.3636,
+      "step": 18472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001784735447686068,
+      "loss": 2.3577,
+      "step": 18480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017845504426913133,
+      "loss": 2.3783,
+      "step": 18488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017843653690400278,
+      "loss": 2.3564,
+      "step": 18496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001784180226750774,
+      "loss": 2.3723,
+      "step": 18504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000178399501584212,
+      "loss": 2.3573,
+      "step": 18512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017838097363326416,
+      "loss": 2.3641,
+      "step": 18520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001783624388240921,
+      "loss": 2.3603,
+      "step": 18528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017834389715855485,
+      "loss": 2.351,
+      "step": 18536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017832534863851195,
+      "loss": 2.3676,
+      "step": 18544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017830679326582373,
+      "loss": 2.3389,
+      "step": 18552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017828823104235118,
+      "loss": 2.368,
+      "step": 18560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017826966196995603,
+      "loss": 2.3504,
+      "step": 18568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001782510860505006,
+      "loss": 2.3597,
+      "step": 18576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017823250328584798,
+      "loss": 2.3658,
+      "step": 18584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001782139136778619,
+      "loss": 2.3667,
+      "step": 18592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001781953172284068,
+      "loss": 2.3673,
+      "step": 18600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001781767139393478,
+      "loss": 2.3695,
+      "step": 18608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001781581038125507,
+      "loss": 2.3714,
+      "step": 18616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000178139486849882,
+      "loss": 2.3517,
+      "step": 18624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017812086305320885,
+      "loss": 2.3468,
+      "step": 18632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017810223242439917,
+      "loss": 2.3598,
+      "step": 18640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017808359496532144,
+      "loss": 2.3608,
+      "step": 18648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017806495067784493,
+      "loss": 2.354,
+      "step": 18656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017804629956383948,
+      "loss": 2.359,
+      "step": 18664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017802764162517582,
+      "loss": 2.3571,
+      "step": 18672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017800897686372517,
+      "loss": 2.3655,
+      "step": 18680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017799030528135949,
+      "loss": 2.3576,
+      "step": 18688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017797162687995143,
+      "loss": 2.348,
+      "step": 18696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017795294166137438,
+      "loss": 2.3568,
+      "step": 18704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001779342496275023,
+      "loss": 2.342,
+      "step": 18712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001779155507802099,
+      "loss": 2.3503,
+      "step": 18720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001778968451213726,
+      "loss": 2.3769,
+      "step": 18728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001778781326528665,
+      "loss": 2.3731,
+      "step": 18736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017785941337656827,
+      "loss": 2.3486,
+      "step": 18744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001778406872943554,
+      "loss": 2.3552,
+      "step": 18752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017782195440810598,
+      "loss": 2.3627,
+      "step": 18760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017780321471969884,
+      "loss": 2.3813,
+      "step": 18768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001777844682310135,
+      "loss": 2.3415,
+      "step": 18776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017776571494393007,
+      "loss": 2.3774,
+      "step": 18784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001777469548603294,
+      "loss": 2.3631,
+      "step": 18792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017772818798209304,
+      "loss": 2.3417,
+      "step": 18800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017770941431110318,
+      "loss": 2.3654,
+      "step": 18808
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017769063384924272,
+      "loss": 2.3642,
+      "step": 18816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001776718465983953,
+      "loss": 2.3471,
+      "step": 18824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000177653052560445,
+      "loss": 2.3692,
+      "step": 18832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017763425173727697,
+      "loss": 2.3617,
+      "step": 18840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001776154441307767,
+      "loss": 2.3598,
+      "step": 18848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001775966297428305,
+      "loss": 2.3466,
+      "step": 18856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001775778085753254,
+      "loss": 2.3757,
+      "step": 18864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000177558980630149,
+      "loss": 2.354,
+      "step": 18872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001775401459091896,
+      "loss": 2.3602,
+      "step": 18880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017752130441433633,
+      "loss": 2.3486,
+      "step": 18888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001775024561474788,
+      "loss": 2.3768,
+      "step": 18896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017748360111050743,
+      "loss": 2.3586,
+      "step": 18904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017746473930531324,
+      "loss": 2.3485,
+      "step": 18912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017744587073378798,
+      "loss": 2.3452,
+      "step": 18920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017742699539782407,
+      "loss": 2.368,
+      "step": 18928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017740811329931453,
+      "loss": 2.3618,
+      "step": 18936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017738922444015322,
+      "loss": 2.3616,
+      "step": 18944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017737032882223454,
+      "loss": 2.3601,
+      "step": 18952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017735142644745362,
+      "loss": 2.3789,
+      "step": 18960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017733251731770627,
+      "loss": 2.3681,
+      "step": 18968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017731360143488897,
+      "loss": 2.3673,
+      "step": 18976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017729467880089883,
+      "loss": 2.3599,
+      "step": 18984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017727574941763373,
+      "loss": 2.3417,
+      "step": 18992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017725681328699215,
+      "loss": 2.3678,
+      "step": 19000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017723787041087335,
+      "loss": 2.352,
+      "step": 19008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017721892079117707,
+      "loss": 2.3614,
+      "step": 19016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017719996442980398,
+      "loss": 2.3612,
+      "step": 19024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017718100132865523,
+      "loss": 2.3615,
+      "step": 19032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017716203148963267,
+      "loss": 2.3527,
+      "step": 19040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017714305491463894,
+      "loss": 2.3635,
+      "step": 19048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017712407160557726,
+      "loss": 2.3474,
+      "step": 19056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017710508156435154,
+      "loss": 2.3694,
+      "step": 19064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001770860847928664,
+      "loss": 2.3654,
+      "step": 19072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017706708129302704,
+      "loss": 2.3675,
+      "step": 19080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017704807106673945,
+      "loss": 2.3614,
+      "step": 19088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017702905411591025,
+      "loss": 2.3551,
+      "step": 19096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017701003044244678,
+      "loss": 2.3585,
+      "step": 19104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017699100004825693,
+      "loss": 2.3714,
+      "step": 19112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017697196293524936,
+      "loss": 2.3416,
+      "step": 19120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017695291910533337,
+      "loss": 2.3811,
+      "step": 19128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017693386856041903,
+      "loss": 2.3385,
+      "step": 19136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001769148113024169,
+      "loss": 2.3523,
+      "step": 19144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017689574733323834,
+      "loss": 2.3547,
+      "step": 19152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017687667665479542,
+      "loss": 2.372,
+      "step": 19160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017685759926900073,
+      "loss": 2.3551,
+      "step": 19168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001768385151777677,
+      "loss": 2.3483,
+      "step": 19176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001768194243830103,
+      "loss": 2.3489,
+      "step": 19184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017680032688664328,
+      "loss": 2.3497,
+      "step": 19192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000176781222690582,
+      "loss": 2.3784,
+      "step": 19200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017676211179674242,
+      "loss": 2.3566,
+      "step": 19208
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001767429942070414,
+      "loss": 2.3713,
+      "step": 19216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001767238699233962,
+      "loss": 2.364,
+      "step": 19224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017670473894772492,
+      "loss": 2.3437,
+      "step": 19232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017668560128194635,
+      "loss": 2.3489,
+      "step": 19240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017666645692797977,
+      "loss": 2.3529,
+      "step": 19248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017664730588774533,
+      "loss": 2.3495,
+      "step": 19256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017662814816316378,
+      "loss": 2.3709,
+      "step": 19264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017660898375615645,
+      "loss": 2.3841,
+      "step": 19272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017658981266864554,
+      "loss": 2.3774,
+      "step": 19280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017657063490255373,
+      "loss": 2.3625,
+      "step": 19288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001765514504598044,
+      "loss": 2.3602,
+      "step": 19296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001765322593423217,
+      "loss": 2.3767,
+      "step": 19304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017651306155203042,
+      "loss": 2.3594,
+      "step": 19312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017649385709085592,
+      "loss": 2.3774,
+      "step": 19320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017647464596072437,
+      "loss": 2.3725,
+      "step": 19328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017645542816356244,
+      "loss": 2.3644,
+      "step": 19336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017643620370129764,
+      "loss": 2.3864,
+      "step": 19344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017641697257585808,
+      "loss": 2.3797,
+      "step": 19352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017639773478917247,
+      "loss": 2.3655,
+      "step": 19360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017637849034317032,
+      "loss": 2.3545,
+      "step": 19368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017635923923978172,
+      "loss": 2.3781,
+      "step": 19376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017633998148093745,
+      "loss": 2.3663,
+      "step": 19384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001763207170685689,
+      "loss": 2.353,
+      "step": 19392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017630144600460824,
+      "loss": 2.361,
+      "step": 19400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001762821682909882,
+      "loss": 2.3603,
+      "step": 19408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001762628839296423,
+      "loss": 2.3727,
+      "step": 19416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017624359292250458,
+      "loss": 2.3707,
+      "step": 19424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017622429527150985,
+      "loss": 2.3632,
+      "step": 19432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017620499097859357,
+      "loss": 2.3458,
+      "step": 19440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017618568004569183,
+      "loss": 2.3635,
+      "step": 19448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001761663624747414,
+      "loss": 2.3594,
+      "step": 19456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017614703826767972,
+      "loss": 2.3644,
+      "step": 19464
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017612770742644491,
+      "loss": 2.3531,
+      "step": 19472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017610836995297577,
+      "loss": 2.363,
+      "step": 19480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001760890258492117,
+      "loss": 2.3618,
+      "step": 19488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017606967511709282,
+      "loss": 2.3622,
+      "step": 19496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017605031775855994,
+      "loss": 2.3803,
+      "step": 19504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001760309537755544,
+      "loss": 2.3765,
+      "step": 19512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017601158317001833,
+      "loss": 2.3459,
+      "step": 19520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017599220594389457,
+      "loss": 2.3813,
+      "step": 19528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017597282209912645,
+      "loss": 2.3752,
+      "step": 19536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017595343163765808,
+      "loss": 2.3742,
+      "step": 19544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017593403456143424,
+      "loss": 2.3635,
+      "step": 19552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017591463087240038,
+      "loss": 2.3537,
+      "step": 19560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001758952205725025,
+      "loss": 2.3463,
+      "step": 19568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017587580366368736,
+      "loss": 2.3652,
+      "step": 19576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017585638014790242,
+      "loss": 2.3784,
+      "step": 19584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017583695002709572,
+      "loss": 2.37,
+      "step": 19592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017581751330321595,
+      "loss": 2.3507,
+      "step": 19600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017579806997821256,
+      "loss": 2.3713,
+      "step": 19608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001757786200540356,
+      "loss": 2.3637,
+      "step": 19616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017575916353263572,
+      "loss": 2.3668,
+      "step": 19624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017573970041596438,
+      "loss": 2.3752,
+      "step": 19632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017572023070597362,
+      "loss": 2.3644,
+      "step": 19640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017570075440461608,
+      "loss": 2.3576,
+      "step": 19648
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001756812715138451,
+      "loss": 2.3859,
+      "step": 19656
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017566178203561487,
+      "loss": 2.3603,
+      "step": 19664
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017564228597187988,
+      "loss": 2.3396,
+      "step": 19672
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001756227833245956,
+      "loss": 2.3573,
+      "step": 19680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017560327409571797,
+      "loss": 2.3511,
+      "step": 19688
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017558375828720366,
+      "loss": 2.3829,
+      "step": 19696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017556423590101003,
+      "loss": 2.3708,
+      "step": 19704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017554470693909507,
+      "loss": 2.3486,
+      "step": 19712
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001755251714034174,
+      "loss": 2.362,
+      "step": 19720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017550562929593628,
+      "loss": 2.3758,
+      "step": 19728
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017548608061861176,
+      "loss": 2.3544,
+      "step": 19736
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001754665253734044,
+      "loss": 2.371,
+      "step": 19744
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001754469635622755,
+      "loss": 2.3826,
+      "step": 19752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017542739518718698,
+      "loss": 2.3571,
+      "step": 19760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001754078202501015,
+      "loss": 2.3608,
+      "step": 19768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017538823875298224,
+      "loss": 2.3789,
+      "step": 19776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001753686506977931,
+      "loss": 2.3562,
+      "step": 19784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017534905608649874,
+      "loss": 2.3731,
+      "step": 19792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017532945492106436,
+      "loss": 2.3503,
+      "step": 19800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001753098472034558,
+      "loss": 2.3792,
+      "step": 19808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017529023293563967,
+      "loss": 2.3617,
+      "step": 19816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001752706121195831,
+      "loss": 2.3721,
+      "step": 19824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017525098475725396,
+      "loss": 2.3495,
+      "step": 19832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001752313508506208,
+      "loss": 2.3503,
+      "step": 19840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017521171040165277,
+      "loss": 2.3765,
+      "step": 19848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001751920634123197,
+      "loss": 2.3723,
+      "step": 19856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017517240988459208,
+      "loss": 2.3673,
+      "step": 19864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017515274982044104,
+      "loss": 2.3582,
+      "step": 19872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017513308322183833,
+      "loss": 2.3634,
+      "step": 19880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017511341009075648,
+      "loss": 2.3637,
+      "step": 19888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001750937304291685,
+      "loss": 2.3578,
+      "step": 19896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017507404423904824,
+      "loss": 2.3792,
+      "step": 19904
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017505435152237006,
+      "loss": 2.3734,
+      "step": 19912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017503465228110905,
+      "loss": 2.3611,
+      "step": 19920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017501494651724094,
+      "loss": 2.3591,
+      "step": 19928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017499523423274204,
+      "loss": 2.3544,
+      "step": 19936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017497551542958946,
+      "loss": 2.3551,
+      "step": 19944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017495579010976087,
+      "loss": 2.3635,
+      "step": 19952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017493605827523457,
+      "loss": 2.3465,
+      "step": 19960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017491631992798957,
+      "loss": 2.3459,
+      "step": 19968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017489657507000552,
+      "loss": 2.3583,
+      "step": 19976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748768237032627,
+      "loss": 2.3568,
+      "step": 19984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748570658297421,
+      "loss": 2.3692,
+      "step": 19992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748373014514253,
+      "loss": 2.351,
+      "step": 20000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748175305702945,
+      "loss": 2.3703,
+      "step": 20008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017479775318833275,
+      "loss": 2.3765,
+      "step": 20016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017477796930752344,
+      "loss": 2.3469,
+      "step": 20024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017475817892985093,
+      "loss": 2.3684,
+      "step": 20032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017473838205729993,
+      "loss": 2.3591,
+      "step": 20040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017471857869185608,
+      "loss": 2.3401,
+      "step": 20048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017469876883550553,
+      "loss": 2.3562,
+      "step": 20056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017467895249023504,
+      "loss": 2.3558,
+      "step": 20064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017465912965803208,
+      "loss": 2.3774,
+      "step": 20072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017463930034088485,
+      "loss": 2.3807,
+      "step": 20080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017461946454078205,
+      "loss": 2.359,
+      "step": 20088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001745996222597131,
+      "loss": 2.3857,
+      "step": 20096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017457977349966805,
+      "loss": 2.3659,
+      "step": 20104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001745599182626377,
+      "loss": 2.3506,
+      "step": 20112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017454005655061333,
+      "loss": 2.3504,
+      "step": 20120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000174520188365587,
+      "loss": 2.3779,
+      "step": 20128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017450031370955138,
+      "loss": 2.3579,
+      "step": 20136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017448043258449977,
+      "loss": 2.3399,
+      "step": 20144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001744605449924261,
+      "loss": 2.3628,
+      "step": 20152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017444065093532505,
+      "loss": 2.3445,
+      "step": 20160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017442075041519187,
+      "loss": 2.3727,
+      "step": 20168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017440084343402245,
+      "loss": 2.3636,
+      "step": 20176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017438092999381332,
+      "loss": 2.3718,
+      "step": 20184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017436101009656174,
+      "loss": 2.355,
+      "step": 20192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017434108374426554,
+      "loss": 2.3559,
+      "step": 20200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001743211509389232,
+      "loss": 2.3768,
+      "step": 20208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017430121168253388,
+      "loss": 2.3529,
+      "step": 20216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017428126597709737,
+      "loss": 2.3641,
+      "step": 20224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017426131382461415,
+      "loss": 2.3489,
+      "step": 20232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017424135522708527,
+      "loss": 2.3587,
+      "step": 20240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017422139018651245,
+      "loss": 2.3844,
+      "step": 20248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001742014187048981,
+      "loss": 2.3504,
+      "step": 20256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017418144078424526,
+      "loss": 2.3695,
+      "step": 20264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001741614564265576,
+      "loss": 2.368,
+      "step": 20272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001741414656338394,
+      "loss": 2.3634,
+      "step": 20280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001741214684080956,
+      "loss": 2.3665,
+      "step": 20288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017410146475133192,
+      "loss": 2.3592,
+      "step": 20296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017408145466555448,
+      "loss": 2.3603,
+      "step": 20304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017406143815277026,
+      "loss": 2.3426,
+      "step": 20312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001740414152149868,
+      "loss": 2.3596,
+      "step": 20320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017402138585421225,
+      "loss": 2.3716,
+      "step": 20328
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001740013500724555,
+      "loss": 2.3583,
+      "step": 20336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017398130787172597,
+      "loss": 2.3814,
+      "step": 20344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001739612592540338,
+      "loss": 2.3608,
+      "step": 20352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017394120422138977,
+      "loss": 2.3618,
+      "step": 20360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017392114277580523,
+      "loss": 2.3501,
+      "step": 20368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001739010749192923,
+      "loss": 2.3855,
+      "step": 20376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017388100065386363,
+      "loss": 2.3531,
+      "step": 20384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017386091998153254,
+      "loss": 2.3401,
+      "step": 20392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017384083290431307,
+      "loss": 2.3539,
+      "step": 20400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017382073942421978,
+      "loss": 2.3492,
+      "step": 20408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017380063954326795,
+      "loss": 2.375,
+      "step": 20416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017378053326347348,
+      "loss": 2.364,
+      "step": 20424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017376042058685296,
+      "loss": 2.3499,
+      "step": 20432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001737403015154235,
+      "loss": 2.3574,
+      "step": 20440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000173720176051203,
+      "loss": 2.3583,
+      "step": 20448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017370004419620992,
+      "loss": 2.3536,
+      "step": 20456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001736799059524633,
+      "loss": 2.3585,
+      "step": 20464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017365976132198297,
+      "loss": 2.3634,
+      "step": 20472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017363961030678927,
+      "loss": 2.3562,
+      "step": 20480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017361945290890333,
+      "loss": 2.3574,
+      "step": 20488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017359928913034667,
+      "loss": 2.3597,
+      "step": 20496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001735791189731417,
+      "loss": 2.3698,
+      "step": 20504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017355894243931136,
+      "loss": 2.3545,
+      "step": 20512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017353875953087926,
+      "loss": 2.3419,
+      "step": 20520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001735185702498696,
+      "loss": 2.3703,
+      "step": 20528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017349837459830727,
+      "loss": 2.3631,
+      "step": 20536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017347817257821774,
+      "loss": 2.3385,
+      "step": 20544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017345796419162722,
+      "loss": 2.3378,
+      "step": 20552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017343774944056245,
+      "loss": 2.3522,
+      "step": 20560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017341752832705087,
+      "loss": 2.3528,
+      "step": 20568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001733973008531205,
+      "loss": 2.3652,
+      "step": 20576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017337706702080015,
+      "loss": 2.3476,
+      "step": 20584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001733568268321191,
+      "loss": 2.3499,
+      "step": 20592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017333658028910727,
+      "loss": 2.3661,
+      "step": 20600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017331632739379533,
+      "loss": 2.3412,
+      "step": 20608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001732960681482145,
+      "loss": 2.3603,
+      "step": 20616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017327580255439674,
+      "loss": 2.3723,
+      "step": 20624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017325553061437452,
+      "loss": 2.3664,
+      "step": 20632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000173235252330181,
+      "loss": 2.3541,
+      "step": 20640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017321496770384995,
+      "loss": 2.3506,
+      "step": 20648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017319467673741586,
+      "loss": 2.3671,
+      "step": 20656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017317437943291378,
+      "loss": 2.366,
+      "step": 20664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001731540757923794,
+      "loss": 2.3544,
+      "step": 20672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017313376581784906,
+      "loss": 2.3636,
+      "step": 20680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017311344951135974,
+      "loss": 2.3882,
+      "step": 20688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017309312687494905,
+      "loss": 2.3519,
+      "step": 20696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001730727979106553,
+      "loss": 2.3672,
+      "step": 20704
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001730524626205172,
+      "loss": 2.3559,
+      "step": 20712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017303212100657446,
+      "loss": 2.3584,
+      "step": 20720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017301177307086713,
+      "loss": 2.3725,
+      "step": 20728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017299141881543602,
+      "loss": 2.3512,
+      "step": 20736
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017297105824232246,
+      "loss": 2.3546,
+      "step": 20744
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017295069135356866,
+      "loss": 2.3544,
+      "step": 20752
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001729303181512172,
+      "loss": 2.3528,
+      "step": 20760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017290993863731136,
+      "loss": 2.3577,
+      "step": 20768
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017288955281389516,
+      "loss": 2.358,
+      "step": 20776
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017286916068301317,
+      "loss": 2.3479,
+      "step": 20784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017284876224671062,
+      "loss": 2.3588,
+      "step": 20792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017282835750703332,
+      "loss": 2.3663,
+      "step": 20800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017280794646602776,
+      "loss": 2.3518,
+      "step": 20808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001727875291257411,
+      "loss": 2.3629,
+      "step": 20816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000172767105488221,
+      "loss": 2.3465,
+      "step": 20824
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001727466755555159,
+      "loss": 2.3512,
+      "step": 20832
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017272623932967476,
+      "loss": 2.3667,
+      "step": 20840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017270579681274723,
+      "loss": 2.3549,
+      "step": 20848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017268534800678363,
+      "loss": 2.3753,
+      "step": 20856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001726648929138348,
+      "loss": 2.3611,
+      "step": 20864
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001726444315359523,
+      "loss": 2.3641,
+      "step": 20872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017262396387518827,
+      "loss": 2.3571,
+      "step": 20880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017260348993359553,
+      "loss": 2.3429,
+      "step": 20888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017258300971322745,
+      "loss": 2.3627,
+      "step": 20896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017256252321613813,
+      "loss": 2.3658,
+      "step": 20904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725420304443822,
+      "loss": 2.3496,
+      "step": 20912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017252153140001503,
+      "loss": 2.367,
+      "step": 20920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725010260850925,
+      "loss": 2.3561,
+      "step": 20928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017248051450167128,
+      "loss": 2.3691,
+      "step": 20936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017245999665180842,
+      "loss": 2.3788,
+      "step": 20944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017243947253756182,
+      "loss": 2.3731,
+      "step": 20952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017241894216098996,
+      "loss": 2.3693,
+      "step": 20960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017239840552415184,
+      "loss": 2.3532,
+      "step": 20968
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017237786262910726,
+      "loss": 2.364,
+      "step": 20976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001723573134779165,
+      "loss": 2.344,
+      "step": 20984
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017233675807264048,
+      "loss": 2.352,
+      "step": 20992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017231619641534088,
+      "loss": 2.3354,
+      "step": 21000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001722956285080799,
+      "loss": 2.3586,
+      "step": 21008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017227505435292036,
+      "loss": 2.3695,
+      "step": 21016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017225447395192574,
+      "loss": 2.3683,
+      "step": 21024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017223388730716014,
+      "loss": 2.3712,
+      "step": 21032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017221329442068824,
+      "loss": 2.3521,
+      "step": 21040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001721926952945755,
+      "loss": 2.3694,
+      "step": 21048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017217208993088775,
+      "loss": 2.3718,
+      "step": 21056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017215147833169172,
+      "loss": 2.359,
+      "step": 21064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017213086049905455,
+      "loss": 2.3636,
+      "step": 21072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017211023643504413,
+      "loss": 2.3312,
+      "step": 21080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017208960614172893,
+      "loss": 2.3831,
+      "step": 21088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001720689696211781,
+      "loss": 2.3657,
+      "step": 21096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017204832687546127,
+      "loss": 2.3433,
+      "step": 21104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017202767790664884,
+      "loss": 2.3646,
+      "step": 21112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017200702271681178,
+      "loss": 2.3587,
+      "step": 21120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017198636130802173,
+      "loss": 2.3693,
+      "step": 21128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017196569368235084,
+      "loss": 2.3746,
+      "step": 21136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017194501984187202,
+      "loss": 2.3536,
+      "step": 21144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001719243397886587,
+      "loss": 2.3687,
+      "step": 21152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017190365352478496,
+      "loss": 2.388,
+      "step": 21160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017188296105232555,
+      "loss": 2.3701,
+      "step": 21168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017186226237335582,
+      "loss": 2.3608,
+      "step": 21176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001718415574899517,
+      "loss": 2.3408,
+      "step": 21184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017182084640418973,
+      "loss": 2.3441,
+      "step": 21192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001718001291181472,
+      "loss": 2.3467,
+      "step": 21200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001717794056339019,
+      "loss": 2.3735,
+      "step": 21208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017175867595353226,
+      "loss": 2.3646,
+      "step": 21216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017173794007911737,
+      "loss": 2.3566,
+      "step": 21224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001717171980127369,
+      "loss": 2.3412,
+      "step": 21232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017169644975647118,
+      "loss": 2.368,
+      "step": 21240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001716756953124012,
+      "loss": 2.3452,
+      "step": 21248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001716549346826084,
+      "loss": 2.3668,
+      "step": 21256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000171634167869175,
+      "loss": 2.3482,
+      "step": 21264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017161339487418385,
+      "loss": 2.3584,
+      "step": 21272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017159261569971824,
+      "loss": 2.3733,
+      "step": 21280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017157183034786238,
+      "loss": 2.3621,
+      "step": 21288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017155103882070073,
+      "loss": 2.3795,
+      "step": 21296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001715302411203187,
+      "loss": 2.3466,
+      "step": 21304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001715094372488021,
+      "loss": 2.3504,
+      "step": 21312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017148862720823752,
+      "loss": 2.36,
+      "step": 21320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000171467811000712,
+      "loss": 2.359,
+      "step": 21328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017144698862831337,
+      "loss": 2.3521,
+      "step": 21336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017142616009312996,
+      "loss": 2.376,
+      "step": 21344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017140532539725075,
+      "loss": 2.3709,
+      "step": 21352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001713844845427653,
+      "loss": 2.3589,
+      "step": 21360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017136363753176393,
+      "loss": 2.3604,
+      "step": 21368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001713427843663374,
+      "loss": 2.3421,
+      "step": 21376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017132192504857715,
+      "loss": 2.3505,
+      "step": 21384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017130105958057531,
+      "loss": 2.368,
+      "step": 21392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017128018796442457,
+      "loss": 2.3544,
+      "step": 21400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017125931020221818,
+      "loss": 2.3421,
+      "step": 21408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017123842629605013,
+      "loss": 2.3609,
+      "step": 21416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001712175362480149,
+      "loss": 2.3649,
+      "step": 21424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017119664006020764,
+      "loss": 2.362,
+      "step": 21432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017117573773472417,
+      "loss": 2.361,
+      "step": 21440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017115482927366083,
+      "loss": 2.3711,
+      "step": 21448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017113391467911464,
+      "loss": 2.3486,
+      "step": 21456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001711129939531832,
+      "loss": 2.3584,
+      "step": 21464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017109206709796477,
+      "loss": 2.361,
+      "step": 21472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017107113411555814,
+      "loss": 2.3648,
+      "step": 21480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017105019500806285,
+      "loss": 2.3331,
+      "step": 21488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017102924977757892,
+      "loss": 2.3595,
+      "step": 21496
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017100829842620706,
+      "loss": 2.3584,
+      "step": 21504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017098734095604855,
+      "loss": 2.3472,
+      "step": 21512
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001709663773692053,
+      "loss": 2.3614,
+      "step": 21520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017094540766777983,
+      "loss": 2.3515,
+      "step": 21528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017092443185387534,
+      "loss": 2.3551,
+      "step": 21536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017090344992959556,
+      "loss": 2.3527,
+      "step": 21544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017088246189704486,
+      "loss": 2.3484,
+      "step": 21552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017086146775832818,
+      "loss": 2.3581,
+      "step": 21560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017084046751555119,
+      "loss": 2.3426,
+      "step": 21568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017081946117082,
+      "loss": 2.3368,
+      "step": 21576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001707984487262415,
+      "loss": 2.3484,
+      "step": 21584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001707774301839231,
+      "loss": 2.3585,
+      "step": 21592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001707564055459728,
+      "loss": 2.3544,
+      "step": 21600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001707353748144993,
+      "loss": 2.3731,
+      "step": 21608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001707143379916119,
+      "loss": 2.3689,
+      "step": 21616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017069329507942034,
+      "loss": 2.354,
+      "step": 21624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017067224608003528,
+      "loss": 2.3552,
+      "step": 21632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017065119099556769,
+      "loss": 2.3687,
+      "step": 21640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001706301298281293,
+      "loss": 2.3573,
+      "step": 21648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017060906257983244,
+      "loss": 2.3725,
+      "step": 21656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017058798925279005,
+      "loss": 2.3824,
+      "step": 21664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017056690984911562,
+      "loss": 2.369,
+      "step": 21672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017054582437092336,
+      "loss": 2.3629,
+      "step": 21680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017052473282032796,
+      "loss": 2.3764,
+      "step": 21688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017050363519944482,
+      "loss": 2.353,
+      "step": 21696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017048253151038988,
+      "loss": 2.3595,
+      "step": 21704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001704614217552798,
+      "loss": 2.3739,
+      "step": 21712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017044030593623167,
+      "loss": 2.3635,
+      "step": 21720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001704191840553633,
+      "loss": 2.3805,
+      "step": 21728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017039805611479317,
+      "loss": 2.3468,
+      "step": 21736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017037692211664026,
+      "loss": 2.3738,
+      "step": 21744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017035578206302411,
+      "loss": 2.3571,
+      "step": 21752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017033463595606506,
+      "loss": 2.3456,
+      "step": 21760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017031348379788392,
+      "loss": 2.3619,
+      "step": 21768
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017029232559060206,
+      "loss": 2.3699,
+      "step": 21776
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017027116133634164,
+      "loss": 2.3569,
+      "step": 21784
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001702499910372252,
+      "loss": 2.3819,
+      "step": 21792
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017022881469537607,
+      "loss": 2.3633,
+      "step": 21800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017020763231291808,
+      "loss": 2.3789,
+      "step": 21808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017018644389197573,
+      "loss": 2.3709,
+      "step": 21816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017016524943467412,
+      "loss": 2.3684,
+      "step": 21824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017014404894313888,
+      "loss": 2.3515,
+      "step": 21832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017012284241949634,
+      "loss": 2.3462,
+      "step": 21840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017010162986587338,
+      "loss": 2.3505,
+      "step": 21848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001700804112843975,
+      "loss": 2.3732,
+      "step": 21856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001700591866771968,
+      "loss": 2.3474,
+      "step": 21864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017003795604639998,
+      "loss": 2.3502,
+      "step": 21872
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017001671939413638,
+      "loss": 2.369,
+      "step": 21880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001699954767225359,
+      "loss": 2.3612,
+      "step": 21888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016997422803372903,
+      "loss": 2.383,
+      "step": 21896
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016995297332984697,
+      "loss": 2.361,
+      "step": 21904
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016993171261302134,
+      "loss": 2.3653,
+      "step": 21912
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016991044588538453,
+      "loss": 2.3646,
+      "step": 21920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016988917314906952,
+      "loss": 2.3656,
+      "step": 21928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016986789440620977,
+      "loss": 2.3614,
+      "step": 21936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016984660965893943,
+      "loss": 2.3484,
+      "step": 21944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016982531890939327,
+      "loss": 2.3467,
+      "step": 21952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001698040221597066,
+      "loss": 2.3713,
+      "step": 21960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016978271941201536,
+      "loss": 2.3651,
+      "step": 21968
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016976141066845613,
+      "loss": 2.3661,
+      "step": 21976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016974009593116604,
+      "loss": 2.3643,
+      "step": 21984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016971877520228284,
+      "loss": 2.3611,
+      "step": 21992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016969744848394484,
+      "loss": 2.3636,
+      "step": 22000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016967611577829104,
+      "loss": 2.3562,
+      "step": 22008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000169654777087461,
+      "loss": 2.3769,
+      "step": 22016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001696334324135948,
+      "loss": 2.3671,
+      "step": 22024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016961208175883324,
+      "loss": 2.3611,
+      "step": 22032
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016959072512531765,
+      "loss": 2.3582,
+      "step": 22040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016956936251519002,
+      "loss": 2.3397,
+      "step": 22048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016954799393059284,
+      "loss": 2.3583,
+      "step": 22056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001695266193736693,
+      "loss": 2.3723,
+      "step": 22064
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016950523884656313,
+      "loss": 2.3695,
+      "step": 22072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016948385235141867,
+      "loss": 2.3528,
+      "step": 22080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016946245989038088,
+      "loss": 2.365,
+      "step": 22088
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001694410614655953,
+      "loss": 2.3471,
+      "step": 22096
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016941965707920807,
+      "loss": 2.3646,
+      "step": 22104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016939824673336596,
+      "loss": 2.3495,
+      "step": 22112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001693768304302162,
+      "loss": 2.3508,
+      "step": 22120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016935540817190686,
+      "loss": 2.3792,
+      "step": 22128
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016933397996058634,
+      "loss": 2.3712,
+      "step": 22136
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001693125457984039,
+      "loss": 2.3504,
+      "step": 22144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016929110568750913,
+      "loss": 2.3512,
+      "step": 22152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001692696596300525,
+      "loss": 2.3535,
+      "step": 22160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001692482076281848,
+      "loss": 2.3641,
+      "step": 22168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016922674968405759,
+      "loss": 2.3936,
+      "step": 22176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000169205285799823,
+      "loss": 2.3403,
+      "step": 22184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001691838159776337,
+      "loss": 2.3827,
+      "step": 22192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016916234021964308,
+      "loss": 2.3686,
+      "step": 22200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016914085852800487,
+      "loss": 2.3668,
+      "step": 22208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016911937090487373,
+      "loss": 2.3742,
+      "step": 22216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016909787735240465,
+      "loss": 2.351,
+      "step": 22224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016907637787275334,
+      "loss": 2.3903,
+      "step": 22232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016905487246807608,
+      "loss": 2.3518,
+      "step": 22240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016903336114052975,
+      "loss": 2.3531,
+      "step": 22248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001690118438922718,
+      "loss": 2.3678,
+      "step": 22256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016899032072546023,
+      "loss": 2.36,
+      "step": 22264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016896879164225378,
+      "loss": 2.3821,
+      "step": 22272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001689472566448117,
+      "loss": 2.374,
+      "step": 22280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016892571573529379,
+      "loss": 2.3602,
+      "step": 22288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016890416891586043,
+      "loss": 2.3696,
+      "step": 22296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001688826161886727,
+      "loss": 2.3799,
+      "step": 22304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016886105755589226,
+      "loss": 2.3509,
+      "step": 22312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016883949301968126,
+      "loss": 2.3687,
+      "step": 22320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016881792258220248,
+      "loss": 2.3816,
+      "step": 22328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001687963462456194,
+      "loss": 2.3513,
+      "step": 22336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016877476401209592,
+      "loss": 2.3722,
+      "step": 22344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016875317588379662,
+      "loss": 2.3862,
+      "step": 22352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016873158186288674,
+      "loss": 2.3885,
+      "step": 22360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016870998195153195,
+      "loss": 2.3662,
+      "step": 22368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016868837615189865,
+      "loss": 2.3477,
+      "step": 22376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001686667644661538,
+      "loss": 2.3491,
+      "step": 22384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016864514689646486,
+      "loss": 2.3741,
+      "step": 22392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000168623523445,
+      "loss": 2.3728,
+      "step": 22400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016860189411392794,
+      "loss": 2.3623,
+      "step": 22408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001685802589054179,
+      "loss": 2.3462,
+      "step": 22416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001685586178216399,
+      "loss": 2.3754,
+      "step": 22424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001685369708647643,
+      "loss": 2.3521,
+      "step": 22432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016851531803696224,
+      "loss": 2.3654,
+      "step": 22440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001684936593404053,
+      "loss": 2.3576,
+      "step": 22448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001684719947772658,
+      "loss": 2.3722,
+      "step": 22456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016845032434971655,
+      "loss": 2.3473,
+      "step": 22464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016842864805993095,
+      "loss": 2.361,
+      "step": 22472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016840696591008302,
+      "loss": 2.3534,
+      "step": 22480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016838527790234738,
+      "loss": 2.3708,
+      "step": 22488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016836358403889918,
+      "loss": 2.3316,
+      "step": 22496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016834188432191422,
+      "loss": 2.3883,
+      "step": 22504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016832017875356885,
+      "loss": 2.3649,
+      "step": 22512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016829846733604,
+      "loss": 2.3415,
+      "step": 22520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001682767500715052,
+      "loss": 2.3627,
+      "step": 22528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001682550269621426,
+      "loss": 2.3526,
+      "step": 22536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001682332980101309,
+      "loss": 2.3444,
+      "step": 22544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001682115632176494,
+      "loss": 2.3688,
+      "step": 22552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001681898225868779,
+      "loss": 2.356,
+      "step": 22560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001681680761199969,
+      "loss": 2.3715,
+      "step": 22568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016814632381918751,
+      "loss": 2.3462,
+      "step": 22576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016812456568663132,
+      "loss": 2.3519,
+      "step": 22584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001681028017245105,
+      "loss": 2.383,
+      "step": 22592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016808103193500793,
+      "loss": 2.3523,
+      "step": 22600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016805925632030696,
+      "loss": 2.3505,
+      "step": 22608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016803747488259156,
+      "loss": 2.3571,
+      "step": 22616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016801568762404628,
+      "loss": 2.366,
+      "step": 22624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016799389454685626,
+      "loss": 2.3635,
+      "step": 22632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016797209565320726,
+      "loss": 2.3707,
+      "step": 22640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016795029094528554,
+      "loss": 2.3546,
+      "step": 22648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016792848042527798,
+      "loss": 2.3388,
+      "step": 22656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001679066640953721,
+      "loss": 2.3613,
+      "step": 22664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001678848419577559,
+      "loss": 2.3609,
+      "step": 22672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016786301401461805,
+      "loss": 2.3542,
+      "step": 22680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001678411802681478,
+      "loss": 2.3655,
+      "step": 22688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001678193407205349,
+      "loss": 2.3626,
+      "step": 22696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016779749537396973,
+      "loss": 2.3452,
+      "step": 22704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016777564423064328,
+      "loss": 2.3502,
+      "step": 22712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016775378729274709,
+      "loss": 2.3817,
+      "step": 22720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001677319245624733,
+      "loss": 2.361,
+      "step": 22728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016771005604201462,
+      "loss": 2.351,
+      "step": 22736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016768818173356427,
+      "loss": 2.3419,
+      "step": 22744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016766630163931622,
+      "loss": 2.3564,
+      "step": 22752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001676444157614648,
+      "loss": 2.3648,
+      "step": 22760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016762252410220519,
+      "loss": 2.3581,
+      "step": 22768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016760062666373288,
+      "loss": 2.347,
+      "step": 22776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016757872344824411,
+      "loss": 2.3686,
+      "step": 22784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016755681445793563,
+      "loss": 2.3588,
+      "step": 22792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001675348996950048,
+      "loss": 2.3629,
+      "step": 22800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016751297916164953,
+      "loss": 2.3817,
+      "step": 22808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016749105286006835,
+      "loss": 2.3339,
+      "step": 22816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016746912079246032,
+      "loss": 2.3611,
+      "step": 22824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016744718296102514,
+      "loss": 2.3589,
+      "step": 22832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016742523936796297,
+      "loss": 2.3602,
+      "step": 22840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001674032900154747,
+      "loss": 2.3483,
+      "step": 22848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016738133490576176,
+      "loss": 2.3679,
+      "step": 22856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016735937404102605,
+      "loss": 2.3581,
+      "step": 22864
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001673374074234701,
+      "loss": 2.3889,
+      "step": 22872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016731543505529707,
+      "loss": 2.3844,
+      "step": 22880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001672934569387107,
+      "loss": 2.3568,
+      "step": 22888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016727147307591523,
+      "loss": 2.3638,
+      "step": 22896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016724948346911553,
+      "loss": 2.354,
+      "step": 22904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016722748812051706,
+      "loss": 2.3641,
+      "step": 22912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016720548703232576,
+      "loss": 2.3637,
+      "step": 22920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001671834802067483,
+      "loss": 2.3566,
+      "step": 22928
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016716146764599174,
+      "loss": 2.3524,
+      "step": 22936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001671394493522639,
+      "loss": 2.332,
+      "step": 22944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016711742532777308,
+      "loss": 2.3691,
+      "step": 22952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001670953955747281,
+      "loss": 2.3513,
+      "step": 22960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016707336009533848,
+      "loss": 2.3489,
+      "step": 22968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016705131889181425,
+      "loss": 2.362,
+      "step": 22976
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000167029271966366,
+      "loss": 2.3763,
+      "step": 22984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016700721932120494,
+      "loss": 2.3639,
+      "step": 22992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016698516095854275,
+      "loss": 2.3535,
+      "step": 23000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016696309688059185,
+      "loss": 2.364,
+      "step": 23008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001669410270895651,
+      "loss": 2.3434,
+      "step": 23016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016691895158767596,
+      "loss": 2.3584,
+      "step": 23024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016689687037713854,
+      "loss": 2.3772,
+      "step": 23032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001668747834601674,
+      "loss": 2.3541,
+      "step": 23040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001668526908389777,
+      "loss": 2.3568,
+      "step": 23048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016683059251578527,
+      "loss": 2.3415,
+      "step": 23056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001668084884928065,
+      "loss": 2.3583,
+      "step": 23064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016678637877225815,
+      "loss": 2.3558,
+      "step": 23072
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016676426335635782,
+      "loss": 2.3473,
+      "step": 23080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001667421422473235,
+      "loss": 2.3499,
+      "step": 23088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016672001544737382,
+      "loss": 2.3647,
+      "step": 23096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016669788295872804,
+      "loss": 2.3281,
+      "step": 23104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001666757447836058,
+      "loss": 2.361,
+      "step": 23112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016665360092422754,
+      "loss": 2.3783,
+      "step": 23120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001666314513828141,
+      "loss": 2.3494,
+      "step": 23128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016660929616158703,
+      "loss": 2.3425,
+      "step": 23136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016658713526276827,
+      "loss": 2.353,
+      "step": 23144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016656496868858048,
+      "loss": 2.3706,
+      "step": 23152
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016654279644124688,
+      "loss": 2.3552,
+      "step": 23160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016652061852299118,
+      "loss": 2.3691,
+      "step": 23168
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016649843493603766,
+      "loss": 2.3605,
+      "step": 23176
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001664762456826113,
+      "loss": 2.3614,
+      "step": 23184
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001664540507649375,
+      "loss": 2.3548,
+      "step": 23192
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016643185018524228,
+      "loss": 2.3568,
+      "step": 23200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016640964394575226,
+      "loss": 2.3591,
+      "step": 23208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001663874320486946,
+      "loss": 2.3771,
+      "step": 23216
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000166365214496297,
+      "loss": 2.3362,
+      "step": 23224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016634299129078775,
+      "loss": 2.3573,
+      "step": 23232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016632076243439575,
+      "loss": 2.3491,
+      "step": 23240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001662985279293504,
+      "loss": 2.3791,
+      "step": 23248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016627628777788174,
+      "loss": 2.3497,
+      "step": 23256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016625404198222026,
+      "loss": 2.3791,
+      "step": 23264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016623179054459715,
+      "loss": 2.3517,
+      "step": 23272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016620953346724405,
+      "loss": 2.37,
+      "step": 23280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016618727075239323,
+      "loss": 2.3703,
+      "step": 23288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016616500240227756,
+      "loss": 2.3514,
+      "step": 23296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001661427284191304,
+      "loss": 2.3604,
+      "step": 23304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016612044880518566,
+      "loss": 2.3929,
+      "step": 23312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016609816356267794,
+      "loss": 2.3677,
+      "step": 23320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016607587269384225,
+      "loss": 2.3498,
+      "step": 23328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001660535762009143,
+      "loss": 2.3682,
+      "step": 23336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016603127408613025,
+      "loss": 2.3444,
+      "step": 23344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016600896635172692,
+      "loss": 2.3704,
+      "step": 23352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016598665299994166,
+      "loss": 2.3707,
+      "step": 23360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001659643340330123,
+      "loss": 2.3491,
+      "step": 23368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016594200945317734,
+      "loss": 2.3676,
+      "step": 23376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016591967926267583,
+      "loss": 2.3657,
+      "step": 23384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016589734346374738,
+      "loss": 2.3764,
+      "step": 23392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016587500205863208,
+      "loss": 2.3708,
+      "step": 23400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016585265504957065,
+      "loss": 2.3501,
+      "step": 23408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016583030243880447,
+      "loss": 2.3565,
+      "step": 23416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016580794422857523,
+      "loss": 2.3661,
+      "step": 23424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016578558042112547,
+      "loss": 2.3465,
+      "step": 23432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016576321101869807,
+      "loss": 2.3455,
+      "step": 23440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016574083602353658,
+      "loss": 2.3842,
+      "step": 23448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016571845543788508,
+      "loss": 2.3526,
+      "step": 23456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001656960692639882,
+      "loss": 2.3627,
+      "step": 23464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016567367750409118,
+      "loss": 2.3439,
+      "step": 23472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016565128016043975,
+      "loss": 2.3557,
+      "step": 23480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016562887723528031,
+      "loss": 2.3578,
+      "step": 23488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016560646873085965,
+      "loss": 2.3724,
+      "step": 23496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016558405464942526,
+      "loss": 2.3647,
+      "step": 23504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016556163499322518,
+      "loss": 2.3543,
+      "step": 23512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016553920976450793,
+      "loss": 2.379,
+      "step": 23520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001655167789655226,
+      "loss": 2.3753,
+      "step": 23528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016549434259851894,
+      "loss": 2.3593,
+      "step": 23536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016547190066574719,
+      "loss": 2.3729,
+      "step": 23544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001654494531694581,
+      "loss": 2.3434,
+      "step": 23552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016542700011190305,
+      "loss": 2.3788,
+      "step": 23560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016540454149533394,
+      "loss": 2.3377,
+      "step": 23568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016538207732200332,
+      "loss": 2.3639,
+      "step": 23576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001653596075941641,
+      "loss": 2.3649,
+      "step": 23584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016533713231406993,
+      "loss": 2.3611,
+      "step": 23592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016531465148397496,
+      "loss": 2.3517,
+      "step": 23600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016529216510613386,
+      "loss": 2.3647,
+      "step": 23608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001652696731828019,
+      "loss": 2.369,
+      "step": 23616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016524717571623492,
+      "loss": 2.3559,
+      "step": 23624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016522467270868924,
+      "loss": 2.3478,
+      "step": 23632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016520216416242184,
+      "loss": 2.3569,
+      "step": 23640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016517965007969014,
+      "loss": 2.362,
+      "step": 23648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016515713046275223,
+      "loss": 2.339,
+      "step": 23656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016513460531386665,
+      "loss": 2.349,
+      "step": 23664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016511207463529255,
+      "loss": 2.3451,
+      "step": 23672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016508953842928967,
+      "loss": 2.3501,
+      "step": 23680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016506699669811823,
+      "loss": 2.3709,
+      "step": 23688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016504444944403905,
+      "loss": 2.3678,
+      "step": 23696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001650218966693135,
+      "loss": 2.3429,
+      "step": 23704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016499933837620348,
+      "loss": 2.3644,
+      "step": 23712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016497677456697144,
+      "loss": 2.3346,
+      "step": 23720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016495420524388046,
+      "loss": 2.3665,
+      "step": 23728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016493163040919405,
+      "loss": 2.354,
+      "step": 23736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016490905006517643,
+      "loss": 2.3642,
+      "step": 23744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016488646421409215,
+      "loss": 2.3749,
+      "step": 23752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016486387285820654,
+      "loss": 2.3801,
+      "step": 23760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016484127599978535,
+      "loss": 2.34,
+      "step": 23768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016481867364109495,
+      "loss": 2.3519,
+      "step": 23776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016479606578440222,
+      "loss": 2.3702,
+      "step": 23784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016477345243197454,
+      "loss": 2.3706,
+      "step": 23792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016475083358607996,
+      "loss": 2.3361,
+      "step": 23800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000164728209248987,
+      "loss": 2.3657,
+      "step": 23808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016470557942296476,
+      "loss": 2.357,
+      "step": 23816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016468294411028288,
+      "loss": 2.3648,
+      "step": 23824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016466030331321156,
+      "loss": 2.3444,
+      "step": 23832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016463765703402154,
+      "loss": 2.3488,
+      "step": 23840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016461500527498415,
+      "loss": 2.3555,
+      "step": 23848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016459234803837113,
+      "loss": 2.3417,
+      "step": 23856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016456968532645497,
+      "loss": 2.3533,
+      "step": 23864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016454701714150859,
+      "loss": 2.3653,
+      "step": 23872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001645243434858055,
+      "loss": 2.3464,
+      "step": 23880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016450166436161968,
+      "loss": 2.3649,
+      "step": 23888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001644789797712257,
+      "loss": 2.3424,
+      "step": 23896
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001644562897168988,
+      "loss": 2.3538,
+      "step": 23904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016443359420091456,
+      "loss": 2.3689,
+      "step": 23912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001644108932255493,
+      "loss": 2.3539,
+      "step": 23920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016438818679307977,
+      "loss": 2.357,
+      "step": 23928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016436547490578327,
+      "loss": 2.3491,
+      "step": 23936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016434275756593768,
+      "loss": 2.378,
+      "step": 23944
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016432003477582144,
+      "loss": 2.359,
+      "step": 23952
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001642973065377135,
+      "loss": 2.365,
+      "step": 23960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016427457285389337,
+      "loss": 2.3656,
+      "step": 23968
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016425183372664113,
+      "loss": 2.3732,
+      "step": 23976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016422908915823738,
+      "loss": 2.3437,
+      "step": 23984
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016420633915096327,
+      "loss": 2.3592,
+      "step": 23992
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001641835837071005,
+      "loss": 2.3625,
+      "step": 24000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001641608228289313,
+      "loss": 2.3549,
+      "step": 24008
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016413805651873843,
+      "loss": 2.3686,
+      "step": 24016
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016411528477880527,
+      "loss": 2.3392,
+      "step": 24024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016409250761141567,
+      "loss": 2.3443,
+      "step": 24032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001640697250188541,
+      "loss": 2.3353,
+      "step": 24040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016404693700340544,
+      "loss": 2.353,
+      "step": 24048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001640241435673553,
+      "loss": 2.3483,
+      "step": 24056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016400134471298964,
+      "loss": 2.3602,
+      "step": 24064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016397854044259512,
+      "loss": 2.3606,
+      "step": 24072
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016395573075845882,
+      "loss": 2.3673,
+      "step": 24080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016393291566286846,
+      "loss": 2.3277,
+      "step": 24088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001639100951581123,
+      "loss": 2.3626,
+      "step": 24096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016388726924647905,
+      "loss": 2.3611,
+      "step": 24104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000163864437930258,
+      "loss": 2.3727,
+      "step": 24112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016384160121173905,
+      "loss": 2.3576,
+      "step": 24120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001638187590932126,
+      "loss": 2.3507,
+      "step": 24128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016379591157696956,
+      "loss": 2.3467,
+      "step": 24136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001637730586653014,
+      "loss": 2.3667,
+      "step": 24144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001637502003605002,
+      "loss": 2.3465,
+      "step": 24152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001637273366648584,
+      "loss": 2.3501,
+      "step": 24160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016370446758066923,
+      "loss": 2.3741,
+      "step": 24168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001636815931102262,
+      "loss": 2.3718,
+      "step": 24176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016365871325582364,
+      "loss": 2.3344,
+      "step": 24184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016363582801975614,
+      "loss": 2.363,
+      "step": 24192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016361293740431904,
+      "loss": 2.3628,
+      "step": 24200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016359004141180803,
+      "loss": 2.3312,
+      "step": 24208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016356714004451962,
+      "loss": 2.3715,
+      "step": 24216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001635442333047505,
+      "loss": 2.347,
+      "step": 24224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016352132119479827,
+      "loss": 2.3299,
+      "step": 24232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001634984037169607,
+      "loss": 2.3644,
+      "step": 24240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016347548087353644,
+      "loss": 2.3767,
+      "step": 24248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016345255266682443,
+      "loss": 2.3614,
+      "step": 24256
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001634296190991243,
+      "loss": 2.3426,
+      "step": 24264
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016340668017273611,
+      "loss": 2.3465,
+      "step": 24272
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016338373588996048,
+      "loss": 2.3713,
+      "step": 24280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016336078625309866,
+      "loss": 2.342,
+      "step": 24288
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016333783126445232,
+      "loss": 2.3604,
+      "step": 24296
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016331487092632375,
+      "loss": 2.3597,
+      "step": 24304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016329190524101573,
+      "loss": 2.3383,
+      "step": 24312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016326893421083155,
+      "loss": 2.3325,
+      "step": 24320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016324595783807517,
+      "loss": 2.3749,
+      "step": 24328
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016322297612505086,
+      "loss": 2.3572,
+      "step": 24336
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016319998907406367,
+      "loss": 2.3476,
+      "step": 24344
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016317699668741897,
+      "loss": 2.3465,
+      "step": 24352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001631539989674229,
+      "loss": 2.3665,
+      "step": 24360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016313099591638185,
+      "loss": 2.3642,
+      "step": 24368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016310798753660296,
+      "loss": 2.3656,
+      "step": 24376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001630849738303939,
+      "loss": 2.3449,
+      "step": 24384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016306195480006273,
+      "loss": 2.3582,
+      "step": 24392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016303893044791818,
+      "loss": 2.3552,
+      "step": 24400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016301590077626944,
+      "loss": 2.3693,
+      "step": 24408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016299286578742622,
+      "loss": 2.3662,
+      "step": 24416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016296982548369887,
+      "loss": 2.3623,
+      "step": 24424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016294677986739817,
+      "loss": 2.3417,
+      "step": 24432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016292372894083544,
+      "loss": 2.3441,
+      "step": 24440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016290067270632261,
+      "loss": 2.3478,
+      "step": 24448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016287761116617203,
+      "loss": 2.3561,
+      "step": 24456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016285454432269665,
+      "loss": 2.3321,
+      "step": 24464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016283147217821,
+      "loss": 2.3695,
+      "step": 24472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016280839473502607,
+      "loss": 2.3564,
+      "step": 24480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016278531199545937,
+      "loss": 2.3608,
+      "step": 24488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016276222396182494,
+      "loss": 2.343,
+      "step": 24496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016273913063643842,
+      "loss": 2.3622,
+      "step": 24504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016271603202161597,
+      "loss": 2.3573,
+      "step": 24512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016269292811967417,
+      "loss": 2.3377,
+      "step": 24520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016266981893293028,
+      "loss": 2.3627,
+      "step": 24528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162646704463702,
+      "loss": 2.3667,
+      "step": 24536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016262358471430758,
+      "loss": 2.3495,
+      "step": 24544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016260045968706572,
+      "loss": 2.3558,
+      "step": 24552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001625773293842959,
+      "loss": 2.3618,
+      "step": 24560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001625541938083178,
+      "loss": 2.3511,
+      "step": 24568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016253105296145187,
+      "loss": 2.3406,
+      "step": 24576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162507906846019,
+      "loss": 2.3462,
+      "step": 24584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001624847554643406,
+      "loss": 2.3325,
+      "step": 24592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001624615988187386,
+      "loss": 2.3366,
+      "step": 24600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001624384369115355,
+      "loss": 2.3308,
+      "step": 24608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016241526974505432,
+      "loss": 2.3752,
+      "step": 24616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001623920973216186,
+      "loss": 2.384,
+      "step": 24624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001623689196435524,
+      "loss": 2.3582,
+      "step": 24632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016234573671318025,
+      "loss": 2.3284,
+      "step": 24640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016232254853282732,
+      "loss": 2.3589,
+      "step": 24648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016229935510481927,
+      "loss": 2.3539,
+      "step": 24656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016227615643148226,
+      "loss": 2.3566,
+      "step": 24664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016225295251514298,
+      "loss": 2.3356,
+      "step": 24672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016222974335812861,
+      "loss": 2.3592,
+      "step": 24680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016220652896276692,
+      "loss": 2.3592,
+      "step": 24688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016218330933138626,
+      "loss": 2.3168,
+      "step": 24696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001621600844663153,
+      "loss": 2.3479,
+      "step": 24704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016213685436988344,
+      "loss": 2.3532,
+      "step": 24712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016211361904442054,
+      "loss": 2.381,
+      "step": 24720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016209037849225687,
+      "loss": 2.3638,
+      "step": 24728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016206713271572348,
+      "loss": 2.3467,
+      "step": 24736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016204388171715164,
+      "loss": 2.3445,
+      "step": 24744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001620206254988734,
+      "loss": 2.3513,
+      "step": 24752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016199736406322118,
+      "loss": 2.3738,
+      "step": 24760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016197409741252797,
+      "loss": 2.3668,
+      "step": 24768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016195082554912727,
+      "loss": 2.352,
+      "step": 24776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001619275484753532,
+      "loss": 2.3323,
+      "step": 24784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001619042661935402,
+      "loss": 2.345,
+      "step": 24792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016188097870602347,
+      "loss": 2.3678,
+      "step": 24800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001618576860151385,
+      "loss": 2.3336,
+      "step": 24808
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016183438812322144,
+      "loss": 2.3462,
+      "step": 24816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000161811085032609,
+      "loss": 2.3648,
+      "step": 24824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001617877767456383,
+      "loss": 2.359,
+      "step": 24832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016176446326464705,
+      "loss": 2.3535,
+      "step": 24840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016174114459197345,
+      "loss": 2.3545,
+      "step": 24848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016171782072995622,
+      "loss": 2.3495,
+      "step": 24856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001616944916809346,
+      "loss": 2.3503,
+      "step": 24864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016167115744724839,
+      "loss": 2.3568,
+      "step": 24872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016164781803123786,
+      "loss": 2.3508,
+      "step": 24880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016162447343524389,
+      "loss": 2.3444,
+      "step": 24888
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001616011236616077,
+      "loss": 2.3645,
+      "step": 24896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001615777687126712,
+      "loss": 2.3424,
+      "step": 24904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016155440859077678,
+      "loss": 2.3547,
+      "step": 24912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016153104329826728,
+      "loss": 2.3337,
+      "step": 24920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001615076728374861,
+      "loss": 2.3787,
+      "step": 24928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016148429721077727,
+      "loss": 2.3323,
+      "step": 24936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016146091642048511,
+      "loss": 2.3654,
+      "step": 24944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016143753046895463,
+      "loss": 2.3683,
+      "step": 24952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001614141393585313,
+      "loss": 2.3359,
+      "step": 24960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016139074309156116,
+      "loss": 2.3483,
+      "step": 24968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016136734167039064,
+      "loss": 2.3603,
+      "step": 24976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001613439350973668,
+      "loss": 2.3502,
+      "step": 24984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016132052337483724,
+      "loss": 2.3342,
+      "step": 24992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016129710650515,
+      "loss": 2.3699,
+      "step": 25000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016127368449065358,
+      "loss": 2.344,
+      "step": 25008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016125025733369717,
+      "loss": 2.3424,
+      "step": 25016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016122682503663037,
+      "loss": 2.3594,
+      "step": 25024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016120338760180328,
+      "loss": 2.36,
+      "step": 25032
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016117994503156652,
+      "loss": 2.3613,
+      "step": 25040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016115649732827128,
+      "loss": 2.3258,
+      "step": 25048
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016113304449426923,
+      "loss": 2.3698,
+      "step": 25056
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016110958653191258,
+      "loss": 2.3586,
+      "step": 25064
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016108612344355397,
+      "loss": 2.3525,
+      "step": 25072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016106265523154666,
+      "loss": 2.3534,
+      "step": 25080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016103918189824438,
+      "loss": 2.3509,
+      "step": 25088
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016101570344600135,
+      "loss": 2.3679,
+      "step": 25096
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016099221987717232,
+      "loss": 2.3677,
+      "step": 25104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001609687311941126,
+      "loss": 2.361,
+      "step": 25112
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016094523739917794,
+      "loss": 2.3361,
+      "step": 25120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016092173849472464,
+      "loss": 2.3544,
+      "step": 25128
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016089823448310947,
+      "loss": 2.3409,
+      "step": 25136
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016087472536668984,
+      "loss": 2.3485,
+      "step": 25144
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001608512111478235,
+      "loss": 2.353,
+      "step": 25152
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016082769182886885,
+      "loss": 2.3376,
+      "step": 25160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001608041674121847,
+      "loss": 2.3499,
+      "step": 25168
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016078063790013044,
+      "loss": 2.3453,
+      "step": 25176
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001607571032950659,
+      "loss": 2.3383,
+      "step": 25184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001607335635993515,
+      "loss": 2.3628,
+      "step": 25192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016071001881534818,
+      "loss": 2.3678,
+      "step": 25200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001606864689454173,
+      "loss": 2.3337,
+      "step": 25208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001606629139919208,
+      "loss": 2.3558,
+      "step": 25216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016063935395722108,
+      "loss": 2.3316,
+      "step": 25224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001606157888436811,
+      "loss": 2.3502,
+      "step": 25232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001605922186536643,
+      "loss": 2.3596,
+      "step": 25240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016056864338953467,
+      "loss": 2.3441,
+      "step": 25248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016054506305365662,
+      "loss": 2.3555,
+      "step": 25256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016052147764839515,
+      "loss": 2.3349,
+      "step": 25264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016049788717611575,
+      "loss": 2.3601,
+      "step": 25272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016047429163918442,
+      "loss": 2.3557,
+      "step": 25280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016045069103996765,
+      "loss": 2.3347,
+      "step": 25288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016042708538083238,
+      "loss": 2.3745,
+      "step": 25296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016040347466414622,
+      "loss": 2.3502,
+      "step": 25304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016037985889227714,
+      "loss": 2.3538,
+      "step": 25312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016035623806759367,
+      "loss": 2.3558,
+      "step": 25320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001603326121924649,
+      "loss": 2.3514,
+      "step": 25328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001603089812692603,
+      "loss": 2.3545,
+      "step": 25336
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016028534530034992,
+      "loss": 2.3417,
+      "step": 25344
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016026170428810437,
+      "loss": 2.3681,
+      "step": 25352
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016023805823489468,
+      "loss": 2.3764,
+      "step": 25360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001602144071430924,
+      "loss": 2.3735,
+      "step": 25368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016019075101506963,
+      "loss": 2.3288,
+      "step": 25376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016016708985319892,
+      "loss": 2.3661,
+      "step": 25384
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016014342365985336,
+      "loss": 2.3376,
+      "step": 25392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016011975243740653,
+      "loss": 2.369,
+      "step": 25400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016009607618823254,
+      "loss": 2.3474,
+      "step": 25408
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016007239491470595,
+      "loss": 2.3394,
+      "step": 25416
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001600487086192019,
+      "loss": 2.3354,
+      "step": 25424
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016002501730409595,
+      "loss": 2.3636,
+      "step": 25432
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016000132097176422,
+      "loss": 2.3228,
+      "step": 25440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001599776196245833,
+      "loss": 2.3397,
+      "step": 25448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015995391326493037,
+      "loss": 2.3327,
+      "step": 25456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015993020189518294,
+      "loss": 2.36,
+      "step": 25464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015990648551771923,
+      "loss": 2.3627,
+      "step": 25472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015988276413491777,
+      "loss": 2.3344,
+      "step": 25480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015985903774915774,
+      "loss": 2.3618,
+      "step": 25488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015983530636281873,
+      "loss": 2.364,
+      "step": 25496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015981156997828087,
+      "loss": 2.3557,
+      "step": 25504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015978782859792483,
+      "loss": 2.3556,
+      "step": 25512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015976408222413165,
+      "loss": 2.3487,
+      "step": 25520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015974033085928302,
+      "loss": 2.3641,
+      "step": 25528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015971657450576104,
+      "loss": 2.3469,
+      "step": 25536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015969281316594836,
+      "loss": 2.3521,
+      "step": 25544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015966904684222807,
+      "loss": 2.3417,
+      "step": 25552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015964527553698386,
+      "loss": 2.3446,
+      "step": 25560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015962149925259977,
+      "loss": 2.3452,
+      "step": 25568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595977179914605,
+      "loss": 2.3637,
+      "step": 25576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015957393175595113,
+      "loss": 2.3543,
+      "step": 25584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015955014054845737,
+      "loss": 2.3542,
+      "step": 25592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015952634437136522,
+      "loss": 2.351,
+      "step": 25600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595025432270614,
+      "loss": 2.3437,
+      "step": 25608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015947873711793294,
+      "loss": 2.3741,
+      "step": 25616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015945492604636755,
+      "loss": 2.3573,
+      "step": 25624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015943111001475333,
+      "loss": 2.3406,
+      "step": 25632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015940728902547882,
+      "loss": 2.362,
+      "step": 25640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001593834630809332,
+      "loss": 2.3443,
+      "step": 25648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015935963218350604,
+      "loss": 2.3741,
+      "step": 25656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015933579633558746,
+      "loss": 2.3339,
+      "step": 25664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015931195553956805,
+      "loss": 2.3407,
+      "step": 25672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015928810979783895,
+      "loss": 2.3564,
+      "step": 25680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015926425911279171,
+      "loss": 2.3782,
+      "step": 25688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015924040348681842,
+      "loss": 2.3478,
+      "step": 25696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015921654292231166,
+      "loss": 2.3544,
+      "step": 25704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015919267742166453,
+      "loss": 2.3387,
+      "step": 25712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015916880698727058,
+      "loss": 2.3613,
+      "step": 25720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015914493162152396,
+      "loss": 2.3464,
+      "step": 25728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001591210513268191,
+      "loss": 2.3402,
+      "step": 25736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015909716610555113,
+      "loss": 2.3524,
+      "step": 25744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015907327596011563,
+      "loss": 2.3427,
+      "step": 25752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001590493808929086,
+      "loss": 2.347,
+      "step": 25760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015902548090632662,
+      "loss": 2.3587,
+      "step": 25768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015900157600276667,
+      "loss": 2.3541,
+      "step": 25776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001589776661846263,
+      "loss": 2.3557,
+      "step": 25784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015895375145430355,
+      "loss": 2.3426,
+      "step": 25792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001589298318141969,
+      "loss": 2.3521,
+      "step": 25800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015890590726670538,
+      "loss": 2.3145,
+      "step": 25808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001588819778142285,
+      "loss": 2.3619,
+      "step": 25816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015885804345916623,
+      "loss": 2.328,
+      "step": 25824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000158834104203919,
+      "loss": 2.3757,
+      "step": 25832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015881016005088786,
+      "loss": 2.329,
+      "step": 25840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001587862110024742,
+      "loss": 2.3689,
+      "step": 25848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001587622570610801,
+      "loss": 2.3612,
+      "step": 25856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015873829822910784,
+      "loss": 2.3493,
+      "step": 25864
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015871433450896045,
+      "loss": 2.3461,
+      "step": 25872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015869036590304133,
+      "loss": 2.3466,
+      "step": 25880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015866639241375442,
+      "loss": 2.3555,
+      "step": 25888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015864241404350412,
+      "loss": 2.3412,
+      "step": 25896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015861843079469532,
+      "loss": 2.3376,
+      "step": 25904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001585944426697334,
+      "loss": 2.3717,
+      "step": 25912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001585704496710242,
+      "loss": 2.3511,
+      "step": 25920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015854645180097415,
+      "loss": 2.3541,
+      "step": 25928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015852244906199,
+      "loss": 2.3503,
+      "step": 25936
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015849844145647923,
+      "loss": 2.3597,
+      "step": 25944
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015847442898684957,
+      "loss": 2.3428,
+      "step": 25952
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015845041165550932,
+      "loss": 2.3713,
+      "step": 25960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015842638946486734,
+      "loss": 2.3428,
+      "step": 25968
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001584023624173329,
+      "loss": 2.3439,
+      "step": 25976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015837833051531575,
+      "loss": 2.3665,
+      "step": 25984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015835429376122618,
+      "loss": 2.3568,
+      "step": 25992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015833025215747498,
+      "loss": 2.3508,
+      "step": 26000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001583062057064733,
+      "loss": 2.37,
+      "step": 26008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015828215441063292,
+      "loss": 2.3413,
+      "step": 26016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000158258098272366,
+      "loss": 2.3753,
+      "step": 26024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001582340372940853,
+      "loss": 2.3664,
+      "step": 26032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015820997147820398,
+      "loss": 2.3182,
+      "step": 26040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015818590082713568,
+      "loss": 2.3373,
+      "step": 26048
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015816182534329455,
+      "loss": 2.3565,
+      "step": 26056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015813774502909525,
+      "loss": 2.3632,
+      "step": 26064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001581136598869529,
+      "loss": 2.3606,
+      "step": 26072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001580895699192831,
+      "loss": 2.3742,
+      "step": 26080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001580654751285019,
+      "loss": 2.3302,
+      "step": 26088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015804137551702588,
+      "loss": 2.3556,
+      "step": 26096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015801727108727217,
+      "loss": 2.3447,
+      "step": 26104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001579931618416582,
+      "loss": 2.3368,
+      "step": 26112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015796904778260209,
+      "loss": 2.374,
+      "step": 26120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015794492891252226,
+      "loss": 2.3352,
+      "step": 26128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015792080523383773,
+      "loss": 2.3634,
+      "step": 26136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000157896676748968,
+      "loss": 2.3441,
+      "step": 26144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015787254346033292,
+      "loss": 2.3276,
+      "step": 26152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015784840537035305,
+      "loss": 2.3567,
+      "step": 26160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015782426248144924,
+      "loss": 2.3533,
+      "step": 26168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015780011479604288,
+      "loss": 2.3451,
+      "step": 26176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015777596231655587,
+      "loss": 2.3481,
+      "step": 26184
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015775180504541054,
+      "loss": 2.3433,
+      "step": 26192
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015772764298502972,
+      "loss": 2.327,
+      "step": 26200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001577034761378368,
+      "loss": 2.3611,
+      "step": 26208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001576793045062555,
+      "loss": 2.3377,
+      "step": 26216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001576551280927101,
+      "loss": 2.3541,
+      "step": 26224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001576309468996254,
+      "loss": 2.3543,
+      "step": 26232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015760676092942661,
+      "loss": 2.3363,
+      "step": 26240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015758257018453947,
+      "loss": 2.365,
+      "step": 26248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015755837466739016,
+      "loss": 2.3608,
+      "step": 26256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015753417438040535,
+      "loss": 2.3474,
+      "step": 26264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015750996932601218,
+      "loss": 2.3553,
+      "step": 26272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015748575950663834,
+      "loss": 2.346,
+      "step": 26280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015746154492471182,
+      "loss": 2.3306,
+      "step": 26288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001574373255826613,
+      "loss": 2.3284,
+      "step": 26296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001574131014829158,
+      "loss": 2.352,
+      "step": 26304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015738887262790494,
+      "loss": 2.3187,
+      "step": 26312
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015736463902005864,
+      "loss": 2.3556,
+      "step": 26320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015734040066180745,
+      "loss": 2.3584,
+      "step": 26328
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015731615755558225,
+      "loss": 2.3371,
+      "step": 26336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015729190970381463,
+      "loss": 2.3758,
+      "step": 26344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001572676571089364,
+      "loss": 2.3543,
+      "step": 26352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015724339977338003,
+      "loss": 2.3462,
+      "step": 26360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001572191376995783,
+      "loss": 2.336,
+      "step": 26368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015719487088996465,
+      "loss": 2.3513,
+      "step": 26376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015717059934697292,
+      "loss": 2.3433,
+      "step": 26384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015714632307303728,
+      "loss": 2.3628,
+      "step": 26392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001571220420705926,
+      "loss": 2.3407,
+      "step": 26400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015709775634207413,
+      "loss": 2.3131,
+      "step": 26408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015707346588991755,
+      "loss": 2.3613,
+      "step": 26416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001570491707165591,
+      "loss": 2.3444,
+      "step": 26424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001570248708244354,
+      "loss": 2.3195,
+      "step": 26432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015700056621598363,
+      "loss": 2.3584,
+      "step": 26440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015697625689364137,
+      "loss": 2.3682,
+      "step": 26448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015695194285984677,
+      "loss": 2.3372,
+      "step": 26456
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015692762411703828,
+      "loss": 2.3702,
+      "step": 26464
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015690330066765504,
+      "loss": 2.3328,
+      "step": 26472
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001568789725141365,
+      "loss": 2.3444,
+      "step": 26480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015685463965892264,
+      "loss": 2.339,
+      "step": 26488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001568303021044539,
+      "loss": 2.3607,
+      "step": 26496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015680595985317127,
+      "loss": 2.3462,
+      "step": 26504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015678161290751604,
+      "loss": 2.3455,
+      "step": 26512
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015675726126993013,
+      "loss": 2.3634,
+      "step": 26520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015673290494285587,
+      "loss": 2.3436,
+      "step": 26528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015670854392873601,
+      "loss": 2.3242,
+      "step": 26536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015668417823001387,
+      "loss": 2.3532,
+      "step": 26544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015665980784913318,
+      "loss": 2.3646,
+      "step": 26552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015663543278853818,
+      "loss": 2.3541,
+      "step": 26560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001566110530506735,
+      "loss": 2.3552,
+      "step": 26568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001565866686379843,
+      "loss": 2.3261,
+      "step": 26576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015656227955291625,
+      "loss": 2.3554,
+      "step": 26584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015653788579791535,
+      "loss": 2.3611,
+      "step": 26592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001565134873754282,
+      "loss": 2.3374,
+      "step": 26600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001564890842879019,
+      "loss": 2.3489,
+      "step": 26608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015646467653778381,
+      "loss": 2.3733,
+      "step": 26616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015644026412752195,
+      "loss": 2.3396,
+      "step": 26624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015641584705956474,
+      "loss": 2.368,
+      "step": 26632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001563914253363611,
+      "loss": 2.3337,
+      "step": 26640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015636699896036036,
+      "loss": 2.3404,
+      "step": 26648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015634256793401233,
+      "loss": 2.3482,
+      "step": 26656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015631813225976734,
+      "loss": 2.3404,
+      "step": 26664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015629369194007618,
+      "loss": 2.3383,
+      "step": 26672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015626924697738993,
+      "loss": 2.3398,
+      "step": 26680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015624479737416048,
+      "loss": 2.3386,
+      "step": 26688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015622034313283987,
+      "loss": 2.3544,
+      "step": 26696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001561958842558807,
+      "loss": 2.367,
+      "step": 26704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015617142074573616,
+      "loss": 2.3442,
+      "step": 26712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015614695260485973,
+      "loss": 2.3491,
+      "step": 26720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001561224798357054,
+      "loss": 2.3523,
+      "step": 26728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001560980024407277,
+      "loss": 2.3193,
+      "step": 26736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015607352042238158,
+      "loss": 2.3346,
+      "step": 26744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015604903378312244,
+      "loss": 2.3652,
+      "step": 26752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015602454252540608,
+      "loss": 2.3316,
+      "step": 26760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015600004665168894,
+      "loss": 2.3361,
+      "step": 26768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015597554616442774,
+      "loss": 2.3498,
+      "step": 26776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015595104106607983,
+      "loss": 2.3353,
+      "step": 26784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015592653135910282,
+      "loss": 2.3532,
+      "step": 26792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015590201704595495,
+      "loss": 2.3302,
+      "step": 26800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001558774981290949,
+      "loss": 2.3484,
+      "step": 26808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001558529746109817,
+      "loss": 2.3495,
+      "step": 26816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000155828446494075,
+      "loss": 2.3646,
+      "step": 26824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015580391378083478,
+      "loss": 2.3281,
+      "step": 26832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015577937647372157,
+      "loss": 2.33,
+      "step": 26840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015575483457519625,
+      "loss": 2.3452,
+      "step": 26848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015573028808772032,
+      "loss": 2.3138,
+      "step": 26856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015570573701375562,
+      "loss": 2.3516,
+      "step": 26864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001556811813557645,
+      "loss": 2.3445,
+      "step": 26872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015565662111620965,
+      "loss": 2.3429,
+      "step": 26880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015563205629755449,
+      "loss": 2.3573,
+      "step": 26888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001556074869022626,
+      "loss": 2.3477,
+      "step": 26896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015558291293279824,
+      "loss": 2.3274,
+      "step": 26904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015555833439162597,
+      "loss": 2.3406,
+      "step": 26912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015553375128121092,
+      "loss": 2.3379,
+      "step": 26920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015550916360401864,
+      "loss": 2.3392,
+      "step": 26928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015548457136251508,
+      "loss": 2.3623,
+      "step": 26936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015545997455916675,
+      "loss": 2.3331,
+      "step": 26944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001554353731964406,
+      "loss": 2.3389,
+      "step": 26952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015541076727680395,
+      "loss": 2.352,
+      "step": 26960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015538615680272465,
+      "loss": 2.3686,
+      "step": 26968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015536154177667096,
+      "loss": 2.3361,
+      "step": 26976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015533692220111173,
+      "loss": 2.3631,
+      "step": 26984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015531229807851606,
+      "loss": 2.3459,
+      "step": 26992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015528766941135365,
+      "loss": 2.3276,
+      "step": 27000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015526303620209462,
+      "loss": 2.3583,
+      "step": 27008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001552383984532095,
+      "loss": 2.3418,
+      "step": 27016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001552137561671694,
+      "loss": 2.3421,
+      "step": 27024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015518910934644574,
+      "loss": 2.3628,
+      "step": 27032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015516445799351045,
+      "loss": 2.3553,
+      "step": 27040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015513980211083594,
+      "loss": 2.3505,
+      "step": 27048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015511514170089504,
+      "loss": 2.362,
+      "step": 27056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015509047676616112,
+      "loss": 2.3502,
+      "step": 27064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001550658073091078,
+      "loss": 2.3353,
+      "step": 27072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015504113333220942,
+      "loss": 2.3458,
+      "step": 27080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001550164548379406,
+      "loss": 2.349,
+      "step": 27088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001549917718287764,
+      "loss": 2.3257,
+      "step": 27096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001549670843071924,
+      "loss": 2.3582,
+      "step": 27104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015494239227566468,
+      "loss": 2.3337,
+      "step": 27112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015491769573666966,
+      "loss": 2.3431,
+      "step": 27120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001548929946926843,
+      "loss": 2.3536,
+      "step": 27128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001548682891461859,
+      "loss": 2.3524,
+      "step": 27136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015484357909965237,
+      "loss": 2.3405,
+      "step": 27144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015481886455556197,
+      "loss": 2.3395,
+      "step": 27152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015479414551639338,
+      "loss": 2.3419,
+      "step": 27160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001547694219846258,
+      "loss": 2.3394,
+      "step": 27168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001547446939627389,
+      "loss": 2.3433,
+      "step": 27176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015471996145321272,
+      "loss": 2.3401,
+      "step": 27184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015469522445852782,
+      "loss": 2.3359,
+      "step": 27192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015467048298116516,
+      "loss": 2.3478,
+      "step": 27200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015464573702360617,
+      "loss": 2.3498,
+      "step": 27208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015462098658833274,
+      "loss": 2.3467,
+      "step": 27216
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015459623167782719,
+      "loss": 2.3392,
+      "step": 27224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001545714722945723,
+      "loss": 2.3436,
+      "step": 27232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015454670844105125,
+      "loss": 2.3436,
+      "step": 27240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001545219401197478,
+      "loss": 2.3522,
+      "step": 27248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015449716733314606,
+      "loss": 2.3591,
+      "step": 27256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015447239008373052,
+      "loss": 2.3432,
+      "step": 27264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015444760837398627,
+      "loss": 2.3186,
+      "step": 27272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015442282220639876,
+      "loss": 2.3482,
+      "step": 27280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001543980315834539,
+      "loss": 2.3542,
+      "step": 27288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000154373236507638,
+      "loss": 2.3371,
+      "step": 27296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015434843698143797,
+      "loss": 2.3645,
+      "step": 27304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015432363300734095,
+      "loss": 2.3462,
+      "step": 27312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015429882458783473,
+      "loss": 2.335,
+      "step": 27320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001542740117254074,
+      "loss": 2.35,
+      "step": 27328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015424919442254758,
+      "loss": 2.3501,
+      "step": 27336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015422437268174425,
+      "loss": 2.3226,
+      "step": 27344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000154199546505487,
+      "loss": 2.3499,
+      "step": 27352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015417471589626564,
+      "loss": 2.3527,
+      "step": 27360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001541498808565706,
+      "loss": 2.3363,
+      "step": 27368
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015412504138889263,
+      "loss": 2.338,
+      "step": 27376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015410019749572307,
+      "loss": 2.3607,
+      "step": 27384
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015407534917955356,
+      "loss": 2.3495,
+      "step": 27392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015405049644287631,
+      "loss": 2.3433,
+      "step": 27400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015402563928818385,
+      "loss": 2.3516,
+      "step": 27408
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015400077771796926,
+      "loss": 2.3483,
+      "step": 27416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015397591173472595,
+      "loss": 2.331,
+      "step": 27424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015395104134094788,
+      "loss": 2.3395,
+      "step": 27432
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015392616653912941,
+      "loss": 2.3596,
+      "step": 27440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015390128733176536,
+      "loss": 2.3341,
+      "step": 27448
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015387640372135092,
+      "loss": 2.3516,
+      "step": 27456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001538515157103818,
+      "loss": 2.352,
+      "step": 27464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015382662330135413,
+      "loss": 2.3459,
+      "step": 27472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001538017264967645,
+      "loss": 2.3475,
+      "step": 27480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001537768252991099,
+      "loss": 2.3473,
+      "step": 27488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015375191971088778,
+      "loss": 2.337,
+      "step": 27496
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015372700973459598,
+      "loss": 2.3546,
+      "step": 27504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001537020953727329,
+      "loss": 2.3639,
+      "step": 27512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001536771766277973,
+      "loss": 2.3299,
+      "step": 27520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015365225350228836,
+      "loss": 2.3784,
+      "step": 27528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015362732599870572,
+      "loss": 2.3445,
+      "step": 27536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015360239411954952,
+      "loss": 2.3425,
+      "step": 27544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015357745786732024,
+      "loss": 2.3485,
+      "step": 27552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015355251724451886,
+      "loss": 2.3412,
+      "step": 27560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001535275722536468,
+      "loss": 2.335,
+      "step": 27568
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015350262289720587,
+      "loss": 2.3613,
+      "step": 27576
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015347766917769834,
+      "loss": 2.336,
+      "step": 27584
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000153452711097627,
+      "loss": 2.34,
+      "step": 27592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015342774865949492,
+      "loss": 2.3729,
+      "step": 27600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001534027818658057,
+      "loss": 2.332,
+      "step": 27608
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015337781071906342,
+      "loss": 2.3427,
+      "step": 27616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015335283522177254,
+      "loss": 2.331,
+      "step": 27624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015332785537643792,
+      "loss": 2.3674,
+      "step": 27632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001533028711855649,
+      "loss": 2.3552,
+      "step": 27640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015327788265165928,
+      "loss": 2.3365,
+      "step": 27648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015325288977722728,
+      "loss": 2.3378,
+      "step": 27656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001532278925647755,
+      "loss": 2.3511,
+      "step": 27664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000153202891016811,
+      "loss": 2.3587,
+      "step": 27672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015317788513584136,
+      "loss": 2.3446,
+      "step": 27680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001531528749243745,
+      "loss": 2.3359,
+      "step": 27688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015312786038491885,
+      "loss": 2.3245,
+      "step": 27696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001531028415199831,
+      "loss": 2.3622,
+      "step": 27704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015307781833207663,
+      "loss": 2.3527,
+      "step": 27712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015305279082370907,
+      "loss": 2.335,
+      "step": 27720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015302775899739052,
+      "loss": 2.3363,
+      "step": 27728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015300272285563156,
+      "loss": 2.3449,
+      "step": 27736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015297768240094318,
+      "loss": 2.3365,
+      "step": 27744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015295263763583677,
+      "loss": 2.3446,
+      "step": 27752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015292758856282418,
+      "loss": 2.3681,
+      "step": 27760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001529025351844177,
+      "loss": 2.3404,
+      "step": 27768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015287747750313005,
+      "loss": 2.3392,
+      "step": 27776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015285241552147436,
+      "loss": 2.3452,
+      "step": 27784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001528273492419642,
+      "loss": 2.3503,
+      "step": 27792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015280227866711364,
+      "loss": 2.3446,
+      "step": 27800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015277720379943698,
+      "loss": 2.357,
+      "step": 27808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015275212464144918,
+      "loss": 2.3229,
+      "step": 27816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015272704119566558,
+      "loss": 2.3273,
+      "step": 27824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001527019534646018,
+      "loss": 2.3513,
+      "step": 27832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015267686145077408,
+      "loss": 2.3468,
+      "step": 27840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015265176515669897,
+      "loss": 2.3324,
+      "step": 27848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015262666458489346,
+      "loss": 2.3574,
+      "step": 27856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015260155973787507,
+      "loss": 2.3569,
+      "step": 27864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015257645061816162,
+      "loss": 2.3308,
+      "step": 27872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001525513372282714,
+      "loss": 2.3383,
+      "step": 27880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015252621957072323,
+      "loss": 2.3512,
+      "step": 27888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015250109764803612,
+      "loss": 2.3349,
+      "step": 27896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001524759714627298,
+      "loss": 2.3295,
+      "step": 27904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001524508410173242,
+      "loss": 2.3294,
+      "step": 27912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015242570631433978,
+      "loss": 2.3564,
+      "step": 27920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015240056735629746,
+      "loss": 2.3491,
+      "step": 27928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015237542414571844,
+      "loss": 2.3508,
+      "step": 27936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015235027668512448,
+      "loss": 2.3542,
+      "step": 27944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015232512497703778,
+      "loss": 2.3519,
+      "step": 27952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015229996902398085,
+      "loss": 2.3242,
+      "step": 27960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015227480882847673,
+      "loss": 2.3295,
+      "step": 27968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001522496443930488,
+      "loss": 2.3675,
+      "step": 27976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015222447572022096,
+      "loss": 2.3498,
+      "step": 27984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015219930281251746,
+      "loss": 2.3399,
+      "step": 27992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015217412567246298,
+      "loss": 2.3291,
+      "step": 28000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001521489443025827,
+      "loss": 2.3343,
+      "step": 28008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015212375870540213,
+      "loss": 2.3622,
+      "step": 28016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015209856888344722,
+      "loss": 2.3427,
+      "step": 28024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001520733748392444,
+      "loss": 2.3513,
+      "step": 28032
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001520481765753205,
+      "loss": 2.3327,
+      "step": 28040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015202297409420274,
+      "loss": 2.3453,
+      "step": 28048
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001519977673984188,
+      "loss": 2.3351,
+      "step": 28056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001519725564904968,
+      "loss": 2.3244,
+      "step": 28064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015194734137296515,
+      "loss": 2.3351,
+      "step": 28072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015192212204835287,
+      "loss": 2.3457,
+      "step": 28080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015189689851918928,
+      "loss": 2.3522,
+      "step": 28088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001518716707880042,
+      "loss": 2.3318,
+      "step": 28096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015184643885732778,
+      "loss": 2.3137,
+      "step": 28104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015182120272969066,
+      "loss": 2.3415,
+      "step": 28112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015179596240762387,
+      "loss": 2.3595,
+      "step": 28120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001517707178936589,
+      "loss": 2.3465,
+      "step": 28128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015174546919032757,
+      "loss": 2.338,
+      "step": 28136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015172021630016226,
+      "loss": 2.3299,
+      "step": 28144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015169495922569566,
+      "loss": 2.3421,
+      "step": 28152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015166969796946087,
+      "loss": 2.3332,
+      "step": 28160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015164443253399152,
+      "loss": 2.3336,
+      "step": 28168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015161916292182154,
+      "loss": 2.3521,
+      "step": 28176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015159388913548537,
+      "loss": 2.3402,
+      "step": 28184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015156861117751778,
+      "loss": 2.3478,
+      "step": 28192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015154332905045403,
+      "loss": 2.3329,
+      "step": 28200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015151804275682977,
+      "loss": 2.3357,
+      "step": 28208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015149275229918107,
+      "loss": 2.3491,
+      "step": 28216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015146745768004444,
+      "loss": 2.3523,
+      "step": 28224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015144215890195678,
+      "loss": 2.3362,
+      "step": 28232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015141685596745542,
+      "loss": 2.3381,
+      "step": 28240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015139154887907805,
+      "loss": 2.3543,
+      "step": 28248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001513662376393629,
+      "loss": 2.3439,
+      "step": 28256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015134092225084847,
+      "loss": 2.345,
+      "step": 28264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001513156027160738,
+      "loss": 2.3319,
+      "step": 28272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001512902790375783,
+      "loss": 2.332,
+      "step": 28280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015126495121790178,
+      "loss": 2.3283,
+      "step": 28288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015123961925958445,
+      "loss": 2.3364,
+      "step": 28296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015121428316516698,
+      "loss": 2.3491,
+      "step": 28304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015118894293719045,
+      "loss": 2.3717,
+      "step": 28312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015116359857819634,
+      "loss": 2.3313,
+      "step": 28320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015113825009072657,
+      "loss": 2.3385,
+      "step": 28328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001511128974773234,
+      "loss": 2.3365,
+      "step": 28336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015108754074052955,
+      "loss": 2.3404,
+      "step": 28344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015106217988288816,
+      "loss": 2.331,
+      "step": 28352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015103681490694282,
+      "loss": 2.3407,
+      "step": 28360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001510114458152375,
+      "loss": 2.3581,
+      "step": 28368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015098607261031653,
+      "loss": 2.3557,
+      "step": 28376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001509606952947247,
+      "loss": 2.3406,
+      "step": 28384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015093531387100725,
+      "loss": 2.3517,
+      "step": 28392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015090992834170975,
+      "loss": 2.3336,
+      "step": 28400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001508845387093783,
+      "loss": 2.3263,
+      "step": 28408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015085914497655923,
+      "loss": 2.3366,
+      "step": 28416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015083374714579942,
+      "loss": 2.3341,
+      "step": 28424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001508083452196462,
+      "loss": 2.3656,
+      "step": 28432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015078293920064718,
+      "loss": 2.3516,
+      "step": 28440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015075752909135038,
+      "loss": 2.3374,
+      "step": 28448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015073211489430446,
+      "loss": 2.3196,
+      "step": 28456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015070669661205817,
+      "loss": 2.3555,
+      "step": 28464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015068127424716085,
+      "loss": 2.3525,
+      "step": 28472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015065584780216224,
+      "loss": 2.35,
+      "step": 28480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015063041727961246,
+      "loss": 2.3504,
+      "step": 28488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015060498268206206,
+      "loss": 2.3407,
+      "step": 28496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000150579544012062,
+      "loss": 2.3414,
+      "step": 28504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015055410127216357,
+      "loss": 2.3395,
+      "step": 28512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015052865446491863,
+      "loss": 2.3395,
+      "step": 28520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015050320359287926,
+      "loss": 2.3479,
+      "step": 28528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001504777486585981,
+      "loss": 2.3378,
+      "step": 28536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001504522896646281,
+      "loss": 2.3331,
+      "step": 28544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001504268266135226,
+      "loss": 2.3431,
+      "step": 28552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015040135950783554,
+      "loss": 2.3502,
+      "step": 28560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015037588835012099,
+      "loss": 2.3253,
+      "step": 28568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015035041314293365,
+      "loss": 2.3419,
+      "step": 28576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001503249338888285,
+      "loss": 2.3567,
+      "step": 28584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000150299450590361,
+      "loss": 2.3198,
+      "step": 28592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015027396325008696,
+      "loss": 2.3482,
+      "step": 28600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015024847187056258,
+      "loss": 2.3629,
+      "step": 28608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015022297645434453,
+      "loss": 2.3581,
+      "step": 28616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015019747700398988,
+      "loss": 2.3516,
+      "step": 28624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015017197352205604,
+      "loss": 2.329,
+      "step": 28632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015014646601110092,
+      "loss": 2.3226,
+      "step": 28640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001501209544736827,
+      "loss": 2.3293,
+      "step": 28648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001500954389123601,
+      "loss": 2.3383,
+      "step": 28656
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015006991932969214,
+      "loss": 2.3345,
+      "step": 28664
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00015004439572823835,
+      "loss": 2.347,
+      "step": 28672
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001500188681105586,
+      "loss": 2.3363,
+      "step": 28680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014999333647921308,
+      "loss": 2.3535,
+      "step": 28688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014996780083676256,
+      "loss": 2.3487,
+      "step": 28696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014994226118576808,
+      "loss": 2.3424,
+      "step": 28704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001499167175287911,
+      "loss": 2.3354,
+      "step": 28712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014989116986839358,
+      "loss": 2.3441,
+      "step": 28720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014986561820713775,
+      "loss": 2.332,
+      "step": 28728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001498400625475863,
+      "loss": 2.3381,
+      "step": 28736
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014981450289230228,
+      "loss": 2.3243,
+      "step": 28744
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014978893924384924,
+      "loss": 2.3433,
+      "step": 28752
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014976337160479107,
+      "loss": 2.3453,
+      "step": 28760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014973779997769207,
+      "loss": 2.334,
+      "step": 28768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014971222436511688,
+      "loss": 2.3379,
+      "step": 28776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014968664476963054,
+      "loss": 2.3375,
+      "step": 28784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001496610611937987,
+      "loss": 2.3479,
+      "step": 28792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014963547364018712,
+      "loss": 2.3551,
+      "step": 28800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014960988211136207,
+      "loss": 2.3474,
+      "step": 28808
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014958428660989036,
+      "loss": 2.3363,
+      "step": 28816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014955868713833895,
+      "loss": 2.3387,
+      "step": 28824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001495330836992754,
+      "loss": 2.327,
+      "step": 28832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014950747629526755,
+      "loss": 2.3386,
+      "step": 28840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001494818649288837,
+      "loss": 2.3362,
+      "step": 28848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001494562496026925,
+      "loss": 2.3549,
+      "step": 28856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014943063031926308,
+      "loss": 2.3443,
+      "step": 28864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014940500708116483,
+      "loss": 2.3407,
+      "step": 28872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014937937989096764,
+      "loss": 2.3372,
+      "step": 28880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014935374875124177,
+      "loss": 2.3329,
+      "step": 28888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001493281136645579,
+      "loss": 2.3524,
+      "step": 28896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001493024746334871,
+      "loss": 2.3366,
+      "step": 28904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001492768316606008,
+      "loss": 2.3344,
+      "step": 28912
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014925118474847082,
+      "loss": 2.3587,
+      "step": 28920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001492255338996694,
+      "loss": 2.3319,
+      "step": 28928
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014919987911676921,
+      "loss": 2.3376,
+      "step": 28936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001491742204023433,
+      "loss": 2.3359,
+      "step": 28944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000149148557758965,
+      "loss": 2.3534,
+      "step": 28952
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014912289118920822,
+      "loss": 2.3306,
+      "step": 28960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014909722069564714,
+      "loss": 2.3442,
+      "step": 28968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014907154628085632,
+      "loss": 2.3387,
+      "step": 28976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014904586794741088,
+      "loss": 2.3303,
+      "step": 28984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014902018569788608,
+      "loss": 2.3572,
+      "step": 28992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014899449953485774,
+      "loss": 2.3339,
+      "step": 29000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014896880946090204,
+      "loss": 2.3472,
+      "step": 29008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001489431154785956,
+      "loss": 2.3409,
+      "step": 29016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014891741759051532,
+      "loss": 2.3505,
+      "step": 29024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001488917157992386,
+      "loss": 2.3222,
+      "step": 29032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014886601010734312,
+      "loss": 2.3337,
+      "step": 29040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001488403005174071,
+      "loss": 2.3479,
+      "step": 29048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014881458703200896,
+      "loss": 2.3243,
+      "step": 29056
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001487888696537277,
+      "loss": 2.3342,
+      "step": 29064
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014876314838514266,
+      "loss": 2.3166,
+      "step": 29072
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001487374232288334,
+      "loss": 2.3249,
+      "step": 29080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014871169418738013,
+      "loss": 2.3302,
+      "step": 29088
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014868596126336328,
+      "loss": 2.3495,
+      "step": 29096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014866022445936372,
+      "loss": 2.3304,
+      "step": 29104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001486344837779627,
+      "loss": 2.3214,
+      "step": 29112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014860873922174187,
+      "loss": 2.3299,
+      "step": 29120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001485829907932833,
+      "loss": 2.3484,
+      "step": 29128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014855723849516934,
+      "loss": 2.3348,
+      "step": 29136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014853148232998284,
+      "loss": 2.3215,
+      "step": 29144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014850572230030702,
+      "loss": 2.3235,
+      "step": 29152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001484799584087254,
+      "loss": 2.3429,
+      "step": 29160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014845419065782203,
+      "loss": 2.35,
+      "step": 29168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001484284190501812,
+      "loss": 2.3447,
+      "step": 29176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001484026435883877,
+      "loss": 2.3307,
+      "step": 29184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001483768642750266,
+      "loss": 2.3297,
+      "step": 29192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014835108111268354,
+      "loss": 2.3107,
+      "step": 29200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014832529410394436,
+      "loss": 2.3137,
+      "step": 29208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001482995032513953,
+      "loss": 2.3262,
+      "step": 29216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001482737085576231,
+      "loss": 2.3533,
+      "step": 29224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014824791002521478,
+      "loss": 2.3325,
+      "step": 29232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014822210765675785,
+      "loss": 2.3406,
+      "step": 29240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014819630145484008,
+      "loss": 2.3257,
+      "step": 29248
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014817049142204969,
+      "loss": 2.3314,
+      "step": 29256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001481446775609753,
+      "loss": 2.3156,
+      "step": 29264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014811885987420592,
+      "loss": 2.3407,
+      "step": 29272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014809303836433086,
+      "loss": 2.3236,
+      "step": 29280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001480672130339399,
+      "loss": 2.3444,
+      "step": 29288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014804138388562317,
+      "loss": 2.3287,
+      "step": 29296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014801555092197122,
+      "loss": 2.3245,
+      "step": 29304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014798971414557488,
+      "loss": 2.3374,
+      "step": 29312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014796387355902547,
+      "loss": 2.3293,
+      "step": 29320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001479380291649147,
+      "loss": 2.3382,
+      "step": 29328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014791218096583456,
+      "loss": 2.3235,
+      "step": 29336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001478863289643775,
+      "loss": 2.3306,
+      "step": 29344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001478604731631363,
+      "loss": 2.3396,
+      "step": 29352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014783461356470415,
+      "loss": 2.3325,
+      "step": 29360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014780875017167468,
+      "loss": 2.3407,
+      "step": 29368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014778288298664181,
+      "loss": 2.3434,
+      "step": 29376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014775701201219982,
+      "loss": 2.3369,
+      "step": 29384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001477311372509435,
+      "loss": 2.3491,
+      "step": 29392
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014770525870546792,
+      "loss": 2.3242,
+      "step": 29400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014767937637836853,
+      "loss": 2.3186,
+      "step": 29408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001476534902722412,
+      "loss": 2.3348,
+      "step": 29416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014762760038968214,
+      "loss": 2.3589,
+      "step": 29424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000147601706733288,
+      "loss": 2.3274,
+      "step": 29432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001475758093056557,
+      "loss": 2.3279,
+      "step": 29440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014754990810938266,
+      "loss": 2.3216,
+      "step": 29448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014752400314706657,
+      "loss": 2.3277,
+      "step": 29456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014749809442130566,
+      "loss": 2.325,
+      "step": 29464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001474721819346983,
+      "loss": 2.318,
+      "step": 29472
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014744626568984344,
+      "loss": 2.3395,
+      "step": 29480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001474203456893403,
+      "loss": 2.3437,
+      "step": 29488
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014739442193578852,
+      "loss": 2.3328,
+      "step": 29496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014736849443178812,
+      "loss": 2.339,
+      "step": 29504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001473425631799395,
+      "loss": 2.3505,
+      "step": 29512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014731662818284332,
+      "loss": 2.3389,
+      "step": 29520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014729068944310082,
+      "loss": 2.3392,
+      "step": 29528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014726474696331347,
+      "loss": 2.3339,
+      "step": 29536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014723880074608316,
+      "loss": 2.3296,
+      "step": 29544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014721285079401212,
+      "loss": 2.3506,
+      "step": 29552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000147186897109703,
+      "loss": 2.3408,
+      "step": 29560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014716093969575886,
+      "loss": 2.3333,
+      "step": 29568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014713497855478299,
+      "loss": 2.3327,
+      "step": 29576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014710901368937922,
+      "loss": 2.3382,
+      "step": 29584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014708304510215162,
+      "loss": 2.3124,
+      "step": 29592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014705707279570477,
+      "loss": 2.3373,
+      "step": 29600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014703109677264344,
+      "loss": 2.338,
+      "step": 29608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014700511703557293,
+      "loss": 2.3312,
+      "step": 29616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014697913358709887,
+      "loss": 2.3422,
+      "step": 29624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014695314642982726,
+      "loss": 2.3347,
+      "step": 29632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014692715556636446,
+      "loss": 2.3351,
+      "step": 29640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014690116099931724,
+      "loss": 2.3259,
+      "step": 29648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014687516273129259,
+      "loss": 2.334,
+      "step": 29656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001468491607648981,
+      "loss": 2.3336,
+      "step": 29664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014682315510274155,
+      "loss": 2.3478,
+      "step": 29672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001467971457474312,
+      "loss": 2.3421,
+      "step": 29680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001467711327015757,
+      "loss": 2.3377,
+      "step": 29688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001467451159677839,
+      "loss": 2.3256,
+      "step": 29696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014671909554866516,
+      "loss": 2.34,
+      "step": 29704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014669307144682923,
+      "loss": 2.3314,
+      "step": 29712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014666704366488612,
+      "loss": 2.3404,
+      "step": 29720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001466410122054463,
+      "loss": 2.3467,
+      "step": 29728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014661497707112062,
+      "loss": 2.3349,
+      "step": 29736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014658893826452018,
+      "loss": 2.3457,
+      "step": 29744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014656289578825654,
+      "loss": 2.3265,
+      "step": 29752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014653684964494162,
+      "loss": 2.3338,
+      "step": 29760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014651079983718775,
+      "loss": 2.3222,
+      "step": 29768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014648474636760753,
+      "loss": 2.3273,
+      "step": 29776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014645868923881397,
+      "loss": 2.3521,
+      "step": 29784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014643262845342049,
+      "loss": 2.3332,
+      "step": 29792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001464065640140408,
+      "loss": 2.3301,
+      "step": 29800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000146380495923289,
+      "loss": 2.3247,
+      "step": 29808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014635442418377964,
+      "loss": 2.3303,
+      "step": 29816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014632834879812755,
+      "loss": 2.3413,
+      "step": 29824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014630226976894789,
+      "loss": 2.3486,
+      "step": 29832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014627618709885627,
+      "loss": 2.3395,
+      "step": 29840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014625010079046864,
+      "loss": 2.325,
+      "step": 29848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001462240108464013,
+      "loss": 2.3321,
+      "step": 29856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001461979172692709,
+      "loss": 2.3157,
+      "step": 29864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014617182006169456,
+      "loss": 2.3328,
+      "step": 29872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014614571922628964,
+      "loss": 2.3355,
+      "step": 29880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014611961476567383,
+      "loss": 2.3394,
+      "step": 29888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014609350668246535,
+      "loss": 2.3394,
+      "step": 29896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014606739497928268,
+      "loss": 2.3201,
+      "step": 29904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014604127965874467,
+      "loss": 2.3392,
+      "step": 29912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014601516072347052,
+      "loss": 2.3644,
+      "step": 29920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014598903817607982,
+      "loss": 2.344,
+      "step": 29928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014596291201919253,
+      "loss": 2.335,
+      "step": 29936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014593678225542894,
+      "loss": 2.3298,
+      "step": 29944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001459106488874097,
+      "loss": 2.3384,
+      "step": 29952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014588451191775587,
+      "loss": 2.3236,
+      "step": 29960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001458583713490888,
+      "loss": 2.3433,
+      "step": 29968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001458322271840303,
+      "loss": 2.3322,
+      "step": 29976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014580607942520242,
+      "loss": 2.3439,
+      "step": 29984
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001457799280752277,
+      "loss": 2.3376,
+      "step": 29992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014575377313672894,
+      "loss": 2.334,
+      "step": 30000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001457276146123293,
+      "loss": 2.346,
+      "step": 30008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014570145250465238,
+      "loss": 2.3478,
+      "step": 30016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014567528681632203,
+      "loss": 2.3316,
+      "step": 30024
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014564911754996258,
+      "loss": 2.3405,
+      "step": 30032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014562294470819864,
+      "loss": 2.348,
+      "step": 30040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001455967682936552,
+      "loss": 2.335,
+      "step": 30048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001455705883089576,
+      "loss": 2.35,
+      "step": 30056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014554440475673155,
+      "loss": 2.3286,
+      "step": 30064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001455182176396031,
+      "loss": 2.3412,
+      "step": 30072
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001454920269601987,
+      "loss": 2.3289,
+      "step": 30080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014546583272114508,
+      "loss": 2.3202,
+      "step": 30088
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001454396349250694,
+      "loss": 2.3399,
+      "step": 30096
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014541343357459915,
+      "loss": 2.3324,
+      "step": 30104
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014538722867236218,
+      "loss": 2.3385,
+      "step": 30112
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001453610202209867,
+      "loss": 2.3492,
+      "step": 30120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014533480822310124,
+      "loss": 2.3365,
+      "step": 30128
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014530859268133475,
+      "loss": 2.3304,
+      "step": 30136
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014528237359831648,
+      "loss": 2.3142,
+      "step": 30144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014525615097667603,
+      "loss": 2.3339,
+      "step": 30152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014522992481904345,
+      "loss": 2.3159,
+      "step": 30160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000145203695128049,
+      "loss": 2.3171,
+      "step": 30168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014517746190632343,
+      "loss": 2.3399,
+      "step": 30176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014515122515649777,
+      "loss": 2.3326,
+      "step": 30184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014512498488120337,
+      "loss": 2.3464,
+      "step": 30192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014509874108307205,
+      "loss": 2.3256,
+      "step": 30200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014507249376473585,
+      "loss": 2.342,
+      "step": 30208
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014504624292882728,
+      "loss": 2.3033,
+      "step": 30216
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014501998857797908,
+      "loss": 2.3319,
+      "step": 30224
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001449937307148245,
+      "loss": 2.322,
+      "step": 30232
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000144967469341997,
+      "loss": 2.3478,
+      "step": 30240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001449412044621305,
+      "loss": 2.3316,
+      "step": 30248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014491493607785916,
+      "loss": 2.3201,
+      "step": 30256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014488866419181754,
+      "loss": 2.3326,
+      "step": 30264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014486238880664063,
+      "loss": 2.3317,
+      "step": 30272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014483610992496363,
+      "loss": 2.3521,
+      "step": 30280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014480982754942222,
+      "loss": 2.353,
+      "step": 30288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014478354168265235,
+      "loss": 2.3476,
+      "step": 30296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014475725232729032,
+      "loss": 2.3191,
+      "step": 30304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014473095948597286,
+      "loss": 2.315,
+      "step": 30312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014470466316133696,
+      "loss": 2.3267,
+      "step": 30320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014467836335601994,
+      "loss": 2.3145,
+      "step": 30328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014465206007265963,
+      "loss": 2.3269,
+      "step": 30336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014462575331389402,
+      "loss": 2.3471,
+      "step": 30344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014459944308236157,
+      "loss": 2.3319,
+      "step": 30352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000144573129380701,
+      "loss": 2.3414,
+      "step": 30360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001445468122115515,
+      "loss": 2.3323,
+      "step": 30368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014452049157755243,
+      "loss": 2.3348,
+      "step": 30376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001444941674813437,
+      "loss": 2.3375,
+      "step": 30384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014446783992556542,
+      "loss": 2.3368,
+      "step": 30392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014444150891285807,
+      "loss": 2.3473,
+      "step": 30400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001444151744458626,
+      "loss": 2.3309,
+      "step": 30408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001443888365272201,
+      "loss": 2.3415,
+      "step": 30416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014436249515957213,
+      "loss": 2.3244,
+      "step": 30424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001443361503455606,
+      "loss": 2.3226,
+      "step": 30432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014430980208782781,
+      "loss": 2.3234,
+      "step": 30440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014428345038901627,
+      "loss": 2.3263,
+      "step": 30448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014425709525176889,
+      "loss": 2.3313,
+      "step": 30456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014423073667872892,
+      "loss": 2.3595,
+      "step": 30464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014420437467254008,
+      "loss": 2.3352,
+      "step": 30472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001441780092358463,
+      "loss": 2.3284,
+      "step": 30480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014415164037129182,
+      "loss": 2.3267,
+      "step": 30488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014412526808152133,
+      "loss": 2.3277,
+      "step": 30496
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014409889236917983,
+      "loss": 2.3253,
+      "step": 30504
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001440725132369126,
+      "loss": 2.3273,
+      "step": 30512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014404613068736539,
+      "loss": 2.3413,
+      "step": 30520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014401974472318417,
+      "loss": 2.3401,
+      "step": 30528
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014399335534701535,
+      "loss": 2.3177,
+      "step": 30536
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001439669625615056,
+      "loss": 2.3336,
+      "step": 30544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014394056636930194,
+      "loss": 2.2879,
+      "step": 30552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014391416677305183,
+      "loss": 2.3249,
+      "step": 30560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014388776377540294,
+      "loss": 2.3318,
+      "step": 30568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014386135737900338,
+      "loss": 2.3436,
+      "step": 30576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001438349475865015,
+      "loss": 2.3322,
+      "step": 30584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014380853440054616,
+      "loss": 2.3301,
+      "step": 30592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014378211782378633,
+      "loss": 2.3403,
+      "step": 30600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014375569785887154,
+      "loss": 2.3475,
+      "step": 30608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014372927450845153,
+      "loss": 2.323,
+      "step": 30616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014370284777517637,
+      "loss": 2.3429,
+      "step": 30624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014367641766169654,
+      "loss": 2.3332,
+      "step": 30632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001436499841706629,
+      "loss": 2.3243,
+      "step": 30640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014362354730472647,
+      "loss": 2.3246,
+      "step": 30648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014359710706653882,
+      "loss": 2.3242,
+      "step": 30656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014357066345875165,
+      "loss": 2.3182,
+      "step": 30664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014354421648401716,
+      "loss": 2.3333,
+      "step": 30672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014351776614498785,
+      "loss": 2.3454,
+      "step": 30680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001434913124443165,
+      "loss": 2.3311,
+      "step": 30688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014346485538465633,
+      "loss": 2.34,
+      "step": 30696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014343839496866074,
+      "loss": 2.3112,
+      "step": 30704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001434119311989836,
+      "loss": 2.3262,
+      "step": 30712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001433854640782791,
+      "loss": 2.3146,
+      "step": 30720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014335899360920176,
+      "loss": 2.335,
+      "step": 30728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001433325197944063,
+      "loss": 2.3279,
+      "step": 30736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014330604263654807,
+      "loss": 2.3299,
+      "step": 30744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014327956213828244,
+      "loss": 2.3252,
+      "step": 30752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001432530783022653,
+      "loss": 2.3326,
+      "step": 30760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014322659113115283,
+      "loss": 2.3351,
+      "step": 30768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014320010062760153,
+      "loss": 2.3142,
+      "step": 30776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014317360679426828,
+      "loss": 2.3354,
+      "step": 30784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014314710963381024,
+      "loss": 2.3315,
+      "step": 30792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014312060914888492,
+      "loss": 2.3304,
+      "step": 30800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014309410534215017,
+      "loss": 2.3285,
+      "step": 30808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014306759821626415,
+      "loss": 2.3338,
+      "step": 30816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014304108777388546,
+      "loss": 2.3328,
+      "step": 30824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014301457401767287,
+      "loss": 2.3262,
+      "step": 30832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014298805695028556,
+      "loss": 2.3694,
+      "step": 30840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014296153657438306,
+      "loss": 2.3391,
+      "step": 30848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001429350128926252,
+      "loss": 2.3307,
+      "step": 30856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014290848590767222,
+      "loss": 2.3472,
+      "step": 30864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001428819556221845,
+      "loss": 2.3372,
+      "step": 30872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000142855422038823,
+      "loss": 2.3245,
+      "step": 30880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014282888516024882,
+      "loss": 2.3338,
+      "step": 30888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001428023449891235,
+      "loss": 2.3499,
+      "step": 30896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001427758015281088,
+      "loss": 2.3538,
+      "step": 30904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014274925477986695,
+      "loss": 2.3487,
+      "step": 30912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014272270474706045,
+      "loss": 2.3325,
+      "step": 30920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014269615143235205,
+      "loss": 2.3428,
+      "step": 30928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001426695948384049,
+      "loss": 2.336,
+      "step": 30936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001426430349678825,
+      "loss": 2.3348,
+      "step": 30944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001426164718234487,
+      "loss": 2.3212,
+      "step": 30952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001425899054077676,
+      "loss": 2.3392,
+      "step": 30960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014256333572350361,
+      "loss": 2.3245,
+      "step": 30968
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001425367627733216,
+      "loss": 2.3066,
+      "step": 30976
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014251018655988661,
+      "loss": 2.336,
+      "step": 30984
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014248360708586413,
+      "loss": 2.3472,
+      "step": 30992
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014245702435391992,
+      "loss": 2.3399,
+      "step": 31000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014243043836672012,
+      "loss": 2.3288,
+      "step": 31008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014240384912693105,
+      "loss": 2.3253,
+      "step": 31016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014237725663721957,
+      "loss": 2.3653,
+      "step": 31024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014235066090025267,
+      "loss": 2.3518,
+      "step": 31032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014232406191869783,
+      "loss": 2.3105,
+      "step": 31040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014229745969522272,
+      "loss": 2.3407,
+      "step": 31048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014227085423249545,
+      "loss": 2.3451,
+      "step": 31056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014224424553318428,
+      "loss": 2.337,
+      "step": 31064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014221763359995805,
+      "loss": 2.3458,
+      "step": 31072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014219101843548568,
+      "loss": 2.3316,
+      "step": 31080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001421644000424366,
+      "loss": 2.3445,
+      "step": 31088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014213777842348043,
+      "loss": 2.3446,
+      "step": 31096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014211115358128722,
+      "loss": 2.3419,
+      "step": 31104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014208452551852722,
+      "loss": 2.3441,
+      "step": 31112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014205789423787113,
+      "loss": 2.3363,
+      "step": 31120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001420312597419899,
+      "loss": 2.3372,
+      "step": 31128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014200462203355478,
+      "loss": 2.3237,
+      "step": 31136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014197798111523747,
+      "loss": 2.3379,
+      "step": 31144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014195133698970984,
+      "loss": 2.3358,
+      "step": 31152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014192468965964412,
+      "loss": 2.3271,
+      "step": 31160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014189803912771293,
+      "loss": 2.3326,
+      "step": 31168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001418713853965892,
+      "loss": 2.3544,
+      "step": 31176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001418447284689461,
+      "loss": 2.3409,
+      "step": 31184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014181806834745713,
+      "loss": 2.3238,
+      "step": 31192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014179140503479623,
+      "loss": 2.3243,
+      "step": 31200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014176473853363755,
+      "loss": 2.3351,
+      "step": 31208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014173806884665557,
+      "loss": 2.334,
+      "step": 31216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014171139597652513,
+      "loss": 2.3482,
+      "step": 31224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014168471992592136,
+      "loss": 2.3243,
+      "step": 31232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014165804069751973,
+      "loss": 2.3527,
+      "step": 31240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014163135829399598,
+      "loss": 2.3315,
+      "step": 31248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014160467271802623,
+      "loss": 2.3231,
+      "step": 31256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014157798397228689,
+      "loss": 2.3578,
+      "step": 31264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001415512920594547,
+      "loss": 2.3275,
+      "step": 31272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001415245969822067,
+      "loss": 2.3344,
+      "step": 31280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014149789874322023,
+      "loss": 2.3109,
+      "step": 31288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000141471197345173,
+      "loss": 2.3177,
+      "step": 31296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014144449279074298,
+      "loss": 2.332,
+      "step": 31304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014141778508260853,
+      "loss": 2.3502,
+      "step": 31312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014139107422344829,
+      "loss": 2.3215,
+      "step": 31320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014136436021594114,
+      "loss": 2.3411,
+      "step": 31328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014133764306276637,
+      "loss": 2.3346,
+      "step": 31336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001413109227666036,
+      "loss": 2.3291,
+      "step": 31344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014128419933013268,
+      "loss": 2.325,
+      "step": 31352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014125747275603386,
+      "loss": 2.3407,
+      "step": 31360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014123074304698757,
+      "loss": 2.3422,
+      "step": 31368
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014120401020567476,
+      "loss": 2.319,
+      "step": 31376
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014117727423477652,
+      "loss": 2.3368,
+      "step": 31384
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014115053513697438,
+      "loss": 2.3233,
+      "step": 31392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014112379291495003,
+      "loss": 2.3315,
+      "step": 31400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014109704757138565,
+      "loss": 2.3328,
+      "step": 31408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014107029910896354,
+      "loss": 2.3139,
+      "step": 31416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014104354753036654,
+      "loss": 2.3181,
+      "step": 31424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001410167928382776,
+      "loss": 2.3383,
+      "step": 31432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014099003503538008,
+      "loss": 2.3449,
+      "step": 31440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014096327412435765,
+      "loss": 2.3204,
+      "step": 31448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014093651010789433,
+      "loss": 2.3385,
+      "step": 31456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014090974298867426,
+      "loss": 2.3069,
+      "step": 31464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014088297276938213,
+      "loss": 2.3453,
+      "step": 31472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014085619945270288,
+      "loss": 2.3164,
+      "step": 31480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001408294230413216,
+      "loss": 2.3504,
+      "step": 31488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014080264353792392,
+      "loss": 2.3526,
+      "step": 31496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014077586094519558,
+      "loss": 2.3364,
+      "step": 31504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001407490752658228,
+      "loss": 2.3284,
+      "step": 31512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014072228650249203,
+      "loss": 2.3399,
+      "step": 31520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014069549465789001,
+      "loss": 2.3218,
+      "step": 31528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001406686997347038,
+      "loss": 2.3116,
+      "step": 31536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014064190173562074,
+      "loss": 2.3341,
+      "step": 31544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001406151006633286,
+      "loss": 2.3373,
+      "step": 31552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014058829652051534,
+      "loss": 2.3532,
+      "step": 31560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014056148930986922,
+      "loss": 2.3232,
+      "step": 31568
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014053467903407893,
+      "loss": 2.3351,
+      "step": 31576
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014050786569583335,
+      "loss": 2.359,
+      "step": 31584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001404810492978217,
+      "loss": 2.3445,
+      "step": 31592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014045422984273353,
+      "loss": 2.3469,
+      "step": 31600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014042740733325866,
+      "loss": 2.3274,
+      "step": 31608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014040058177208726,
+      "loss": 2.345,
+      "step": 31616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014037375316190974,
+      "loss": 2.3238,
+      "step": 31624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014034692150541696,
+      "loss": 2.3124,
+      "step": 31632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001403200868052998,
+      "loss": 2.3365,
+      "step": 31640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014029324906424982,
+      "loss": 2.3329,
+      "step": 31648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014026640828495855,
+      "loss": 2.317,
+      "step": 31656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001402395644701181,
+      "loss": 2.3134,
+      "step": 31664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001402127176224206,
+      "loss": 2.3412,
+      "step": 31672
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014018586774455874,
+      "loss": 2.3381,
+      "step": 31680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014015901483922539,
+      "loss": 2.3443,
+      "step": 31688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014013215890911377,
+      "loss": 2.3181,
+      "step": 31696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001401052999569173,
+      "loss": 2.3193,
+      "step": 31704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014007843798532989,
+      "loss": 2.3436,
+      "step": 31712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014005157299704557,
+      "loss": 2.3367,
+      "step": 31720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014002470499475874,
+      "loss": 2.3267,
+      "step": 31728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013999783398116415,
+      "loss": 2.3431,
+      "step": 31736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013997095995895676,
+      "loss": 2.3532,
+      "step": 31744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013994408293083193,
+      "loss": 2.3009,
+      "step": 31752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013991720289948525,
+      "loss": 2.334,
+      "step": 31760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013989031986761264,
+      "loss": 2.3417,
+      "step": 31768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013986343383791028,
+      "loss": 2.336,
+      "step": 31776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001398365448130748,
+      "loss": 2.3165,
+      "step": 31784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013980965279580286,
+      "loss": 2.3326,
+      "step": 31792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001397827577887917,
+      "loss": 2.3231,
+      "step": 31800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013975585979473872,
+      "loss": 2.3396,
+      "step": 31808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013972895881634155,
+      "loss": 2.3193,
+      "step": 31816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013970205485629825,
+      "loss": 2.3241,
+      "step": 31824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001396751479173072,
+      "loss": 2.3285,
+      "step": 31832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013964823800206698,
+      "loss": 2.3439,
+      "step": 31840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013962132511327643,
+      "loss": 2.3081,
+      "step": 31848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013959440925363482,
+      "loss": 2.3452,
+      "step": 31856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013956749042584167,
+      "loss": 2.3261,
+      "step": 31864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013954056863259677,
+      "loss": 2.323,
+      "step": 31872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013951364387660022,
+      "loss": 2.3443,
+      "step": 31880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001394867161605524,
+      "loss": 2.3251,
+      "step": 31888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001394597854871541,
+      "loss": 2.3408,
+      "step": 31896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001394328518591062,
+      "loss": 2.3175,
+      "step": 31904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013940591527911,
+      "loss": 2.3375,
+      "step": 31912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013937897574986714,
+      "loss": 2.3308,
+      "step": 31920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001393520332740795,
+      "loss": 2.3262,
+      "step": 31928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013932508785444926,
+      "loss": 2.3275,
+      "step": 31936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013929813949367884,
+      "loss": 2.3295,
+      "step": 31944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013927118819447105,
+      "loss": 2.3114,
+      "step": 31952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013924423395952893,
+      "loss": 2.3261,
+      "step": 31960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013921727679155584,
+      "loss": 2.3264,
+      "step": 31968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013919031669325547,
+      "loss": 2.3457,
+      "step": 31976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013916335366733176,
+      "loss": 2.3238,
+      "step": 31984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013913638771648892,
+      "loss": 2.3341,
+      "step": 31992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013910941884343143,
+      "loss": 2.3246,
+      "step": 32000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013908244705086422,
+      "loss": 2.3329,
+      "step": 32008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013905547234149235,
+      "loss": 2.3328,
+      "step": 32016
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013902849471802127,
+      "loss": 2.3505,
+      "step": 32024
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013900151418315662,
+      "loss": 2.3239,
+      "step": 32032
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001389745307396044,
+      "loss": 2.325,
+      "step": 32040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000138947544390071,
+      "loss": 2.3463,
+      "step": 32048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013892055513726285,
+      "loss": 2.3161,
+      "step": 32056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013889356298388693,
+      "loss": 2.3579,
+      "step": 32064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013886656793265037,
+      "loss": 2.3374,
+      "step": 32072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013883956998626059,
+      "loss": 2.3137,
+      "step": 32080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013881256914742538,
+      "loss": 2.3509,
+      "step": 32088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013878556541885274,
+      "loss": 2.3303,
+      "step": 32096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000138758558803251,
+      "loss": 2.3307,
+      "step": 32104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013873154930332872,
+      "loss": 2.3458,
+      "step": 32112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001387045369217949,
+      "loss": 2.3421,
+      "step": 32120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013867752166135867,
+      "loss": 2.3328,
+      "step": 32128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001386505035247295,
+      "loss": 2.334,
+      "step": 32136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013862348251461715,
+      "loss": 2.3206,
+      "step": 32144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013859645863373169,
+      "loss": 2.3406,
+      "step": 32152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013856943188478353,
+      "loss": 2.3326,
+      "step": 32160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001385424022704832,
+      "loss": 2.3113,
+      "step": 32168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013851536979354162,
+      "loss": 2.3309,
+      "step": 32176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013848833445667007,
+      "loss": 2.3261,
+      "step": 32184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013846129626257997,
+      "loss": 2.3318,
+      "step": 32192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013843425521398316,
+      "loss": 2.3548,
+      "step": 32200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013840721131359166,
+      "loss": 2.3158,
+      "step": 32208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001383801645641178,
+      "loss": 2.3342,
+      "step": 32216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013835311496827423,
+      "loss": 2.3466,
+      "step": 32224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013832606252877394,
+      "loss": 2.3309,
+      "step": 32232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013829900724833007,
+      "loss": 2.3539,
+      "step": 32240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001382719491296561,
+      "loss": 2.3271,
+      "step": 32248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013824488817546583,
+      "loss": 2.3247,
+      "step": 32256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013821782438847336,
+      "loss": 2.3313,
+      "step": 32264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013819075777139297,
+      "loss": 2.3321,
+      "step": 32272
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013816368832693932,
+      "loss": 2.3606,
+      "step": 32280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001381366160578273,
+      "loss": 2.3352,
+      "step": 32288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001381095409667721,
+      "loss": 2.3224,
+      "step": 32296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001380824630564893,
+      "loss": 2.3289,
+      "step": 32304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013805538232969448,
+      "loss": 2.3364,
+      "step": 32312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013802829878910384,
+      "loss": 2.3251,
+      "step": 32320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013800121243743365,
+      "loss": 2.3284,
+      "step": 32328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001379741232774005,
+      "loss": 2.3512,
+      "step": 32336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013794703131172128,
+      "loss": 2.326,
+      "step": 32344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013791993654311318,
+      "loss": 2.3519,
+      "step": 32352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001378928389742937,
+      "loss": 2.3033,
+      "step": 32360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013786573860798048,
+      "loss": 2.3303,
+      "step": 32368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013783863544689158,
+      "loss": 2.3365,
+      "step": 32376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013781152949374526,
+      "loss": 2.306,
+      "step": 32384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013778442075126018,
+      "loss": 2.3207,
+      "step": 32392
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013775730922215509,
+      "loss": 2.3471,
+      "step": 32400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013773019490914918,
+      "loss": 2.3217,
+      "step": 32408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001377030778149618,
+      "loss": 2.3427,
+      "step": 32416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013767595794231277,
+      "loss": 2.3208,
+      "step": 32424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013764883529392193,
+      "loss": 2.3332,
+      "step": 32432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013762170987250959,
+      "loss": 2.3168,
+      "step": 32440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013759458168079627,
+      "loss": 2.3292,
+      "step": 32448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013756745072150274,
+      "loss": 2.3347,
+      "step": 32456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013754031699735015,
+      "loss": 2.3264,
+      "step": 32464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001375131805110598,
+      "loss": 2.3475,
+      "step": 32472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001374860412653533,
+      "loss": 2.3212,
+      "step": 32480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013745889926295265,
+      "loss": 2.3351,
+      "step": 32488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013743175450658,
+      "loss": 2.3362,
+      "step": 32496
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013740460699895779,
+      "loss": 2.3455,
+      "step": 32504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001373774567428088,
+      "loss": 2.3302,
+      "step": 32512
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000137350303740856,
+      "loss": 2.3088,
+      "step": 32520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013732314799582275,
+      "loss": 2.3471,
+      "step": 32528
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013729598951043256,
+      "loss": 2.3461,
+      "step": 32536
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013726882828740927,
+      "loss": 2.3278,
+      "step": 32544
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013724166432947706,
+      "loss": 2.3329,
+      "step": 32552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013721449763936023,
+      "loss": 2.342,
+      "step": 32560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001371873282197835,
+      "loss": 2.3192,
+      "step": 32568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013716015607347182,
+      "loss": 2.3524,
+      "step": 32576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013713298120315037,
+      "loss": 2.3314,
+      "step": 32584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013710580361154463,
+      "loss": 2.3047,
+      "step": 32592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013707862330138046,
+      "loss": 2.3408,
+      "step": 32600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013705144027538374,
+      "loss": 2.3439,
+      "step": 32608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013702425453628085,
+      "loss": 2.3183,
+      "step": 32616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013699706608679834,
+      "loss": 2.3365,
+      "step": 32624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001369698749296631,
+      "loss": 2.3345,
+      "step": 32632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013694268106760225,
+      "loss": 2.317,
+      "step": 32640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001369154845033431,
+      "loss": 2.3474,
+      "step": 32648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001368882852396134,
+      "loss": 2.3258,
+      "step": 32656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013686108327914102,
+      "loss": 2.305,
+      "step": 32664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013683387862465423,
+      "loss": 2.3404,
+      "step": 32672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013680667127888144,
+      "loss": 2.3367,
+      "step": 32680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001367794612445514,
+      "loss": 2.3033,
+      "step": 32688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001367522485243932,
+      "loss": 2.3386,
+      "step": 32696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000136725033121136,
+      "loss": 2.3398,
+      "step": 32704
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013669781503750943,
+      "loss": 2.3317,
+      "step": 32712
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013667059427624326,
+      "loss": 2.3333,
+      "step": 32720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013664337084006765,
+      "loss": 2.3498,
+      "step": 32728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001366161447317129,
+      "loss": 2.3159,
+      "step": 32736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013658891595390966,
+      "loss": 2.3382,
+      "step": 32744
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013656168450938878,
+      "loss": 2.3327,
+      "step": 32752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013653445040088147,
+      "loss": 2.3289,
+      "step": 32760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013650721363111913,
+      "loss": 2.3335,
+      "step": 32768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013647997420283348,
+      "loss": 2.335,
+      "step": 32776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013645273211875645,
+      "loss": 2.3143,
+      "step": 32784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013642548738162027,
+      "loss": 2.3518,
+      "step": 32792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013639823999415745,
+      "loss": 2.3477,
+      "step": 32800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013637098995910075,
+      "loss": 2.3158,
+      "step": 32808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013634373727918316,
+      "loss": 2.3237,
+      "step": 32816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000136316481957138,
+      "loss": 2.3383,
+      "step": 32824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001362892239956988,
+      "loss": 2.335,
+      "step": 32832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013626196339759945,
+      "loss": 2.336,
+      "step": 32840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013623470016557393,
+      "loss": 2.3121,
+      "step": 32848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013620743430235664,
+      "loss": 2.3398,
+      "step": 32856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013618016581068223,
+      "loss": 2.3294,
+      "step": 32864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001361528946932855,
+      "loss": 2.3147,
+      "step": 32872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001361256209529016,
+      "loss": 2.3243,
+      "step": 32880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013609834459226601,
+      "loss": 2.3484,
+      "step": 32888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001360710656141143,
+      "loss": 2.3221,
+      "step": 32896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013604378402118247,
+      "loss": 2.3189,
+      "step": 32904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013601649981620663,
+      "loss": 2.3377,
+      "step": 32912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001359892130019233,
+      "loss": 2.3243,
+      "step": 32920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001359619235810692,
+      "loss": 2.3366,
+      "step": 32928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001359346315563812,
+      "loss": 2.3385,
+      "step": 32936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013590733693059668,
+      "loss": 2.3185,
+      "step": 32944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013588003970645303,
+      "loss": 2.3202,
+      "step": 32952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013585273988668802,
+      "loss": 2.3385,
+      "step": 32960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013582543747403972,
+      "loss": 2.3334,
+      "step": 32968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013579813247124636,
+      "loss": 2.3069,
+      "step": 32976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001357708248810465,
+      "loss": 2.3268,
+      "step": 32984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013574351470617895,
+      "loss": 2.3412,
+      "step": 32992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013571620194938275,
+      "loss": 2.3255,
+      "step": 33000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001356888866133972,
+      "loss": 2.3275,
+      "step": 33008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013566156870096184,
+      "loss": 2.3415,
+      "step": 33016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013563424821481658,
+      "loss": 2.316,
+      "step": 33024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001356069251577015,
+      "loss": 2.3338,
+      "step": 33032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001355795995323569,
+      "loss": 2.3173,
+      "step": 33040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001355522713415234,
+      "loss": 2.3389,
+      "step": 33048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013552494058794188,
+      "loss": 2.3337,
+      "step": 33056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013549760727435346,
+      "loss": 2.3549,
+      "step": 33064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001354702714034995,
+      "loss": 2.3263,
+      "step": 33072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013544293297812165,
+      "loss": 2.3393,
+      "step": 33080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013541559200096176,
+      "loss": 2.3174,
+      "step": 33088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013538824847476205,
+      "loss": 2.3373,
+      "step": 33096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013536090240226484,
+      "loss": 2.3121,
+      "step": 33104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001353335537862128,
+      "loss": 2.329,
+      "step": 33112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013530620262934892,
+      "loss": 2.3358,
+      "step": 33120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013527884893441632,
+      "loss": 2.3492,
+      "step": 33128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013525149270415837,
+      "loss": 2.3306,
+      "step": 33136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001352241339413188,
+      "loss": 2.3263,
+      "step": 33144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001351967726486415,
+      "loss": 2.3263,
+      "step": 33152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013516940882887067,
+      "loss": 2.3302,
+      "step": 33160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013514204248475079,
+      "loss": 2.3144,
+      "step": 33168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001351146736190265,
+      "loss": 2.3273,
+      "step": 33176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013508730223444273,
+      "loss": 2.3389,
+      "step": 33184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001350599283337447,
+      "loss": 2.3215,
+      "step": 33192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001350325519196779,
+      "loss": 2.3227,
+      "step": 33200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001350051729949879,
+      "loss": 2.3158,
+      "step": 33208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001349777915624208,
+      "loss": 2.3082,
+      "step": 33216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013495040762472277,
+      "loss": 2.3326,
+      "step": 33224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013492302118464018,
+      "loss": 2.3257,
+      "step": 33232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001348956322449198,
+      "loss": 2.3187,
+      "step": 33240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013486824080830855,
+      "loss": 2.3479,
+      "step": 33248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013484084687755368,
+      "loss": 2.3246,
+      "step": 33256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013481345045540265,
+      "loss": 2.2993,
+      "step": 33264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001347860515446031,
+      "loss": 2.3429,
+      "step": 33272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013475865014790306,
+      "loss": 2.3235,
+      "step": 33280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013473124626805065,
+      "loss": 2.3162,
+      "step": 33288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001347038399077944,
+      "loss": 2.3467,
+      "step": 33296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013467643106988302,
+      "loss": 2.346,
+      "step": 33304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001346490197570654,
+      "loss": 2.3182,
+      "step": 33312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013462160597209075,
+      "loss": 2.3337,
+      "step": 33320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013459418971770854,
+      "loss": 2.3292,
+      "step": 33328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013456677099666844,
+      "loss": 2.3165,
+      "step": 33336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001345393498117205,
+      "loss": 2.3189,
+      "step": 33344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013451192616561471,
+      "loss": 2.3316,
+      "step": 33352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001344845000611017,
+      "loss": 2.3297,
+      "step": 33360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013445707150093206,
+      "loss": 2.3286,
+      "step": 33368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001344296404878567,
+      "loss": 2.3315,
+      "step": 33376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013440220702462684,
+      "loss": 2.3369,
+      "step": 33384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013437477111399388,
+      "loss": 2.3528,
+      "step": 33392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001343473327587095,
+      "loss": 2.3434,
+      "step": 33400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001343198919615256,
+      "loss": 2.312,
+      "step": 33408
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013429244872519434,
+      "loss": 2.3376,
+      "step": 33416
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013426500305246808,
+      "loss": 2.3139,
+      "step": 33424
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001342375549460996,
+      "loss": 2.3187,
+      "step": 33432
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013421010440884163,
+      "loss": 2.3292,
+      "step": 33440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013418265144344736,
+      "loss": 2.3267,
+      "step": 33448
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001341551960526702,
+      "loss": 2.315,
+      "step": 33456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013412773823926375,
+      "loss": 2.3303,
+      "step": 33464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013410027800598185,
+      "loss": 2.3138,
+      "step": 33472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001340728153555787,
+      "loss": 2.3246,
+      "step": 33480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001340453502908085,
+      "loss": 2.3218,
+      "step": 33488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013401788281442592,
+      "loss": 2.3378,
+      "step": 33496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013399041292918578,
+      "loss": 2.3445,
+      "step": 33504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013396294063784317,
+      "loss": 2.3175,
+      "step": 33512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013393546594315343,
+      "loss": 2.3263,
+      "step": 33520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013390798884787205,
+      "loss": 2.3311,
+      "step": 33528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013388050935475487,
+      "loss": 2.2972,
+      "step": 33536
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001338530274665579,
+      "loss": 2.3509,
+      "step": 33544
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013382554318603745,
+      "loss": 2.3398,
+      "step": 33552
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013379805651594998,
+      "loss": 2.3255,
+      "step": 33560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013377056745905236,
+      "loss": 2.3228,
+      "step": 33568
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001337430760181015,
+      "loss": 2.337,
+      "step": 33576
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013371558219585464,
+      "loss": 2.3389,
+      "step": 33584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013368808599506922,
+      "loss": 2.32,
+      "step": 33592
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013366058741850302,
+      "loss": 2.3347,
+      "step": 33600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000133633086468914,
+      "loss": 2.3055,
+      "step": 33608
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013360558314906034,
+      "loss": 2.3301,
+      "step": 33616
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013357807746170036,
+      "loss": 2.3413,
+      "step": 33624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001335505694095929,
+      "loss": 2.315,
+      "step": 33632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013352305899549672,
+      "loss": 2.3321,
+      "step": 33640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013349554622217103,
+      "loss": 2.3298,
+      "step": 33648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013346803109237518,
+      "loss": 2.3372,
+      "step": 33656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001334405136088688,
+      "loss": 2.3442,
+      "step": 33664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001334129937744117,
+      "loss": 2.3291,
+      "step": 33672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000133385471591764,
+      "loss": 2.2941,
+      "step": 33680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013335794706368606,
+      "loss": 2.325,
+      "step": 33688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013333042019293833,
+      "loss": 2.3194,
+      "step": 33696
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013330289098228165,
+      "loss": 2.3252,
+      "step": 33704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013327535943447706,
+      "loss": 2.3262,
+      "step": 33712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013324782555228581,
+      "loss": 2.3263,
+      "step": 33720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013322028933846936,
+      "loss": 2.319,
+      "step": 33728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001331927507957895,
+      "loss": 2.3209,
+      "step": 33736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013316520992700813,
+      "loss": 2.3131,
+      "step": 33744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013313766673488748,
+      "loss": 2.335,
+      "step": 33752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013311012122218995,
+      "loss": 2.3272,
+      "step": 33760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001330825733916782,
+      "loss": 2.3284,
+      "step": 33768
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013305502324611516,
+      "loss": 2.3199,
+      "step": 33776
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013302747078826395,
+      "loss": 2.3329,
+      "step": 33784
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013299991602088782,
+      "loss": 2.3292,
+      "step": 33792
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001329723589467505,
+      "loss": 2.3425,
+      "step": 33800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001329447995686157,
+      "loss": 2.3359,
+      "step": 33808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013291723788924758,
+      "loss": 2.3153,
+      "step": 33816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013288967391141029,
+      "loss": 2.3149,
+      "step": 33824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013286210763786843,
+      "loss": 2.3068,
+      "step": 33832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013283453907138673,
+      "loss": 2.3255,
+      "step": 33840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013280696821473014,
+      "loss": 2.3269,
+      "step": 33848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013277939507066387,
+      "loss": 2.2989,
+      "step": 33856
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001327518196419533,
+      "loss": 2.3182,
+      "step": 33864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013272424193136424,
+      "loss": 2.3271,
+      "step": 33872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013269666194166242,
+      "loss": 2.3189,
+      "step": 33880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000132669079675614,
+      "loss": 2.3362,
+      "step": 33888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013264149513598537,
+      "loss": 2.3303,
+      "step": 33896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013261390832554305,
+      "loss": 2.3359,
+      "step": 33904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001325863192470539,
+      "loss": 2.3111,
+      "step": 33912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013255872790328486,
+      "loss": 2.3387,
+      "step": 33920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013253113429700326,
+      "loss": 2.3154,
+      "step": 33928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013250353843097655,
+      "loss": 2.3182,
+      "step": 33936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013247594030797244,
+      "loss": 2.328,
+      "step": 33944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001324483399307589,
+      "loss": 2.3183,
+      "step": 33952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013242073730210404,
+      "loss": 2.3307,
+      "step": 33960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013239313242477632,
+      "loss": 2.3219,
+      "step": 33968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013236552530154424,
+      "loss": 2.3249,
+      "step": 33976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013233791593517675,
+      "loss": 2.3335,
+      "step": 33984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001323103043284428,
+      "loss": 2.3166,
+      "step": 33992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001322826904841118,
+      "loss": 2.3181,
+      "step": 34000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013225507440495321,
+      "loss": 2.3349,
+      "step": 34008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013222745609373676,
+      "loss": 2.3305,
+      "step": 34016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013219983555323237,
+      "loss": 2.3511,
+      "step": 34024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001321722127862103,
+      "loss": 2.3142,
+      "step": 34032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013214458779544093,
+      "loss": 2.3216,
+      "step": 34040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013211696058369486,
+      "loss": 2.3294,
+      "step": 34048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013208933115374297,
+      "loss": 2.3095,
+      "step": 34056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001320616995083563,
+      "loss": 2.3049,
+      "step": 34064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013203406565030624,
+      "loss": 2.3436,
+      "step": 34072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001320064295823642,
+      "loss": 2.3446,
+      "step": 34080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000131978791307302,
+      "loss": 2.309,
+      "step": 34088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013195115082789154,
+      "loss": 2.3312,
+      "step": 34096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013192350814690505,
+      "loss": 2.3356,
+      "step": 34104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013189586326711492,
+      "loss": 2.3002,
+      "step": 34112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013186821619129378,
+      "loss": 2.3259,
+      "step": 34120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013184056692221443,
+      "loss": 2.3184,
+      "step": 34128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013181291546265003,
+      "loss": 2.321,
+      "step": 34136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013178526181537376,
+      "loss": 2.332,
+      "step": 34144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001317576059831592,
+      "loss": 2.3391,
+      "step": 34152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013172994796878002,
+      "loss": 2.3083,
+      "step": 34160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013170228777501018,
+      "loss": 2.3083,
+      "step": 34168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001316746254046239,
+      "loss": 2.3349,
+      "step": 34176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001316469608603955,
+      "loss": 2.323,
+      "step": 34184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013161929414509955,
+      "loss": 2.3084,
+      "step": 34192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001315916252615109,
+      "loss": 2.3286,
+      "step": 34200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013156395421240458,
+      "loss": 2.3202,
+      "step": 34208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013153628100055585,
+      "loss": 2.3184,
+      "step": 34216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013150860562874018,
+      "loss": 2.3427,
+      "step": 34224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013148092809973325,
+      "loss": 2.3231,
+      "step": 34232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013145324841631092,
+      "loss": 2.3042,
+      "step": 34240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013142556658124935,
+      "loss": 2.3215,
+      "step": 34248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013139788259732484,
+      "loss": 2.3244,
+      "step": 34256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013137019646731395,
+      "loss": 2.3211,
+      "step": 34264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013134250819399348,
+      "loss": 2.3224,
+      "step": 34272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013131481778014037,
+      "loss": 2.3353,
+      "step": 34280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013128712522853176,
+      "loss": 2.3068,
+      "step": 34288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013125943054194514,
+      "loss": 2.3377,
+      "step": 34296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001312317337231581,
+      "loss": 2.3393,
+      "step": 34304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001312040347749485,
+      "loss": 2.3399,
+      "step": 34312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013117633370009432,
+      "loss": 2.3153,
+      "step": 34320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013114863050137386,
+      "loss": 2.3328,
+      "step": 34328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013112092518156561,
+      "loss": 2.3037,
+      "step": 34336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013109321774344826,
+      "loss": 2.352,
+      "step": 34344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013106550818980066,
+      "loss": 2.3285,
+      "step": 34352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013103779652340198,
+      "loss": 2.3373,
+      "step": 34360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001310100827470315,
+      "loss": 2.3117,
+      "step": 34368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013098236686346878,
+      "loss": 2.3458,
+      "step": 34376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013095464887549353,
+      "loss": 2.3368,
+      "step": 34384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013092692878588578,
+      "loss": 2.3383,
+      "step": 34392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001308992065974256,
+      "loss": 2.3247,
+      "step": 34400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013087148231289347,
+      "loss": 2.3282,
+      "step": 34408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013084375593506986,
+      "loss": 2.3085,
+      "step": 34416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013081602746673572,
+      "loss": 2.3116,
+      "step": 34424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001307882969106719,
+      "loss": 2.3122,
+      "step": 34432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013076056426965973,
+      "loss": 2.3413,
+      "step": 34440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001307328295464806,
+      "loss": 2.3342,
+      "step": 34448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013070509274391615,
+      "loss": 2.3329,
+      "step": 34456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001306773538647482,
+      "loss": 2.3173,
+      "step": 34464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013064961291175882,
+      "loss": 2.336,
+      "step": 34472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001306218698877303,
+      "loss": 2.3235,
+      "step": 34480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013059412479544507,
+      "loss": 2.2992,
+      "step": 34488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001305663776376858,
+      "loss": 2.3191,
+      "step": 34496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013053862841723538,
+      "loss": 2.3185,
+      "step": 34504
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013051087713687696,
+      "loss": 2.3143,
+      "step": 34512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013048312379939373,
+      "loss": 2.3187,
+      "step": 34520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013045536840756926,
+      "loss": 2.3176,
+      "step": 34528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013042761096418727,
+      "loss": 2.3311,
+      "step": 34536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013039985147203165,
+      "loss": 2.3234,
+      "step": 34544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001303720899338865,
+      "loss": 2.344,
+      "step": 34552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013034432635253615,
+      "loss": 2.3544,
+      "step": 34560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013031656073076516,
+      "loss": 2.3232,
+      "step": 34568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013028879307135823,
+      "loss": 2.306,
+      "step": 34576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013026102337710036,
+      "loss": 2.3266,
+      "step": 34584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013023325165077663,
+      "loss": 2.3034,
+      "step": 34592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001302054778951724,
+      "loss": 2.3197,
+      "step": 34600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001301777021130732,
+      "loss": 2.3139,
+      "step": 34608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013014992430726485,
+      "loss": 2.334,
+      "step": 34616
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001301221444805333,
+      "loss": 2.3284,
+      "step": 34624
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001300943626356646,
+      "loss": 2.3146,
+      "step": 34632
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013006657877544524,
+      "loss": 2.3201,
+      "step": 34640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013003879290266173,
+      "loss": 2.3053,
+      "step": 34648
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013001100502010082,
+      "loss": 2.3236,
+      "step": 34656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001299832151305495,
+      "loss": 2.3267,
+      "step": 34664
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012995542323679493,
+      "loss": 2.3376,
+      "step": 34672
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012992762934162447,
+      "loss": 2.3303,
+      "step": 34680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012989983344782575,
+      "loss": 2.3228,
+      "step": 34688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012987203555818645,
+      "loss": 2.3266,
+      "step": 34696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001298442356754946,
+      "loss": 2.3292,
+      "step": 34704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001298164338025384,
+      "loss": 2.3109,
+      "step": 34712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012978862994210609,
+      "loss": 2.3226,
+      "step": 34720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012976082409698635,
+      "loss": 2.3199,
+      "step": 34728
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012973301626996797,
+      "loss": 2.3239,
+      "step": 34736
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012970520646383985,
+      "loss": 2.3247,
+      "step": 34744
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012967739468139116,
+      "loss": 2.3299,
+      "step": 34752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012964958092541128,
+      "loss": 2.3095,
+      "step": 34760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012962176519868977,
+      "loss": 2.3212,
+      "step": 34768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012959394750401644,
+      "loss": 2.3029,
+      "step": 34776
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012956612784418114,
+      "loss": 2.3238,
+      "step": 34784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001295383062219741,
+      "loss": 2.3168,
+      "step": 34792
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001295104826401857,
+      "loss": 2.3121,
+      "step": 34800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012948265710160644,
+      "loss": 2.3227,
+      "step": 34808
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012945482960902705,
+      "loss": 2.343,
+      "step": 34816
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012942700016523842,
+      "loss": 2.3074,
+      "step": 34824
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012939916877303183,
+      "loss": 2.3166,
+      "step": 34832
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012937133543519853,
+      "loss": 2.3095,
+      "step": 34840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012934350015453004,
+      "loss": 2.309,
+      "step": 34848
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001293156629338181,
+      "loss": 2.3113,
+      "step": 34856
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001292878237758546,
+      "loss": 2.3229,
+      "step": 34864
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012925998268343165,
+      "loss": 2.3305,
+      "step": 34872
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012923213965934159,
+      "loss": 2.315,
+      "step": 34880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001292042947063769,
+      "loss": 2.3205,
+      "step": 34888
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012917644782733026,
+      "loss": 2.33,
+      "step": 34896
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012914859902499456,
+      "loss": 2.3137,
+      "step": 34904
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012912074830216287,
+      "loss": 2.3306,
+      "step": 34912
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001290928956616285,
+      "loss": 2.3163,
+      "step": 34920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012906504110618487,
+      "loss": 2.317,
+      "step": 34928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012903718463862563,
+      "loss": 2.3251,
+      "step": 34936
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012900932626174467,
+      "loss": 2.3335,
+      "step": 34944
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012898146597833599,
+      "loss": 2.3228,
+      "step": 34952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012895360379119384,
+      "loss": 2.3115,
+      "step": 34960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012892573970311257,
+      "loss": 2.3396,
+      "step": 34968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001288978737168869,
+      "loss": 2.3162,
+      "step": 34976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012887000583531157,
+      "loss": 2.3303,
+      "step": 34984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012884213606118155,
+      "loss": 2.3148,
+      "step": 34992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012881426439729208,
+      "loss": 2.3229,
+      "step": 35000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012878639084643848,
+      "loss": 2.3267,
+      "step": 35008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012875851541141635,
+      "loss": 2.3249,
+      "step": 35016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012873063809502142,
+      "loss": 2.3268,
+      "step": 35024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012870275890004956,
+      "loss": 2.3257,
+      "step": 35032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012867487782929703,
+      "loss": 2.3137,
+      "step": 35040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012864699488556003,
+      "loss": 2.3177,
+      "step": 35048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012861911007163508,
+      "loss": 2.3147,
+      "step": 35056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012859122339031894,
+      "loss": 2.3161,
+      "step": 35064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012856333484440846,
+      "loss": 2.3207,
+      "step": 35072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012853544443670062,
+      "loss": 2.3339,
+      "step": 35080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012850755216999276,
+      "loss": 2.3327,
+      "step": 35088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012847965804708232,
+      "loss": 2.3168,
+      "step": 35096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012845176207076684,
+      "loss": 2.3467,
+      "step": 35104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012842386424384417,
+      "loss": 2.3303,
+      "step": 35112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012839596456911236,
+      "loss": 2.3164,
+      "step": 35120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001283680630493695,
+      "loss": 2.3109,
+      "step": 35128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012834015968741396,
+      "loss": 2.3139,
+      "step": 35136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001283122544860444,
+      "loss": 2.3147,
+      "step": 35144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012828434744805947,
+      "loss": 2.3319,
+      "step": 35152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001282564385762581,
+      "loss": 2.3231,
+      "step": 35160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001282285278734393,
+      "loss": 2.321,
+      "step": 35168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012820061534240252,
+      "loss": 2.3155,
+      "step": 35176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012817270098594715,
+      "loss": 2.3178,
+      "step": 35184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012814478480687288,
+      "loss": 2.3227,
+      "step": 35192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001281168668079794,
+      "loss": 2.3269,
+      "step": 35200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001280889469920669,
+      "loss": 2.3199,
+      "step": 35208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001280610253619355,
+      "loss": 2.3257,
+      "step": 35216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012803310192038562,
+      "loss": 2.3252,
+      "step": 35224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012800517667021778,
+      "loss": 2.3276,
+      "step": 35232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012797724961423275,
+      "loss": 2.3214,
+      "step": 35240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012794932075523146,
+      "loss": 2.3358,
+      "step": 35248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012792139009601498,
+      "loss": 2.3073,
+      "step": 35256
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012789345763938463,
+      "loss": 2.3399,
+      "step": 35264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012786552338814184,
+      "loss": 2.3255,
+      "step": 35272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012783758734508834,
+      "loss": 2.3135,
+      "step": 35280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001278096495130259,
+      "loss": 2.3236,
+      "step": 35288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001277817098947565,
+      "loss": 2.3276,
+      "step": 35296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012775376849308233,
+      "loss": 2.3406,
+      "step": 35304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001277258253108058,
+      "loss": 2.3086,
+      "step": 35312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012769788035072948,
+      "loss": 2.3288,
+      "step": 35320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000127669933615656,
+      "loss": 2.3275,
+      "step": 35328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001276419851083883,
+      "loss": 2.3064,
+      "step": 35336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012761403483172945,
+      "loss": 2.3304,
+      "step": 35344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012758608278848273,
+      "loss": 2.3202,
+      "step": 35352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012755812898145155,
+      "loss": 2.3175,
+      "step": 35360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012753017341343953,
+      "loss": 2.3092,
+      "step": 35368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012750221608725046,
+      "loss": 2.3127,
+      "step": 35376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012747425700568828,
+      "loss": 2.3398,
+      "step": 35384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001274462961715571,
+      "loss": 2.3146,
+      "step": 35392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001274183335876613,
+      "loss": 2.3411,
+      "step": 35400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012739036925680532,
+      "loss": 2.3546,
+      "step": 35408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001273624031817939,
+      "loss": 2.3257,
+      "step": 35416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012733443536543177,
+      "loss": 2.3158,
+      "step": 35424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012730646581052401,
+      "loss": 2.3337,
+      "step": 35432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012727849451987577,
+      "loss": 2.3402,
+      "step": 35440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012725052149629247,
+      "loss": 2.3223,
+      "step": 35448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012722254674257956,
+      "loss": 2.3288,
+      "step": 35456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001271945702615429,
+      "loss": 2.3428,
+      "step": 35464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012716659205598817,
+      "loss": 2.3239,
+      "step": 35472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012713861212872155,
+      "loss": 2.3195,
+      "step": 35480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012711063048254927,
+      "loss": 2.3424,
+      "step": 35488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001270826471202777,
+      "loss": 2.3361,
+      "step": 35496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012705466204471344,
+      "loss": 2.3074,
+      "step": 35504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001270266752586632,
+      "loss": 2.3243,
+      "step": 35512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001269986867649339,
+      "loss": 2.3278,
+      "step": 35520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012697069656633265,
+      "loss": 2.3328,
+      "step": 35528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012694270466566664,
+      "loss": 2.3323,
+      "step": 35536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001269147110657434,
+      "loss": 2.3209,
+      "step": 35544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001268867157693705,
+      "loss": 2.3379,
+      "step": 35552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012685871877935568,
+      "loss": 2.3276,
+      "step": 35560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001268307200985069,
+      "loss": 2.3087,
+      "step": 35568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012680271972963225,
+      "loss": 2.339,
+      "step": 35576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012677471767554,
+      "loss": 2.3296,
+      "step": 35584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001267467139390387,
+      "loss": 2.3006,
+      "step": 35592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001267187085229368,
+      "loss": 2.3167,
+      "step": 35600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012669070143004317,
+      "loss": 2.3297,
+      "step": 35608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012666269266316683,
+      "loss": 2.3163,
+      "step": 35616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012663468222511677,
+      "loss": 2.3302,
+      "step": 35624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012660667011870236,
+      "loss": 2.3274,
+      "step": 35632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012657865634673303,
+      "loss": 2.3218,
+      "step": 35640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001265506409120184,
+      "loss": 2.3262,
+      "step": 35648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001265226238173683,
+      "loss": 2.3393,
+      "step": 35656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012649460506559263,
+      "loss": 2.3246,
+      "step": 35664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012646658465950152,
+      "loss": 2.3288,
+      "step": 35672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012643856260190531,
+      "loss": 2.326,
+      "step": 35680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012641053889561442,
+      "loss": 2.3293,
+      "step": 35688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001263825135434394,
+      "loss": 2.3186,
+      "step": 35696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012635448654819118,
+      "loss": 2.331,
+      "step": 35704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012632645791268058,
+      "loss": 2.3387,
+      "step": 35712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012629842763971878,
+      "loss": 2.3084,
+      "step": 35720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012627039573211705,
+      "loss": 2.3067,
+      "step": 35728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001262423621926868,
+      "loss": 2.3111,
+      "step": 35736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012621432702423965,
+      "loss": 2.3362,
+      "step": 35744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012618629022958738,
+      "loss": 2.3327,
+      "step": 35752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012615825181154196,
+      "loss": 2.3232,
+      "step": 35760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012613021177291538,
+      "loss": 2.3142,
+      "step": 35768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012610217011652,
+      "loss": 2.3136,
+      "step": 35776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012607412684516818,
+      "loss": 2.3051,
+      "step": 35784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012604608196167252,
+      "loss": 2.3207,
+      "step": 35792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012601803546884575,
+      "loss": 2.3219,
+      "step": 35800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012598998736950082,
+      "loss": 2.3193,
+      "step": 35808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012596193766645076,
+      "loss": 2.3212,
+      "step": 35816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012593388636250873,
+      "loss": 2.3263,
+      "step": 35824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012590583346048826,
+      "loss": 2.3116,
+      "step": 35832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012587777896320277,
+      "loss": 2.2988,
+      "step": 35840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000125849722873466,
+      "loss": 2.3228,
+      "step": 35848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012582166519409192,
+      "loss": 2.3214,
+      "step": 35856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012579360592789441,
+      "loss": 2.3232,
+      "step": 35864
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012576554507768775,
+      "loss": 2.3058,
+      "step": 35872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012573748264628622,
+      "loss": 2.3431,
+      "step": 35880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001257094186365044,
+      "loss": 2.3248,
+      "step": 35888
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012568135305115685,
+      "loss": 2.3134,
+      "step": 35896
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012565328589305848,
+      "loss": 2.2908,
+      "step": 35904
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012562521716502422,
+      "loss": 2.312,
+      "step": 35912
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001255971468698692,
+      "loss": 2.2978,
+      "step": 35920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012556907501040876,
+      "loss": 2.3208,
+      "step": 35928
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012554100158945828,
+      "loss": 2.3214,
+      "step": 35936
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012551292660983342,
+      "loss": 2.2977,
+      "step": 35944
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012548485007434995,
+      "loss": 2.3021,
+      "step": 35952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001254567719858237,
+      "loss": 2.3519,
+      "step": 35960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012542869234707083,
+      "loss": 2.33,
+      "step": 35968
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012540061116090754,
+      "loss": 2.3284,
+      "step": 35976
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012537252843015023,
+      "loss": 2.3012,
+      "step": 35984
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012534444415761544,
+      "loss": 2.3089,
+      "step": 35992
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001253163583461198,
+      "loss": 2.3059,
+      "step": 36000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001252882709984802,
+      "loss": 2.3114,
+      "step": 36008
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001252601821175137,
+      "loss": 2.3095,
+      "step": 36016
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012523209170603737,
+      "loss": 2.3257,
+      "step": 36024
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012520399976686858,
+      "loss": 2.3157,
+      "step": 36032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012517590630282472,
+      "loss": 2.323,
+      "step": 36040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012514781131672348,
+      "loss": 2.3198,
+      "step": 36048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001251197148113826,
+      "loss": 2.3166,
+      "step": 36056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012509161678961995,
+      "loss": 2.2942,
+      "step": 36064
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012506351725425372,
+      "loss": 2.2966,
+      "step": 36072
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000125035416208102,
+      "loss": 2.3076,
+      "step": 36080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012500731365398327,
+      "loss": 2.333,
+      "step": 36088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000124979209594716,
+      "loss": 2.3147,
+      "step": 36096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012495110403311892,
+      "loss": 2.3291,
+      "step": 36104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012492299697201077,
+      "loss": 2.3327,
+      "step": 36112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012489488841421062,
+      "loss": 2.3247,
+      "step": 36120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012486677836253752,
+      "loss": 2.3318,
+      "step": 36128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012483866681981085,
+      "loss": 2.3198,
+      "step": 36136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012481055378884992,
+      "loss": 2.3276,
+      "step": 36144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001247824392724744,
+      "loss": 2.321,
+      "step": 36152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012475432327350395,
+      "loss": 2.3244,
+      "step": 36160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012472620579475857,
+      "loss": 2.3513,
+      "step": 36168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012469808683905807,
+      "loss": 2.3104,
+      "step": 36176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012466996640922282,
+      "loss": 2.3317,
+      "step": 36184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012464184450807303,
+      "loss": 2.3242,
+      "step": 36192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012461372113842923,
+      "loss": 2.3207,
+      "step": 36200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012458559630311198,
+      "loss": 2.33,
+      "step": 36208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012455747000494208,
+      "loss": 2.313,
+      "step": 36216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012452934224674043,
+      "loss": 2.3112,
+      "step": 36224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012450121303132807,
+      "loss": 2.3176,
+      "step": 36232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001244730823615262,
+      "loss": 2.3316,
+      "step": 36240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012444495024015618,
+      "loss": 2.3208,
+      "step": 36248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012441681667003953,
+      "loss": 2.3303,
+      "step": 36256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012438868165399783,
+      "loss": 2.2997,
+      "step": 36264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012436054519485292,
+      "loss": 2.2914,
+      "step": 36272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012433240729542666,
+      "loss": 2.3033,
+      "step": 36280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012430426795854117,
+      "loss": 2.3171,
+      "step": 36288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012427612718701872,
+      "loss": 2.3284,
+      "step": 36296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012424798498368152,
+      "loss": 2.327,
+      "step": 36304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012421984135135218,
+      "loss": 2.3266,
+      "step": 36312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012419169629285335,
+      "loss": 2.3162,
+      "step": 36320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001241635498110078,
+      "loss": 2.3251,
+      "step": 36328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012413540190863844,
+      "loss": 2.332,
+      "step": 36336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012410725258856837,
+      "loss": 2.317,
+      "step": 36344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012407910185362078,
+      "loss": 2.312,
+      "step": 36352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012405094970661907,
+      "loss": 2.3103,
+      "step": 36360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012402279615038675,
+      "loss": 2.328,
+      "step": 36368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001239946411877474,
+      "loss": 2.3279,
+      "step": 36376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012396648482152484,
+      "loss": 2.3065,
+      "step": 36384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000123938327054543,
+      "loss": 2.3227,
+      "step": 36392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012391016788962592,
+      "loss": 2.3109,
+      "step": 36400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012388200732959783,
+      "loss": 2.3046,
+      "step": 36408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012385384537728304,
+      "loss": 2.3268,
+      "step": 36416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012382568203550608,
+      "loss": 2.3244,
+      "step": 36424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001237975173070916,
+      "loss": 2.3162,
+      "step": 36432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012376935119486426,
+      "loss": 2.3322,
+      "step": 36440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000123741183701649,
+      "loss": 2.3115,
+      "step": 36448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012371301483027092,
+      "loss": 2.3171,
+      "step": 36456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012368484458355513,
+      "loss": 2.3138,
+      "step": 36464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012365667296432702,
+      "loss": 2.3149,
+      "step": 36472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001236284999754119,
+      "loss": 2.3253,
+      "step": 36480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001236003256196355,
+      "loss": 2.3028,
+      "step": 36488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012357214989982355,
+      "loss": 2.3456,
+      "step": 36496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012354397281880183,
+      "loss": 2.3143,
+      "step": 36504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012351579437939634,
+      "loss": 2.3296,
+      "step": 36512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012348761458443334,
+      "loss": 2.3229,
+      "step": 36520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012345943343673895,
+      "loss": 2.3292,
+      "step": 36528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001234312509391397,
+      "loss": 2.316,
+      "step": 36536
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012340306709446208,
+      "loss": 2.2996,
+      "step": 36544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012337488190553275,
+      "loss": 2.3416,
+      "step": 36552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001233466953751786,
+      "loss": 2.3318,
+      "step": 36560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001233185075062265,
+      "loss": 2.3093,
+      "step": 36568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012329031830150353,
+      "loss": 2.3301,
+      "step": 36576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012326212776383699,
+      "loss": 2.3024,
+      "step": 36584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012323393589605417,
+      "loss": 2.3384,
+      "step": 36592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012320574270098254,
+      "loss": 2.3186,
+      "step": 36600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001231775481814498,
+      "loss": 2.3222,
+      "step": 36608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001231493523402836,
+      "loss": 2.3406,
+      "step": 36616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001231211551803119,
+      "loss": 2.3087,
+      "step": 36624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012309295670436264,
+      "loss": 2.3262,
+      "step": 36632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012306475691526407,
+      "loss": 2.3166,
+      "step": 36640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001230365558158444,
+      "loss": 2.3204,
+      "step": 36648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012300835340893203,
+      "loss": 2.3239,
+      "step": 36656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001229801496973555,
+      "loss": 2.3336,
+      "step": 36664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012295194468394353,
+      "loss": 2.3312,
+      "step": 36672
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012292373837152485,
+      "loss": 2.3209,
+      "step": 36680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012289553076292852,
+      "loss": 2.3186,
+      "step": 36688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001228673218609835,
+      "loss": 2.311,
+      "step": 36696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012283911166851902,
+      "loss": 2.3076,
+      "step": 36704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012281090018836433,
+      "loss": 2.3019,
+      "step": 36712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000122782687423349,
+      "loss": 2.3257,
+      "step": 36720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012275447337630252,
+      "loss": 2.347,
+      "step": 36728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012272625805005466,
+      "loss": 2.3258,
+      "step": 36736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001226980414474352,
+      "loss": 2.3135,
+      "step": 36744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012266982357127414,
+      "loss": 2.3088,
+      "step": 36752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001226416044244016,
+      "loss": 2.3362,
+      "step": 36760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012261338400964774,
+      "loss": 2.3336,
+      "step": 36768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001225851623298429,
+      "loss": 2.3109,
+      "step": 36776
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012255693938781768,
+      "loss": 2.3124,
+      "step": 36784
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001225287151864025,
+      "loss": 2.3143,
+      "step": 36792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012250048972842822,
+      "loss": 2.3266,
+      "step": 36800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012247226301672566,
+      "loss": 2.3098,
+      "step": 36808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012244403505412576,
+      "loss": 2.2957,
+      "step": 36816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001224158058434597,
+      "loss": 2.3267,
+      "step": 36824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012238757538755862,
+      "loss": 2.3403,
+      "step": 36832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012235934368925395,
+      "loss": 2.3172,
+      "step": 36840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012233111075137714,
+      "loss": 2.3234,
+      "step": 36848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012230287657675977,
+      "loss": 2.311,
+      "step": 36856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001222746411682336,
+      "loss": 2.3063,
+      "step": 36864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001222464045286305,
+      "loss": 2.3026,
+      "step": 36872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001222181666607824,
+      "loss": 2.3283,
+      "step": 36880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012218992756752139,
+      "loss": 2.3204,
+      "step": 36888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012216168725167974,
+      "loss": 2.3205,
+      "step": 36896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001221334457160898,
+      "loss": 2.304,
+      "step": 36904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012210520296358394,
+      "loss": 2.293,
+      "step": 36912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012207695899699487,
+      "loss": 2.3304,
+      "step": 36920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012204871381915521,
+      "loss": 2.3389,
+      "step": 36928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012202046743289785,
+      "loss": 2.3319,
+      "step": 36936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012199221984105571,
+      "loss": 2.311,
+      "step": 36944
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012196397104646187,
+      "loss": 2.337,
+      "step": 36952
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012193572105194953,
+      "loss": 2.3229,
+      "step": 36960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000121907469860352,
+      "loss": 2.3131,
+      "step": 36968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012187921747450271,
+      "loss": 2.3223,
+      "step": 36976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001218509638972352,
+      "loss": 2.3264,
+      "step": 36984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012182270913138319,
+      "loss": 2.3202,
+      "step": 36992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012179445317978044,
+      "loss": 2.3259,
+      "step": 37000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012176619604526088,
+      "loss": 2.3078,
+      "step": 37008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001217379377306585,
+      "loss": 2.3269,
+      "step": 37016
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001217096782388075,
+      "loss": 2.3322,
+      "step": 37024
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001216814175725421,
+      "loss": 2.3154,
+      "step": 37032
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012165315573469671,
+      "loss": 2.3052,
+      "step": 37040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012162489272810586,
+      "loss": 2.309,
+      "step": 37048
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001215966285556041,
+      "loss": 2.3113,
+      "step": 37056
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012156836322002623,
+      "loss": 2.3136,
+      "step": 37064
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012154009672420707,
+      "loss": 2.3161,
+      "step": 37072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012151182907098162,
+      "loss": 2.3354,
+      "step": 37080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012148356026318493,
+      "loss": 2.3048,
+      "step": 37088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012145529030365222,
+      "loss": 2.3229,
+      "step": 37096
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012142701919521877,
+      "loss": 2.3097,
+      "step": 37104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012139874694072007,
+      "loss": 2.3118,
+      "step": 37112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012137047354299163,
+      "loss": 2.3223,
+      "step": 37120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012134219900486913,
+      "loss": 2.3229,
+      "step": 37128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012131392332918836,
+      "loss": 2.3164,
+      "step": 37136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012128564651878515,
+      "loss": 2.323,
+      "step": 37144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012125736857649554,
+      "loss": 2.3151,
+      "step": 37152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012122908950515568,
+      "loss": 2.3147,
+      "step": 37160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012120080930760179,
+      "loss": 2.3124,
+      "step": 37168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012117252798667016,
+      "loss": 2.3009,
+      "step": 37176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012114424554519726,
+      "loss": 2.3105,
+      "step": 37184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012111596198601971,
+      "loss": 2.2959,
+      "step": 37192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012108767731197414,
+      "loss": 2.3256,
+      "step": 37200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012105939152589738,
+      "loss": 2.3349,
+      "step": 37208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012103110463062631,
+      "loss": 2.2895,
+      "step": 37216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012100281662899795,
+      "loss": 2.2992,
+      "step": 37224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012097452752384942,
+      "loss": 2.3281,
+      "step": 37232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012094623731801797,
+      "loss": 2.3133,
+      "step": 37240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012091794601434094,
+      "loss": 2.3121,
+      "step": 37248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012088965361565578,
+      "loss": 2.332,
+      "step": 37256
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012086136012480012,
+      "loss": 2.3212,
+      "step": 37264
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012083306554461154,
+      "loss": 2.3227,
+      "step": 37272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012080476987792787,
+      "loss": 2.2924,
+      "step": 37280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012077647312758698,
+      "loss": 2.3191,
+      "step": 37288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012074817529642693,
+      "loss": 2.3228,
+      "step": 37296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012071987638728579,
+      "loss": 2.3182,
+      "step": 37304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012069157640300181,
+      "loss": 2.3124,
+      "step": 37312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012066327534641327,
+      "loss": 2.322,
+      "step": 37320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012063497322035864,
+      "loss": 2.3227,
+      "step": 37328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012060667002767649,
+      "loss": 2.3065,
+      "step": 37336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012057836577120541,
+      "loss": 2.3055,
+      "step": 37344
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012055006045378417,
+      "loss": 2.3391,
+      "step": 37352
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012052175407825166,
+      "loss": 2.3318,
+      "step": 37360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012049344664744684,
+      "loss": 2.3272,
+      "step": 37368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012046513816420879,
+      "loss": 2.3359,
+      "step": 37376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012043682863137665,
+      "loss": 2.322,
+      "step": 37384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012040851805178974,
+      "loss": 2.328,
+      "step": 37392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012038020642828749,
+      "loss": 2.3279,
+      "step": 37400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012035189376370932,
+      "loss": 2.3141,
+      "step": 37408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012032358006089486,
+      "loss": 2.3221,
+      "step": 37416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012029526532268383,
+      "loss": 2.3143,
+      "step": 37424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012026694955191602,
+      "loss": 2.2934,
+      "step": 37432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012023863275143137,
+      "loss": 2.3392,
+      "step": 37440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001202103149240698,
+      "loss": 2.3168,
+      "step": 37448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012018199607267155,
+      "loss": 2.3223,
+      "step": 37456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012015367620007678,
+      "loss": 2.3083,
+      "step": 37464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012012535530912584,
+      "loss": 2.3219,
+      "step": 37472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001200970334026591,
+      "loss": 2.3158,
+      "step": 37480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012006871048351716,
+      "loss": 2.3126,
+      "step": 37488
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001200403865545406,
+      "loss": 2.3439,
+      "step": 37496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012001206161857013,
+      "loss": 2.311,
+      "step": 37504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011998373567844665,
+      "loss": 2.3219,
+      "step": 37512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000119955408737011,
+      "loss": 2.3139,
+      "step": 37520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011992708079710435,
+      "loss": 2.3104,
+      "step": 37528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001198987518615677,
+      "loss": 2.3077,
+      "step": 37536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011987042193324232,
+      "loss": 2.3353,
+      "step": 37544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011984209101496952,
+      "loss": 2.3075,
+      "step": 37552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011981375910959078,
+      "loss": 2.3164,
+      "step": 37560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011978542621994763,
+      "loss": 2.326,
+      "step": 37568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001197570923488817,
+      "loss": 2.3389,
+      "step": 37576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011972875749923462,
+      "loss": 2.3217,
+      "step": 37584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011970042167384834,
+      "loss": 2.323,
+      "step": 37592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011967208487556474,
+      "loss": 2.3195,
+      "step": 37600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011964374710722581,
+      "loss": 2.295,
+      "step": 37608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011961540837167372,
+      "loss": 2.3128,
+      "step": 37616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001195870686717506,
+      "loss": 2.3232,
+      "step": 37624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011955872801029886,
+      "loss": 2.3186,
+      "step": 37632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011953038639016084,
+      "loss": 2.3133,
+      "step": 37640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011950204381417906,
+      "loss": 2.3324,
+      "step": 37648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011947370028519615,
+      "loss": 2.3105,
+      "step": 37656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001194453558060548,
+      "loss": 2.3042,
+      "step": 37664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011941701037959776,
+      "loss": 2.3289,
+      "step": 37672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011938866400866793,
+      "loss": 2.3219,
+      "step": 37680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011936031669610827,
+      "loss": 2.2825,
+      "step": 37688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011933196844476194,
+      "loss": 2.3304,
+      "step": 37696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011930361925747205,
+      "loss": 2.3094,
+      "step": 37704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011927526913708184,
+      "loss": 2.3282,
+      "step": 37712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011924691808643469,
+      "loss": 2.3044,
+      "step": 37720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011921856610837407,
+      "loss": 2.328,
+      "step": 37728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011919021320574348,
+      "loss": 2.3204,
+      "step": 37736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001191618593813866,
+      "loss": 2.319,
+      "step": 37744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011913350463814715,
+      "loss": 2.3122,
+      "step": 37752
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001191051489788689,
+      "loss": 2.3164,
+      "step": 37760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011907679240639584,
+      "loss": 2.3157,
+      "step": 37768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011904843492357192,
+      "loss": 2.314,
+      "step": 37776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011902007653324123,
+      "loss": 2.3011,
+      "step": 37784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000118991717238248,
+      "loss": 2.3152,
+      "step": 37792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011896335704143646,
+      "loss": 2.3033,
+      "step": 37800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000118934995945651,
+      "loss": 2.3245,
+      "step": 37808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011890663395373611,
+      "loss": 2.2987,
+      "step": 37816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011887827106853624,
+      "loss": 2.3125,
+      "step": 37824
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011884990729289611,
+      "loss": 2.3229,
+      "step": 37832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001188215426296605,
+      "loss": 2.3146,
+      "step": 37840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011879317708167406,
+      "loss": 2.3037,
+      "step": 37848
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011876481065178182,
+      "loss": 2.3309,
+      "step": 37856
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011873644334282871,
+      "loss": 2.3224,
+      "step": 37864
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011870807515765984,
+      "loss": 2.3274,
+      "step": 37872
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011867970609912042,
+      "loss": 2.3,
+      "step": 37880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011865133617005562,
+      "loss": 2.3231,
+      "step": 37888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011862296537331084,
+      "loss": 2.3028,
+      "step": 37896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011859459371173147,
+      "loss": 2.3247,
+      "step": 37904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011856622118816307,
+      "loss": 2.328,
+      "step": 37912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011853784780545123,
+      "loss": 2.3142,
+      "step": 37920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011850947356644162,
+      "loss": 2.3235,
+      "step": 37928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011848109847398003,
+      "loss": 2.3164,
+      "step": 37936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011845272253091234,
+      "loss": 2.3243,
+      "step": 37944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011842434574008445,
+      "loss": 2.3184,
+      "step": 37952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011839596810434239,
+      "loss": 2.3332,
+      "step": 37960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011836758962653236,
+      "loss": 2.3311,
+      "step": 37968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011833921030950048,
+      "loss": 2.283,
+      "step": 37976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011831083015609306,
+      "loss": 2.305,
+      "step": 37984
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011828244916915645,
+      "loss": 2.3127,
+      "step": 37992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011825406735153712,
+      "loss": 2.3114,
+      "step": 38000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011822568470608164,
+      "loss": 2.2898,
+      "step": 38008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001181973012356366,
+      "loss": 2.294,
+      "step": 38016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011816891694304865,
+      "loss": 2.3258,
+      "step": 38024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011814053183116463,
+      "loss": 2.3146,
+      "step": 38032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011811214590283141,
+      "loss": 2.3118,
+      "step": 38040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011808375916089592,
+      "loss": 2.319,
+      "step": 38048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011805537160820517,
+      "loss": 2.3343,
+      "step": 38056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001180269832476063,
+      "loss": 2.3238,
+      "step": 38064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011799859408194651,
+      "loss": 2.3064,
+      "step": 38072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011797020411407305,
+      "loss": 2.2943,
+      "step": 38080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011794181334683327,
+      "loss": 2.3138,
+      "step": 38088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001179134217830746,
+      "loss": 2.3178,
+      "step": 38096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001178850294256446,
+      "loss": 2.3185,
+      "step": 38104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001178566362773908,
+      "loss": 2.3068,
+      "step": 38112
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011782824234116092,
+      "loss": 2.3225,
+      "step": 38120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011779984761980268,
+      "loss": 2.3259,
+      "step": 38128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011777145211616392,
+      "loss": 2.3123,
+      "step": 38136
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011774305583309257,
+      "loss": 2.3154,
+      "step": 38144
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011771465877343658,
+      "loss": 2.3124,
+      "step": 38152
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000117686260940044,
+      "loss": 2.3111,
+      "step": 38160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011765786233576302,
+      "loss": 2.2979,
+      "step": 38168
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011762946296344184,
+      "loss": 2.3212,
+      "step": 38176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011760106282592876,
+      "loss": 2.333,
+      "step": 38184
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011757266192607212,
+      "loss": 2.2995,
+      "step": 38192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011754426026672045,
+      "loss": 2.3076,
+      "step": 38200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011751585785072218,
+      "loss": 2.3159,
+      "step": 38208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011748745468092599,
+      "loss": 2.3275,
+      "step": 38216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011745905076018048,
+      "loss": 2.3247,
+      "step": 38224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011743064609133448,
+      "loss": 2.3027,
+      "step": 38232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011740224067723678,
+      "loss": 2.3414,
+      "step": 38240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011737383452073628,
+      "loss": 2.3372,
+      "step": 38248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011734542762468193,
+      "loss": 2.3117,
+      "step": 38256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011731701999192282,
+      "loss": 2.3179,
+      "step": 38264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011728861162530809,
+      "loss": 2.3148,
+      "step": 38272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011726020252768691,
+      "loss": 2.3098,
+      "step": 38280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011723179270190857,
+      "loss": 2.3142,
+      "step": 38288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011720338215082234,
+      "loss": 2.3194,
+      "step": 38296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011717497087727773,
+      "loss": 2.3348,
+      "step": 38304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011714655888412421,
+      "loss": 2.3006,
+      "step": 38312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011711814617421132,
+      "loss": 2.3151,
+      "step": 38320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001170897327503887,
+      "loss": 2.3295,
+      "step": 38328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011706131861550607,
+      "loss": 2.3098,
+      "step": 38336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011703290377241318,
+      "loss": 2.3393,
+      "step": 38344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011700448822395988,
+      "loss": 2.3019,
+      "step": 38352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011697607197299611,
+      "loss": 2.2937,
+      "step": 38360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011694765502237188,
+      "loss": 2.3271,
+      "step": 38368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001169192373749372,
+      "loss": 2.327,
+      "step": 38376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011689081903354221,
+      "loss": 2.3241,
+      "step": 38384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001168624000010371,
+      "loss": 2.3222,
+      "step": 38392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011683398028027217,
+      "loss": 2.323,
+      "step": 38400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011680555987409776,
+      "loss": 2.315,
+      "step": 38408
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011677713878536425,
+      "loss": 2.3028,
+      "step": 38416
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011674871701692209,
+      "loss": 2.3153,
+      "step": 38424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011672029457162188,
+      "loss": 2.3182,
+      "step": 38432
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011669187145231421,
+      "loss": 2.3148,
+      "step": 38440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011666344766184973,
+      "loss": 2.303,
+      "step": 38448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011663502320307923,
+      "loss": 2.3174,
+      "step": 38456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011660659807885348,
+      "loss": 2.331,
+      "step": 38464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011657817229202339,
+      "loss": 2.3161,
+      "step": 38472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011654974584543989,
+      "loss": 2.3191,
+      "step": 38480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000116521318741954,
+      "loss": 2.3164,
+      "step": 38488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011649289098441678,
+      "loss": 2.3036,
+      "step": 38496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001164644625756794,
+      "loss": 2.3209,
+      "step": 38504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011643603351859302,
+      "loss": 2.3136,
+      "step": 38512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011640760381600899,
+      "loss": 2.3291,
+      "step": 38520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011637917347077856,
+      "loss": 2.3013,
+      "step": 38528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001163507424857532,
+      "loss": 2.2992,
+      "step": 38536
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011632231086378438,
+      "loss": 2.3215,
+      "step": 38544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011629387860772358,
+      "loss": 2.3334,
+      "step": 38552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011626544572042239,
+      "loss": 2.3236,
+      "step": 38560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011623701220473255,
+      "loss": 2.32,
+      "step": 38568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001162085780635057,
+      "loss": 2.3021,
+      "step": 38576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011618014329959366,
+      "loss": 2.3234,
+      "step": 38584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011615170791584824,
+      "loss": 2.2908,
+      "step": 38592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001161232719151214,
+      "loss": 2.3176,
+      "step": 38600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011609483530026508,
+      "loss": 2.3169,
+      "step": 38608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011606639807413129,
+      "loss": 2.3297,
+      "step": 38616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011603796023957218,
+      "loss": 2.3018,
+      "step": 38624
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011600952179943987,
+      "loss": 2.2942,
+      "step": 38632
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011598108275658655,
+      "loss": 2.3208,
+      "step": 38640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011595264311386453,
+      "loss": 2.3259,
+      "step": 38648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011592420287412616,
+      "loss": 2.3124,
+      "step": 38656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011589576204022377,
+      "loss": 2.3131,
+      "step": 38664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011586732061500986,
+      "loss": 2.325,
+      "step": 38672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011583887860133697,
+      "loss": 2.3,
+      "step": 38680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011581043600205761,
+      "loss": 2.295,
+      "step": 38688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011578199282002447,
+      "loss": 2.3361,
+      "step": 38696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011575354905809017,
+      "loss": 2.2889,
+      "step": 38704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011572510471910755,
+      "loss": 2.3189,
+      "step": 38712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011569665980592936,
+      "loss": 2.2983,
+      "step": 38720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011566821432140844,
+      "loss": 2.2947,
+      "step": 38728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011563976826839775,
+      "loss": 2.3158,
+      "step": 38736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011561132164975028,
+      "loss": 2.3284,
+      "step": 38744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011558287446831901,
+      "loss": 2.3124,
+      "step": 38752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011555442672695711,
+      "loss": 2.3056,
+      "step": 38760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011552597842851767,
+      "loss": 2.3164,
+      "step": 38768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011549752957585387,
+      "loss": 2.3147,
+      "step": 38776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011546908017181906,
+      "loss": 2.308,
+      "step": 38784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011544063021926645,
+      "loss": 2.3277,
+      "step": 38792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011541217972104946,
+      "loss": 2.3112,
+      "step": 38800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011538372868002154,
+      "loss": 2.2954,
+      "step": 38808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011535527709903611,
+      "loss": 2.3225,
+      "step": 38816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011532682498094676,
+      "loss": 2.315,
+      "step": 38824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011529837232860701,
+      "loss": 2.3046,
+      "step": 38832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011526991914487058,
+      "loss": 2.3362,
+      "step": 38840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011524146543259111,
+      "loss": 2.3121,
+      "step": 38848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011521301119462236,
+      "loss": 2.3073,
+      "step": 38856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011518455643381809,
+      "loss": 2.3177,
+      "step": 38864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011515610115303222,
+      "loss": 2.3036,
+      "step": 38872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011512764535511861,
+      "loss": 2.3274,
+      "step": 38880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011509918904293122,
+      "loss": 2.3093,
+      "step": 38888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011507073221932407,
+      "loss": 2.3014,
+      "step": 38896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011504227488715122,
+      "loss": 2.3243,
+      "step": 38904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011501381704926678,
+      "loss": 2.3211,
+      "step": 38912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011498535870852485,
+      "loss": 2.3233,
+      "step": 38920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011495689986777973,
+      "loss": 2.3115,
+      "step": 38928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011492844052988565,
+      "loss": 2.319,
+      "step": 38936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011489998069769688,
+      "loss": 2.3161,
+      "step": 38944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011487152037406783,
+      "loss": 2.3085,
+      "step": 38952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011484305956185288,
+      "loss": 2.3102,
+      "step": 38960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011481459826390648,
+      "loss": 2.3186,
+      "step": 38968
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011478613648308319,
+      "loss": 2.3204,
+      "step": 38976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011475767422223753,
+      "loss": 2.3064,
+      "step": 38984
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011472921148422411,
+      "loss": 2.3127,
+      "step": 38992
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011470074827189752,
+      "loss": 2.3052,
+      "step": 39000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011467228458811257,
+      "loss": 2.3219,
+      "step": 39008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011464382043572392,
+      "loss": 2.3244,
+      "step": 39016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011461535581758643,
+      "loss": 2.3035,
+      "step": 39024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011458689073655484,
+      "loss": 2.3157,
+      "step": 39032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011455842519548416,
+      "loss": 2.3044,
+      "step": 39040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011452995919722924,
+      "loss": 2.3036,
+      "step": 39048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011450149274464507,
+      "loss": 2.3129,
+      "step": 39056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001144730258405867,
+      "loss": 2.3054,
+      "step": 39064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011444455848790918,
+      "loss": 2.3219,
+      "step": 39072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011441609068946764,
+      "loss": 2.3145,
+      "step": 39080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011438762244811722,
+      "loss": 2.3103,
+      "step": 39088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011435915376671313,
+      "loss": 2.3282,
+      "step": 39096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011433068464811058,
+      "loss": 2.3122,
+      "step": 39104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011430221509516495,
+      "loss": 2.344,
+      "step": 39112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011427374511073152,
+      "loss": 2.3051,
+      "step": 39120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011424527469766566,
+      "loss": 2.324,
+      "step": 39128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011421680385882279,
+      "loss": 2.3402,
+      "step": 39136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011418833259705839,
+      "loss": 2.3191,
+      "step": 39144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011415986091522797,
+      "loss": 2.29,
+      "step": 39152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011413138881618707,
+      "loss": 2.306,
+      "step": 39160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011410291630279127,
+      "loss": 2.3219,
+      "step": 39168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011407444337789624,
+      "loss": 2.3063,
+      "step": 39176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011404597004435762,
+      "loss": 2.3045,
+      "step": 39184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011401749630503109,
+      "loss": 2.2954,
+      "step": 39192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001139890221627725,
+      "loss": 2.3349,
+      "step": 39200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011396054762043758,
+      "loss": 2.3137,
+      "step": 39208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011393207268088215,
+      "loss": 2.3245,
+      "step": 39216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011390359734696212,
+      "loss": 2.2768,
+      "step": 39224
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011387512162153338,
+      "loss": 2.3144,
+      "step": 39232
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011384664550745193,
+      "loss": 2.318,
+      "step": 39240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011381816900757371,
+      "loss": 2.3161,
+      "step": 39248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011378969212475481,
+      "loss": 2.302,
+      "step": 39256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011376121486185121,
+      "loss": 2.3046,
+      "step": 39264
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011373273722171908,
+      "loss": 2.3222,
+      "step": 39272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011370425920721459,
+      "loss": 2.3058,
+      "step": 39280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011367578082119387,
+      "loss": 2.3222,
+      "step": 39288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011364730206651318,
+      "loss": 2.315,
+      "step": 39296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011361882294602873,
+      "loss": 2.3385,
+      "step": 39304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011359034346259687,
+      "loss": 2.3115,
+      "step": 39312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011356186361907388,
+      "loss": 2.3016,
+      "step": 39320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011353338341831619,
+      "loss": 2.3079,
+      "step": 39328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011350490286318014,
+      "loss": 2.3184,
+      "step": 39336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001134764219565222,
+      "loss": 2.3163,
+      "step": 39344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011344794070119882,
+      "loss": 2.3195,
+      "step": 39352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011341945910006655,
+      "loss": 2.3059,
+      "step": 39360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011339097715598186,
+      "loss": 2.3092,
+      "step": 39368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011336249487180145,
+      "loss": 2.311,
+      "step": 39376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001133340122503818,
+      "loss": 2.293,
+      "step": 39384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011330552929457966,
+      "loss": 2.3154,
+      "step": 39392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011327704600725158,
+      "loss": 2.3334,
+      "step": 39400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011324856239125441,
+      "loss": 2.3094,
+      "step": 39408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011322007844944485,
+      "loss": 2.3285,
+      "step": 39416
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011319159418467966,
+      "loss": 2.3027,
+      "step": 39424
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011316310959981562,
+      "loss": 2.337,
+      "step": 39432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011313462469770966,
+      "loss": 2.3262,
+      "step": 39440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011310613948121859,
+      "loss": 2.3312,
+      "step": 39448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011307765395319932,
+      "loss": 2.2959,
+      "step": 39456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011304916811650879,
+      "loss": 2.3287,
+      "step": 39464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000113020681974004,
+      "loss": 2.3087,
+      "step": 39472
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011299219552854192,
+      "loss": 2.3186,
+      "step": 39480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011296370878297957,
+      "loss": 2.2957,
+      "step": 39488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011293522174017404,
+      "loss": 2.3167,
+      "step": 39496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011290673440298236,
+      "loss": 2.3186,
+      "step": 39504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011287824677426175,
+      "loss": 2.3165,
+      "step": 39512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011284975885686926,
+      "loss": 2.2917,
+      "step": 39520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001128212706536621,
+      "loss": 2.3281,
+      "step": 39528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011279278216749749,
+      "loss": 2.2977,
+      "step": 39536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011276429340123265,
+      "loss": 2.3101,
+      "step": 39544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011273580435772486,
+      "loss": 2.2923,
+      "step": 39552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011270731503983141,
+      "loss": 2.3002,
+      "step": 39560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011267882545040953,
+      "loss": 2.3354,
+      "step": 39568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001126503355923167,
+      "loss": 2.3142,
+      "step": 39576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001126218454684102,
+      "loss": 2.2916,
+      "step": 39584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011259335508154747,
+      "loss": 2.2948,
+      "step": 39592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011256486443458591,
+      "loss": 2.325,
+      "step": 39600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000112536373530383,
+      "loss": 2.3238,
+      "step": 39608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011250788237179617,
+      "loss": 2.3006,
+      "step": 39616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011247939096168297,
+      "loss": 2.3277,
+      "step": 39624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011245089930290089,
+      "loss": 2.3121,
+      "step": 39632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011242240739830749,
+      "loss": 2.3125,
+      "step": 39640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001123939152507604,
+      "loss": 2.307,
+      "step": 39648
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011236542286311713,
+      "loss": 2.283,
+      "step": 39656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011233693023823539,
+      "loss": 2.2993,
+      "step": 39664
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011230843737897274,
+      "loss": 2.3332,
+      "step": 39672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011227994428818691,
+      "loss": 2.3127,
+      "step": 39680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011225145096873561,
+      "loss": 2.3107,
+      "step": 39688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011222295742347653,
+      "loss": 2.3024,
+      "step": 39696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011219446365526742,
+      "loss": 2.3242,
+      "step": 39704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011216596966696602,
+      "loss": 2.3285,
+      "step": 39712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011213747546143014,
+      "loss": 2.2696,
+      "step": 39720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011210898104151761,
+      "loss": 2.2993,
+      "step": 39728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011208048641008621,
+      "loss": 2.273,
+      "step": 39736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011205199156999381,
+      "loss": 2.3094,
+      "step": 39744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001120234965240983,
+      "loss": 2.316,
+      "step": 39752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011199500127525753,
+      "loss": 2.3134,
+      "step": 39760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011196650582632946,
+      "loss": 2.3003,
+      "step": 39768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011193801018017198,
+      "loss": 2.3021,
+      "step": 39776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011190951433964305,
+      "loss": 2.3161,
+      "step": 39784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011188101830760067,
+      "loss": 2.3021,
+      "step": 39792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011185252208690282,
+      "loss": 2.324,
+      "step": 39800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011182402568040749,
+      "loss": 2.3121,
+      "step": 39808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001117955290909727,
+      "loss": 2.302,
+      "step": 39816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011176703232145655,
+      "loss": 2.3075,
+      "step": 39824
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011173853537471709,
+      "loss": 2.3194,
+      "step": 39832
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011171003825361233,
+      "loss": 2.3229,
+      "step": 39840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011168154096100046,
+      "loss": 2.3115,
+      "step": 39848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001116530434997395,
+      "loss": 2.3026,
+      "step": 39856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011162454587268774,
+      "loss": 2.3199,
+      "step": 39864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011159604808270314,
+      "loss": 2.3086,
+      "step": 39872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011156755013264401,
+      "loss": 2.3299,
+      "step": 39880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001115390520253685,
+      "loss": 2.2996,
+      "step": 39888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011151055376373476,
+      "loss": 2.308,
+      "step": 39896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011148205535060106,
+      "loss": 2.324,
+      "step": 39904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011145355678882558,
+      "loss": 2.3061,
+      "step": 39912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001114250580812666,
+      "loss": 2.3108,
+      "step": 39920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011139655923078238,
+      "loss": 2.2693,
+      "step": 39928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011136806024023119,
+      "loss": 2.3138,
+      "step": 39936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011133956111247126,
+      "loss": 2.312,
+      "step": 39944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011131106185036099,
+      "loss": 2.3129,
+      "step": 39952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011128256245675863,
+      "loss": 2.3095,
+      "step": 39960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011125406293452252,
+      "loss": 2.2999,
+      "step": 39968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011122556328651099,
+      "loss": 2.3004,
+      "step": 39976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011119706351558243,
+      "loss": 2.3131,
+      "step": 39984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011116856362459517,
+      "loss": 2.3082,
+      "step": 39992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011114006361640766,
+      "loss": 2.3248,
+      "step": 40000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011111156349387812,
+      "loss": 2.3204,
+      "step": 40008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011108306325986513,
+      "loss": 2.3051,
+      "step": 40016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000111054562917227,
+      "loss": 2.2978,
+      "step": 40024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001110260624688222,
+      "loss": 2.3146,
+      "step": 40032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011099756191750917,
+      "loss": 2.2796,
+      "step": 40040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011096906126614632,
+      "loss": 2.3304,
+      "step": 40048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011094056051759212,
+      "loss": 2.3289,
+      "step": 40056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011091205967470502,
+      "loss": 2.2966,
+      "step": 40064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011088355874034346,
+      "loss": 2.327,
+      "step": 40072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011085505771736602,
+      "loss": 2.3346,
+      "step": 40080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011082655660863113,
+      "loss": 2.3055,
+      "step": 40088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011079805541699728,
+      "loss": 2.3125,
+      "step": 40096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000110769554145323,
+      "loss": 2.3026,
+      "step": 40104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001107410527964668,
+      "loss": 2.3164,
+      "step": 40112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001107125513732872,
+      "loss": 2.2912,
+      "step": 40120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011068404987864276,
+      "loss": 2.3167,
+      "step": 40128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011065554831539195,
+      "loss": 2.2963,
+      "step": 40136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011062704668639338,
+      "loss": 2.306,
+      "step": 40144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001105985449945056,
+      "loss": 2.2963,
+      "step": 40152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001105700432425871,
+      "loss": 2.3024,
+      "step": 40160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011054154143349656,
+      "loss": 2.2982,
+      "step": 40168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011051303957009242,
+      "loss": 2.3268,
+      "step": 40176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011048453765523337,
+      "loss": 2.2912,
+      "step": 40184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011045603569177789,
+      "loss": 2.3089,
+      "step": 40192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011042753368258465,
+      "loss": 2.2939,
+      "step": 40200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011039903163051217,
+      "loss": 2.3153,
+      "step": 40208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011037052953841914,
+      "loss": 2.3241,
+      "step": 40216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011034202740916404,
+      "loss": 2.326,
+      "step": 40224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011031352524560554,
+      "loss": 2.3158,
+      "step": 40232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011028502305060222,
+      "loss": 2.3167,
+      "step": 40240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011025652082701273,
+      "loss": 2.3003,
+      "step": 40248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011022801857769565,
+      "loss": 2.3132,
+      "step": 40256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011019951630550962,
+      "loss": 2.3101,
+      "step": 40264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011017101401331317,
+      "loss": 2.3139,
+      "step": 40272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011014251170396504,
+      "loss": 2.3024,
+      "step": 40280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011011400938032377,
+      "loss": 2.2991,
+      "step": 40288
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000110085507045248,
+      "loss": 2.3013,
+      "step": 40296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011005700470159639,
+      "loss": 2.3226,
+      "step": 40304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011002850235222751,
+      "loss": 2.308,
+      "step": 40312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 2.3367,
+      "step": 40320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001099714976477725,
+      "loss": 2.324,
+      "step": 40328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010994299529840364,
+      "loss": 2.3037,
+      "step": 40336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000109914492954752,
+      "loss": 2.3332,
+      "step": 40344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010988599061967626,
+      "loss": 2.2995,
+      "step": 40352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000109857488296035,
+      "loss": 2.3043,
+      "step": 40360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010982898598668685,
+      "loss": 2.3345,
+      "step": 40368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010980048369449041,
+      "loss": 2.3242,
+      "step": 40376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010977198142230436,
+      "loss": 2.3171,
+      "step": 40384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010974347917298731,
+      "loss": 2.3113,
+      "step": 40392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001097149769493978,
+      "loss": 2.3076,
+      "step": 40400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010968647475439448,
+      "loss": 2.2891,
+      "step": 40408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010965797259083599,
+      "loss": 2.316,
+      "step": 40416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010962947046158092,
+      "loss": 2.3197,
+      "step": 40424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010960096836948784,
+      "loss": 2.296,
+      "step": 40432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010957246631741537,
+      "loss": 2.2941,
+      "step": 40440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010954396430822213,
+      "loss": 2.3092,
+      "step": 40448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010951546234476666,
+      "loss": 2.3177,
+      "step": 40456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010948696042990758,
+      "loss": 2.3084,
+      "step": 40464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010945845856650348,
+      "loss": 2.335,
+      "step": 40472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001094299567574129,
+      "loss": 2.3034,
+      "step": 40480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010940145500549444,
+      "loss": 2.312,
+      "step": 40488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010937295331360664,
+      "loss": 2.3106,
+      "step": 40496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010934445168460808,
+      "loss": 2.3026,
+      "step": 40504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010931595012135727,
+      "loss": 2.2865,
+      "step": 40512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001092874486267128,
+      "loss": 2.3102,
+      "step": 40520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010925894720353322,
+      "loss": 2.3152,
+      "step": 40528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010923044585467703,
+      "loss": 2.293,
+      "step": 40536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010920194458300273,
+      "loss": 2.3052,
+      "step": 40544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001091734433913689,
+      "loss": 2.2823,
+      "step": 40552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010914494228263398,
+      "loss": 2.3042,
+      "step": 40560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010911644125965657,
+      "loss": 2.2974,
+      "step": 40568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010908794032529501,
+      "loss": 2.2982,
+      "step": 40576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010905943948240793,
+      "loss": 2.3155,
+      "step": 40584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010903093873385373,
+      "loss": 2.3153,
+      "step": 40592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010900243808249087,
+      "loss": 2.3206,
+      "step": 40600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010897393753117782,
+      "loss": 2.3006,
+      "step": 40608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000108945437082773,
+      "loss": 2.2978,
+      "step": 40616
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001089169367401349,
+      "loss": 2.3088,
+      "step": 40624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001088884365061219,
+      "loss": 2.3071,
+      "step": 40632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010885993638359242,
+      "loss": 2.3006,
+      "step": 40640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010883143637540482,
+      "loss": 2.2962,
+      "step": 40648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001088029364844176,
+      "loss": 2.283,
+      "step": 40656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010877443671348903,
+      "loss": 2.3001,
+      "step": 40664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010874593706547753,
+      "loss": 2.3058,
+      "step": 40672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001087174375432414,
+      "loss": 2.3164,
+      "step": 40680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010868893814963903,
+      "loss": 2.3095,
+      "step": 40688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010866043888752874,
+      "loss": 2.3285,
+      "step": 40696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010863193975976889,
+      "loss": 2.3068,
+      "step": 40704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010860344076921765,
+      "loss": 2.287,
+      "step": 40712
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010857494191873343,
+      "loss": 2.3192,
+      "step": 40720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010854644321117444,
+      "loss": 2.3142,
+      "step": 40728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010851794464939898,
+      "loss": 2.313,
+      "step": 40736
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010848944623626526,
+      "loss": 2.3191,
+      "step": 40744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010846094797463156,
+      "loss": 2.3046,
+      "step": 40752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010843244986735601,
+      "loss": 2.2815,
+      "step": 40760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010840395191729686,
+      "loss": 2.3175,
+      "step": 40768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010837545412731231,
+      "loss": 2.286,
+      "step": 40776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010834695650026048,
+      "loss": 2.2999,
+      "step": 40784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001083184590389996,
+      "loss": 2.3086,
+      "step": 40792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010828996174638771,
+      "loss": 2.3154,
+      "step": 40800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010826146462528298,
+      "loss": 2.3211,
+      "step": 40808
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010823296767854345,
+      "loss": 2.3023,
+      "step": 40816
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001082044709090273,
+      "loss": 2.3119,
+      "step": 40824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010817597431959255,
+      "loss": 2.3004,
+      "step": 40832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010814747791309723,
+      "loss": 2.2792,
+      "step": 40840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010811898169239934,
+      "loss": 2.3028,
+      "step": 40848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010809048566035696,
+      "loss": 2.3169,
+      "step": 40856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010806198981982805,
+      "loss": 2.2997,
+      "step": 40864
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010803349417367057,
+      "loss": 2.3118,
+      "step": 40872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001080049987247425,
+      "loss": 2.3168,
+      "step": 40880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010797650347590174,
+      "loss": 2.2823,
+      "step": 40888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010794800843000621,
+      "loss": 2.2986,
+      "step": 40896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010791951358991381,
+      "loss": 2.3033,
+      "step": 40904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010789101895848241,
+      "loss": 2.2938,
+      "step": 40912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010786252453856989,
+      "loss": 2.3143,
+      "step": 40920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000107834030333034,
+      "loss": 2.2907,
+      "step": 40928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001078055363447326,
+      "loss": 2.2963,
+      "step": 40936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001077770425765235,
+      "loss": 2.2969,
+      "step": 40944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001077485490312644,
+      "loss": 2.3158,
+      "step": 40952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001077200557118131,
+      "loss": 2.3117,
+      "step": 40960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010769156262102729,
+      "loss": 2.2931,
+      "step": 40968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010766306976176467,
+      "loss": 2.3069,
+      "step": 40976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010763457713688287,
+      "loss": 2.3004,
+      "step": 40984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010760608474923964,
+      "loss": 2.3145,
+      "step": 40992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001075775926016925,
+      "loss": 2.2891,
+      "step": 41000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010754910069709913,
+      "loss": 2.3175,
+      "step": 41008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010752060903831707,
+      "loss": 2.3085,
+      "step": 41016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010749211762820384,
+      "loss": 2.3009,
+      "step": 41024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010746362646961702,
+      "loss": 2.3174,
+      "step": 41032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010743513556541411,
+      "loss": 2.3065,
+      "step": 41040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010740664491845254,
+      "loss": 2.3182,
+      "step": 41048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010737815453158981,
+      "loss": 2.3174,
+      "step": 41056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010734966440768335,
+      "loss": 2.315,
+      "step": 41064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010732117454959048,
+      "loss": 2.2915,
+      "step": 41072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010729268496016864,
+      "loss": 2.3255,
+      "step": 41080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010726419564227516,
+      "loss": 2.2975,
+      "step": 41088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010723570659876738,
+      "loss": 2.3002,
+      "step": 41096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010720721783250254,
+      "loss": 2.2941,
+      "step": 41104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010717872934633792,
+      "loss": 2.3163,
+      "step": 41112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010715024114313078,
+      "loss": 2.3005,
+      "step": 41120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010712175322573827,
+      "loss": 2.3059,
+      "step": 41128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010709326559701765,
+      "loss": 2.3077,
+      "step": 41136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000107064778259826,
+      "loss": 2.2911,
+      "step": 41144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010703629121702045,
+      "loss": 2.3138,
+      "step": 41152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001070078044714581,
+      "loss": 2.3275,
+      "step": 41160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010697931802599603,
+      "loss": 2.309,
+      "step": 41168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010695083188349122,
+      "loss": 2.2899,
+      "step": 41176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010692234604680072,
+      "loss": 2.2874,
+      "step": 41184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010689386051878146,
+      "loss": 2.2992,
+      "step": 41192
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010686537530229037,
+      "loss": 2.3091,
+      "step": 41200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010683689040018439,
+      "loss": 2.3312,
+      "step": 41208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010680840581532036,
+      "loss": 2.3282,
+      "step": 41216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010677992155055516,
+      "loss": 2.3134,
+      "step": 41224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010675143760874562,
+      "loss": 2.3041,
+      "step": 41232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010672295399274843,
+      "loss": 2.3146,
+      "step": 41240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001066944707054204,
+      "loss": 2.3119,
+      "step": 41248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010666598774961822,
+      "loss": 2.2882,
+      "step": 41256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010663750512819859,
+      "loss": 2.284,
+      "step": 41264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010660902284401816,
+      "loss": 2.3088,
+      "step": 41272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010658054089993349,
+      "loss": 2.2812,
+      "step": 41280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010655205929880121,
+      "loss": 2.3095,
+      "step": 41288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010652357804347784,
+      "loss": 2.3062,
+      "step": 41296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010649509713681988,
+      "loss": 2.3074,
+      "step": 41304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010646661658168384,
+      "loss": 2.3275,
+      "step": 41312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010643813638092612,
+      "loss": 2.3013,
+      "step": 41320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010640965653740316,
+      "loss": 2.2992,
+      "step": 41328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010638117705397128,
+      "loss": 2.3256,
+      "step": 41336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010635269793348686,
+      "loss": 2.3019,
+      "step": 41344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010632421917880613,
+      "loss": 2.313,
+      "step": 41352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010629574079278544,
+      "loss": 2.2922,
+      "step": 41360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010626726277828094,
+      "loss": 2.312,
+      "step": 41368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010623878513814883,
+      "loss": 2.3184,
+      "step": 41376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010621030787524522,
+      "loss": 2.297,
+      "step": 41384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010618183099242628,
+      "loss": 2.2943,
+      "step": 41392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010615335449254808,
+      "loss": 2.3047,
+      "step": 41400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010612487837846666,
+      "loss": 2.3133,
+      "step": 41408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001060964026530379,
+      "loss": 2.3041,
+      "step": 41416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010606792731911788,
+      "loss": 2.3138,
+      "step": 41424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010603945237956247,
+      "loss": 2.2992,
+      "step": 41432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010601097783722753,
+      "loss": 2.2993,
+      "step": 41440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010598250369496892,
+      "loss": 2.2912,
+      "step": 41448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010595402995564243,
+      "loss": 2.3078,
+      "step": 41456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001059255566221038,
+      "loss": 2.2973,
+      "step": 41464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010589708369720874,
+      "loss": 2.2894,
+      "step": 41472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010586861118381297,
+      "loss": 2.2959,
+      "step": 41480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010584013908477206,
+      "loss": 2.2747,
+      "step": 41488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010581166740294164,
+      "loss": 2.297,
+      "step": 41496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010578319614117726,
+      "loss": 2.2968,
+      "step": 41504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010575472530233439,
+      "loss": 2.295,
+      "step": 41512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010572625488926853,
+      "loss": 2.3006,
+      "step": 41520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010569778490483506,
+      "loss": 2.3169,
+      "step": 41528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010566931535188944,
+      "loss": 2.3021,
+      "step": 41536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010564084623328692,
+      "loss": 2.304,
+      "step": 41544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010561237755188283,
+      "loss": 2.2904,
+      "step": 41552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001055839093105324,
+      "loss": 2.2943,
+      "step": 41560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010555544151209082,
+      "loss": 2.289,
+      "step": 41568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010552697415941333,
+      "loss": 2.3049,
+      "step": 41576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010549850725535495,
+      "loss": 2.3005,
+      "step": 41584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010547004080277078,
+      "loss": 2.3021,
+      "step": 41592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010544157480451587,
+      "loss": 2.3002,
+      "step": 41600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010541310926344516,
+      "loss": 2.297,
+      "step": 41608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010538464418241361,
+      "loss": 2.2845,
+      "step": 41616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001053561795642761,
+      "loss": 2.2973,
+      "step": 41624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010532771541188746,
+      "loss": 2.2888,
+      "step": 41632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010529925172810248,
+      "loss": 2.2998,
+      "step": 41640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010527078851577595,
+      "loss": 2.2833,
+      "step": 41648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001052423257777625,
+      "loss": 2.2866,
+      "step": 41656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010521386351691682,
+      "loss": 2.3078,
+      "step": 41664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010518540173609355,
+      "loss": 2.3371,
+      "step": 41672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010515694043814716,
+      "loss": 2.2941,
+      "step": 41680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010512847962593219,
+      "loss": 2.3026,
+      "step": 41688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010510001930230313,
+      "loss": 2.3105,
+      "step": 41696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010507155947011437,
+      "loss": 2.3073,
+      "step": 41704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010504310013222032,
+      "loss": 2.3004,
+      "step": 41712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010501464129147516,
+      "loss": 2.3072,
+      "step": 41720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010498618295073326,
+      "loss": 2.3063,
+      "step": 41728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001049577251128488,
+      "loss": 2.2903,
+      "step": 41736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010492926778067595,
+      "loss": 2.3036,
+      "step": 41744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001049008109570688,
+      "loss": 2.2932,
+      "step": 41752
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010487235464488141,
+      "loss": 2.2999,
+      "step": 41760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010484389884696782,
+      "loss": 2.2961,
+      "step": 41768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010481544356618193,
+      "loss": 2.2944,
+      "step": 41776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010478698880537769,
+      "loss": 2.3114,
+      "step": 41784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010475853456740892,
+      "loss": 2.3058,
+      "step": 41792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010473008085512946,
+      "loss": 2.3159,
+      "step": 41800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000104701627671393,
+      "loss": 2.3001,
+      "step": 41808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010467317501905327,
+      "loss": 2.2894,
+      "step": 41816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001046447229009639,
+      "loss": 2.2993,
+      "step": 41824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010461627131997849,
+      "loss": 2.305,
+      "step": 41832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010458782027895055,
+      "loss": 2.279,
+      "step": 41840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001045593697807336,
+      "loss": 2.2947,
+      "step": 41848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000104530919828181,
+      "loss": 2.2992,
+      "step": 41856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010450247042414614,
+      "loss": 2.3024,
+      "step": 41864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010447402157148238,
+      "loss": 2.2991,
+      "step": 41872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010444557327304292,
+      "loss": 2.3166,
+      "step": 41880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000104417125531681,
+      "loss": 2.3056,
+      "step": 41888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010438867835024974,
+      "loss": 2.2767,
+      "step": 41896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010436023173160227,
+      "loss": 2.29,
+      "step": 41904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010433178567859159,
+      "loss": 2.3076,
+      "step": 41912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010430334019407066,
+      "loss": 2.3071,
+      "step": 41920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010427489528089247,
+      "loss": 2.2919,
+      "step": 41928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010424645094190983,
+      "loss": 2.2968,
+      "step": 41936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010421800717997556,
+      "loss": 2.2938,
+      "step": 41944
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001041895639979424,
+      "loss": 2.2932,
+      "step": 41952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010416112139866305,
+      "loss": 2.3062,
+      "step": 41960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010413267938499014,
+      "loss": 2.3143,
+      "step": 41968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010410423795977629,
+      "loss": 2.3132,
+      "step": 41976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010407579712587388,
+      "loss": 2.2922,
+      "step": 41984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001040473568861355,
+      "loss": 2.2885,
+      "step": 41992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010401891724341346,
+      "loss": 2.2924,
+      "step": 42000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010399047820056016,
+      "loss": 2.303,
+      "step": 42008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010396203976042784,
+      "loss": 2.2758,
+      "step": 42016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010393360192586872,
+      "loss": 2.3252,
+      "step": 42024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010390516469973495,
+      "loss": 2.2892,
+      "step": 42032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010387672808487862,
+      "loss": 2.3077,
+      "step": 42040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010384829208415179,
+      "loss": 2.3002,
+      "step": 42048
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010381985670040635,
+      "loss": 2.3017,
+      "step": 42056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010379142193649434,
+      "loss": 2.3172,
+      "step": 42064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001037629877952675,
+      "loss": 2.2812,
+      "step": 42072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001037345542795776,
+      "loss": 2.2807,
+      "step": 42080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010370612139227644,
+      "loss": 2.3093,
+      "step": 42088
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010367768913621565,
+      "loss": 2.2823,
+      "step": 42096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010364925751424681,
+      "loss": 2.3032,
+      "step": 42104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010362082652922148,
+      "loss": 2.3072,
+      "step": 42112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010359239618399103,
+      "loss": 2.317,
+      "step": 42120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010356396648140699,
+      "loss": 2.3177,
+      "step": 42128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010353553742432063,
+      "loss": 2.2968,
+      "step": 42136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010350710901558325,
+      "loss": 2.3242,
+      "step": 42144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010347868125804603,
+      "loss": 2.2912,
+      "step": 42152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010345025415456013,
+      "loss": 2.3084,
+      "step": 42160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010342182770797663,
+      "loss": 2.3047,
+      "step": 42168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010339340192114654,
+      "loss": 2.2915,
+      "step": 42176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010336497679692081,
+      "loss": 2.3056,
+      "step": 42184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010333655233815028,
+      "loss": 2.3089,
+      "step": 42192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010330812854768584,
+      "loss": 2.3013,
+      "step": 42200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010327970542837814,
+      "loss": 2.2917,
+      "step": 42208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010325128298307793,
+      "loss": 2.3007,
+      "step": 42216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010322286121463577,
+      "loss": 2.312,
+      "step": 42224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010319444012590225,
+      "loss": 2.2998,
+      "step": 42232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010316601971972785,
+      "loss": 2.2962,
+      "step": 42240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010313759999896293,
+      "loss": 2.309,
+      "step": 42248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010310918096645783,
+      "loss": 2.2772,
+      "step": 42256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010308076262506283,
+      "loss": 2.3025,
+      "step": 42264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010305234497762817,
+      "loss": 2.3131,
+      "step": 42272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001030239280270039,
+      "loss": 2.3079,
+      "step": 42280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010299551177604014,
+      "loss": 2.2963,
+      "step": 42288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010296709622758684,
+      "loss": 2.3078,
+      "step": 42296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010293868138449399,
+      "loss": 2.283,
+      "step": 42304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010291026724961132,
+      "loss": 2.304,
+      "step": 42312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010288185382578872,
+      "loss": 2.3008,
+      "step": 42320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010285344111587583,
+      "loss": 2.2938,
+      "step": 42328
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010282502912272229,
+      "loss": 2.2929,
+      "step": 42336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010279661784917769,
+      "loss": 2.3177,
+      "step": 42344
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001027682072980915,
+      "loss": 2.3039,
+      "step": 42352
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001027397974723131,
+      "loss": 2.301,
+      "step": 42360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010271138837469194,
+      "loss": 2.3047,
+      "step": 42368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010268298000807719,
+      "loss": 2.2946,
+      "step": 42376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001026545723753181,
+      "loss": 2.2901,
+      "step": 42384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010262616547926376,
+      "loss": 2.3042,
+      "step": 42392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010259775932276326,
+      "loss": 2.3173,
+      "step": 42400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010256935390866556,
+      "loss": 2.3031,
+      "step": 42408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010254094923981954,
+      "loss": 2.2917,
+      "step": 42416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010251254531907406,
+      "loss": 2.294,
+      "step": 42424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010248414214927781,
+      "loss": 2.3146,
+      "step": 42432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001024557397332796,
+      "loss": 2.289,
+      "step": 42440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001024273380739279,
+      "loss": 2.3096,
+      "step": 42448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010239893717407128,
+      "loss": 2.3096,
+      "step": 42456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010237053703655819,
+      "loss": 2.2851,
+      "step": 42464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010234213766423699,
+      "loss": 2.2732,
+      "step": 42472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010231373905995605,
+      "loss": 2.3076,
+      "step": 42480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010228534122656349,
+      "loss": 2.3163,
+      "step": 42488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010225694416690748,
+      "loss": 2.2813,
+      "step": 42496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010222854788383612,
+      "loss": 2.3012,
+      "step": 42504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010220015238019733,
+      "loss": 2.3096,
+      "step": 42512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010217175765883913,
+      "loss": 2.3145,
+      "step": 42520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010214336372260922,
+      "loss": 2.297,
+      "step": 42528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010211497057435542,
+      "loss": 2.2937,
+      "step": 42536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010208657821692541,
+      "loss": 2.2999,
+      "step": 42544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010205818665316675,
+      "loss": 2.2955,
+      "step": 42552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010202979588592698,
+      "loss": 2.2889,
+      "step": 42560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001020014059180535,
+      "loss": 2.3177,
+      "step": 42568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001019730167523937,
+      "loss": 2.2985,
+      "step": 42576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010194462839179486,
+      "loss": 2.3037,
+      "step": 42584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010191624083910412,
+      "loss": 2.3042,
+      "step": 42592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010188785409716864,
+      "loss": 2.3035,
+      "step": 42600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010185946816883538,
+      "loss": 2.2998,
+      "step": 42608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001018310830569514,
+      "loss": 2.3073,
+      "step": 42616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010180269876436345,
+      "loss": 2.3003,
+      "step": 42624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010177431529391837,
+      "loss": 2.2804,
+      "step": 42632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010174593264846288,
+      "loss": 2.2982,
+      "step": 42640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010171755083084356,
+      "loss": 2.3084,
+      "step": 42648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000101689169843907,
+      "loss": 2.3019,
+      "step": 42656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010166078969049957,
+      "loss": 2.2963,
+      "step": 42664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010163241037346766,
+      "loss": 2.3163,
+      "step": 42672
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010160403189565764,
+      "loss": 2.3219,
+      "step": 42680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001015756542599156,
+      "loss": 2.2902,
+      "step": 42688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010154727746908773,
+      "loss": 2.3088,
+      "step": 42696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010151890152602001,
+      "loss": 2.315,
+      "step": 42704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010149052643355841,
+      "loss": 2.2957,
+      "step": 42712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010146215219454885,
+      "loss": 2.3088,
+      "step": 42720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010143377881183695,
+      "loss": 2.2971,
+      "step": 42728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010140540628826855,
+      "loss": 2.2777,
+      "step": 42736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010137703462668917,
+      "loss": 2.2804,
+      "step": 42744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010134866382994442,
+      "loss": 2.2992,
+      "step": 42752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010132029390087962,
+      "loss": 2.2948,
+      "step": 42760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010129192484234015,
+      "loss": 2.311,
+      "step": 42768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010126355665717134,
+      "loss": 2.3109,
+      "step": 42776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001012351893482182,
+      "loss": 2.2969,
+      "step": 42784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010120682291832598,
+      "loss": 2.2997,
+      "step": 42792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010117845737033957,
+      "loss": 2.3081,
+      "step": 42800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010115009270710387,
+      "loss": 2.2822,
+      "step": 42808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010112172893146378,
+      "loss": 2.3076,
+      "step": 42816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010109336604626394,
+      "loss": 2.3113,
+      "step": 42824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010106500405434902,
+      "loss": 2.2999,
+      "step": 42832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010103664295856359,
+      "loss": 2.2949,
+      "step": 42840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010100828276175202,
+      "loss": 2.3018,
+      "step": 42848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010097992346675881,
+      "loss": 2.2996,
+      "step": 42856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010095156507642809,
+      "loss": 2.2856,
+      "step": 42864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010092320759360418,
+      "loss": 2.3038,
+      "step": 42872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010089485102113114,
+      "loss": 2.2979,
+      "step": 42880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010086649536185289,
+      "loss": 2.2989,
+      "step": 42888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010083814061861343,
+      "loss": 2.2923,
+      "step": 42896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010080978679425653,
+      "loss": 2.3045,
+      "step": 42904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010078143389162598,
+      "loss": 2.2918,
+      "step": 42912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010075308191356532,
+      "loss": 2.3095,
+      "step": 42920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010072473086291818,
+      "loss": 2.3025,
+      "step": 42928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010069638074252801,
+      "loss": 2.3297,
+      "step": 42936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010066803155523807,
+      "loss": 2.3128,
+      "step": 42944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010063968330389175,
+      "loss": 2.3206,
+      "step": 42952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001006113359913321,
+      "loss": 2.3323,
+      "step": 42960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010058298962040229,
+      "loss": 2.2972,
+      "step": 42968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010055464419394527,
+      "loss": 2.3056,
+      "step": 42976
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010052629971480387,
+      "loss": 2.3087,
+      "step": 42984
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010049795618582096,
+      "loss": 2.3108,
+      "step": 42992
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010046961360983917,
+      "loss": 2.3072,
+      "step": 43000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010044127198970117,
+      "loss": 2.2862,
+      "step": 43008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010041293132824945,
+      "loss": 2.3101,
+      "step": 43016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010038459162832635,
+      "loss": 2.3118,
+      "step": 43024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010035625289277421,
+      "loss": 2.3137,
+      "step": 43032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010032791512443527,
+      "loss": 2.3008,
+      "step": 43040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001002995783261517,
+      "loss": 2.2978,
+      "step": 43048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010027124250076539,
+      "loss": 2.28,
+      "step": 43056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010024290765111835,
+      "loss": 2.2969,
+      "step": 43064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010021457378005239,
+      "loss": 2.294,
+      "step": 43072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010018624089040922,
+      "loss": 2.3237,
+      "step": 43080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001001579089850305,
+      "loss": 2.3093,
+      "step": 43088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010012957806675771,
+      "loss": 2.3071,
+      "step": 43096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010010124813843236,
+      "loss": 2.2992,
+      "step": 43104
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010007291920289572,
+      "loss": 2.2821,
+      "step": 43112
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000100044591262989,
+      "loss": 2.304,
+      "step": 43120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010001626432155337,
+      "loss": 2.2977,
+      "step": 43128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.998793838142986e-05,
+      "loss": 2.2894,
+      "step": 43136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.995961344545944e-05,
+      "loss": 2.3071,
+      "step": 43144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.993128951648289e-05,
+      "loss": 2.3119,
+      "step": 43152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.990296659734091e-05,
+      "loss": 2.2908,
+      "step": 43160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.98746446908742e-05,
+      "loss": 2.2991,
+      "step": 43168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.984632379992323e-05,
+      "loss": 2.2774,
+      "step": 43176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.981800392732845e-05,
+      "loss": 2.3025,
+      "step": 43184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.97896850759302e-05,
+      "loss": 2.2995,
+      "step": 43192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.976136724856866e-05,
+      "loss": 2.3057,
+      "step": 43200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.973305044808401e-05,
+      "loss": 2.3014,
+      "step": 43208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.970473467731618e-05,
+      "loss": 2.3051,
+      "step": 43216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.967641993910518e-05,
+      "loss": 2.3001,
+      "step": 43224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.964810623629071e-05,
+      "loss": 2.2885,
+      "step": 43232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.961979357171254e-05,
+      "loss": 2.2805,
+      "step": 43240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.959148194821027e-05,
+      "loss": 2.3048,
+      "step": 43248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.956317136862336e-05,
+      "loss": 2.2945,
+      "step": 43256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.953486183579124e-05,
+      "loss": 2.3017,
+      "step": 43264
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.950655335255317e-05,
+      "loss": 2.2784,
+      "step": 43272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.947824592174835e-05,
+      "loss": 2.3053,
+      "step": 43280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.944993954621587e-05,
+      "loss": 2.2954,
+      "step": 43288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.942163422879463e-05,
+      "loss": 2.2982,
+      "step": 43296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.939332997232356e-05,
+      "loss": 2.3051,
+      "step": 43304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.936502677964135e-05,
+      "loss": 2.3075,
+      "step": 43312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.933672465358677e-05,
+      "loss": 2.3024,
+      "step": 43320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.930842359699824e-05,
+      "loss": 2.3206,
+      "step": 43328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.928012361271422e-05,
+      "loss": 2.2904,
+      "step": 43336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.925182470357311e-05,
+      "loss": 2.2906,
+      "step": 43344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.922352687241303e-05,
+      "loss": 2.2915,
+      "step": 43352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.91952301220722e-05,
+      "loss": 2.2974,
+      "step": 43360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.916693445538852e-05,
+      "loss": 2.3028,
+      "step": 43368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.913863987519992e-05,
+      "loss": 2.3058,
+      "step": 43376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.911034638434424e-05,
+      "loss": 2.2847,
+      "step": 43384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.908205398565907e-05,
+      "loss": 2.302,
+      "step": 43392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.905376268198205e-05,
+      "loss": 2.2857,
+      "step": 43400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.902547247615062e-05,
+      "loss": 2.3047,
+      "step": 43408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.899718337100207e-05,
+      "loss": 2.3228,
+      "step": 43416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.896889536937375e-05,
+      "loss": 2.296,
+      "step": 43424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.894060847410264e-05,
+      "loss": 2.3182,
+      "step": 43432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.891232268802588e-05,
+      "loss": 2.2831,
+      "step": 43440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.88840380139803e-05,
+      "loss": 2.3262,
+      "step": 43448
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.885575445480278e-05,
+      "loss": 2.3164,
+      "step": 43456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.882747201332989e-05,
+      "loss": 2.3014,
+      "step": 43464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.879919069239825e-05,
+      "loss": 2.3126,
+      "step": 43472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.877091049484435e-05,
+      "loss": 2.2903,
+      "step": 43480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.874263142350444e-05,
+      "loss": 2.2932,
+      "step": 43488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.871435348121487e-05,
+      "loss": 2.3037,
+      "step": 43496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.868607667081169e-05,
+      "loss": 2.3064,
+      "step": 43504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.865780099513089e-05,
+      "loss": 2.3046,
+      "step": 43512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.86295264570084e-05,
+      "loss": 2.2858,
+      "step": 43520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.860125305927994e-05,
+      "loss": 2.2735,
+      "step": 43528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.857298080478123e-05,
+      "loss": 2.2981,
+      "step": 43536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.854470969634784e-05,
+      "loss": 2.3059,
+      "step": 43544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.851643973681511e-05,
+      "loss": 2.3115,
+      "step": 43552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.848817092901843e-05,
+      "loss": 2.2911,
+      "step": 43560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.845990327579295e-05,
+      "loss": 2.2974,
+      "step": 43568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.843163677997378e-05,
+      "loss": 2.3143,
+      "step": 43576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.840337144439593e-05,
+      "loss": 2.2975,
+      "step": 43584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.837510727189419e-05,
+      "loss": 2.3232,
+      "step": 43592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.834684426530332e-05,
+      "loss": 2.2983,
+      "step": 43600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.831858242745791e-05,
+      "loss": 2.2911,
+      "step": 43608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.829032176119256e-05,
+      "loss": 2.3037,
+      "step": 43616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.826206226934151e-05,
+      "loss": 2.2845,
+      "step": 43624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.823380395473916e-05,
+      "loss": 2.3093,
+      "step": 43632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.820554682021958e-05,
+      "loss": 2.3247,
+      "step": 43640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.817729086861682e-05,
+      "loss": 2.2972,
+      "step": 43648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.814903610276482e-05,
+      "loss": 2.2789,
+      "step": 43656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.812078252549732e-05,
+      "loss": 2.3011,
+      "step": 43664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.809253013964804e-05,
+      "loss": 2.2929,
+      "step": 43672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.80642789480505e-05,
+      "loss": 2.2947,
+      "step": 43680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.803602895353816e-05,
+      "loss": 2.294,
+      "step": 43688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.800778015894434e-05,
+      "loss": 2.2883,
+      "step": 43696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.797953256710214e-05,
+      "loss": 2.2903,
+      "step": 43704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.79512861808448e-05,
+      "loss": 2.3099,
+      "step": 43712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.792304100300518e-05,
+      "loss": 2.3017,
+      "step": 43720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.789479703641607e-05,
+      "loss": 2.2888,
+      "step": 43728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.786655428391025e-05,
+      "loss": 2.3041,
+      "step": 43736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.783831274832025e-05,
+      "loss": 2.2966,
+      "step": 43744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.781007243247864e-05,
+      "loss": 2.3075,
+      "step": 43752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.778183333921766e-05,
+      "loss": 2.3012,
+      "step": 43760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.775359547136953e-05,
+      "loss": 2.3098,
+      "step": 43768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.772535883176641e-05,
+      "loss": 2.295,
+      "step": 43776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.769712342324023e-05,
+      "loss": 2.3145,
+      "step": 43784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.76688892486229e-05,
+      "loss": 2.3048,
+      "step": 43792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.764065631074606e-05,
+      "loss": 2.3006,
+      "step": 43800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.761242461244137e-05,
+      "loss": 2.3134,
+      "step": 43808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.758419415654034e-05,
+      "loss": 2.2953,
+      "step": 43816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.755596494587425e-05,
+      "loss": 2.3063,
+      "step": 43824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.752773698327435e-05,
+      "loss": 2.3052,
+      "step": 43832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.749951027157178e-05,
+      "loss": 2.3159,
+      "step": 43840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.747128481359751e-05,
+      "loss": 2.2961,
+      "step": 43848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.74430606121824e-05,
+      "loss": 2.2969,
+      "step": 43856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.741483767015709e-05,
+      "loss": 2.2936,
+      "step": 43864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.738661599035228e-05,
+      "loss": 2.2882,
+      "step": 43872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.735839557559843e-05,
+      "loss": 2.2976,
+      "step": 43880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.733017642872585e-05,
+      "loss": 2.3021,
+      "step": 43888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.73019585525648e-05,
+      "loss": 2.3144,
+      "step": 43896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.727374194994535e-05,
+      "loss": 2.2869,
+      "step": 43904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.724552662369748e-05,
+      "loss": 2.2863,
+      "step": 43912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.7217312576651e-05,
+      "loss": 2.2926,
+      "step": 43920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.718909981163569e-05,
+      "loss": 2.298,
+      "step": 43928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.716088833148105e-05,
+      "loss": 2.3022,
+      "step": 43936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.713267813901652e-05,
+      "loss": 2.279,
+      "step": 43944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.71044692370715e-05,
+      "loss": 2.2969,
+      "step": 43952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.707626162847511e-05,
+      "loss": 2.3199,
+      "step": 43960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.70480553160565e-05,
+      "loss": 2.3056,
+      "step": 43968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.701985030264451e-05,
+      "loss": 2.3002,
+      "step": 43976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.699164659106801e-05,
+      "loss": 2.3045,
+      "step": 43984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.696344418415566e-05,
+      "loss": 2.3102,
+      "step": 43992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.693524308473595e-05,
+      "loss": 2.2897,
+      "step": 44000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.690704329563735e-05,
+      "loss": 2.3133,
+      "step": 44008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.687884481968811e-05,
+      "loss": 2.2975,
+      "step": 44016
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.685064765971641e-05,
+      "loss": 2.2806,
+      "step": 44024
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.682245181855023e-05,
+      "loss": 2.3249,
+      "step": 44032
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.679425729901743e-05,
+      "loss": 2.2984,
+      "step": 44040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.676606410394584e-05,
+      "loss": 2.2948,
+      "step": 44048
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.673787223616302e-05,
+      "loss": 2.3128,
+      "step": 44056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.670968169849648e-05,
+      "loss": 2.3133,
+      "step": 44064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.668149249377354e-05,
+      "loss": 2.3145,
+      "step": 44072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.665330462482143e-05,
+      "loss": 2.2746,
+      "step": 44080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.662511809446726e-05,
+      "loss": 2.2985,
+      "step": 44088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.659693290553793e-05,
+      "loss": 2.2929,
+      "step": 44096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.656874906086031e-05,
+      "loss": 2.3128,
+      "step": 44104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.654056656326108e-05,
+      "loss": 2.3123,
+      "step": 44112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.651238541556669e-05,
+      "loss": 2.2935,
+      "step": 44120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.648420562060368e-05,
+      "loss": 2.286,
+      "step": 44128
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.645602718119821e-05,
+      "loss": 2.3084,
+      "step": 44136
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.64278501001765e-05,
+      "loss": 2.3011,
+      "step": 44144
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.639967438036448e-05,
+      "loss": 2.3173,
+      "step": 44152
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.637150002458813e-05,
+      "loss": 2.3074,
+      "step": 44160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.634332703567306e-05,
+      "loss": 2.2841,
+      "step": 44168
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.631515541644487e-05,
+      "loss": 2.3107,
+      "step": 44176
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.628698516972911e-05,
+      "loss": 2.2944,
+      "step": 44184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.6258816298351e-05,
+      "loss": 2.2882,
+      "step": 44192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.623064880513579e-05,
+      "loss": 2.3073,
+      "step": 44200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.620248269290846e-05,
+      "loss": 2.3152,
+      "step": 44208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.617431796449392e-05,
+      "loss": 2.3081,
+      "step": 44216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.614615462271698e-05,
+      "loss": 2.2906,
+      "step": 44224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.61179926704022e-05,
+      "loss": 2.2955,
+      "step": 44232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.60898321103741e-05,
+      "loss": 2.2945,
+      "step": 44240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.606167294545705e-05,
+      "loss": 2.3038,
+      "step": 44248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.603351517847519e-05,
+      "loss": 2.2973,
+      "step": 44256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.600535881225266e-05,
+      "loss": 2.2903,
+      "step": 44264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.59772038496133e-05,
+      "loss": 2.2653,
+      "step": 44272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.594905029338095e-05,
+      "loss": 2.311,
+      "step": 44280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.592089814637925e-05,
+      "loss": 2.3064,
+      "step": 44288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.589274741143166e-05,
+      "loss": 2.3186,
+      "step": 44296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.58645980913616e-05,
+      "loss": 2.2997,
+      "step": 44304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.583645018899223e-05,
+      "loss": 2.2999,
+      "step": 44312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.580830370714669e-05,
+      "loss": 2.299,
+      "step": 44320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.578015864864782e-05,
+      "loss": 2.2914,
+      "step": 44328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.575201501631853e-05,
+      "loss": 2.3105,
+      "step": 44336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.572387281298135e-05,
+      "loss": 2.3056,
+      "step": 44344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.569573204145881e-05,
+      "loss": 2.293,
+      "step": 44352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.566759270457337e-05,
+      "loss": 2.2826,
+      "step": 44360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.563945480514711e-05,
+      "loss": 2.2935,
+      "step": 44368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.561131834600216e-05,
+      "loss": 2.3018,
+      "step": 44376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.55831833299605e-05,
+      "loss": 2.2963,
+      "step": 44384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.555504975984382e-05,
+      "loss": 2.2936,
+      "step": 44392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.552691763847384e-05,
+      "loss": 2.2907,
+      "step": 44400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.549878696867196e-05,
+      "loss": 2.3125,
+      "step": 44408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.54706577532596e-05,
+      "loss": 2.3076,
+      "step": 44416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.544252999505796e-05,
+      "loss": 2.2936,
+      "step": 44424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.541440369688803e-05,
+      "loss": 2.3006,
+      "step": 44432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.538627886157081e-05,
+      "loss": 2.3292,
+      "step": 44440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.535815549192697e-05,
+      "loss": 2.2957,
+      "step": 44448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.533003359077721e-05,
+      "loss": 2.3082,
+      "step": 44456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.530191316094196e-05,
+      "loss": 2.2839,
+      "step": 44464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.52737942052415e-05,
+      "loss": 2.302,
+      "step": 44472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.524567672649605e-05,
+      "loss": 2.307,
+      "step": 44480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.521756072752562e-05,
+      "loss": 2.2939,
+      "step": 44488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.518944621115009e-05,
+      "loss": 2.3133,
+      "step": 44496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.516133318018919e-05,
+      "loss": 2.3084,
+      "step": 44504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.513322163746248e-05,
+      "loss": 2.2941,
+      "step": 44512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.510511158578943e-05,
+      "loss": 2.2944,
+      "step": 44520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.507700302798925e-05,
+      "loss": 2.2914,
+      "step": 44528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.504889596688113e-05,
+      "loss": 2.2982,
+      "step": 44536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.502079040528398e-05,
+      "loss": 2.2999,
+      "step": 44544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.499268634601675e-05,
+      "loss": 2.3109,
+      "step": 44552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.496458379189804e-05,
+      "loss": 2.2933,
+      "step": 44560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.493648274574632e-05,
+      "loss": 2.284,
+      "step": 44568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.490838321038005e-05,
+      "loss": 2.2871,
+      "step": 44576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.488028518861742e-05,
+      "loss": 2.3025,
+      "step": 44584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.485218868327654e-05,
+      "loss": 2.3163,
+      "step": 44592
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.482409369717533e-05,
+      "loss": 2.3007,
+      "step": 44600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.479600023313144e-05,
+      "loss": 2.2959,
+      "step": 44608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.476790829396265e-05,
+      "loss": 2.2752,
+      "step": 44616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.473981788248632e-05,
+      "loss": 2.3065,
+      "step": 44624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.471172900151982e-05,
+      "loss": 2.3096,
+      "step": 44632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.468364165388024e-05,
+      "loss": 2.3138,
+      "step": 44640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.465555584238459e-05,
+      "loss": 2.3157,
+      "step": 44648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.46274715698498e-05,
+      "loss": 2.286,
+      "step": 44656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.459938883909246e-05,
+      "loss": 2.304,
+      "step": 44664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.457130765292919e-05,
+      "loss": 2.3226,
+      "step": 44672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.45432280141763e-05,
+      "loss": 2.2986,
+      "step": 44680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.451514992565009e-05,
+      "loss": 2.2969,
+      "step": 44688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.44870733901666e-05,
+      "loss": 2.2838,
+      "step": 44696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.445899841054174e-05,
+      "loss": 2.3107,
+      "step": 44704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.443092498959127e-05,
+      "loss": 2.3009,
+      "step": 44712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.44028531301308e-05,
+      "loss": 2.2919,
+      "step": 44720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.437478283497583e-05,
+      "loss": 2.2823,
+      "step": 44728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.434671410694156e-05,
+      "loss": 2.3057,
+      "step": 44736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.431864694884316e-05,
+      "loss": 2.3022,
+      "step": 44744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.429058136349564e-05,
+      "loss": 2.328,
+      "step": 44752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.426251735371378e-05,
+      "loss": 2.2939,
+      "step": 44760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.423445492231228e-05,
+      "loss": 2.3189,
+      "step": 44768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.420639407210561e-05,
+      "loss": 2.2837,
+      "step": 44776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.41783348059081e-05,
+      "loss": 2.3248,
+      "step": 44784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.415027712653399e-05,
+      "loss": 2.2949,
+      "step": 44792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.412222103679724e-05,
+      "loss": 2.308,
+      "step": 44800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.409416653951178e-05,
+      "loss": 2.2958,
+      "step": 44808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.40661136374913e-05,
+      "loss": 2.283,
+      "step": 44816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.40380623335493e-05,
+      "loss": 2.2944,
+      "step": 44824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.401001263049924e-05,
+      "loss": 2.2861,
+      "step": 44832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.398196453115427e-05,
+      "loss": 2.3139,
+      "step": 44840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.395391803832749e-05,
+      "loss": 2.3026,
+      "step": 44848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.392587315483184e-05,
+      "loss": 2.29,
+      "step": 44856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.389782988348001e-05,
+      "loss": 2.2968,
+      "step": 44864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.386978822708463e-05,
+      "loss": 2.2976,
+      "step": 44872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.384174818845806e-05,
+      "loss": 2.2958,
+      "step": 44880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.381370977041264e-05,
+      "loss": 2.2975,
+      "step": 44888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.378567297576035e-05,
+      "loss": 2.2957,
+      "step": 44896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.375763780731323e-05,
+      "loss": 2.2832,
+      "step": 44904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.3729604267883e-05,
+      "loss": 2.2923,
+      "step": 44912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.370157236028123e-05,
+      "loss": 2.3051,
+      "step": 44920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.367354208731945e-05,
+      "loss": 2.3089,
+      "step": 44928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.364551345180885e-05,
+      "loss": 2.2897,
+      "step": 44936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.361748645656059e-05,
+      "loss": 2.2766,
+      "step": 44944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.358946110438565e-05,
+      "loss": 2.3074,
+      "step": 44952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.356143739809471e-05,
+      "loss": 2.2981,
+      "step": 44960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.353341534049849e-05,
+      "loss": 2.2768,
+      "step": 44968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.35053949344074e-05,
+      "loss": 2.2961,
+      "step": 44976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.347737618263173e-05,
+      "loss": 2.2824,
+      "step": 44984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.344935908798162e-05,
+      "loss": 2.2931,
+      "step": 44992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.3421343653267e-05,
+      "loss": 2.2919,
+      "step": 45000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.339332988129766e-05,
+      "loss": 2.2981,
+      "step": 45008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.336531777488324e-05,
+      "loss": 2.2888,
+      "step": 45016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.33373073368332e-05,
+      "loss": 2.2865,
+      "step": 45024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.33092985699568e-05,
+      "loss": 2.2976,
+      "step": 45032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.328129147706323e-05,
+      "loss": 2.2725,
+      "step": 45040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.325328606096136e-05,
+      "loss": 2.3004,
+      "step": 45048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.322528232445997e-05,
+      "loss": 2.3152,
+      "step": 45056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.319728027036778e-05,
+      "loss": 2.277,
+      "step": 45064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.316927990149311e-05,
+      "loss": 2.3114,
+      "step": 45072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.314128122064431e-05,
+      "loss": 2.2914,
+      "step": 45080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.311328423062953e-05,
+      "loss": 2.2849,
+      "step": 45088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.308528893425659e-05,
+      "loss": 2.2992,
+      "step": 45096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.305729533433336e-05,
+      "loss": 2.2908,
+      "step": 45104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.302930343366739e-05,
+      "loss": 2.2913,
+      "step": 45112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.300131323506614e-05,
+      "loss": 2.3015,
+      "step": 45120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.297332474133684e-05,
+      "loss": 2.2934,
+      "step": 45128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.29453379552866e-05,
+      "loss": 2.288,
+      "step": 45136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.29173528797223e-05,
+      "loss": 2.2725,
+      "step": 45144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.288936951745071e-05,
+      "loss": 2.3112,
+      "step": 45152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.286138787127845e-05,
+      "loss": 2.2873,
+      "step": 45160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.283340794401187e-05,
+      "loss": 2.3076,
+      "step": 45168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.280542973845716e-05,
+      "loss": 2.3029,
+      "step": 45176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.277745325742044e-05,
+      "loss": 2.2746,
+      "step": 45184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.274947850370754e-05,
+      "loss": 2.2837,
+      "step": 45192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.272150548012427e-05,
+      "loss": 2.3092,
+      "step": 45200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.269353418947602e-05,
+      "loss": 2.2926,
+      "step": 45208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.266556463456826e-05,
+      "loss": 2.2868,
+      "step": 45216
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.263759681820616e-05,
+      "loss": 2.3081,
+      "step": 45224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.260963074319469e-05,
+      "loss": 2.2923,
+      "step": 45232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.258166641233874e-05,
+      "loss": 2.2925,
+      "step": 45240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.255370382844292e-05,
+      "loss": 2.3034,
+      "step": 45248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.252574299431176e-05,
+      "loss": 2.2927,
+      "step": 45256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.249778391274962e-05,
+      "loss": 2.2903,
+      "step": 45264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.24698265865605e-05,
+      "loss": 2.2981,
+      "step": 45272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.244187101854847e-05,
+      "loss": 2.2874,
+      "step": 45280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.241391721151729e-05,
+      "loss": 2.3192,
+      "step": 45288
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.238596516827056e-05,
+      "loss": 2.2886,
+      "step": 45296
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.235801489161176e-05,
+      "loss": 2.2944,
+      "step": 45304
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.233006638434403e-05,
+      "loss": 2.2936,
+      "step": 45312
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.230211964927055e-05,
+      "loss": 2.3177,
+      "step": 45320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.22741746891942e-05,
+      "loss": 2.2883,
+      "step": 45328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.224623150691769e-05,
+      "loss": 2.286,
+      "step": 45336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.221829010524354e-05,
+      "loss": 2.2937,
+      "step": 45344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.219035048697413e-05,
+      "loss": 2.2754,
+      "step": 45352
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.21624126549117e-05,
+      "loss": 2.2949,
+      "step": 45360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.213447661185814e-05,
+      "loss": 2.2937,
+      "step": 45368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.21065423606154e-05,
+      "loss": 2.2918,
+      "step": 45376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.207860990398504e-05,
+      "loss": 2.2954,
+      "step": 45384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.205067924476857e-05,
+      "loss": 2.2767,
+      "step": 45392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.202275038576728e-05,
+      "loss": 2.2897,
+      "step": 45400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.199482332978224e-05,
+      "loss": 2.3068,
+      "step": 45408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.19668980796144e-05,
+      "loss": 2.2811,
+      "step": 45416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.19389746380645e-05,
+      "loss": 2.2993,
+      "step": 45424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.191105300793311e-05,
+      "loss": 2.3043,
+      "step": 45432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.188313319202062e-05,
+      "loss": 2.3052,
+      "step": 45440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.185521519312719e-05,
+      "loss": 2.29,
+      "step": 45448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.182729901405287e-05,
+      "loss": 2.286,
+      "step": 45456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.179938465759747e-05,
+      "loss": 2.304,
+      "step": 45464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.177147212656069e-05,
+      "loss": 2.294,
+      "step": 45472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.174356142374196e-05,
+      "loss": 2.2973,
+      "step": 45480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.171565255194054e-05,
+      "loss": 2.3022,
+      "step": 45488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.168774551395563e-05,
+      "loss": 2.3157,
+      "step": 45496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.1659840312586e-05,
+      "loss": 2.3247,
+      "step": 45504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.163193695063054e-05,
+      "loss": 2.2842,
+      "step": 45512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.160403543088769e-05,
+      "loss": 2.2827,
+      "step": 45520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.157613575615585e-05,
+      "loss": 2.2944,
+      "step": 45528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.154823792923321e-05,
+      "loss": 2.2814,
+      "step": 45536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.152034195291771e-05,
+      "loss": 2.2961,
+      "step": 45544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.149244783000726e-05,
+      "loss": 2.2884,
+      "step": 45552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.146455556329938e-05,
+      "loss": 2.3171,
+      "step": 45560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.143666515559157e-05,
+      "loss": 2.2987,
+      "step": 45568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.140877660968108e-05,
+      "loss": 2.3025,
+      "step": 45576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.138088992836488e-05,
+      "loss": 2.2871,
+      "step": 45584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.135300511444e-05,
+      "loss": 2.2779,
+      "step": 45592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.1325122170703e-05,
+      "loss": 2.2873,
+      "step": 45600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.129724109995043e-05,
+      "loss": 2.2971,
+      "step": 45608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.126936190497864e-05,
+      "loss": 2.3215,
+      "step": 45616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.124148458858366e-05,
+      "loss": 2.2876,
+      "step": 45624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.121360915356155e-05,
+      "loss": 2.2955,
+      "step": 45632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.118573560270794e-05,
+      "loss": 2.2958,
+      "step": 45640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.115786393881844e-05,
+      "loss": 2.3108,
+      "step": 45648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.112999416468848e-05,
+      "loss": 2.3128,
+      "step": 45656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.110212628311313e-05,
+      "loss": 2.2928,
+      "step": 45664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.107426029688745e-05,
+      "loss": 2.2802,
+      "step": 45672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.10463962088062e-05,
+      "loss": 2.3134,
+      "step": 45680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.101853402166402e-05,
+      "loss": 2.3189,
+      "step": 45688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.099067373825537e-05,
+      "loss": 2.325,
+      "step": 45696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.096281536137437e-05,
+      "loss": 2.3141,
+      "step": 45704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.093495889381516e-05,
+      "loss": 2.3007,
+      "step": 45712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.090710433837151e-05,
+      "loss": 2.2985,
+      "step": 45720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.087925169783713e-05,
+      "loss": 2.2951,
+      "step": 45728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.085140097500544e-05,
+      "loss": 2.2774,
+      "step": 45736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.082355217266978e-05,
+      "loss": 2.2902,
+      "step": 45744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.079570529362313e-05,
+      "loss": 2.2977,
+      "step": 45752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.076786034065842e-05,
+      "loss": 2.3039,
+      "step": 45760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.074001731656839e-05,
+      "loss": 2.3134,
+      "step": 45768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.071217622414543e-05,
+      "loss": 2.291,
+      "step": 45776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.068433706618193e-05,
+      "loss": 2.3016,
+      "step": 45784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.065649984547e-05,
+      "loss": 2.2941,
+      "step": 45792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.06286645648015e-05,
+      "loss": 2.2863,
+      "step": 45800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.06008312269682e-05,
+      "loss": 2.288,
+      "step": 45808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.057299983476157e-05,
+      "loss": 2.2841,
+      "step": 45816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.054517039097301e-05,
+      "loss": 2.3119,
+      "step": 45824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.051734289839363e-05,
+      "loss": 2.3046,
+      "step": 45832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.048951735981433e-05,
+      "loss": 2.2852,
+      "step": 45840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.046169377802589e-05,
+      "loss": 2.2851,
+      "step": 45848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.043387215581884e-05,
+      "loss": 2.2761,
+      "step": 45856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.04060524959836e-05,
+      "loss": 2.3052,
+      "step": 45864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.037823480131023e-05,
+      "loss": 2.3008,
+      "step": 45872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.035041907458874e-05,
+      "loss": 2.264,
+      "step": 45880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.032260531860888e-05,
+      "loss": 2.284,
+      "step": 45888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.029479353616018e-05,
+      "loss": 2.2915,
+      "step": 45896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.026698373003208e-05,
+      "loss": 2.2846,
+      "step": 45904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.023917590301365e-05,
+      "loss": 2.3085,
+      "step": 45912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.021137005789394e-05,
+      "loss": 2.2704,
+      "step": 45920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.018356619746166e-05,
+      "loss": 2.2944,
+      "step": 45928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.015576432450541e-05,
+      "loss": 2.2966,
+      "step": 45936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.012796444181357e-05,
+      "loss": 2.2991,
+      "step": 45944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.010016655217428e-05,
+      "loss": 2.2947,
+      "step": 45952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.007237065837552e-05,
+      "loss": 2.2893,
+      "step": 45960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.00445767632051e-05,
+      "loss": 2.3064,
+      "step": 45968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.001678486945053e-05,
+      "loss": 2.3045,
+      "step": 45976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.99889949798992e-05,
+      "loss": 2.3081,
+      "step": 45984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.996120709733829e-05,
+      "loss": 2.2962,
+      "step": 45992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.99334212245548e-05,
+      "loss": 2.2801,
+      "step": 46000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.990563736433541e-05,
+      "loss": 2.2993,
+      "step": 46008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.987785551946675e-05,
+      "loss": 2.2789,
+      "step": 46016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.985007569273516e-05,
+      "loss": 2.2743,
+      "step": 46024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.982229788692678e-05,
+      "loss": 2.291,
+      "step": 46032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.979452210482763e-05,
+      "loss": 2.2813,
+      "step": 46040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.976674834922339e-05,
+      "loss": 2.3083,
+      "step": 46048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.973897662289966e-05,
+      "loss": 2.3035,
+      "step": 46056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.971120692864179e-05,
+      "loss": 2.2905,
+      "step": 46064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.968343926923485e-05,
+      "loss": 2.3,
+      "step": 46072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.965567364746388e-05,
+      "loss": 2.2914,
+      "step": 46080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.962791006611355e-05,
+      "loss": 2.2907,
+      "step": 46088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.960014852796839e-05,
+      "loss": 2.2903,
+      "step": 46096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.957238903581278e-05,
+      "loss": 2.2917,
+      "step": 46104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.954463159243073e-05,
+      "loss": 2.2701,
+      "step": 46112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.95168762006063e-05,
+      "loss": 2.3019,
+      "step": 46120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.948912286312306e-05,
+      "loss": 2.2809,
+      "step": 46128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.946137158276461e-05,
+      "loss": 2.3004,
+      "step": 46136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.943362236231425e-05,
+      "loss": 2.3145,
+      "step": 46144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.940587520455496e-05,
+      "loss": 2.2967,
+      "step": 46152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.937813011226972e-05,
+      "loss": 2.294,
+      "step": 46160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.935038708824117e-05,
+      "loss": 2.2884,
+      "step": 46168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.932264613525183e-05,
+      "loss": 2.3068,
+      "step": 46176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.929490725608389e-05,
+      "loss": 2.2923,
+      "step": 46184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.926717045351942e-05,
+      "loss": 2.2819,
+      "step": 46192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.923943573034031e-05,
+      "loss": 2.3054,
+      "step": 46200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.921170308932811e-05,
+      "loss": 2.2758,
+      "step": 46208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.918397253326433e-05,
+      "loss": 2.2963,
+      "step": 46216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.915624406493016e-05,
+      "loss": 2.2893,
+      "step": 46224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.912851768710657e-05,
+      "loss": 2.285,
+      "step": 46232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.910079340257444e-05,
+      "loss": 2.3097,
+      "step": 46240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.907307121411426e-05,
+      "loss": 2.2888,
+      "step": 46248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.904535112450649e-05,
+      "loss": 2.283,
+      "step": 46256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.901763313653125e-05,
+      "loss": 2.2802,
+      "step": 46264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.898991725296852e-05,
+      "loss": 2.2998,
+      "step": 46272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.896220347659807e-05,
+      "loss": 2.292,
+      "step": 46280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.893449181019935e-05,
+      "loss": 2.3059,
+      "step": 46288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.890678225655178e-05,
+      "loss": 2.2975,
+      "step": 46296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.887907481843438e-05,
+      "loss": 2.2791,
+      "step": 46304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.885136949862616e-05,
+      "loss": 2.2875,
+      "step": 46312
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.882366629990571e-05,
+      "loss": 2.2904,
+      "step": 46320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.879596522505153e-05,
+      "loss": 2.3126,
+      "step": 46328
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.876826627684191e-05,
+      "loss": 2.2879,
+      "step": 46336
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.874056945805487e-05,
+      "loss": 2.2791,
+      "step": 46344
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.871287477146826e-05,
+      "loss": 2.2887,
+      "step": 46352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.868518221985968e-05,
+      "loss": 2.2978,
+      "step": 46360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.865749180600653e-05,
+      "loss": 2.3063,
+      "step": 46368
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.862980353268606e-05,
+      "loss": 2.2812,
+      "step": 46376
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.860211740267516e-05,
+      "loss": 2.2669,
+      "step": 46384
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.857443341875067e-05,
+      "loss": 2.302,
+      "step": 46392
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.854675158368909e-05,
+      "loss": 2.2953,
+      "step": 46400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.851907190026678e-05,
+      "loss": 2.2824,
+      "step": 46408
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.849139437125985e-05,
+      "loss": 2.3172,
+      "step": 46416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.846371899944413e-05,
+      "loss": 2.2976,
+      "step": 46424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.843604578759544e-05,
+      "loss": 2.3026,
+      "step": 46432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.840837473848911e-05,
+      "loss": 2.306,
+      "step": 46440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.838070585490048e-05,
+      "loss": 2.2931,
+      "step": 46448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.835303913960455e-05,
+      "loss": 2.2862,
+      "step": 46456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.832537459537611e-05,
+      "loss": 2.3008,
+      "step": 46464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.829771222498981e-05,
+      "loss": 2.2791,
+      "step": 46472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.827005203121998e-05,
+      "loss": 2.2934,
+      "step": 46480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.824239401684083e-05,
+      "loss": 2.2906,
+      "step": 46488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.821473818462627e-05,
+      "loss": 2.3209,
+      "step": 46496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.818708453735001e-05,
+      "loss": 2.3005,
+      "step": 46504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.815943307778561e-05,
+      "loss": 2.3194,
+      "step": 46512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.813178380870625e-05,
+      "loss": 2.2927,
+      "step": 46520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.810413673288511e-05,
+      "loss": 2.2937,
+      "step": 46528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.807649185309498e-05,
+      "loss": 2.2878,
+      "step": 46536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.804884917210849e-05,
+      "loss": 2.2774,
+      "step": 46544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.802120869269805e-05,
+      "loss": 2.2864,
+      "step": 46552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.799357041763581e-05,
+      "loss": 2.2788,
+      "step": 46560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.79659343496938e-05,
+      "loss": 2.293,
+      "step": 46568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.793830049164369e-05,
+      "loss": 2.2979,
+      "step": 46576
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.791066884625707e-05,
+      "loss": 2.2982,
+      "step": 46584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.788303941630518e-05,
+      "loss": 2.2859,
+      "step": 46592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.78554122045591e-05,
+      "loss": 2.2981,
+      "step": 46600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.782778721378973e-05,
+      "loss": 2.2858,
+      "step": 46608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.780016444676764e-05,
+      "loss": 2.2771,
+      "step": 46616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.777254390626329e-05,
+      "loss": 2.3085,
+      "step": 46624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.774492559504683e-05,
+      "loss": 2.2926,
+      "step": 46632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.77173095158882e-05,
+      "loss": 2.2986,
+      "step": 46640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.76896956715572e-05,
+      "loss": 2.2903,
+      "step": 46648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.766208406482327e-05,
+      "loss": 2.3169,
+      "step": 46656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.763447469845576e-05,
+      "loss": 2.2932,
+      "step": 46664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.760686757522374e-05,
+      "loss": 2.2838,
+      "step": 46672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.757926269789596e-05,
+      "loss": 2.2914,
+      "step": 46680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.755166006924112e-05,
+      "loss": 2.2918,
+      "step": 46688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.752405969202755e-05,
+      "loss": 2.2939,
+      "step": 46696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.749646156902347e-05,
+      "loss": 2.3129,
+      "step": 46704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.74688657029968e-05,
+      "loss": 2.2885,
+      "step": 46712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.744127209671517e-05,
+      "loss": 2.3108,
+      "step": 46720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.741368075294616e-05,
+      "loss": 2.2734,
+      "step": 46728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.738609167445695e-05,
+      "loss": 2.2924,
+      "step": 46736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.735850486401466e-05,
+      "loss": 2.2828,
+      "step": 46744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.733092032438601e-05,
+      "loss": 2.2692,
+      "step": 46752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.730333805833759e-05,
+      "loss": 2.2952,
+      "step": 46760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.727575806863581e-05,
+      "loss": 2.3156,
+      "step": 46768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.724818035804667e-05,
+      "loss": 2.2924,
+      "step": 46776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.722060492933616e-05,
+      "loss": 2.2943,
+      "step": 46784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.71930317852699e-05,
+      "loss": 2.2949,
+      "step": 46792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.716546092861331e-05,
+      "loss": 2.3124,
+      "step": 46800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.71378923621316e-05,
+      "loss": 2.2969,
+      "step": 46808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.711032608858974e-05,
+      "loss": 2.2802,
+      "step": 46816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.708276211075248e-05,
+      "loss": 2.2939,
+      "step": 46824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.705520043138431e-05,
+      "loss": 2.2812,
+      "step": 46832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.702764105324954e-05,
+      "loss": 2.3016,
+      "step": 46840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.700008397911221e-05,
+      "loss": 2.2774,
+      "step": 46848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.697252921173611e-05,
+      "loss": 2.2791,
+      "step": 46856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.694497675388487e-05,
+      "loss": 2.2931,
+      "step": 46864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.691742660832178e-05,
+      "loss": 2.2903,
+      "step": 46872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.688987877781007e-05,
+      "loss": 2.2856,
+      "step": 46880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.686233326511255e-05,
+      "loss": 2.3065,
+      "step": 46888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.683479007299187e-05,
+      "loss": 2.3015,
+      "step": 46896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.680724920421053e-05,
+      "loss": 2.2848,
+      "step": 46904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.677971066153063e-05,
+      "loss": 2.2821,
+      "step": 46912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.675217444771422e-05,
+      "loss": 2.3013,
+      "step": 46920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.672464056552294e-05,
+      "loss": 2.29,
+      "step": 46928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.669710901771834e-05,
+      "loss": 2.2922,
+      "step": 46936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.666957980706172e-05,
+      "loss": 2.2894,
+      "step": 46944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.664205293631397e-05,
+      "loss": 2.3048,
+      "step": 46952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.661452840823598e-05,
+      "loss": 2.2729,
+      "step": 46960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.658700622558829e-05,
+      "loss": 2.2823,
+      "step": 46968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.655948639113122e-05,
+      "loss": 2.3191,
+      "step": 46976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.653196890762486e-05,
+      "loss": 2.3,
+      "step": 46984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.650445377782899e-05,
+      "loss": 2.2593,
+      "step": 46992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.64769410045033e-05,
+      "loss": 2.2855,
+      "step": 47000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.644943059040713e-05,
+      "loss": 2.2902,
+      "step": 47008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.642192253829964e-05,
+      "loss": 2.2908,
+      "step": 47016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.639441685093973e-05,
+      "loss": 2.2773,
+      "step": 47024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.636691353108602e-05,
+      "loss": 2.2824,
+      "step": 47032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.633941258149698e-05,
+      "loss": 2.2896,
+      "step": 47040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.631191400493079e-05,
+      "loss": 2.2833,
+      "step": 47048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.628441780414542e-05,
+      "loss": 2.275,
+      "step": 47056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.625692398189856e-05,
+      "loss": 2.2652,
+      "step": 47064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.622943254094766e-05,
+      "loss": 2.289,
+      "step": 47072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.620194348405003e-05,
+      "loss": 2.2838,
+      "step": 47080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.617445681396257e-05,
+      "loss": 2.3225,
+      "step": 47088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.614697253344211e-05,
+      "loss": 2.3051,
+      "step": 47096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.611949064524514e-05,
+      "loss": 2.287,
+      "step": 47104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.609201115212797e-05,
+      "loss": 2.2978,
+      "step": 47112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.606453405684662e-05,
+      "loss": 2.307,
+      "step": 47120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.603705936215682e-05,
+      "loss": 2.3043,
+      "step": 47128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.600958707081424e-05,
+      "loss": 2.2614,
+      "step": 47136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.59821171855741e-05,
+      "loss": 2.281,
+      "step": 47144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.595464970919154e-05,
+      "loss": 2.2977,
+      "step": 47152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.592718464442135e-05,
+      "loss": 2.286,
+      "step": 47160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.589972199401812e-05,
+      "loss": 2.2882,
+      "step": 47168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.587226176073626e-05,
+      "loss": 2.2814,
+      "step": 47176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.584480394732978e-05,
+      "loss": 2.2769,
+      "step": 47184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.581734855655266e-05,
+      "loss": 2.2947,
+      "step": 47192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.578989559115842e-05,
+      "loss": 2.2816,
+      "step": 47200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.576244505390045e-05,
+      "loss": 2.2802,
+      "step": 47208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.573499694753194e-05,
+      "loss": 2.2707,
+      "step": 47216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.57075512748057e-05,
+      "loss": 2.281,
+      "step": 47224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.568010803847444e-05,
+      "loss": 2.2832,
+      "step": 47232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.565266724129055e-05,
+      "loss": 2.2764,
+      "step": 47240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.562522888600614e-05,
+      "loss": 2.3093,
+      "step": 47248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.559779297537323e-05,
+      "loss": 2.2952,
+      "step": 47256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.557035951214334e-05,
+      "loss": 2.2961,
+      "step": 47264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.554292849906797e-05,
+      "loss": 2.2849,
+      "step": 47272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.551549993889831e-05,
+      "loss": 2.3083,
+      "step": 47280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.548807383438528e-05,
+      "loss": 2.2844,
+      "step": 47288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.546065018827956e-05,
+      "loss": 2.2628,
+      "step": 47296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.543322900333152e-05,
+      "loss": 2.2998,
+      "step": 47304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.540581028229149e-05,
+      "loss": 2.3172,
+      "step": 47312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.537839402790928e-05,
+      "loss": 2.2872,
+      "step": 47320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.535098024293466e-05,
+      "loss": 2.2958,
+      "step": 47328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.532356893011704e-05,
+      "loss": 2.2943,
+      "step": 47336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.52961600922056e-05,
+      "loss": 2.2966,
+      "step": 47344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.526875373194938e-05,
+      "loss": 2.2766,
+      "step": 47352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.524134985209698e-05,
+      "loss": 2.2856,
+      "step": 47360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.521394845539692e-05,
+      "loss": 2.2947,
+      "step": 47368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.518654954459739e-05,
+      "loss": 2.3038,
+      "step": 47376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.515915312244631e-05,
+      "loss": 2.2959,
+      "step": 47384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.513175919169147e-05,
+      "loss": 2.2896,
+      "step": 47392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.510436775508022e-05,
+      "loss": 2.2947,
+      "step": 47400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.507697881535983e-05,
+      "loss": 2.3049,
+      "step": 47408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.50495923752773e-05,
+      "loss": 2.2986,
+      "step": 47416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.50222084375792e-05,
+      "loss": 2.2964,
+      "step": 47424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.499482700501209e-05,
+      "loss": 2.2985,
+      "step": 47432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.496744808032213e-05,
+      "loss": 2.3122,
+      "step": 47440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.494007166625531e-05,
+      "loss": 2.3028,
+      "step": 47448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.491269776555728e-05,
+      "loss": 2.2938,
+      "step": 47456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.488532638097353e-05,
+      "loss": 2.2895,
+      "step": 47464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.485795751524925e-05,
+      "loss": 2.2876,
+      "step": 47472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.483059117112934e-05,
+      "loss": 2.2789,
+      "step": 47480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.480322735135855e-05,
+      "loss": 2.275,
+      "step": 47488
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.477586605868125e-05,
+      "loss": 2.296,
+      "step": 47496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.474850729584166e-05,
+      "loss": 2.2917,
+      "step": 47504
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.472115106558374e-05,
+      "loss": 2.282,
+      "step": 47512
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.469379737065108e-05,
+      "loss": 2.2934,
+      "step": 47520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.466644621378718e-05,
+      "loss": 2.2988,
+      "step": 47528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.463909759773516e-05,
+      "loss": 2.2761,
+      "step": 47536
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.461175152523799e-05,
+      "loss": 2.2779,
+      "step": 47544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.458440799903826e-05,
+      "loss": 2.2903,
+      "step": 47552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.455706702187837e-05,
+      "loss": 2.3057,
+      "step": 47560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.452972859650052e-05,
+      "loss": 2.2912,
+      "step": 47568
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.450239272564656e-05,
+      "loss": 2.3204,
+      "step": 47576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.447505941205814e-05,
+      "loss": 2.2923,
+      "step": 47584
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.444772865847661e-05,
+      "loss": 2.2888,
+      "step": 47592
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.44204004676431e-05,
+      "loss": 2.2744,
+      "step": 47600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.439307484229853e-05,
+      "loss": 2.2943,
+      "step": 47608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.43657517851834e-05,
+      "loss": 2.2959,
+      "step": 47616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.433843129903818e-05,
+      "loss": 2.2875,
+      "step": 47624
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.431111338660283e-05,
+      "loss": 2.2976,
+      "step": 47632
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.428379805061725e-05,
+      "loss": 2.2787,
+      "step": 47640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.425648529382107e-05,
+      "loss": 2.2982,
+      "step": 47648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.42291751189535e-05,
+      "loss": 2.2897,
+      "step": 47656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.420186752875363e-05,
+      "loss": 2.2981,
+      "step": 47664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.417456252596027e-05,
+      "loss": 2.2776,
+      "step": 47672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.414726011331197e-05,
+      "loss": 2.2887,
+      "step": 47680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.411996029354703e-05,
+      "loss": 2.2873,
+      "step": 47688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.409266306940336e-05,
+      "loss": 2.2809,
+      "step": 47696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.406536844361881e-05,
+      "loss": 2.2981,
+      "step": 47704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.403807641893082e-05,
+      "loss": 2.2948,
+      "step": 47712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.401078699807672e-05,
+      "loss": 2.2959,
+      "step": 47720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.398350018379339e-05,
+      "loss": 2.2763,
+      "step": 47728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.395621597881756e-05,
+      "loss": 2.2872,
+      "step": 47736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.392893438588572e-05,
+      "loss": 2.286,
+      "step": 47744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.390165540773401e-05,
+      "loss": 2.2588,
+      "step": 47752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.38743790470984e-05,
+      "loss": 2.2955,
+      "step": 47760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.384710530671453e-05,
+      "loss": 2.2967,
+      "step": 47768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.38198341893178e-05,
+      "loss": 2.2893,
+      "step": 47776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.379256569764338e-05,
+      "loss": 2.2892,
+      "step": 47784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.37652998344261e-05,
+      "loss": 2.293,
+      "step": 47792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.373803660240058e-05,
+      "loss": 2.3,
+      "step": 47800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.371077600430118e-05,
+      "loss": 2.2913,
+      "step": 47808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.3683518042862e-05,
+      "loss": 2.3001,
+      "step": 47816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.365626272081688e-05,
+      "loss": 2.3142,
+      "step": 47824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.362901004089927e-05,
+      "loss": 2.282,
+      "step": 47832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.360176000584256e-05,
+      "loss": 2.2974,
+      "step": 47840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.357451261837975e-05,
+      "loss": 2.2852,
+      "step": 47848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.354726788124359e-05,
+      "loss": 2.2567,
+      "step": 47856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.352002579716655e-05,
+      "loss": 2.299,
+      "step": 47864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.349278636888087e-05,
+      "loss": 2.2992,
+      "step": 47872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.346554959911856e-05,
+      "loss": 2.2775,
+      "step": 47880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.343831549061123e-05,
+      "loss": 2.2681,
+      "step": 47888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.341108404609039e-05,
+      "loss": 2.282,
+      "step": 47896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.338385526828714e-05,
+      "loss": 2.29,
+      "step": 47904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.335662915993237e-05,
+      "loss": 2.2939,
+      "step": 47912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.332940572375676e-05,
+      "loss": 2.2872,
+      "step": 47920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.33021849624906e-05,
+      "loss": 2.2936,
+      "step": 47928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.327496687886403e-05,
+      "loss": 2.2779,
+      "step": 47936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.324775147560686e-05,
+      "loss": 2.2987,
+      "step": 47944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.32205387554486e-05,
+      "loss": 2.2954,
+      "step": 47952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.31933287211186e-05,
+      "loss": 2.2972,
+      "step": 47960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.31661213753458e-05,
+      "loss": 2.3026,
+      "step": 47968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.313891672085899e-05,
+      "loss": 2.2858,
+      "step": 47976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.311171476038662e-05,
+      "loss": 2.2796,
+      "step": 47984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.308451549665692e-05,
+      "loss": 2.2926,
+      "step": 47992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.305731893239781e-05,
+      "loss": 2.2662,
+      "step": 48000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.303012507033692e-05,
+      "loss": 2.306,
+      "step": 48008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.300293391320169e-05,
+      "loss": 2.3121,
+      "step": 48016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.297574546371918e-05,
+      "loss": 2.2752,
+      "step": 48024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.29485597246163e-05,
+      "loss": 2.2915,
+      "step": 48032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.292137669861962e-05,
+      "loss": 2.299,
+      "step": 48040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.289419638845535e-05,
+      "loss": 2.2983,
+      "step": 48048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.286701879684966e-05,
+      "loss": 2.2676,
+      "step": 48056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.283984392652821e-05,
+      "loss": 2.3034,
+      "step": 48064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.281267178021652e-05,
+      "loss": 2.3008,
+      "step": 48072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.278550236063982e-05,
+      "loss": 2.307,
+      "step": 48080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.275833567052298e-05,
+      "loss": 2.305,
+      "step": 48088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.273117171259077e-05,
+      "loss": 2.2898,
+      "step": 48096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.270401048956748e-05,
+      "loss": 2.2883,
+      "step": 48104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.267685200417727e-05,
+      "loss": 2.2793,
+      "step": 48112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.264969625914402e-05,
+      "loss": 2.2816,
+      "step": 48120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.262254325719121e-05,
+      "loss": 2.2954,
+      "step": 48128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.259539300104222e-05,
+      "loss": 2.2756,
+      "step": 48136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.256824549342e-05,
+      "loss": 2.276,
+      "step": 48144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.254110073704735e-05,
+      "loss": 2.28,
+      "step": 48152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.251395873464669e-05,
+      "loss": 2.3125,
+      "step": 48160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.248681948894023e-05,
+      "loss": 2.2883,
+      "step": 48168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.24596830026499e-05,
+      "loss": 2.2917,
+      "step": 48176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.243254927849727e-05,
+      "loss": 2.2725,
+      "step": 48184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.240541831920376e-05,
+      "loss": 2.3109,
+      "step": 48192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.237829012749041e-05,
+      "loss": 2.2707,
+      "step": 48200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.235116470607809e-05,
+      "loss": 2.2996,
+      "step": 48208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.232404205768727e-05,
+      "loss": 2.2963,
+      "step": 48216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.22969221850382e-05,
+      "loss": 2.2683,
+      "step": 48224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.226980509085087e-05,
+      "loss": 2.2854,
+      "step": 48232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.224269077784492e-05,
+      "loss": 2.2784,
+      "step": 48240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.221557924873986e-05,
+      "loss": 2.2799,
+      "step": 48248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.218847050625476e-05,
+      "loss": 2.2843,
+      "step": 48256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.216136455310843e-05,
+      "loss": 2.3051,
+      "step": 48264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.213426139201956e-05,
+      "loss": 2.2747,
+      "step": 48272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.210716102570632e-05,
+      "loss": 2.278,
+      "step": 48280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.208006345688681e-05,
+      "loss": 2.2908,
+      "step": 48288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.205296868827877e-05,
+      "loss": 2.3108,
+      "step": 48296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.202587672259951e-05,
+      "loss": 2.2913,
+      "step": 48304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.199878756256639e-05,
+      "loss": 2.2628,
+      "step": 48312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.197170121089617e-05,
+      "loss": 2.3054,
+      "step": 48320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.194461767030554e-05,
+      "loss": 2.2987,
+      "step": 48328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.191753694351074e-05,
+      "loss": 2.2575,
+      "step": 48336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.18904590332279e-05,
+      "loss": 2.2703,
+      "step": 48344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.186338394217274e-05,
+      "loss": 2.2846,
+      "step": 48352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.183631167306071e-05,
+      "loss": 2.28,
+      "step": 48360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.180924222860706e-05,
+      "loss": 2.2915,
+      "step": 48368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.178217561152666e-05,
+      "loss": 2.2846,
+      "step": 48376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.175511182453418e-05,
+      "loss": 2.2694,
+      "step": 48384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.172805087034394e-05,
+      "loss": 2.2944,
+      "step": 48392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.170099275166995e-05,
+      "loss": 2.2867,
+      "step": 48400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.16739374712261e-05,
+      "loss": 2.2867,
+      "step": 48408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.164688503172575e-05,
+      "loss": 2.2918,
+      "step": 48416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.161983543588225e-05,
+      "loss": 2.2513,
+      "step": 48424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.159278868640838e-05,
+      "loss": 2.2913,
+      "step": 48432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.156574478601686e-05,
+      "loss": 2.2958,
+      "step": 48440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.153870373742005e-05,
+      "loss": 2.287,
+      "step": 48448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.151166554332995e-05,
+      "loss": 2.2893,
+      "step": 48456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.14846302064584e-05,
+      "loss": 2.2622,
+      "step": 48464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.145759772951685e-05,
+      "loss": 2.3132,
+      "step": 48472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.14305681152165e-05,
+      "loss": 2.2758,
+      "step": 48480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.140354136626831e-05,
+      "loss": 2.3069,
+      "step": 48488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.137651748538286e-05,
+      "loss": 2.2954,
+      "step": 48496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.134949647527053e-05,
+      "loss": 2.2896,
+      "step": 48504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.132247833864135e-05,
+      "loss": 2.2764,
+      "step": 48512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.129546307820512e-05,
+      "loss": 2.3077,
+      "step": 48520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.12684506966713e-05,
+      "loss": 2.2945,
+      "step": 48528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.124144119674903e-05,
+      "loss": 2.2856,
+      "step": 48536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.12144345811473e-05,
+      "loss": 2.2969,
+      "step": 48544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.118743085257463e-05,
+      "loss": 2.2788,
+      "step": 48552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.116043001373944e-05,
+      "loss": 2.2875,
+      "step": 48560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.113343206734967e-05,
+      "loss": 2.2855,
+      "step": 48568
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.11064370161131e-05,
+      "loss": 2.2914,
+      "step": 48576
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.107944486273717e-05,
+      "loss": 2.2845,
+      "step": 48584
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.105245560992904e-05,
+      "loss": 2.2944,
+      "step": 48592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.102546926039563e-05,
+      "loss": 2.2795,
+      "step": 48600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.099848581684345e-05,
+      "loss": 2.2873,
+      "step": 48608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.097150528197878e-05,
+      "loss": 2.2982,
+      "step": 48616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.094452765850768e-05,
+      "loss": 2.2899,
+      "step": 48624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.09175529491358e-05,
+      "loss": 2.2689,
+      "step": 48632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.089058115656858e-05,
+      "loss": 2.304,
+      "step": 48640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.086361228351114e-05,
+      "loss": 2.2851,
+      "step": 48648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.083664633266825e-05,
+      "loss": 2.2791,
+      "step": 48656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.080968330674454e-05,
+      "loss": 2.2945,
+      "step": 48664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.078272320844415e-05,
+      "loss": 2.2925,
+      "step": 48672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.075576604047108e-05,
+      "loss": 2.2793,
+      "step": 48680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.072881180552896e-05,
+      "loss": 2.2949,
+      "step": 48688
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.070186050632119e-05,
+      "loss": 2.2985,
+      "step": 48696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.067491214555079e-05,
+      "loss": 2.3011,
+      "step": 48704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.06479667259205e-05,
+      "loss": 2.2753,
+      "step": 48712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.062102425013287e-05,
+      "loss": 2.3052,
+      "step": 48720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.059408472089e-05,
+      "loss": 2.2969,
+      "step": 48728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.056714814089386e-05,
+      "loss": 2.2744,
+      "step": 48736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.054021451284595e-05,
+      "loss": 2.2753,
+      "step": 48744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.051328383944758e-05,
+      "loss": 2.2866,
+      "step": 48752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.048635612339981e-05,
+      "loss": 2.287,
+      "step": 48760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.045943136740323e-05,
+      "loss": 2.2632,
+      "step": 48768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.043250957415833e-05,
+      "loss": 2.2922,
+      "step": 48776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.040559074636522e-05,
+      "loss": 2.3004,
+      "step": 48784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.03786748867236e-05,
+      "loss": 2.2882,
+      "step": 48792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.035176199793309e-05,
+      "loss": 2.2954,
+      "step": 48800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.03248520826928e-05,
+      "loss": 2.2685,
+      "step": 48808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.029794514370174e-05,
+      "loss": 2.2691,
+      "step": 48816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.027104118365851e-05,
+      "loss": 2.2626,
+      "step": 48824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.024414020526134e-05,
+      "loss": 2.3043,
+      "step": 48832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.021724221120832e-05,
+      "loss": 2.2813,
+      "step": 48840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.019034720419713e-05,
+      "loss": 2.2723,
+      "step": 48848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.016345518692525e-05,
+      "loss": 2.3087,
+      "step": 48856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.013656616208968e-05,
+      "loss": 2.2964,
+      "step": 48864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.01096801323874e-05,
+      "loss": 2.2818,
+      "step": 48872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.008279710051478e-05,
+      "loss": 2.2719,
+      "step": 48880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.005591706916809e-05,
+      "loss": 2.2977,
+      "step": 48888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.002904004104325e-05,
+      "loss": 2.3191,
+      "step": 48896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.000216601883588e-05,
+      "loss": 2.2987,
+      "step": 48904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.997529500524126e-05,
+      "loss": 2.292,
+      "step": 48912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.994842700295447e-05,
+      "loss": 2.2781,
+      "step": 48920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.992156201467011e-05,
+      "loss": 2.2619,
+      "step": 48928
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.989470004308271e-05,
+      "loss": 2.2873,
+      "step": 48936
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.986784109088623e-05,
+      "loss": 2.2732,
+      "step": 48944
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.984098516077461e-05,
+      "loss": 2.2924,
+      "step": 48952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.98141322554413e-05,
+      "loss": 2.2995,
+      "step": 48960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.978728237757942e-05,
+      "loss": 2.2665,
+      "step": 48968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.976043552988194e-05,
+      "loss": 2.2926,
+      "step": 48976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.973359171504144e-05,
+      "loss": 2.311,
+      "step": 48984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.970675093575021e-05,
+      "loss": 2.2715,
+      "step": 48992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.967991319470023e-05,
+      "loss": 2.2823,
+      "step": 49000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.965307849458313e-05,
+      "loss": 2.2758,
+      "step": 49008
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.962624683809028e-05,
+      "loss": 2.2909,
+      "step": 49016
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.959941822791275e-05,
+      "loss": 2.278,
+      "step": 49024
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.957259266674136e-05,
+      "loss": 2.2647,
+      "step": 49032
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.954577015726648e-05,
+      "loss": 2.2941,
+      "step": 49040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.95189507021783e-05,
+      "loss": 2.2716,
+      "step": 49048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.949213430416667e-05,
+      "loss": 2.2819,
+      "step": 49056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.946532096592108e-05,
+      "loss": 2.2656,
+      "step": 49064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.94385106901308e-05,
+      "loss": 2.3214,
+      "step": 49072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.941170347948468e-05,
+      "loss": 2.2943,
+      "step": 49080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.938489933667144e-05,
+      "loss": 2.2697,
+      "step": 49088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.93580982643793e-05,
+      "loss": 2.2941,
+      "step": 49096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.933130026529625e-05,
+      "loss": 2.29,
+      "step": 49104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.930450534211002e-05,
+      "loss": 2.3026,
+      "step": 49112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.927771349750796e-05,
+      "loss": 2.2832,
+      "step": 49120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.925092473417719e-05,
+      "loss": 2.3065,
+      "step": 49128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.922413905480444e-05,
+      "loss": 2.285,
+      "step": 49136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.919735646207609e-05,
+      "loss": 2.285,
+      "step": 49144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.917057695867843e-05,
+      "loss": 2.2637,
+      "step": 49152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.914380054729715e-05,
+      "loss": 2.2922,
+      "step": 49160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.911702723061787e-05,
+      "loss": 2.299,
+      "step": 49168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.909025701132577e-05,
+      "loss": 2.2707,
+      "step": 49176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.906348989210571e-05,
+      "loss": 2.3058,
+      "step": 49184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.903672587564237e-05,
+      "loss": 2.2814,
+      "step": 49192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.900996496461992e-05,
+      "loss": 2.2585,
+      "step": 49200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.898320716172243e-05,
+      "loss": 2.2833,
+      "step": 49208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.895645246963348e-05,
+      "loss": 2.2823,
+      "step": 49216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.892970089103646e-05,
+      "loss": 2.3014,
+      "step": 49224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.890295242861441e-05,
+      "loss": 2.2855,
+      "step": 49232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.887620708504999e-05,
+      "loss": 2.2654,
+      "step": 49240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.884946486302564e-05,
+      "loss": 2.2692,
+      "step": 49248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.882272576522344e-05,
+      "loss": 2.2812,
+      "step": 49256
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.879598979432526e-05,
+      "loss": 2.2812,
+      "step": 49264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.876925695301245e-05,
+      "loss": 2.2949,
+      "step": 49272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.874252724396618e-05,
+      "loss": 2.2824,
+      "step": 49280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.871580066986734e-05,
+      "loss": 2.2542,
+      "step": 49288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.868907723339643e-05,
+      "loss": 2.2893,
+      "step": 49296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.866235693723365e-05,
+      "loss": 2.2784,
+      "step": 49304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.86356397840589e-05,
+      "loss": 2.2868,
+      "step": 49312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.860892577655175e-05,
+      "loss": 2.2769,
+      "step": 49320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.858221491739149e-05,
+      "loss": 2.2822,
+      "step": 49328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.855550720925701e-05,
+      "loss": 2.2797,
+      "step": 49336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.852880265482702e-05,
+      "loss": 2.2764,
+      "step": 49344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.850210125677982e-05,
+      "loss": 2.2706,
+      "step": 49352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.847540301779333e-05,
+      "loss": 2.2817,
+      "step": 49360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.844870794054536e-05,
+      "loss": 2.2736,
+      "step": 49368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.842201602771314e-05,
+      "loss": 2.2831,
+      "step": 49376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.839532728197379e-05,
+      "loss": 2.2632,
+      "step": 49384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.836864170600401e-05,
+      "loss": 2.313,
+      "step": 49392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.83419593024803e-05,
+      "loss": 2.2779,
+      "step": 49400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.831528007407865e-05,
+      "loss": 2.2927,
+      "step": 49408
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.828860402347487e-05,
+      "loss": 2.2876,
+      "step": 49416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.826193115334444e-05,
+      "loss": 2.2898,
+      "step": 49424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.823526146636246e-05,
+      "loss": 2.2773,
+      "step": 49432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.820859496520378e-05,
+      "loss": 2.2946,
+      "step": 49440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.818193165254289e-05,
+      "loss": 2.2693,
+      "step": 49448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.815527153105393e-05,
+      "loss": 2.3034,
+      "step": 49456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.812861460341083e-05,
+      "loss": 2.2945,
+      "step": 49464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.810196087228705e-05,
+      "loss": 2.2733,
+      "step": 49472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.807531034035589e-05,
+      "loss": 2.2989,
+      "step": 49480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.804866301029022e-05,
+      "loss": 2.2804,
+      "step": 49488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.802201888476255e-05,
+      "loss": 2.2679,
+      "step": 49496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.799537796644525e-05,
+      "loss": 2.2695,
+      "step": 49504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.796874025801013e-05,
+      "loss": 2.3016,
+      "step": 49512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.79421057621289e-05,
+      "loss": 2.2896,
+      "step": 49520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.79154744814728e-05,
+      "loss": 2.2851,
+      "step": 49528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.788884641871281e-05,
+      "loss": 2.2932,
+      "step": 49536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.786222157651959e-05,
+      "loss": 2.2895,
+      "step": 49544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.78355999575634e-05,
+      "loss": 2.2882,
+      "step": 49552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.780898156451433e-05,
+      "loss": 2.2759,
+      "step": 49560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.778236640004197e-05,
+      "loss": 2.2738,
+      "step": 49568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.775575446681574e-05,
+      "loss": 2.3054,
+      "step": 49576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.772914576750461e-05,
+      "loss": 2.2918,
+      "step": 49584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.770254030477729e-05,
+      "loss": 2.2742,
+      "step": 49592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.767593808130218e-05,
+      "loss": 2.2668,
+      "step": 49600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.764933909974732e-05,
+      "loss": 2.304,
+      "step": 49608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.762274336278044e-05,
+      "loss": 2.2829,
+      "step": 49616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.759615087306897e-05,
+      "loss": 2.3082,
+      "step": 49624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.756956163327991e-05,
+      "loss": 2.283,
+      "step": 49632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.75429756460801e-05,
+      "loss": 2.2905,
+      "step": 49640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.751639291413589e-05,
+      "loss": 2.2738,
+      "step": 49648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.748981344011341e-05,
+      "loss": 2.2819,
+      "step": 49656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.746323722667846e-05,
+      "loss": 2.2842,
+      "step": 49664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.743666427649641e-05,
+      "loss": 2.2713,
+      "step": 49672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.741009459223244e-05,
+      "loss": 2.2873,
+      "step": 49680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.73835281765513e-05,
+      "loss": 2.2688,
+      "step": 49688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.735696503211751e-05,
+      "loss": 2.2603,
+      "step": 49696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.733040516159516e-05,
+      "loss": 2.2864,
+      "step": 49704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.730384856764801e-05,
+      "loss": 2.2808,
+      "step": 49712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.727729525293961e-05,
+      "loss": 2.2821,
+      "step": 49720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.725074522013305e-05,
+      "loss": 2.2922,
+      "step": 49728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.722419847189122e-05,
+      "loss": 2.2855,
+      "step": 49736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.719765501087652e-05,
+      "loss": 2.2881,
+      "step": 49744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.717111483975119e-05,
+      "loss": 2.2889,
+      "step": 49752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.714457796117705e-05,
+      "loss": 2.26,
+      "step": 49760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.711804437781551e-05,
+      "loss": 2.2757,
+      "step": 49768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.709151409232784e-05,
+      "loss": 2.2863,
+      "step": 49776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.70649871073748e-05,
+      "loss": 2.3026,
+      "step": 49784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.703846342561696e-05,
+      "loss": 2.2784,
+      "step": 49792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.701194304971449e-05,
+      "loss": 2.2677,
+      "step": 49800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.698542598232717e-05,
+      "loss": 2.2586,
+      "step": 49808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.695891222611456e-05,
+      "loss": 2.2885,
+      "step": 49816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.693240178373581e-05,
+      "loss": 2.2854,
+      "step": 49824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.690589465784985e-05,
+      "loss": 2.2772,
+      "step": 49832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.687939085111512e-05,
+      "loss": 2.2681,
+      "step": 49840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.68528903661898e-05,
+      "loss": 2.291,
+      "step": 49848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.682639320573173e-05,
+      "loss": 2.2708,
+      "step": 49856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.679989937239846e-05,
+      "loss": 2.2966,
+      "step": 49864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.67734088688472e-05,
+      "loss": 2.305,
+      "step": 49872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.674692169773472e-05,
+      "loss": 2.2856,
+      "step": 49880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.672043786171758e-05,
+      "loss": 2.2915,
+      "step": 49888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.669395736345198e-05,
+      "loss": 2.2819,
+      "step": 49896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.666748020559369e-05,
+      "loss": 2.2512,
+      "step": 49904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.664100639079829e-05,
+      "loss": 2.2822,
+      "step": 49912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.661453592172092e-05,
+      "loss": 2.2865,
+      "step": 49920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.65880688010164e-05,
+      "loss": 2.2882,
+      "step": 49928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.656160503133931e-05,
+      "loss": 2.2555,
+      "step": 49936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.653514461534372e-05,
+      "loss": 2.2797,
+      "step": 49944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.65086875556835e-05,
+      "loss": 2.2911,
+      "step": 49952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.648223385501214e-05,
+      "loss": 2.2704,
+      "step": 49960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.645578351598286e-05,
+      "loss": 2.2787,
+      "step": 49968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.64293365412484e-05,
+      "loss": 2.2889,
+      "step": 49976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.640289293346124e-05,
+      "loss": 2.2693,
+      "step": 49984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.637645269527354e-05,
+      "loss": 2.2801,
+      "step": 49992
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.635001582933712e-05,
+      "loss": 2.2737,
+      "step": 50000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.632358233830345e-05,
+      "loss": 2.2853,
+      "step": 50008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.629715222482365e-05,
+      "loss": 2.2999,
+      "step": 50016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.62707254915485e-05,
+      "loss": 2.2692,
+      "step": 50024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.624430214112849e-05,
+      "loss": 2.2808,
+      "step": 50032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.621788217621366e-05,
+      "loss": 2.2785,
+      "step": 50040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.619146559945389e-05,
+      "loss": 2.3031,
+      "step": 50048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.616505241349847e-05,
+      "loss": 2.2807,
+      "step": 50056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.613864262099665e-05,
+      "loss": 2.2801,
+      "step": 50064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.611223622459709e-05,
+      "loss": 2.263,
+      "step": 50072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.60858332269482e-05,
+      "loss": 2.2784,
+      "step": 50080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.605943363069808e-05,
+      "loss": 2.2842,
+      "step": 50088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.603303743849441e-05,
+      "loss": 2.282,
+      "step": 50096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.600664465298466e-05,
+      "loss": 2.2884,
+      "step": 50104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.598025527681585e-05,
+      "loss": 2.2678,
+      "step": 50112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.59538693126346e-05,
+      "loss": 2.2912,
+      "step": 50120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.59274867630874e-05,
+      "loss": 2.2705,
+      "step": 50128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.59011076308202e-05,
+      "loss": 2.2812,
+      "step": 50136
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.587473191847869e-05,
+      "loss": 2.2687,
+      "step": 50144
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.58483596287082e-05,
+      "loss": 2.2838,
+      "step": 50152
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.582199076415371e-05,
+      "loss": 2.291,
+      "step": 50160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.579562532745991e-05,
+      "loss": 2.2623,
+      "step": 50168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.576926332127105e-05,
+      "loss": 2.2933,
+      "step": 50176
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.574290474823116e-05,
+      "loss": 2.2897,
+      "step": 50184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.57165496109838e-05,
+      "loss": 2.2824,
+      "step": 50192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.56901979121722e-05,
+      "loss": 2.2773,
+      "step": 50200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.56638496544394e-05,
+      "loss": 2.2895,
+      "step": 50208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.563750484042789e-05,
+      "loss": 2.2807,
+      "step": 50216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.561116347277994e-05,
+      "loss": 2.2932,
+      "step": 50224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.558482555413743e-05,
+      "loss": 2.2614,
+      "step": 50232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.555849108714192e-05,
+      "loss": 2.2808,
+      "step": 50240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.55321600744346e-05,
+      "loss": 2.2882,
+      "step": 50248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.550583251865631e-05,
+      "loss": 2.2964,
+      "step": 50256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.54795084224476e-05,
+      "loss": 2.2811,
+      "step": 50264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.545318778844854e-05,
+      "loss": 2.2666,
+      "step": 50272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.542687061929903e-05,
+      "loss": 2.2777,
+      "step": 50280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.540055691763847e-05,
+      "loss": 2.2693,
+      "step": 50288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.5374246686106e-05,
+      "loss": 2.3016,
+      "step": 50296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.534793992734042e-05,
+      "loss": 2.3021,
+      "step": 50304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.532163664398007e-05,
+      "loss": 2.2787,
+      "step": 50312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.529533683866307e-05,
+      "loss": 2.2508,
+      "step": 50320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.526904051402719e-05,
+      "loss": 2.2961,
+      "step": 50328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.52427476727097e-05,
+      "loss": 2.2905,
+      "step": 50336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.521645831734769e-05,
+      "loss": 2.2832,
+      "step": 50344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.51901724505778e-05,
+      "loss": 2.2688,
+      "step": 50352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.516389007503639e-05,
+      "loss": 2.2739,
+      "step": 50360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.513761119335943e-05,
+      "loss": 2.2749,
+      "step": 50368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.511133580818248e-05,
+      "loss": 2.284,
+      "step": 50376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.508506392214087e-05,
+      "loss": 2.2799,
+      "step": 50384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.505879553786952e-05,
+      "loss": 2.2635,
+      "step": 50392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.5032530658003e-05,
+      "loss": 2.2772,
+      "step": 50400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.500626928517549e-05,
+      "loss": 2.2578,
+      "step": 50408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.498001142202094e-05,
+      "loss": 2.2858,
+      "step": 50416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.495375707117276e-05,
+      "loss": 2.2848,
+      "step": 50424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.492750623526417e-05,
+      "loss": 2.2847,
+      "step": 50432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.490125891692798e-05,
+      "loss": 2.3052,
+      "step": 50440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.487501511879664e-05,
+      "loss": 2.2923,
+      "step": 50448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.484877484350225e-05,
+      "loss": 2.2606,
+      "step": 50456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.482253809367659e-05,
+      "loss": 2.2817,
+      "step": 50464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.4796304871951e-05,
+      "loss": 2.2803,
+      "step": 50472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.477007518095658e-05,
+      "loss": 2.2906,
+      "step": 50480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.474384902332396e-05,
+      "loss": 2.2884,
+      "step": 50488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.471762640168354e-05,
+      "loss": 2.2819,
+      "step": 50496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.469140731866528e-05,
+      "loss": 2.2925,
+      "step": 50504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.466519177689877e-05,
+      "loss": 2.2674,
+      "step": 50512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.463897977901333e-05,
+      "loss": 2.2873,
+      "step": 50520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.461277132763782e-05,
+      "loss": 2.2626,
+      "step": 50528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.458656642540087e-05,
+      "loss": 2.2828,
+      "step": 50536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.456036507493064e-05,
+      "loss": 2.2733,
+      "step": 50544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.453416727885495e-05,
+      "loss": 2.3076,
+      "step": 50552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.450797303980134e-05,
+      "loss": 2.3074,
+      "step": 50560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.44817823603969e-05,
+      "loss": 2.2839,
+      "step": 50568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.445559524326848e-05,
+      "loss": 2.261,
+      "step": 50576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.442941169104241e-05,
+      "loss": 2.2978,
+      "step": 50584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.440323170634482e-05,
+      "loss": 2.2911,
+      "step": 50592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.43770552918014e-05,
+      "loss": 2.2726,
+      "step": 50600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.435088245003743e-05,
+      "loss": 2.2753,
+      "step": 50608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.432471318367802e-05,
+      "loss": 2.2742,
+      "step": 50616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.429854749534765e-05,
+      "loss": 2.2814,
+      "step": 50624
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.427238538767072e-05,
+      "loss": 2.2731,
+      "step": 50632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.424622686327112e-05,
+      "loss": 2.2716,
+      "step": 50640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.422007192477232e-05,
+      "loss": 2.2924,
+      "step": 50648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.419392057479757e-05,
+      "loss": 2.2891,
+      "step": 50656
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.416777281596972e-05,
+      "loss": 2.2673,
+      "step": 50664
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.414162865091121e-05,
+      "loss": 2.274,
+      "step": 50672
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.41154880822442e-05,
+      "loss": 2.2897,
+      "step": 50680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.408935111259034e-05,
+      "loss": 2.2947,
+      "step": 50688
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.40632177445711e-05,
+      "loss": 2.2702,
+      "step": 50696
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.403708798080749e-05,
+      "loss": 2.2882,
+      "step": 50704
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.40109618239202e-05,
+      "loss": 2.2869,
+      "step": 50712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.39848392765295e-05,
+      "loss": 2.2868,
+      "step": 50720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.395872034125535e-05,
+      "loss": 2.2703,
+      "step": 50728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.393260502071733e-05,
+      "loss": 2.2806,
+      "step": 50736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.390649331753463e-05,
+      "loss": 2.2814,
+      "step": 50744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.388038523432619e-05,
+      "loss": 2.2717,
+      "step": 50752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.38542807737104e-05,
+      "loss": 2.2858,
+      "step": 50760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.382817993830544e-05,
+      "loss": 2.2936,
+      "step": 50768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.380208273072911e-05,
+      "loss": 2.2609,
+      "step": 50776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.377598915359873e-05,
+      "loss": 2.2859,
+      "step": 50784
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.37498992095314e-05,
+      "loss": 2.2694,
+      "step": 50792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.372381290114374e-05,
+      "loss": 2.2618,
+      "step": 50800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.369773023105215e-05,
+      "loss": 2.2814,
+      "step": 50808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.367165120187252e-05,
+      "loss": 2.2567,
+      "step": 50816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.364557581622038e-05,
+      "loss": 2.2814,
+      "step": 50824
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.361950407671101e-05,
+      "loss": 2.2868,
+      "step": 50832
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.359343598595922e-05,
+      "loss": 2.2778,
+      "step": 50840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.356737154657954e-05,
+      "loss": 2.286,
+      "step": 50848
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.354131076118605e-05,
+      "loss": 2.2811,
+      "step": 50856
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.351525363239246e-05,
+      "loss": 2.2797,
+      "step": 50864
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.348920016281227e-05,
+      "loss": 2.2872,
+      "step": 50872
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.346315035505837e-05,
+      "loss": 2.2826,
+      "step": 50880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.343710421174348e-05,
+      "loss": 2.2785,
+      "step": 50888
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.341106173547987e-05,
+      "loss": 2.2672,
+      "step": 50896
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.33850229288794e-05,
+      "loss": 2.2885,
+      "step": 50904
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.335898779455372e-05,
+      "loss": 2.281,
+      "step": 50912
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.33329563351139e-05,
+      "loss": 2.297,
+      "step": 50920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.33069285531708e-05,
+      "loss": 2.2994,
+      "step": 50928
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.328090445133485e-05,
+      "loss": 2.2789,
+      "step": 50936
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.325488403221614e-05,
+      "loss": 2.2551,
+      "step": 50944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.322886729842435e-05,
+      "loss": 2.2858,
+      "step": 50952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.320285425256877e-05,
+      "loss": 2.2884,
+      "step": 50960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.317684489725847e-05,
+      "loss": 2.2933,
+      "step": 50968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.315083923510191e-05,
+      "loss": 2.2718,
+      "step": 50976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.312483726870745e-05,
+      "loss": 2.2781,
+      "step": 50984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.309883900068282e-05,
+      "loss": 2.2918,
+      "step": 50992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.307284443363553e-05,
+      "loss": 2.2843,
+      "step": 51000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.304685357017273e-05,
+      "loss": 2.2558,
+      "step": 51008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.302086641290112e-05,
+      "loss": 2.3004,
+      "step": 51016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.299488296442706e-05,
+      "loss": 2.2812,
+      "step": 51024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.29689032273566e-05,
+      "loss": 2.305,
+      "step": 51032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.294292720429527e-05,
+      "loss": 2.2794,
+      "step": 51040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.29169548978484e-05,
+      "loss": 2.2841,
+      "step": 51048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.289098631062081e-05,
+      "loss": 2.281,
+      "step": 51056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.286502144521701e-05,
+      "loss": 2.2823,
+      "step": 51064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.283906030424119e-05,
+      "loss": 2.2798,
+      "step": 51072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.281310289029699e-05,
+      "loss": 2.2783,
+      "step": 51080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.278714920598788e-05,
+      "loss": 2.2764,
+      "step": 51088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.276119925391685e-05,
+      "loss": 2.287,
+      "step": 51096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.273525303668653e-05,
+      "loss": 2.287,
+      "step": 51104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.270931055689918e-05,
+      "loss": 2.2818,
+      "step": 51112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.26833718171567e-05,
+      "loss": 2.2844,
+      "step": 51120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.265743682006054e-05,
+      "loss": 2.2911,
+      "step": 51128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.26315055682119e-05,
+      "loss": 2.2866,
+      "step": 51136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.260557806421152e-05,
+      "loss": 2.3046,
+      "step": 51144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.257965431065971e-05,
+      "loss": 2.2969,
+      "step": 51152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.25537343101566e-05,
+      "loss": 2.2809,
+      "step": 51160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.252781806530173e-05,
+      "loss": 2.2655,
+      "step": 51168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.25019055786944e-05,
+      "loss": 2.3136,
+      "step": 51176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.247599685293345e-05,
+      "loss": 2.2816,
+      "step": 51184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.245009189061736e-05,
+      "loss": 2.2842,
+      "step": 51192
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.242419069434434e-05,
+      "loss": 2.2568,
+      "step": 51200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.239829326671206e-05,
+      "loss": 2.2993,
+      "step": 51208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.237239961031785e-05,
+      "loss": 2.2776,
+      "step": 51216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.234650972775885e-05,
+      "loss": 2.2768,
+      "step": 51224
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.232062362163145e-05,
+      "loss": 2.2978,
+      "step": 51232
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.229474129453212e-05,
+      "loss": 2.2981,
+      "step": 51240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.226886274905653e-05,
+      "loss": 2.2841,
+      "step": 51248
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.224298798780016e-05,
+      "loss": 2.2849,
+      "step": 51256
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.221711701335825e-05,
+      "loss": 2.2789,
+      "step": 51264
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.219124982832532e-05,
+      "loss": 2.2816,
+      "step": 51272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.216538643529584e-05,
+      "loss": 2.2631,
+      "step": 51280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.213952683686371e-05,
+      "loss": 2.2641,
+      "step": 51288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.211367103562253e-05,
+      "loss": 2.2966,
+      "step": 51296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.208781903416548e-05,
+      "loss": 2.2889,
+      "step": 51304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.20619708350853e-05,
+      "loss": 2.2868,
+      "step": 51312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.203612644097452e-05,
+      "loss": 2.2752,
+      "step": 51320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.201028585442513e-05,
+      "loss": 2.2658,
+      "step": 51328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.198444907802881e-05,
+      "loss": 2.28,
+      "step": 51336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.195861611437685e-05,
+      "loss": 2.2611,
+      "step": 51344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.193278696606014e-05,
+      "loss": 2.2846,
+      "step": 51352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.190696163566919e-05,
+      "loss": 2.286,
+      "step": 51360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.188114012579412e-05,
+      "loss": 2.2942,
+      "step": 51368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.185532243902474e-05,
+      "loss": 2.264,
+      "step": 51376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.182950857795035e-05,
+      "loss": 2.2863,
+      "step": 51384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.180369854515995e-05,
+      "loss": 2.2785,
+      "step": 51392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.177789234324219e-05,
+      "loss": 2.2961,
+      "step": 51400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.175208997478525e-05,
+      "loss": 2.2723,
+      "step": 51408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.172629144237694e-05,
+      "loss": 2.2939,
+      "step": 51416
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.170049674860473e-05,
+      "loss": 2.2835,
+      "step": 51424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.16747058960557e-05,
+      "loss": 2.2826,
+      "step": 51432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.164891888731649e-05,
+      "loss": 2.2873,
+      "step": 51440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.162313572497336e-05,
+      "loss": 2.2834,
+      "step": 51448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.159735641161234e-05,
+      "loss": 2.2841,
+      "step": 51456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.157158094981882e-05,
+      "loss": 2.2986,
+      "step": 51464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.154580934217803e-05,
+      "loss": 2.2891,
+      "step": 51472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.152004159127463e-05,
+      "loss": 2.2804,
+      "step": 51480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.149427769969301e-05,
+      "loss": 2.2863,
+      "step": 51488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.14685176700172e-05,
+      "loss": 2.2873,
+      "step": 51496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.144276150483067e-05,
+      "loss": 2.252,
+      "step": 51504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.141700920671673e-05,
+      "loss": 2.2907,
+      "step": 51512
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.139126077825817e-05,
+      "loss": 2.278,
+      "step": 51520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.13655162220373e-05,
+      "loss": 2.2674,
+      "step": 51528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.133977554063634e-05,
+      "loss": 2.258,
+      "step": 51536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.131403873663673e-05,
+      "loss": 2.2839,
+      "step": 51544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.128830581261987e-05,
+      "loss": 2.2718,
+      "step": 51552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.126257677116663e-05,
+      "loss": 2.2713,
+      "step": 51560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.123685161485738e-05,
+      "loss": 2.2934,
+      "step": 51568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.121113034627229e-05,
+      "loss": 2.2814,
+      "step": 51576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.118541296799104e-05,
+      "loss": 2.2717,
+      "step": 51584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.115969948259293e-05,
+      "loss": 2.2928,
+      "step": 51592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.113398989265691e-05,
+      "loss": 2.2868,
+      "step": 51600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.110828420076143e-05,
+      "loss": 2.2616,
+      "step": 51608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.108258240948469e-05,
+      "loss": 2.2741,
+      "step": 51616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.105688452140442e-05,
+      "loss": 2.2704,
+      "step": 51624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.103119053909797e-05,
+      "loss": 2.2696,
+      "step": 51632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.10055004651423e-05,
+      "loss": 2.2927,
+      "step": 51640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.097981430211395e-05,
+      "loss": 2.2569,
+      "step": 51648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.095413205258916e-05,
+      "loss": 2.284,
+      "step": 51656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.092845371914367e-05,
+      "loss": 2.2688,
+      "step": 51664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.09027793043529e-05,
+      "loss": 2.2819,
+      "step": 51672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.087710881079179e-05,
+      "loss": 2.2877,
+      "step": 51680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.085144224103499e-05,
+      "loss": 2.2501,
+      "step": 51688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.082577959765672e-05,
+      "loss": 2.2829,
+      "step": 51696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.08001208832308e-05,
+      "loss": 2.2829,
+      "step": 51704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.077446610033061e-05,
+      "loss": 2.2597,
+      "step": 51712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.074881525152919e-05,
+      "loss": 2.2776,
+      "step": 51720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.072316833939923e-05,
+      "loss": 2.2745,
+      "step": 51728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.069752536651291e-05,
+      "loss": 2.2633,
+      "step": 51736
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.067188633544211e-05,
+      "loss": 2.2577,
+      "step": 51744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.064625124875825e-05,
+      "loss": 2.271,
+      "step": 51752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.062062010903239e-05,
+      "loss": 2.2964,
+      "step": 51760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.059499291883524e-05,
+      "loss": 2.2872,
+      "step": 51768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.056936968073697e-05,
+      "loss": 2.2943,
+      "step": 51776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.054375039730749e-05,
+      "loss": 2.2775,
+      "step": 51784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.051813507111634e-05,
+      "loss": 2.3028,
+      "step": 51792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.049252370473245e-05,
+      "loss": 2.2656,
+      "step": 51800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.046691630072464e-05,
+      "loss": 2.2737,
+      "step": 51808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.044131286166104e-05,
+      "loss": 2.2872,
+      "step": 51816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.041571339010966e-05,
+      "loss": 2.2812,
+      "step": 51824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.039011788863796e-05,
+      "loss": 2.2974,
+      "step": 51832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.036452635981292e-05,
+      "loss": 2.2546,
+      "step": 51840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.033893880620132e-05,
+      "loss": 2.2738,
+      "step": 51848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.031335523036944e-05,
+      "loss": 2.2607,
+      "step": 51856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.028777563488316e-05,
+      "loss": 2.2566,
+      "step": 51864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.026220002230799e-05,
+      "loss": 2.2674,
+      "step": 51872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.023662839520892e-05,
+      "loss": 2.2842,
+      "step": 51880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.021106075615076e-05,
+      "loss": 2.2684,
+      "step": 51888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.018549710769773e-05,
+      "loss": 2.2813,
+      "step": 51896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.015993745241374e-05,
+      "loss": 2.2962,
+      "step": 51904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.013438179286229e-05,
+      "loss": 2.3061,
+      "step": 51912
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.010883013160645e-05,
+      "loss": 2.2899,
+      "step": 51920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.00832824712089e-05,
+      "loss": 2.2693,
+      "step": 51928
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.005773881423197e-05,
+      "loss": 2.267,
+      "step": 51936
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.003219916323747e-05,
+      "loss": 2.2926,
+      "step": 51944
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.000666352078695e-05,
+      "loss": 2.2925,
+      "step": 51952
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.998113188944146e-05,
+      "loss": 2.2763,
+      "step": 51960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.995560427176166e-05,
+      "loss": 2.2742,
+      "step": 51968
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.993008067030788e-05,
+      "loss": 2.2603,
+      "step": 51976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.990456108763995e-05,
+      "loss": 2.2819,
+      "step": 51984
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.987904552631731e-05,
+      "loss": 2.292,
+      "step": 51992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.985353398889913e-05,
+      "loss": 2.2897,
+      "step": 52000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.982802647794398e-05,
+      "loss": 2.2672,
+      "step": 52008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.980252299601015e-05,
+      "loss": 2.2743,
+      "step": 52016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.977702354565549e-05,
+      "loss": 2.2691,
+      "step": 52024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.975152812943742e-05,
+      "loss": 2.2864,
+      "step": 52032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.972603674991308e-05,
+      "loss": 2.287,
+      "step": 52040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.970054940963903e-05,
+      "loss": 2.2924,
+      "step": 52048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.967506611117147e-05,
+      "loss": 2.2605,
+      "step": 52056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.964958685706637e-05,
+      "loss": 2.284,
+      "step": 52064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.962411164987898e-05,
+      "loss": 2.2652,
+      "step": 52072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.959864049216451e-05,
+      "loss": 2.2705,
+      "step": 52080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.957317338647742e-05,
+      "loss": 2.2751,
+      "step": 52088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.954771033537195e-05,
+      "loss": 2.2833,
+      "step": 52096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.952225134140197e-05,
+      "loss": 2.2667,
+      "step": 52104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.949679640712074e-05,
+      "loss": 2.2702,
+      "step": 52112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.947134553508138e-05,
+      "loss": 2.2823,
+      "step": 52120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.944589872783645e-05,
+      "loss": 2.2729,
+      "step": 52128
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.942045598793801e-05,
+      "loss": 2.2863,
+      "step": 52136
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.939501731793796e-05,
+      "loss": 2.2566,
+      "step": 52144
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.936958272038753e-05,
+      "loss": 2.2851,
+      "step": 52152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.934415219783777e-05,
+      "loss": 2.2822,
+      "step": 52160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.931872575283916e-05,
+      "loss": 2.2661,
+      "step": 52168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.929330338794186e-05,
+      "loss": 2.2544,
+      "step": 52176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.926788510569558e-05,
+      "loss": 2.276,
+      "step": 52184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.92424709086496e-05,
+      "loss": 2.2695,
+      "step": 52192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.921706079935284e-05,
+      "loss": 2.2744,
+      "step": 52200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.919165478035382e-05,
+      "loss": 2.249,
+      "step": 52208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.916625285420057e-05,
+      "loss": 2.2891,
+      "step": 52216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.91408550234408e-05,
+      "loss": 2.2749,
+      "step": 52224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.911546129062174e-05,
+      "loss": 2.2731,
+      "step": 52232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.909007165829024e-05,
+      "loss": 2.2604,
+      "step": 52240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.906468612899278e-05,
+      "loss": 2.2928,
+      "step": 52248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.903930470527531e-05,
+      "loss": 2.2838,
+      "step": 52256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.901392738968349e-05,
+      "loss": 2.2608,
+      "step": 52264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.898855418476252e-05,
+      "loss": 2.268,
+      "step": 52272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.896318509305718e-05,
+      "loss": 2.2768,
+      "step": 52280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.893782011711182e-05,
+      "loss": 2.295,
+      "step": 52288
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.891245925947048e-05,
+      "loss": 2.2872,
+      "step": 52296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.888710252267665e-05,
+      "loss": 2.2949,
+      "step": 52304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.886174990927348e-05,
+      "loss": 2.2843,
+      "step": 52312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.883640142180368e-05,
+      "loss": 2.2809,
+      "step": 52320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.881105706280953e-05,
+      "loss": 2.2787,
+      "step": 52328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.878571683483304e-05,
+      "loss": 2.283,
+      "step": 52336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.876038074041555e-05,
+      "loss": 2.2846,
+      "step": 52344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.873504878209823e-05,
+      "loss": 2.2898,
+      "step": 52352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.870972096242174e-05,
+      "loss": 2.2833,
+      "step": 52360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.86843972839262e-05,
+      "loss": 2.2726,
+      "step": 52368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.865907774915157e-05,
+      "loss": 2.2843,
+      "step": 52376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.863376236063712e-05,
+      "loss": 2.2916,
+      "step": 52384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.860845112092195e-05,
+      "loss": 2.2659,
+      "step": 52392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.858314403254464e-05,
+      "loss": 2.2758,
+      "step": 52400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.855784109804323e-05,
+      "loss": 2.2861,
+      "step": 52408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.853254231995556e-05,
+      "loss": 2.2619,
+      "step": 52416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.850724770081894e-05,
+      "loss": 2.259,
+      "step": 52424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.848195724317024e-05,
+      "loss": 2.2812,
+      "step": 52432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.845667094954601e-05,
+      "loss": 2.269,
+      "step": 52440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.843138882248225e-05,
+      "loss": 2.2673,
+      "step": 52448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.840611086451467e-05,
+      "loss": 2.2668,
+      "step": 52456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.838083707817848e-05,
+      "loss": 2.2826,
+      "step": 52464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.83555674660085e-05,
+      "loss": 2.282,
+      "step": 52472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.833030203053913e-05,
+      "loss": 2.2647,
+      "step": 52480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.830504077430437e-05,
+      "loss": 2.2569,
+      "step": 52488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.827978369983776e-05,
+      "loss": 2.2892,
+      "step": 52496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.825453080967244e-05,
+      "loss": 2.2619,
+      "step": 52504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.822928210634114e-05,
+      "loss": 2.2944,
+      "step": 52512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.820403759237615e-05,
+      "loss": 2.2842,
+      "step": 52520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.817879727030938e-05,
+      "loss": 2.2615,
+      "step": 52528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.815356114267224e-05,
+      "loss": 2.2687,
+      "step": 52536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.812832921199584e-05,
+      "loss": 2.2757,
+      "step": 52544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.810310148081074e-05,
+      "loss": 2.2693,
+      "step": 52552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.807787795164712e-05,
+      "loss": 2.2876,
+      "step": 52560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.805265862703487e-05,
+      "loss": 2.2892,
+      "step": 52568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.802744350950326e-05,
+      "loss": 2.2786,
+      "step": 52576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.80022326015812e-05,
+      "loss": 2.2513,
+      "step": 52584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.797702590579726e-05,
+      "loss": 2.2718,
+      "step": 52592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.795182342467948e-05,
+      "loss": 2.2647,
+      "step": 52600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.792662516075561e-05,
+      "loss": 2.2937,
+      "step": 52608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.79014311165528e-05,
+      "loss": 2.2727,
+      "step": 52616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.787624129459788e-05,
+      "loss": 2.2667,
+      "step": 52624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.785105569741734e-05,
+      "loss": 2.268,
+      "step": 52632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.7825874327537e-05,
+      "loss": 2.2886,
+      "step": 52640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.780069718748256e-05,
+      "loss": 2.2785,
+      "step": 52648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.777552427977906e-05,
+      "loss": 2.2725,
+      "step": 52656
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.775035560695117e-05,
+      "loss": 2.2701,
+      "step": 52664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.77251911715233e-05,
+      "loss": 2.2621,
+      "step": 52672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.770003097601913e-05,
+      "loss": 2.2788,
+      "step": 52680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.767487502296223e-05,
+      "loss": 2.2999,
+      "step": 52688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.764972331487551e-05,
+      "loss": 2.272,
+      "step": 52696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.762457585428157e-05,
+      "loss": 2.2705,
+      "step": 52704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.75994326437026e-05,
+      "loss": 2.2687,
+      "step": 52712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.757429368566022e-05,
+      "loss": 2.2663,
+      "step": 52720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.754915898267579e-05,
+      "loss": 2.2976,
+      "step": 52728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.752402853727023e-05,
+      "loss": 2.2605,
+      "step": 52736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.749890235196387e-05,
+      "loss": 2.2931,
+      "step": 52744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.747378042927681e-05,
+      "loss": 2.2759,
+      "step": 52752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.74486627717286e-05,
+      "loss": 2.267,
+      "step": 52760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.74235493818384e-05,
+      "loss": 2.2635,
+      "step": 52768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.739844026212494e-05,
+      "loss": 2.2744,
+      "step": 52776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.737333541510656e-05,
+      "loss": 2.2937,
+      "step": 52784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.734823484330106e-05,
+      "loss": 2.2804,
+      "step": 52792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.732313854922594e-05,
+      "loss": 2.2481,
+      "step": 52800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.729804653539823e-05,
+      "loss": 2.2934,
+      "step": 52808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.727295880433446e-05,
+      "loss": 2.2371,
+      "step": 52816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.724787535855081e-05,
+      "loss": 2.2699,
+      "step": 52824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.722279620056306e-05,
+      "loss": 2.2522,
+      "step": 52832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.719772133288642e-05,
+      "loss": 2.254,
+      "step": 52840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.717265075803581e-05,
+      "loss": 2.2726,
+      "step": 52848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.714758447852566e-05,
+      "loss": 2.286,
+      "step": 52856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.712252249686996e-05,
+      "loss": 2.2798,
+      "step": 52864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.709746481558228e-05,
+      "loss": 2.2871,
+      "step": 52872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.707241143717584e-05,
+      "loss": 2.2921,
+      "step": 52880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.704736236416326e-05,
+      "loss": 2.2625,
+      "step": 52888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.702231759905682e-05,
+      "loss": 2.2784,
+      "step": 52896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.699727714436846e-05,
+      "loss": 2.3037,
+      "step": 52904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.697224100260947e-05,
+      "loss": 2.285,
+      "step": 52912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.694720917629096e-05,
+      "loss": 2.2763,
+      "step": 52920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.692218166792339e-05,
+      "loss": 2.276,
+      "step": 52928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.689715848001688e-05,
+      "loss": 2.2715,
+      "step": 52936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.687213961508121e-05,
+      "loss": 2.2522,
+      "step": 52944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.684712507562548e-05,
+      "loss": 2.2733,
+      "step": 52952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.682211486415862e-05,
+      "loss": 2.2649,
+      "step": 52960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.679710898318903e-05,
+      "loss": 2.3027,
+      "step": 52968
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.677210743522453e-05,
+      "loss": 2.276,
+      "step": 52976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.674711022277278e-05,
+      "loss": 2.2645,
+      "step": 52984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.672211734834072e-05,
+      "loss": 2.256,
+      "step": 52992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.66971288144351e-05,
+      "loss": 2.2754,
+      "step": 53000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.667214462356212e-05,
+      "loss": 2.2688,
+      "step": 53008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.664716477822747e-05,
+      "loss": 2.2851,
+      "step": 53016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.662218928093656e-05,
+      "loss": 2.2589,
+      "step": 53024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.65972181341943e-05,
+      "loss": 2.2824,
+      "step": 53032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.65722513405051e-05,
+      "loss": 2.2812,
+      "step": 53040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.654728890237302e-05,
+      "loss": 2.2744,
+      "step": 53048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.652233082230167e-05,
+      "loss": 2.2756,
+      "step": 53056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.649737710279415e-05,
+      "loss": 2.2789,
+      "step": 53064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.647242774635323e-05,
+      "loss": 2.2598,
+      "step": 53072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.644748275548113e-05,
+      "loss": 2.2871,
+      "step": 53080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.642254213267978e-05,
+      "loss": 2.2648,
+      "step": 53088
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.63976058804505e-05,
+      "loss": 2.2792,
+      "step": 53096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.637267400129428e-05,
+      "loss": 2.2696,
+      "step": 53104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.634774649771166e-05,
+      "loss": 2.2694,
+      "step": 53112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.632282337220273e-05,
+      "loss": 2.2551,
+      "step": 53120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.629790462726712e-05,
+      "loss": 2.2857,
+      "step": 53128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.627299026540404e-05,
+      "loss": 2.2881,
+      "step": 53136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.624808028911227e-05,
+      "loss": 2.2535,
+      "step": 53144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.622317470089013e-05,
+      "loss": 2.2861,
+      "step": 53152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.619827350323551e-05,
+      "loss": 2.2669,
+      "step": 53160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.617337669864585e-05,
+      "loss": 2.2894,
+      "step": 53168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.614848428961822e-05,
+      "loss": 2.2506,
+      "step": 53176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.612359627864911e-05,
+      "loss": 2.2699,
+      "step": 53184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.609871266823469e-05,
+      "loss": 2.2736,
+      "step": 53192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.607383346087062e-05,
+      "loss": 2.2744,
+      "step": 53200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.604895865905211e-05,
+      "loss": 2.2612,
+      "step": 53208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.602408826527408e-05,
+      "loss": 2.2646,
+      "step": 53216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.599922228203076e-05,
+      "loss": 2.2841,
+      "step": 53224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.597436071181616e-05,
+      "loss": 2.2816,
+      "step": 53232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.594950355712372e-05,
+      "loss": 2.2712,
+      "step": 53240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.592465082044642e-05,
+      "loss": 2.2515,
+      "step": 53248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.589980250427697e-05,
+      "loss": 2.2737,
+      "step": 53256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.587495861110738e-05,
+      "loss": 2.2694,
+      "step": 53264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.585011914342944e-05,
+      "loss": 2.2542,
+      "step": 53272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.582528410373441e-05,
+      "loss": 2.2778,
+      "step": 53280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.580045349451302e-05,
+      "loss": 2.2696,
+      "step": 53288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.577562731825573e-05,
+      "loss": 2.2844,
+      "step": 53296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.575080557745244e-05,
+      "loss": 2.2577,
+      "step": 53304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.57259882745926e-05,
+      "loss": 2.272,
+      "step": 53312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.570117541216532e-05,
+      "loss": 2.2651,
+      "step": 53320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.567636699265904e-05,
+      "loss": 2.2732,
+      "step": 53328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.565156301856206e-05,
+      "loss": 2.2705,
+      "step": 53336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.562676349236201e-05,
+      "loss": 2.2632,
+      "step": 53344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.560196841654613e-05,
+      "loss": 2.2637,
+      "step": 53352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.557717779360127e-05,
+      "loss": 2.2885,
+      "step": 53360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.555239162601375e-05,
+      "loss": 2.302,
+      "step": 53368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.552760991626949e-05,
+      "loss": 2.2661,
+      "step": 53376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.550283266685399e-05,
+      "loss": 2.2732,
+      "step": 53384
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.547805988025222e-05,
+      "loss": 2.2884,
+      "step": 53392
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.545329155894872e-05,
+      "loss": 2.2969,
+      "step": 53400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.542852770542775e-05,
+      "loss": 2.2764,
+      "step": 53408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.540376832217285e-05,
+      "loss": 2.2663,
+      "step": 53416
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.53790134116673e-05,
+      "loss": 2.2682,
+      "step": 53424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.535426297639386e-05,
+      "loss": 2.2653,
+      "step": 53432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.532951701883483e-05,
+      "loss": 2.2732,
+      "step": 53440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.530477554147221e-05,
+      "loss": 2.2731,
+      "step": 53448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.52800385467873e-05,
+      "loss": 2.2818,
+      "step": 53456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.525530603726108e-05,
+      "loss": 2.2946,
+      "step": 53464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.523057801537422e-05,
+      "loss": 2.2812,
+      "step": 53472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.520585448360664e-05,
+      "loss": 2.2719,
+      "step": 53480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.51811354444381e-05,
+      "loss": 2.2912,
+      "step": 53488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.515642090034768e-05,
+      "loss": 2.2606,
+      "step": 53496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.51317108538141e-05,
+      "loss": 2.2778,
+      "step": 53504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.510700530731577e-05,
+      "loss": 2.2814,
+      "step": 53512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.508230426333035e-05,
+      "loss": 2.2885,
+      "step": 53520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.505760772433533e-05,
+      "loss": 2.2791,
+      "step": 53528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.503291569280763e-05,
+      "loss": 2.2717,
+      "step": 53536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.500822817122363e-05,
+      "loss": 2.2665,
+      "step": 53544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.498354516205947e-05,
+      "loss": 2.27,
+      "step": 53552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.495886666779059e-05,
+      "loss": 2.279,
+      "step": 53560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.49341926908922e-05,
+      "loss": 2.2633,
+      "step": 53568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.490952323383892e-05,
+      "loss": 2.2409,
+      "step": 53576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.488485829910495e-05,
+      "loss": 2.2768,
+      "step": 53584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.486019788916409e-05,
+      "loss": 2.2809,
+      "step": 53592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.483554200648957e-05,
+      "loss": 2.2848,
+      "step": 53600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.481089065355428e-05,
+      "loss": 2.2872,
+      "step": 53608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.478624383283062e-05,
+      "loss": 2.2808,
+      "step": 53616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.47616015467905e-05,
+      "loss": 2.2836,
+      "step": 53624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.473696379790541e-05,
+      "loss": 2.2655,
+      "step": 53632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.471233058864637e-05,
+      "loss": 2.2716,
+      "step": 53640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.468770192148396e-05,
+      "loss": 2.2783,
+      "step": 53648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.466307779888829e-05,
+      "loss": 2.2704,
+      "step": 53656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.463845822332903e-05,
+      "loss": 2.2637,
+      "step": 53664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.461384319727537e-05,
+      "loss": 2.2746,
+      "step": 53672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.458923272319608e-05,
+      "loss": 2.2786,
+      "step": 53680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.456462680355944e-05,
+      "loss": 2.2658,
+      "step": 53688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.454002544083327e-05,
+      "loss": 2.2716,
+      "step": 53696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.451542863748494e-05,
+      "loss": 2.2689,
+      "step": 53704
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.449083639598142e-05,
+      "loss": 2.2737,
+      "step": 53712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.44662487187891e-05,
+      "loss": 2.2613,
+      "step": 53720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.444166560837406e-05,
+      "loss": 2.2867,
+      "step": 53728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.441708706720177e-05,
+      "loss": 2.2574,
+      "step": 53736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.439251309773742e-05,
+      "loss": 2.2724,
+      "step": 53744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.436794370244552e-05,
+      "loss": 2.2693,
+      "step": 53752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.434337888379037e-05,
+      "loss": 2.2665,
+      "step": 53760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.431881864423557e-05,
+      "loss": 2.269,
+      "step": 53768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.429426298624439e-05,
+      "loss": 2.2772,
+      "step": 53776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.42697119122797e-05,
+      "loss": 2.2908,
+      "step": 53784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.424516542480374e-05,
+      "loss": 2.2569,
+      "step": 53792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.422062352627844e-05,
+      "loss": 2.2412,
+      "step": 53800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.419608621916526e-05,
+      "loss": 2.2548,
+      "step": 53808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.4171553505925e-05,
+      "loss": 2.2814,
+      "step": 53816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.414702538901833e-05,
+      "loss": 2.2739,
+      "step": 53824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.412250187090511e-05,
+      "loss": 2.2709,
+      "step": 53832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.409798295404506e-05,
+      "loss": 2.2543,
+      "step": 53840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.407346864089724e-05,
+      "loss": 2.2834,
+      "step": 53848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.40489589339202e-05,
+      "loss": 2.286,
+      "step": 53856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.402445383557225e-05,
+      "loss": 2.2441,
+      "step": 53864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.399995334831108e-05,
+      "loss": 2.265,
+      "step": 53872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.397545747459393e-05,
+      "loss": 2.2759,
+      "step": 53880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.395096621687761e-05,
+      "loss": 2.2908,
+      "step": 53888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.392647957761846e-05,
+      "loss": 2.2623,
+      "step": 53896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.390199755927231e-05,
+      "loss": 2.2807,
+      "step": 53904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.387752016429462e-05,
+      "loss": 2.2465,
+      "step": 53912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.38530473951403e-05,
+      "loss": 2.276,
+      "step": 53920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.382857925426387e-05,
+      "loss": 2.2514,
+      "step": 53928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.380411574411931e-05,
+      "loss": 2.2681,
+      "step": 53936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.377965686716017e-05,
+      "loss": 2.2814,
+      "step": 53944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.375520262583954e-05,
+      "loss": 2.2729,
+      "step": 53952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.373075302261006e-05,
+      "loss": 2.2882,
+      "step": 53960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.370630805992387e-05,
+      "loss": 2.2634,
+      "step": 53968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.368186774023268e-05,
+      "loss": 2.2844,
+      "step": 53976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.36574320659877e-05,
+      "loss": 2.2724,
+      "step": 53984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.363300103963967e-05,
+      "loss": 2.26,
+      "step": 53992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.360857466363893e-05,
+      "loss": 2.2622,
+      "step": 54000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.358415294043525e-05,
+      "loss": 2.2565,
+      "step": 54008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.355973587247807e-05,
+      "loss": 2.2832,
+      "step": 54016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.353532346221622e-05,
+      "loss": 2.2691,
+      "step": 54024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.351091571209816e-05,
+      "loss": 2.2593,
+      "step": 54032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.348651262457179e-05,
+      "loss": 2.2622,
+      "step": 54040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.346211420208463e-05,
+      "loss": 2.2536,
+      "step": 54048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.343772044708381e-05,
+      "loss": 2.2908,
+      "step": 54056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.341333136201568e-05,
+      "loss": 2.2839,
+      "step": 54064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.338894694932651e-05,
+      "loss": 2.2628,
+      "step": 54072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.336456721146185e-05,
+      "loss": 2.273,
+      "step": 54080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.334019215086682e-05,
+      "loss": 2.2552,
+      "step": 54088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.331582176998616e-05,
+      "loss": 2.2683,
+      "step": 54096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.3291456071264e-05,
+      "loss": 2.2824,
+      "step": 54104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.326709505714417e-05,
+      "loss": 2.2882,
+      "step": 54112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.324273873006992e-05,
+      "loss": 2.2738,
+      "step": 54120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.321838709248397e-05,
+      "loss": 2.2767,
+      "step": 54128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.319404014682875e-05,
+      "loss": 2.2539,
+      "step": 54136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.316969789554607e-05,
+      "loss": 2.2655,
+      "step": 54144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.314536034107736e-05,
+      "loss": 2.2856,
+      "step": 54152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.312102748586355e-05,
+      "loss": 2.2747,
+      "step": 54160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.309669933234498e-05,
+      "loss": 2.2792,
+      "step": 54168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.307237588296173e-05,
+      "loss": 2.2684,
+      "step": 54176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.304805714015327e-05,
+      "loss": 2.2682,
+      "step": 54184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.302374310635863e-05,
+      "loss": 2.2747,
+      "step": 54192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.29994337840164e-05,
+      "loss": 2.2831,
+      "step": 54200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.297512917556461e-05,
+      "loss": 2.2631,
+      "step": 54208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.295082928344092e-05,
+      "loss": 2.2917,
+      "step": 54216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.292653411008246e-05,
+      "loss": 2.281,
+      "step": 54224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.290224365792588e-05,
+      "loss": 2.2734,
+      "step": 54232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.287795792940737e-05,
+      "loss": 2.2574,
+      "step": 54240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.285367692696273e-05,
+      "loss": 2.2587,
+      "step": 54248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.282940065302714e-05,
+      "loss": 2.2796,
+      "step": 54256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.280512911003536e-05,
+      "loss": 2.264,
+      "step": 54264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.278086230042172e-05,
+      "loss": 2.2584,
+      "step": 54272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.275660022661999e-05,
+      "loss": 2.2865,
+      "step": 54280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.273234289106363e-05,
+      "loss": 2.274,
+      "step": 54288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.27080902961854e-05,
+      "loss": 2.2589,
+      "step": 54296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.268384244441773e-05,
+      "loss": 2.2627,
+      "step": 54304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.265959933819261e-05,
+      "loss": 2.2744,
+      "step": 54312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.263536097994136e-05,
+      "loss": 2.2822,
+      "step": 54320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.26111273720951e-05,
+      "loss": 2.2866,
+      "step": 54328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.258689851708421e-05,
+      "loss": 2.2814,
+      "step": 54336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.25626744173387e-05,
+      "loss": 2.2806,
+      "step": 54344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.253845507528822e-05,
+      "loss": 2.2825,
+      "step": 54352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.25142404933617e-05,
+      "loss": 2.2819,
+      "step": 54360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.249003067398782e-05,
+      "loss": 2.2882,
+      "step": 54368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.246582561959468e-05,
+      "loss": 2.2593,
+      "step": 54376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.244162533260984e-05,
+      "loss": 2.2862,
+      "step": 54384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.241742981546058e-05,
+      "loss": 2.2794,
+      "step": 54392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.239323907057337e-05,
+      "loss": 2.2554,
+      "step": 54400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.23690531003746e-05,
+      "loss": 2.2451,
+      "step": 54408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.234487190728991e-05,
+      "loss": 2.2504,
+      "step": 54416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.232069549374453e-05,
+      "loss": 2.2822,
+      "step": 54424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.229652386216323e-05,
+      "loss": 2.2772,
+      "step": 54432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.227235701497027e-05,
+      "loss": 2.2761,
+      "step": 54440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.224819495458949e-05,
+      "loss": 2.2802,
+      "step": 54448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.222403768344415e-05,
+      "loss": 2.2857,
+      "step": 54456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.219988520395714e-05,
+      "loss": 2.2694,
+      "step": 54464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.21757375185508e-05,
+      "loss": 2.2837,
+      "step": 54472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.215159462964698e-05,
+      "loss": 2.2831,
+      "step": 54480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.21274565396671e-05,
+      "loss": 2.2381,
+      "step": 54488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.210332325103206e-05,
+      "loss": 2.2809,
+      "step": 54496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.20791947661623e-05,
+      "loss": 2.272,
+      "step": 54504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.205507108747778e-05,
+      "loss": 2.2464,
+      "step": 54512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.203095221739795e-05,
+      "loss": 2.2502,
+      "step": 54520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.200683815834182e-05,
+      "loss": 2.2613,
+      "step": 54528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.198272891272787e-05,
+      "loss": 2.2576,
+      "step": 54536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.195862448297411e-05,
+      "loss": 2.2672,
+      "step": 54544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.193452487149813e-05,
+      "loss": 2.2606,
+      "step": 54552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.191043008071693e-05,
+      "loss": 2.2686,
+      "step": 54560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.188634011304712e-05,
+      "loss": 2.2691,
+      "step": 54568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.186225497090473e-05,
+      "loss": 2.2628,
+      "step": 54576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.183817465670547e-05,
+      "loss": 2.2769,
+      "step": 54584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.181409917286431e-05,
+      "loss": 2.2716,
+      "step": 54592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.179002852179605e-05,
+      "loss": 2.2835,
+      "step": 54600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.176596270591473e-05,
+      "loss": 2.2653,
+      "step": 54608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.174190172763397e-05,
+      "loss": 2.2832,
+      "step": 54616
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.171784558936713e-05,
+      "loss": 2.2436,
+      "step": 54624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.169379429352671e-05,
+      "loss": 2.2732,
+      "step": 54632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.166974784252504e-05,
+      "loss": 2.261,
+      "step": 54640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.164570623877384e-05,
+      "loss": 2.2606,
+      "step": 54648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.162166948468425e-05,
+      "loss": 2.257,
+      "step": 54656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.159763758266715e-05,
+      "loss": 2.2697,
+      "step": 54664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.157361053513266e-05,
+      "loss": 2.2682,
+      "step": 54672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.154958834449068e-05,
+      "loss": 2.268,
+      "step": 54680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.152557101315049e-05,
+      "loss": 2.2837,
+      "step": 54688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.15015585435208e-05,
+      "loss": 2.2711,
+      "step": 54696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.147755093800997e-05,
+      "loss": 2.2931,
+      "step": 54704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.145354819902587e-05,
+      "loss": 2.2788,
+      "step": 54712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.142955032897581e-05,
+      "loss": 2.2638,
+      "step": 54720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.140555733026667e-05,
+      "loss": 2.2666,
+      "step": 54728
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.138156920530469e-05,
+      "loss": 2.2791,
+      "step": 54736
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.135758595649588e-05,
+      "loss": 2.2696,
+      "step": 54744
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.133360758624557e-05,
+      "loss": 2.2699,
+      "step": 54752
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.130963409695866e-05,
+      "loss": 2.2608,
+      "step": 54760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.128566549103957e-05,
+      "loss": 2.2895,
+      "step": 54768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.126170177089219e-05,
+      "loss": 2.2695,
+      "step": 54776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.123774293891995e-05,
+      "loss": 2.2903,
+      "step": 54784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.12137889975258e-05,
+      "loss": 2.2801,
+      "step": 54792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.118983994911218e-05,
+      "loss": 2.2504,
+      "step": 54800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.116589579608103e-05,
+      "loss": 2.2872,
+      "step": 54808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.114195654083382e-05,
+      "loss": 2.2883,
+      "step": 54816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.111802218577154e-05,
+      "loss": 2.2679,
+      "step": 54824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.109409273329463e-05,
+      "loss": 2.2644,
+      "step": 54832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.107016818580312e-05,
+      "loss": 2.2748,
+      "step": 54840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.104624854569645e-05,
+      "loss": 2.2744,
+      "step": 54848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.1022333815373746e-05,
+      "loss": 2.2944,
+      "step": 54856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.099842399723338e-05,
+      "loss": 2.2649,
+      "step": 54864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.097451909367343e-05,
+      "loss": 2.2415,
+      "step": 54872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.095061910709143e-05,
+      "loss": 2.2486,
+      "step": 54880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0926724039884376e-05,
+      "loss": 2.2646,
+      "step": 54888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.090283389444891e-05,
+      "loss": 2.2895,
+      "step": 54896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0878948673180935e-05,
+      "loss": 2.2561,
+      "step": 54904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.085506837847608e-05,
+      "loss": 2.2644,
+      "step": 54912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.083119301272944e-05,
+      "loss": 2.27,
+      "step": 54920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.080732257833548e-05,
+      "loss": 2.2816,
+      "step": 54928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.078345707768838e-05,
+      "loss": 2.2742,
+      "step": 54936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0759596513181595e-05,
+      "loss": 2.286,
+      "step": 54944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.07357408872083e-05,
+      "loss": 2.2715,
+      "step": 54952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.071189020216109e-05,
+      "loss": 2.2711,
+      "step": 54960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0688044460431944e-05,
+      "loss": 2.2957,
+      "step": 54968
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0664203664412545e-05,
+      "loss": 2.275,
+      "step": 54976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.064036781649397e-05,
+      "loss": 2.2658,
+      "step": 54984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0616536919066814e-05,
+      "loss": 2.2451,
+      "step": 54992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.059271097452124e-05,
+      "loss": 2.2754,
+      "step": 55000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.05688899852467e-05,
+      "loss": 2.2712,
+      "step": 55008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.054507395363245e-05,
+      "loss": 2.2475,
+      "step": 55016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.052126288206705e-05,
+      "loss": 2.2858,
+      "step": 55024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.049745677293863e-05,
+      "loss": 2.2794,
+      "step": 55032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0473655628634786e-05,
+      "loss": 2.2723,
+      "step": 55040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0449859451542666e-05,
+      "loss": 2.2859,
+      "step": 55048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.042606824404886e-05,
+      "loss": 2.2635,
+      "step": 55056
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.04022820085395e-05,
+      "loss": 2.2732,
+      "step": 55064
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.037850074740025e-05,
+      "loss": 2.2564,
+      "step": 55072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0354724463016174e-05,
+      "loss": 2.2786,
+      "step": 55080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0330953157771955e-05,
+      "loss": 2.2669,
+      "step": 55088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.030718683405168e-05,
+      "loss": 2.2575,
+      "step": 55096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0283425494238996e-05,
+      "loss": 2.2893,
+      "step": 55104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.025966914071701e-05,
+      "loss": 2.2761,
+      "step": 55112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0235917775868346e-05,
+      "loss": 2.278,
+      "step": 55120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0212171402075224e-05,
+      "loss": 2.2733,
+      "step": 55128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.018843002171914e-05,
+      "loss": 2.2978,
+      "step": 55136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.016469363718127e-05,
+      "loss": 2.2715,
+      "step": 55144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.014096225084229e-05,
+      "loss": 2.2603,
+      "step": 55152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.011723586508222e-05,
+      "loss": 2.2685,
+      "step": 55160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.009351448228081e-05,
+      "loss": 2.2771,
+      "step": 55168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.0069798104817075e-05,
+      "loss": 2.266,
+      "step": 55176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.004608673506965e-05,
+      "loss": 2.2639,
+      "step": 55184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.002238037541672e-05,
+      "loss": 2.2898,
+      "step": 55192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.999867902823579e-05,
+      "loss": 2.2684,
+      "step": 55200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.997498269590407e-05,
+      "loss": 2.2887,
+      "step": 55208
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.995129138079816e-05,
+      "loss": 2.289,
+      "step": 55216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.992760508529405e-05,
+      "loss": 2.2609,
+      "step": 55224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.990392381176751e-05,
+      "loss": 2.2702,
+      "step": 55232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.988024756259348e-05,
+      "loss": 2.2953,
+      "step": 55240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.985657634014665e-05,
+      "loss": 2.2795,
+      "step": 55248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9832910146801125e-05,
+      "loss": 2.2674,
+      "step": 55256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.980924898493039e-05,
+      "loss": 2.2614,
+      "step": 55264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9785592856907614e-05,
+      "loss": 2.2792,
+      "step": 55272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.976194176510533e-05,
+      "loss": 2.2681,
+      "step": 55280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.973829571189564e-05,
+      "loss": 2.2708,
+      "step": 55288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9714654699650075e-05,
+      "loss": 2.2622,
+      "step": 55296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9691018730739726e-05,
+      "loss": 2.2725,
+      "step": 55304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.966738780753514e-05,
+      "loss": 2.2564,
+      "step": 55312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.964376193240634e-05,
+      "loss": 2.276,
+      "step": 55320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.962014110772289e-05,
+      "loss": 2.2865,
+      "step": 55328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9596525335853815e-05,
+      "loss": 2.2658,
+      "step": 55336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.957291461916765e-05,
+      "loss": 2.2823,
+      "step": 55344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.95493089600324e-05,
+      "loss": 2.2641,
+      "step": 55352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.952570836081561e-05,
+      "loss": 2.2554,
+      "step": 55360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.950211282388427e-05,
+      "loss": 2.2621,
+      "step": 55368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.947852235160486e-05,
+      "loss": 2.2807,
+      "step": 55376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.945493694634341e-05,
+      "loss": 2.2777,
+      "step": 55384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.943135661046536e-05,
+      "loss": 2.2646,
+      "step": 55392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.940778134633571e-05,
+      "loss": 2.2526,
+      "step": 55400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.938421115631892e-05,
+      "loss": 2.2508,
+      "step": 55408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.936064604277891e-05,
+      "loss": 2.2786,
+      "step": 55416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.933708600807924e-05,
+      "loss": 2.2653,
+      "step": 55424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.931353105458273e-05,
+      "loss": 2.2778,
+      "step": 55432
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.928998118465185e-05,
+      "loss": 2.2677,
+      "step": 55440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.926643640064852e-05,
+      "loss": 2.2546,
+      "step": 55448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.924289670493413e-05,
+      "loss": 2.2786,
+      "step": 55456
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.921936209986962e-05,
+      "loss": 2.2825,
+      "step": 55464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.919583258781531e-05,
+      "loss": 2.2567,
+      "step": 55472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9172308171131173e-05,
+      "loss": 2.2801,
+      "step": 55480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.914878885217653e-05,
+      "loss": 2.2859,
+      "step": 55488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.912527463331016e-05,
+      "loss": 2.2617,
+      "step": 55496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.910176551689055e-05,
+      "loss": 2.2625,
+      "step": 55504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.907826150527538e-05,
+      "loss": 2.2537,
+      "step": 55512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.905476260082208e-05,
+      "loss": 2.2667,
+      "step": 55520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.903126880588744e-05,
+      "loss": 2.2586,
+      "step": 55528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.900778012282767e-05,
+      "loss": 2.2778,
+      "step": 55536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.8984296553998666e-05,
+      "loss": 2.2604,
+      "step": 55544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.8960818101755636e-05,
+      "loss": 2.2686,
+      "step": 55552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.893734476845334e-05,
+      "loss": 2.2647,
+      "step": 55560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.891387655644607e-05,
+      "loss": 2.2729,
+      "step": 55568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.889041346808743e-05,
+      "loss": 2.2707,
+      "step": 55576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.886695550573076e-05,
+      "loss": 2.2921,
+      "step": 55584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.884350267172874e-05,
+      "loss": 2.2302,
+      "step": 55592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.88200549684335e-05,
+      "loss": 2.282,
+      "step": 55600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.879661239819676e-05,
+      "loss": 2.2733,
+      "step": 55608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.877317496336965e-05,
+      "loss": 2.2822,
+      "step": 55616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.874974266630284e-05,
+      "loss": 2.2679,
+      "step": 55624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.872631550934643e-05,
+      "loss": 2.262,
+      "step": 55632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.870289349485004e-05,
+      "loss": 2.2876,
+      "step": 55640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.867947662516278e-05,
+      "loss": 2.2759,
+      "step": 55648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.8656064902633204e-05,
+      "loss": 2.258,
+      "step": 55656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.863265832960939e-05,
+      "loss": 2.265,
+      "step": 55664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.860925690843888e-05,
+      "loss": 2.2727,
+      "step": 55672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.858586064146872e-05,
+      "loss": 2.2719,
+      "step": 55680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.8562469531045364e-05,
+      "loss": 2.2517,
+      "step": 55688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.853908357951492e-05,
+      "loss": 2.2744,
+      "step": 55696
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.851570278922276e-05,
+      "loss": 2.2578,
+      "step": 55704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.849232716251389e-05,
+      "loss": 2.2551,
+      "step": 55712
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.846895670173276e-05,
+      "loss": 2.277,
+      "step": 55720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.844559140922323e-05,
+      "loss": 2.282,
+      "step": 55728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.842223128732882e-05,
+      "loss": 2.2623,
+      "step": 55736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.839887633839234e-05,
+      "loss": 2.2765,
+      "step": 55744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.837552656475614e-05,
+      "loss": 2.2713,
+      "step": 55752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.835218196876217e-05,
+      "loss": 2.2692,
+      "step": 55760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.832884255275163e-05,
+      "loss": 2.2497,
+      "step": 55768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.830550831906546e-05,
+      "loss": 2.2845,
+      "step": 55776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.828217927004386e-05,
+      "loss": 2.2795,
+      "step": 55784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.825885540802658e-05,
+      "loss": 2.2622,
+      "step": 55792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.8235536735353e-05,
+      "loss": 2.2656,
+      "step": 55800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.821222325436172e-05,
+      "loss": 2.2807,
+      "step": 55808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.8188914967391016e-05,
+      "loss": 2.2577,
+      "step": 55816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.816561187677856e-05,
+      "loss": 2.2703,
+      "step": 55824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.814231398486154e-05,
+      "loss": 2.2639,
+      "step": 55832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.8119021293976596e-05,
+      "loss": 2.2767,
+      "step": 55840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.809573380645981e-05,
+      "loss": 2.2713,
+      "step": 55848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.807245152464682e-05,
+      "loss": 2.2572,
+      "step": 55856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.8049174450872726e-05,
+      "loss": 2.2706,
+      "step": 55864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.802590258747205e-05,
+      "loss": 2.2631,
+      "step": 55872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.8002635936778834e-05,
+      "loss": 2.271,
+      "step": 55880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.7979374501126624e-05,
+      "loss": 2.266,
+      "step": 55888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.795611828284837e-05,
+      "loss": 2.2911,
+      "step": 55896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.793286728427656e-05,
+      "loss": 2.2669,
+      "step": 55904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.790962150774313e-05,
+      "loss": 2.2657,
+      "step": 55912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.78863809555795e-05,
+      "loss": 2.2661,
+      "step": 55920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.786314563011658e-05,
+      "loss": 2.2503,
+      "step": 55928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.783991553368473e-05,
+      "loss": 2.2751,
+      "step": 55936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.781669066861378e-05,
+      "loss": 2.2949,
+      "step": 55944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.7793471037233084e-05,
+      "loss": 2.2709,
+      "step": 55952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.7770256641871414e-05,
+      "loss": 2.2662,
+      "step": 55960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.7747047484857065e-05,
+      "loss": 2.2744,
+      "step": 55968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.772384356851777e-05,
+      "loss": 2.2479,
+      "step": 55976
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.7700644895180714e-05,
+      "loss": 2.2681,
+      "step": 55984
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.767745146717269e-05,
+      "loss": 2.2463,
+      "step": 55992
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.7654263286819756e-05,
+      "loss": 2.2881,
+      "step": 56000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.763108035644766e-05,
+      "loss": 2.2432,
+      "step": 56008
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.760790267838143e-05,
+      "loss": 2.2554,
+      "step": 56016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.7584730254945675e-05,
+      "loss": 2.2619,
+      "step": 56024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.756156308846452e-05,
+      "loss": 2.28,
+      "step": 56032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.753840118126141e-05,
+      "loss": 2.2596,
+      "step": 56040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.7515244535659416e-05,
+      "loss": 2.239,
+      "step": 56048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.749209315398103e-05,
+      "loss": 2.2789,
+      "step": 56056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.746894703854813e-05,
+      "loss": 2.2553,
+      "step": 56064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.744580619168223e-05,
+      "loss": 2.2766,
+      "step": 56072
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.742267061570412e-05,
+      "loss": 2.2847,
+      "step": 56080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.739954031293428e-05,
+      "loss": 2.2683,
+      "step": 56088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.7376415285692507e-05,
+      "loss": 2.2767,
+      "step": 56096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.735329553629802e-05,
+      "loss": 2.2476,
+      "step": 56104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.7330181067069746e-05,
+      "loss": 2.2882,
+      "step": 56112
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.730707188032585e-05,
+      "loss": 2.2634,
+      "step": 56120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.7283967978384066e-05,
+      "loss": 2.2764,
+      "step": 56128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.726086936356162e-05,
+      "loss": 2.2598,
+      "step": 56136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.723777603817508e-05,
+      "loss": 2.2751,
+      "step": 56144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.721468800454067e-05,
+      "loss": 2.2564,
+      "step": 56152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.719160526497396e-05,
+      "loss": 2.2599,
+      "step": 56160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.716852782179001e-05,
+      "loss": 2.2691,
+      "step": 56168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.714545567730334e-05,
+      "loss": 2.2672,
+      "step": 56176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.7122388833828e-05,
+      "loss": 2.2682,
+      "step": 56184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.709932729367744e-05,
+      "loss": 2.2751,
+      "step": 56192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.707627105916459e-05,
+      "loss": 2.2817,
+      "step": 56200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.705322013260186e-05,
+      "loss": 2.2707,
+      "step": 56208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.703017451630115e-05,
+      "loss": 2.2448,
+      "step": 56216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.70071342125738e-05,
+      "loss": 2.2641,
+      "step": 56224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.69840992237306e-05,
+      "loss": 2.2874,
+      "step": 56232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.6961069552081846e-05,
+      "loss": 2.2596,
+      "step": 56240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.693804519993728e-05,
+      "loss": 2.2775,
+      "step": 56248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.69150261696061e-05,
+      "loss": 2.2741,
+      "step": 56256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.689201246339705e-05,
+      "loss": 2.294,
+      "step": 56264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.6869004083618196e-05,
+      "loss": 2.2568,
+      "step": 56272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.684600103257716e-05,
+      "loss": 2.2354,
+      "step": 56280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.682300331258104e-05,
+      "loss": 2.2833,
+      "step": 56288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.680001092593634e-05,
+      "loss": 2.2695,
+      "step": 56296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.6777023874949166e-05,
+      "loss": 2.2639,
+      "step": 56304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.675404216192489e-05,
+      "loss": 2.2588,
+      "step": 56312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.673106578916845e-05,
+      "loss": 2.2589,
+      "step": 56320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.670809475898432e-05,
+      "loss": 2.2599,
+      "step": 56328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.668512907367626e-05,
+      "loss": 2.2749,
+      "step": 56336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.6662168735547704e-05,
+      "loss": 2.2416,
+      "step": 56344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.6639213746901354e-05,
+      "loss": 2.2671,
+      "step": 56352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.661626411003953e-05,
+      "loss": 2.2778,
+      "step": 56360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.6593319827263955e-05,
+      "loss": 2.261,
+      "step": 56368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.6570380900875716e-05,
+      "loss": 2.2711,
+      "step": 56376
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.6547447333175574e-05,
+      "loss": 2.2877,
+      "step": 56384
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.6524519126463573e-05,
+      "loss": 2.2705,
+      "step": 56392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.65015962830393e-05,
+      "loss": 2.2777,
+      "step": 56400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.647867880520181e-05,
+      "loss": 2.2828,
+      "step": 56408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.6455766695249505e-05,
+      "loss": 2.2532,
+      "step": 56416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.643285995548042e-05,
+      "loss": 2.2847,
+      "step": 56424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.640995858819196e-05,
+      "loss": 2.2827,
+      "step": 56432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.6387062595681006e-05,
+      "loss": 2.2645,
+      "step": 56440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.6364171980243886e-05,
+      "loss": 2.2786,
+      "step": 56448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.634128674417638e-05,
+      "loss": 2.281,
+      "step": 56456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.631840688977379e-05,
+      "loss": 2.245,
+      "step": 56464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.629553241933079e-05,
+      "loss": 2.2676,
+      "step": 56472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.627266333514161e-05,
+      "loss": 2.2848,
+      "step": 56480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.6249799639499835e-05,
+      "loss": 2.2738,
+      "step": 56488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.622694133469861e-05,
+      "loss": 2.2708,
+      "step": 56496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.6204088423030466e-05,
+      "loss": 2.259,
+      "step": 56504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.618124090678742e-05,
+      "loss": 2.2692,
+      "step": 56512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.6158398788260965e-05,
+      "loss": 2.2709,
+      "step": 56520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.6135562069742e-05,
+      "loss": 2.2678,
+      "step": 56528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.6112730753521e-05,
+      "loss": 2.2614,
+      "step": 56536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.608990484188774e-05,
+      "loss": 2.2572,
+      "step": 56544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.606708433713155e-05,
+      "loss": 2.2752,
+      "step": 56552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.6044269241541206e-05,
+      "loss": 2.2738,
+      "step": 56560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.60214595574049e-05,
+      "loss": 2.2668,
+      "step": 56568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5998655287010385e-05,
+      "loss": 2.2562,
+      "step": 56576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.597585643264474e-05,
+      "loss": 2.2596,
+      "step": 56584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.595306299659454e-05,
+      "loss": 2.2523,
+      "step": 56592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5930274981145935e-05,
+      "loss": 2.2567,
+      "step": 56600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.590749238858432e-05,
+      "loss": 2.2804,
+      "step": 56608
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.588471522119476e-05,
+      "loss": 2.2616,
+      "step": 56616
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5861943481261616e-05,
+      "loss": 2.2752,
+      "step": 56624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5839177171068735e-05,
+      "loss": 2.2702,
+      "step": 56632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.581641629289956e-05,
+      "loss": 2.2637,
+      "step": 56640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.579366084903675e-05,
+      "loss": 2.2654,
+      "step": 56648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5770910841762626e-05,
+      "loss": 2.2566,
+      "step": 56656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.57481662733589e-05,
+      "loss": 2.2758,
+      "step": 56664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5725427146106625e-05,
+      "loss": 2.2694,
+      "step": 56672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5702693462286546e-05,
+      "loss": 2.2783,
+      "step": 56680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.567996522417858e-05,
+      "loss": 2.2866,
+      "step": 56688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.565724243406234e-05,
+      "loss": 2.2718,
+      "step": 56696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.563452509421674e-05,
+      "loss": 2.2679,
+      "step": 56704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5611813206920246e-05,
+      "loss": 2.2636,
+      "step": 56712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5589106774450697e-05,
+      "loss": 2.2554,
+      "step": 56720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.556640579908543e-05,
+      "loss": 2.2621,
+      "step": 56728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.554371028310122e-05,
+      "loss": 2.2629,
+      "step": 56736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.552102022877431e-05,
+      "loss": 2.2645,
+      "step": 56744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.549833563838037e-05,
+      "loss": 2.2709,
+      "step": 56752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.547565651419455e-05,
+      "loss": 2.2756,
+      "step": 56760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.545298285849143e-05,
+      "loss": 2.2616,
+      "step": 56768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.543031467354505e-05,
+      "loss": 2.2743,
+      "step": 56776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.540765196162888e-05,
+      "loss": 2.2647,
+      "step": 56784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5384994725015896e-05,
+      "loss": 2.2598,
+      "step": 56792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.536234296597846e-05,
+      "loss": 2.2608,
+      "step": 56800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5339696686788466e-05,
+      "loss": 2.2827,
+      "step": 56808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.531705588971714e-05,
+      "loss": 2.2723,
+      "step": 56816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5294420577035244e-05,
+      "loss": 2.2785,
+      "step": 56824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.527179075101303e-05,
+      "loss": 2.2646,
+      "step": 56832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5249166413920085e-05,
+      "loss": 2.2802,
+      "step": 56840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.52265475680255e-05,
+      "loss": 2.2693,
+      "step": 56848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.520393421559783e-05,
+      "loss": 2.2755,
+      "step": 56856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.518132635890504e-05,
+      "loss": 2.26,
+      "step": 56864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5158724000214655e-05,
+      "loss": 2.2723,
+      "step": 56872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.513612714179345e-05,
+      "loss": 2.2801,
+      "step": 56880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.511353578590785e-05,
+      "loss": 2.2602,
+      "step": 56888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.509094993482362e-05,
+      "loss": 2.2683,
+      "step": 56896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.506836959080593e-05,
+      "loss": 2.2729,
+      "step": 56904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.504579475611957e-05,
+      "loss": 2.2713,
+      "step": 56912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.502322543302856e-05,
+      "loss": 2.258,
+      "step": 56920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5000661623796537e-05,
+      "loss": 2.2705,
+      "step": 56928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.497810333068654e-05,
+      "loss": 2.2685,
+      "step": 56936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4955550555960954e-05,
+      "loss": 2.2703,
+      "step": 56944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.493300330188178e-05,
+      "loss": 2.2767,
+      "step": 56952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.491046157071034e-05,
+      "loss": 2.2705,
+      "step": 56960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.488792536470746e-05,
+      "loss": 2.2525,
+      "step": 56968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.486539468613341e-05,
+      "loss": 2.2978,
+      "step": 56976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4842869537247796e-05,
+      "loss": 2.2742,
+      "step": 56984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4820349920309864e-05,
+      "loss": 2.2796,
+      "step": 56992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4797835837578184e-05,
+      "loss": 2.2635,
+      "step": 57000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.477532729131076e-05,
+      "loss": 2.2512,
+      "step": 57008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.475282428376509e-05,
+      "loss": 2.2744,
+      "step": 57016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.473032681719809e-05,
+      "loss": 2.2679,
+      "step": 57024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.470783489386615e-05,
+      "loss": 2.2667,
+      "step": 57032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.468534851602507e-05,
+      "loss": 2.2784,
+      "step": 57040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.46628676859301e-05,
+      "loss": 2.2628,
+      "step": 57048
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.464039240583594e-05,
+      "loss": 2.2626,
+      "step": 57056
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4617922677996726e-05,
+      "loss": 2.2883,
+      "step": 57064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.459545850466607e-05,
+      "loss": 2.2652,
+      "step": 57072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4572999888096974e-05,
+      "loss": 2.255,
+      "step": 57080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.455054683054194e-05,
+      "loss": 2.2658,
+      "step": 57088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.452809933425282e-05,
+      "loss": 2.266,
+      "step": 57096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.450565740148108e-05,
+      "loss": 2.2655,
+      "step": 57104
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.448322103447743e-05,
+      "loss": 2.2552,
+      "step": 57112
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.446079023549212e-05,
+      "loss": 2.262,
+      "step": 57120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.443836500677486e-05,
+      "loss": 2.2891,
+      "step": 57128
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4415945350574724e-05,
+      "loss": 2.2806,
+      "step": 57136
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4393531269140374e-05,
+      "loss": 2.2686,
+      "step": 57144
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.437112276471974e-05,
+      "loss": 2.2849,
+      "step": 57152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4348719839560235e-05,
+      "loss": 2.2605,
+      "step": 57160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.432632249590885e-05,
+      "loss": 2.269,
+      "step": 57168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.430393073601181e-05,
+      "loss": 2.2482,
+      "step": 57176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.428154456211496e-05,
+      "loss": 2.2798,
+      "step": 57184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.425916397646347e-05,
+      "loss": 2.264,
+      "step": 57192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.423678898130194e-05,
+      "loss": 2.2343,
+      "step": 57200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.421441957887457e-05,
+      "loss": 2.2502,
+      "step": 57208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.419205577142476e-05,
+      "loss": 2.2836,
+      "step": 57216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.416969756119556e-05,
+      "loss": 2.2618,
+      "step": 57224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.414734495042935e-05,
+      "loss": 2.2657,
+      "step": 57232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.412499794136794e-05,
+      "loss": 2.2765,
+      "step": 57240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.410265653625268e-05,
+      "loss": 2.2722,
+      "step": 57248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.408032073732416e-05,
+      "loss": 2.2637,
+      "step": 57256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.4057990546822654e-05,
+      "loss": 2.2567,
+      "step": 57264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.4035665966987726e-05,
+      "loss": 2.2322,
+      "step": 57272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.4013347000058366e-05,
+      "loss": 2.2844,
+      "step": 57280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.399103364827307e-05,
+      "loss": 2.272,
+      "step": 57288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.396872591386974e-05,
+      "loss": 2.2665,
+      "step": 57296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3946423799085724e-05,
+      "loss": 2.2653,
+      "step": 57304
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.392412730615777e-05,
+      "loss": 2.2621,
+      "step": 57312
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.390183643732209e-05,
+      "loss": 2.2858,
+      "step": 57320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.387955119481436e-05,
+      "loss": 2.2604,
+      "step": 57328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.385727158086965e-05,
+      "loss": 2.2699,
+      "step": 57336
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.383499759772248e-05,
+      "loss": 2.2819,
+      "step": 57344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3812729247606785e-05,
+      "loss": 2.2717,
+      "step": 57352
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.379046653275599e-05,
+      "loss": 2.2726,
+      "step": 57360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3768209455402896e-05,
+      "loss": 2.2539,
+      "step": 57368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.374595801777977e-05,
+      "loss": 2.2691,
+      "step": 57376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.372371222211829e-05,
+      "loss": 2.2676,
+      "step": 57384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3701472070649606e-05,
+      "loss": 2.2384,
+      "step": 57392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3679237565604266e-05,
+      "loss": 2.2666,
+      "step": 57400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3657008709212243e-05,
+      "loss": 2.2604,
+      "step": 57408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3634785503703046e-05,
+      "loss": 2.2726,
+      "step": 57416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.361256795130544e-05,
+      "loss": 2.2547,
+      "step": 57424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.359035605424774e-05,
+      "loss": 2.2619,
+      "step": 57432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.356814981475774e-05,
+      "loss": 2.2696,
+      "step": 57440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.354594923506251e-05,
+      "loss": 2.2559,
+      "step": 57448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.352375431738874e-05,
+      "loss": 2.2776,
+      "step": 57456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.350156506396236e-05,
+      "loss": 2.2533,
+      "step": 57464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.347938147700885e-05,
+      "loss": 2.2783,
+      "step": 57472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3457203558753166e-05,
+      "loss": 2.259,
+      "step": 57480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3435031311419514e-05,
+      "loss": 2.2464,
+      "step": 57488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.341286473723175e-05,
+      "loss": 2.2748,
+      "step": 57496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.339070383841304e-05,
+      "loss": 2.2657,
+      "step": 57504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3368548617185895e-05,
+      "loss": 2.265,
+      "step": 57512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.33463990757725e-05,
+      "loss": 2.2693,
+      "step": 57520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3324255216394194e-05,
+      "loss": 2.2854,
+      "step": 57528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.330211704127199e-05,
+      "loss": 2.2615,
+      "step": 57536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3279984552626203e-05,
+      "loss": 2.2776,
+      "step": 57544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.325785775267652e-05,
+      "loss": 2.2665,
+      "step": 57552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.323573664364221e-05,
+      "loss": 2.2507,
+      "step": 57560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.321362122774186e-05,
+      "loss": 2.2547,
+      "step": 57568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.319151150719355e-05,
+      "loss": 2.2663,
+      "step": 57576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.316940748421473e-05,
+      "loss": 2.2796,
+      "step": 57584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.314730916102231e-05,
+      "loss": 2.2959,
+      "step": 57592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3125216539832645e-05,
+      "loss": 2.2668,
+      "step": 57600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.31031296228615e-05,
+      "loss": 2.2696,
+      "step": 57608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3081048412324045e-05,
+      "loss": 2.2499,
+      "step": 57616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.305897291043492e-05,
+      "loss": 2.2821,
+      "step": 57624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.303690311940817e-05,
+      "loss": 2.2699,
+      "step": 57632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3014839041457254e-05,
+      "loss": 2.2495,
+      "step": 57640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2992780678795105e-05,
+      "loss": 2.2517,
+      "step": 57648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.297072803363402e-05,
+      "loss": 2.261,
+      "step": 57656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2948681108185763e-05,
+      "loss": 2.256,
+      "step": 57664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2926639904661534e-05,
+      "loss": 2.2452,
+      "step": 57672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.290460442527192e-05,
+      "loss": 2.2529,
+      "step": 57680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.288257467222697e-05,
+      "loss": 2.2699,
+      "step": 57688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.286055064773613e-05,
+      "loss": 2.2599,
+      "step": 57696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.283853235400825e-05,
+      "loss": 2.2443,
+      "step": 57704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.281651979325175e-05,
+      "loss": 2.2547,
+      "step": 57712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.279451296767426e-05,
+      "loss": 2.27,
+      "step": 57720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2772511879482976e-05,
+      "loss": 2.2542,
+      "step": 57728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.27505165308845e-05,
+      "loss": 2.2607,
+      "step": 57736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.272852692408476e-05,
+      "loss": 2.2777,
+      "step": 57744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.270654306128934e-05,
+      "loss": 2.2735,
+      "step": 57752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.268456494470293e-05,
+      "loss": 2.2563,
+      "step": 57760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2662592576529924e-05,
+      "loss": 2.2809,
+      "step": 57768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.264062595897403e-05,
+      "loss": 2.2711,
+      "step": 57776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.261866509423827e-05,
+      "loss": 2.2724,
+      "step": 57784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.259670998452533e-05,
+      "loss": 2.2523,
+      "step": 57792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.257476063203703e-05,
+      "loss": 2.2774,
+      "step": 57800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.255281703897489e-05,
+      "loss": 2.2759,
+      "step": 57808
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.253087920753972e-05,
+      "loss": 2.2686,
+      "step": 57816
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.250894713993166e-05,
+      "loss": 2.2651,
+      "step": 57824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.248702083835049e-05,
+      "loss": 2.2669,
+      "step": 57832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.246510030499522e-05,
+      "loss": 2.2789,
+      "step": 57840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.244318554206439e-05,
+      "loss": 2.267,
+      "step": 57848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.242127655175595e-05,
+      "loss": 2.2503,
+      "step": 57856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2399373336267145e-05,
+      "loss": 2.2873,
+      "step": 57864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2377475897794835e-05,
+      "loss": 2.2617,
+      "step": 57872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.235558423853519e-05,
+      "loss": 2.2638,
+      "step": 57880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2333698360683825e-05,
+      "loss": 2.2769,
+      "step": 57888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2311818266435744e-05,
+      "loss": 2.2754,
+      "step": 57896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.228994395798542e-05,
+      "loss": 2.2696,
+      "step": 57904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.226807543752672e-05,
+      "loss": 2.2594,
+      "step": 57912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2246212707252915e-05,
+      "loss": 2.25,
+      "step": 57920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.222435576935674e-05,
+      "loss": 2.2746,
+      "step": 57928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.220250462603026e-05,
+      "loss": 2.2557,
+      "step": 57936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.218065927946514e-05,
+      "loss": 2.2697,
+      "step": 57944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2158819731852226e-05,
+      "loss": 2.2683,
+      "step": 57952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.213698598538195e-05,
+      "loss": 2.2706,
+      "step": 57960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2115158042244125e-05,
+      "loss": 2.2706,
+      "step": 57968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2093335904627915e-05,
+      "loss": 2.2562,
+      "step": 57976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.207151957472205e-05,
+      "loss": 2.2473,
+      "step": 57984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2049709054714515e-05,
+      "loss": 2.2694,
+      "step": 57992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2027904346792766e-05,
+      "loss": 2.2525,
+      "step": 58000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.200610545314377e-05,
+      "loss": 2.2458,
+      "step": 58008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.198431237595374e-05,
+      "loss": 2.2497,
+      "step": 58016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.196252511740849e-05,
+      "loss": 2.2732,
+      "step": 58024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1940743679693084e-05,
+      "loss": 2.2718,
+      "step": 58032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.191896806499208e-05,
+      "loss": 2.2768,
+      "step": 58040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1897198275489534e-05,
+      "loss": 2.2617,
+      "step": 58048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.187543431336871e-05,
+      "loss": 2.2784,
+      "step": 58056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1853676180812493e-05,
+      "loss": 2.2603,
+      "step": 58064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1831923880003106e-05,
+      "loss": 2.2429,
+      "step": 58072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1810177413122144e-05,
+      "loss": 2.2496,
+      "step": 58080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.178843678235068e-05,
+      "loss": 2.2738,
+      "step": 58088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1766701989869106e-05,
+      "loss": 2.273,
+      "step": 58096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1744973037857405e-05,
+      "loss": 2.2488,
+      "step": 58104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1723249928494796e-05,
+      "loss": 2.2882,
+      "step": 58112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.170153266396003e-05,
+      "loss": 2.2522,
+      "step": 58120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.167982124643118e-05,
+      "loss": 2.2537,
+      "step": 58128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.165811567808578e-05,
+      "loss": 2.2554,
+      "step": 58136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.163641596110083e-05,
+      "loss": 2.2703,
+      "step": 58144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.161472209765263e-05,
+      "loss": 2.2505,
+      "step": 58152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.159303408991699e-05,
+      "loss": 2.2432,
+      "step": 58160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.157135194006908e-05,
+      "loss": 2.2623,
+      "step": 58168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.154967565028348e-05,
+      "loss": 2.2529,
+      "step": 58176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1528005222734224e-05,
+      "loss": 2.2724,
+      "step": 58184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.150634065959472e-05,
+      "loss": 2.2669,
+      "step": 58192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.148468196303781e-05,
+      "loss": 2.2745,
+      "step": 58200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1463029135235726e-05,
+      "loss": 2.2594,
+      "step": 58208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1441382178360144e-05,
+      "loss": 2.2645,
+      "step": 58216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1419741094582095e-05,
+      "loss": 2.2529,
+      "step": 58224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.13981058860721e-05,
+      "loss": 2.2685,
+      "step": 58232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.137647655500002e-05,
+      "loss": 2.2719,
+      "step": 58240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.135485310353513e-05,
+      "loss": 2.2649,
+      "step": 58248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1333235533846236e-05,
+      "loss": 2.2593,
+      "step": 58256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1311623848101355e-05,
+      "loss": 2.2612,
+      "step": 58264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1290018048468046e-05,
+      "loss": 2.2477,
+      "step": 58272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1268418137113304e-05,
+      "loss": 2.2554,
+      "step": 58280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1246824116203384e-05,
+      "loss": 2.2758,
+      "step": 58288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.122523598790413e-05,
+      "loss": 2.2683,
+      "step": 58296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.120365375438065e-05,
+      "loss": 2.2543,
+      "step": 58304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.11820774177975e-05,
+      "loss": 2.2582,
+      "step": 58312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.116050698031878e-05,
+      "loss": 2.2769,
+      "step": 58320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.113894244410775e-05,
+      "loss": 2.2634,
+      "step": 58328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.111738381132728e-05,
+      "loss": 2.2719,
+      "step": 58336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.109583108413962e-05,
+      "loss": 2.259,
+      "step": 58344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.107428426470625e-05,
+      "loss": 2.2601,
+      "step": 58352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.105274335518835e-05,
+      "loss": 2.2699,
+      "step": 58360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.10312083577462e-05,
+      "loss": 2.2618,
+      "step": 58368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1009679274539766e-05,
+      "loss": 2.2558,
+      "step": 58376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0988156107728266e-05,
+      "loss": 2.258,
+      "step": 58384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.096663885947026e-05,
+      "loss": 2.2656,
+      "step": 58392
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.094512753192393e-05,
+      "loss": 2.2553,
+      "step": 58400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0923622127246665e-05,
+      "loss": 2.2579,
+      "step": 58408
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.090212264759536e-05,
+      "loss": 2.2642,
+      "step": 58416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.08806290951263e-05,
+      "loss": 2.2649,
+      "step": 58424
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.085914147199513e-05,
+      "loss": 2.2518,
+      "step": 58432
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.083765978035697e-05,
+      "loss": 2.2428,
+      "step": 58440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.08161840223663e-05,
+      "loss": 2.2589,
+      "step": 58448
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.079471420017701e-05,
+      "loss": 2.2557,
+      "step": 58456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.077325031594242e-05,
+      "loss": 2.2665,
+      "step": 58464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.075179237181523e-05,
+      "loss": 2.26,
+      "step": 58472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.073034036994755e-05,
+      "loss": 2.251,
+      "step": 58480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.070889431249087e-05,
+      "loss": 2.2564,
+      "step": 58488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.068745420159615e-05,
+      "loss": 2.2553,
+      "step": 58496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.066602003941368e-05,
+      "loss": 2.2559,
+      "step": 58504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.064459182809319e-05,
+      "loss": 2.2671,
+      "step": 58512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.062316956978381e-05,
+      "loss": 2.2548,
+      "step": 58520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.06017532666341e-05,
+      "loss": 2.2666,
+      "step": 58528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.058034292079195e-05,
+      "loss": 2.2443,
+      "step": 58536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.055893853440468e-05,
+      "loss": 2.2851,
+      "step": 58544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.053754010961913e-05,
+      "loss": 2.264,
+      "step": 58552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0516147648581345e-05,
+      "loss": 2.2671,
+      "step": 58560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.04947611534369e-05,
+      "loss": 2.2406,
+      "step": 58568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0473380626330736e-05,
+      "loss": 2.2823,
+      "step": 58576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.045200606940717e-05,
+      "loss": 2.2757,
+      "step": 58584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0430637484810027e-05,
+      "loss": 2.2609,
+      "step": 58592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0409274874682346e-05,
+      "loss": 2.2437,
+      "step": 58600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0387918241166775e-05,
+      "loss": 2.2704,
+      "step": 58608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.036656758640524e-05,
+      "loss": 2.2557,
+      "step": 58616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.034522291253903e-05,
+      "loss": 2.2417,
+      "step": 58624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0323884221708994e-05,
+      "loss": 2.2608,
+      "step": 58632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.030255151605516e-05,
+      "loss": 2.2723,
+      "step": 58640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.02812247977172e-05,
+      "loss": 2.2558,
+      "step": 58648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0259904068834004e-05,
+      "loss": 2.2481,
+      "step": 58656
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.023858933154388e-05,
+      "loss": 2.2676,
+      "step": 58664
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.021728058798465e-05,
+      "loss": 2.2517,
+      "step": 58672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.019597784029342e-05,
+      "loss": 2.2712,
+      "step": 58680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.017468109060674e-05,
+      "loss": 2.254,
+      "step": 58688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0153390341060614e-05,
+      "loss": 2.2729,
+      "step": 58696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0132105593790256e-05,
+      "loss": 2.2615,
+      "step": 58704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0110826850930504e-05,
+      "loss": 2.279,
+      "step": 58712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.008955411461547e-05,
+      "loss": 2.2556,
+      "step": 58720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0068287386978685e-05,
+      "loss": 2.2685,
+      "step": 58728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0047026670153076e-05,
+      "loss": 2.2605,
+      "step": 58736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.002577196627099e-05,
+      "loss": 2.2489,
+      "step": 58744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.000452327746413e-05,
+      "loss": 2.2541,
+      "step": 58752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9983280605863644e-05,
+      "loss": 2.2741,
+      "step": 58760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.996204395360004e-05,
+      "loss": 2.2745,
+      "step": 58768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9940813322803196e-05,
+      "loss": 2.2645,
+      "step": 58776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9919588715602534e-05,
+      "loss": 2.2517,
+      "step": 58784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.989837013412665e-05,
+      "loss": 2.2707,
+      "step": 58792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.987715758050368e-05,
+      "loss": 2.2534,
+      "step": 58800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.985595105686114e-05,
+      "loss": 2.2723,
+      "step": 58808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9834750565325885e-05,
+      "loss": 2.2646,
+      "step": 58816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.981355610802429e-05,
+      "loss": 2.2831,
+      "step": 58824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9792367687081954e-05,
+      "loss": 2.2833,
+      "step": 58832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.977118530462395e-05,
+      "loss": 2.254,
+      "step": 58840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9750008962774855e-05,
+      "loss": 2.2512,
+      "step": 58848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.972883866365839e-05,
+      "loss": 2.2677,
+      "step": 58856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9707674409397967e-05,
+      "loss": 2.2542,
+      "step": 58864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9686516202116125e-05,
+      "loss": 2.2701,
+      "step": 58872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.966536404393493e-05,
+      "loss": 2.2454,
+      "step": 58880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.96442179369759e-05,
+      "loss": 2.264,
+      "step": 58888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.962307788335976e-05,
+      "loss": 2.2557,
+      "step": 58896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9601943885206826e-05,
+      "loss": 2.2735,
+      "step": 58904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.958081594463671e-05,
+      "loss": 2.2647,
+      "step": 58912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.955969406376835e-05,
+      "loss": 2.264,
+      "step": 58920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9538578244720246e-05,
+      "loss": 2.267,
+      "step": 58928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.951746848961011e-05,
+      "loss": 2.2647,
+      "step": 58936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.949636480055519e-05,
+      "loss": 2.2571,
+      "step": 58944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.947526717967204e-05,
+      "loss": 2.2709,
+      "step": 58952
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.945417562907665e-05,
+      "loss": 2.2724,
+      "step": 58960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9433090150884376e-05,
+      "loss": 2.254,
+      "step": 58968
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.941201074720997e-05,
+      "loss": 2.26,
+      "step": 58976
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.939093742016757e-05,
+      "loss": 2.2725,
+      "step": 58984
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.936987017187072e-05,
+      "loss": 2.2711,
+      "step": 58992
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9348809004432336e-05,
+      "loss": 2.2531,
+      "step": 59000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9327753919964744e-05,
+      "loss": 2.2927,
+      "step": 59008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9306704920579643e-05,
+      "loss": 2.2562,
+      "step": 59016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.928566200838814e-05,
+      "loss": 2.2695,
+      "step": 59024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.92646251855007e-05,
+      "loss": 2.2561,
+      "step": 59032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.924359445402722e-05,
+      "loss": 2.2558,
+      "step": 59040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.922256981607694e-05,
+      "loss": 2.257,
+      "step": 59048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9201551273758534e-05,
+      "loss": 2.266,
+      "step": 59056
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9180538829180025e-05,
+      "loss": 2.2699,
+      "step": 59064
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9159532484448856e-05,
+      "loss": 2.256,
+      "step": 59072
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9138532241671834e-05,
+      "loss": 2.2541,
+      "step": 59080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.911753810295516e-05,
+      "loss": 2.2645,
+      "step": 59088
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.909655007040445e-05,
+      "loss": 2.2523,
+      "step": 59096
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.907556814612466e-05,
+      "loss": 2.2626,
+      "step": 59104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.905459233222014e-05,
+      "loss": 2.2592,
+      "step": 59112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.903362263079474e-05,
+      "loss": 2.2509,
+      "step": 59120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.901265904395147e-05,
+      "loss": 2.2672,
+      "step": 59128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.899170157379298e-05,
+      "loss": 2.2619,
+      "step": 59136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.89707502224211e-05,
+      "loss": 2.2566,
+      "step": 59144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.894980499193715e-05,
+      "loss": 2.2711,
+      "step": 59152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.892886588444187e-05,
+      "loss": 2.2482,
+      "step": 59160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.890793290203524e-05,
+      "loss": 2.2656,
+      "step": 59168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8887006046816796e-05,
+      "loss": 2.268,
+      "step": 59176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.88660853208854e-05,
+      "loss": 2.2701,
+      "step": 59184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.884517072633918e-05,
+      "loss": 2.2545,
+      "step": 59192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8824262265275874e-05,
+      "loss": 2.2785,
+      "step": 59200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.880335993979236e-05,
+      "loss": 2.2756,
+      "step": 59208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8782463751985125e-05,
+      "loss": 2.27,
+      "step": 59216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876157370394991e-05,
+      "loss": 2.2473,
+      "step": 59224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.874068979778181e-05,
+      "loss": 2.2614,
+      "step": 59232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.871981203557544e-05,
+      "loss": 2.2744,
+      "step": 59240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.869894041942468e-05,
+      "loss": 2.2613,
+      "step": 59248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8678074951422846e-05,
+      "loss": 2.2461,
+      "step": 59256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.865721563366266e-05,
+      "loss": 2.2814,
+      "step": 59264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.86363624682361e-05,
+      "loss": 2.2585,
+      "step": 59272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8615515457234704e-05,
+      "loss": 2.2589,
+      "step": 59280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.859467460274929e-05,
+      "loss": 2.2708,
+      "step": 59288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8573839906870064e-05,
+      "loss": 2.252,
+      "step": 59296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.855301137168664e-05,
+      "loss": 2.2657,
+      "step": 59304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8532188999288004e-05,
+      "loss": 2.2588,
+      "step": 59312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8511372791762504e-05,
+      "loss": 2.2502,
+      "step": 59320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.849056275119791e-05,
+      "loss": 2.2678,
+      "step": 59328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8469758879681324e-05,
+      "loss": 2.2414,
+      "step": 59336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.844896117929929e-05,
+      "loss": 2.2691,
+      "step": 59344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8428169652137666e-05,
+      "loss": 2.2592,
+      "step": 59352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.840738430028176e-05,
+      "loss": 2.245,
+      "step": 59360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.838660512581619e-05,
+      "loss": 2.2691,
+      "step": 59368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.836583213082501e-05,
+      "loss": 2.261,
+      "step": 59376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8345065317391604e-05,
+      "loss": 2.2605,
+      "step": 59384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.832430468759885e-05,
+      "loss": 2.246,
+      "step": 59392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.830355024352882e-05,
+      "loss": 2.2747,
+      "step": 59400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.828280198726312e-05,
+      "loss": 2.2543,
+      "step": 59408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.826205992088267e-05,
+      "loss": 2.2804,
+      "step": 59416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8241324046467754e-05,
+      "loss": 2.2683,
+      "step": 59424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.822059436609815e-05,
+      "loss": 2.2423,
+      "step": 59432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819987088185284e-05,
+      "loss": 2.2525,
+      "step": 59440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817915359581028e-05,
+      "loss": 2.2587,
+      "step": 59448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.815844251004836e-05,
+      "loss": 2.2466,
+      "step": 59456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8137737626644196e-05,
+      "loss": 2.2624,
+      "step": 59464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.811703894767447e-05,
+      "loss": 2.2626,
+      "step": 59472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.809634647521504e-05,
+      "loss": 2.2623,
+      "step": 59480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.807566021134131e-05,
+      "loss": 2.2652,
+      "step": 59488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.805498015812802e-05,
+      "loss": 2.2524,
+      "step": 59496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.803430631764915e-05,
+      "loss": 2.2492,
+      "step": 59504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.801363869197829e-05,
+      "loss": 2.25,
+      "step": 59512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.79929772831882e-05,
+      "loss": 2.2514,
+      "step": 59520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.797232209335117e-05,
+      "loss": 2.2737,
+      "step": 59528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.795167312453878e-05,
+      "loss": 2.2604,
+      "step": 59536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.793103037882193e-05,
+      "loss": 2.2621,
+      "step": 59544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.791039385827108e-05,
+      "loss": 2.2785,
+      "step": 59552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.788976356495588e-05,
+      "loss": 2.2312,
+      "step": 59560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.786913950094547e-05,
+      "loss": 2.2622,
+      "step": 59568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.784852166830831e-05,
+      "loss": 2.2517,
+      "step": 59576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.782791006911227e-05,
+      "loss": 2.2701,
+      "step": 59584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.780730470542455e-05,
+      "loss": 2.2509,
+      "step": 59592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.778670557931177e-05,
+      "loss": 2.2628,
+      "step": 59600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.77661126928399e-05,
+      "loss": 2.2622,
+      "step": 59608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7745526048074293e-05,
+      "loss": 2.2653,
+      "step": 59616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7724945647079664e-05,
+      "loss": 2.2535,
+      "step": 59624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.770437149192012e-05,
+      "loss": 2.2442,
+      "step": 59632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.768380358465913e-05,
+      "loss": 2.2534,
+      "step": 59640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.766324192735954e-05,
+      "loss": 2.2732,
+      "step": 59648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.764268652208353e-05,
+      "loss": 2.2776,
+      "step": 59656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.762213737089279e-05,
+      "loss": 2.2534,
+      "step": 59664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.760159447584819e-05,
+      "loss": 2.2387,
+      "step": 59672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7581057839010054e-05,
+      "loss": 2.2604,
+      "step": 59680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.756052746243821e-05,
+      "loss": 2.261,
+      "step": 59688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.754000334819158e-05,
+      "loss": 2.2597,
+      "step": 59696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.751948549832877e-05,
+      "loss": 2.2471,
+      "step": 59704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.74989739149075e-05,
+      "loss": 2.2674,
+      "step": 59712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.747846859998496e-05,
+      "loss": 2.2687,
+      "step": 59720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7457969555617815e-05,
+      "loss": 2.2469,
+      "step": 59728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.743747678386188e-05,
+      "loss": 2.2625,
+      "step": 59736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.741699028677256e-05,
+      "loss": 2.2473,
+      "step": 59744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7396510066404523e-05,
+      "loss": 2.2488,
+      "step": 59752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7376036124811735e-05,
+      "loss": 2.256,
+      "step": 59760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.735556846404774e-05,
+      "loss": 2.2509,
+      "step": 59768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.733510708616521e-05,
+      "loss": 2.2807,
+      "step": 59776
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.731465199321638e-05,
+      "loss": 2.2686,
+      "step": 59784
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.72942031872528e-05,
+      "loss": 2.2668,
+      "step": 59792
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.727376067032527e-05,
+      "loss": 2.258,
+      "step": 59800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.725332444448414e-05,
+      "loss": 2.2759,
+      "step": 59808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.723289451177903e-05,
+      "loss": 2.249,
+      "step": 59816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.721247087425894e-05,
+      "loss": 2.2611,
+      "step": 59824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.719205353397224e-05,
+      "loss": 2.2587,
+      "step": 59832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.71716424929667e-05,
+      "loss": 2.2451,
+      "step": 59840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.71512377532894e-05,
+      "loss": 2.2624,
+      "step": 59848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.713083931698683e-05,
+      "loss": 2.2421,
+      "step": 59856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.711044718610486e-05,
+      "loss": 2.2491,
+      "step": 59864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.709006136268866e-05,
+      "loss": 2.2604,
+      "step": 59872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.706968184878285e-05,
+      "loss": 2.2763,
+      "step": 59880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.704930864643137e-05,
+      "loss": 2.23,
+      "step": 59888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7028941757677536e-05,
+      "loss": 2.2531,
+      "step": 59896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7008581184564024e-05,
+      "loss": 2.2575,
+      "step": 59904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6988226929132886e-05,
+      "loss": 2.2693,
+      "step": 59912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6967878993425544e-05,
+      "loss": 2.2584,
+      "step": 59920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6947537379482794e-05,
+      "loss": 2.2517,
+      "step": 59928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6927202089344756e-05,
+      "loss": 2.2583,
+      "step": 59936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.690687312505096e-05,
+      "loss": 2.2711,
+      "step": 59944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6886550488640266e-05,
+      "loss": 2.2534,
+      "step": 59952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.686623418215098e-05,
+      "loss": 2.2674,
+      "step": 59960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.684592420762065e-05,
+      "loss": 2.2434,
+      "step": 59968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.682562056708627e-05,
+      "loss": 2.2703,
+      "step": 59976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6805323262584185e-05,
+      "loss": 2.2588,
+      "step": 59984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6785032296150055e-05,
+      "loss": 2.2544,
+      "step": 59992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.676474766981906e-05,
+      "loss": 2.2517,
+      "step": 60000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.67444693856255e-05,
+      "loss": 2.2716,
+      "step": 60008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.672419744560326e-05,
+      "loss": 2.2362,
+      "step": 60016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6703931851785503e-05,
+      "loss": 2.2652,
+      "step": 60024
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6683672606204674e-05,
+      "loss": 2.26,
+      "step": 60032
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.666341971089277e-05,
+      "loss": 2.2544,
+      "step": 60040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6643173167880934e-05,
+      "loss": 2.2547,
+      "step": 60048
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6622932979199846e-05,
+      "loss": 2.2591,
+      "step": 60056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6602699146879495e-05,
+      "loss": 2.2613,
+      "step": 60064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.658247167294915e-05,
+      "loss": 2.259,
+      "step": 60072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.656225055943756e-05,
+      "loss": 2.2496,
+      "step": 60080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.654203580837279e-05,
+      "loss": 2.2542,
+      "step": 60088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.652182742178226e-05,
+      "loss": 2.2591,
+      "step": 60096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6501625401692775e-05,
+      "loss": 2.2519,
+      "step": 60104
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6481429750130404e-05,
+      "loss": 2.2558,
+      "step": 60112
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6461240469120746e-05,
+      "loss": 2.2639,
+      "step": 60120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.644105756068864e-05,
+      "loss": 2.2614,
+      "step": 60128
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6420881026858306e-05,
+      "loss": 2.2494,
+      "step": 60136
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.640071086965335e-05,
+      "loss": 2.2375,
+      "step": 60144
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.638054709109672e-05,
+      "loss": 2.2415,
+      "step": 60152
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6360389693210735e-05,
+      "loss": 2.238,
+      "step": 60160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.634023867801705e-05,
+      "loss": 2.2491,
+      "step": 60168
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6320094047536716e-05,
+      "loss": 2.2502,
+      "step": 60176
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.629995580379013e-05,
+      "loss": 2.2501,
+      "step": 60184
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6279823948797024e-05,
+      "loss": 2.2437,
+      "step": 60192
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6259698484576505e-05,
+      "loss": 2.2367,
+      "step": 60200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.623957941314707e-05,
+      "loss": 2.2303,
+      "step": 60208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6219466736526536e-05,
+      "loss": 2.2475,
+      "step": 60216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.619936045673205e-05,
+      "loss": 2.2527,
+      "step": 60224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.617926057578026e-05,
+      "loss": 2.2613,
+      "step": 60232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6159167095686974e-05,
+      "loss": 2.268,
+      "step": 60240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.613908001846747e-05,
+      "loss": 2.2523,
+      "step": 60248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.611899934613641e-05,
+      "loss": 2.2471,
+      "step": 60256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.609892508070771e-05,
+      "loss": 2.2448,
+      "step": 60264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6078857224194794e-05,
+      "loss": 2.2537,
+      "step": 60272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.605879577861027e-05,
+      "loss": 2.2655,
+      "step": 60280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6038740745966195e-05,
+      "loss": 2.2713,
+      "step": 60288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.601869212827405e-05,
+      "loss": 2.2501,
+      "step": 60296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5998649927544504e-05,
+      "loss": 2.2539,
+      "step": 60304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5978614145787766e-05,
+      "loss": 2.2493,
+      "step": 60312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.595858478501323e-05,
+      "loss": 2.2695,
+      "step": 60320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.593856184722974e-05,
+      "loss": 2.2677,
+      "step": 60328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.591854533444556e-05,
+      "loss": 2.2654,
+      "step": 60336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.589853524866812e-05,
+      "loss": 2.2458,
+      "step": 60344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5878531591904404e-05,
+      "loss": 2.2535,
+      "step": 60352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.585853436616064e-05,
+      "loss": 2.2374,
+      "step": 60360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5838543573442426e-05,
+      "loss": 2.2734,
+      "step": 60368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.581855921575477e-05,
+      "loss": 2.2392,
+      "step": 60376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.579858129510189e-05,
+      "loss": 2.25,
+      "step": 60384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5778609813487556e-05,
+      "loss": 2.2607,
+      "step": 60392
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5758644772914745e-05,
+      "loss": 2.2629,
+      "step": 60400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.573868617538587e-05,
+      "loss": 2.228,
+      "step": 60408
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.571873402290263e-05,
+      "loss": 2.239,
+      "step": 60416
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.569878831746614e-05,
+      "loss": 2.2614,
+      "step": 60424
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.567884906107683e-05,
+      "loss": 2.2629,
+      "step": 60432
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5658916255734495e-05,
+      "loss": 2.2453,
+      "step": 60440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.563898990343828e-05,
+      "loss": 2.2656,
+      "step": 60448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.561907000618669e-05,
+      "loss": 2.2579,
+      "step": 60456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.559915656597757e-05,
+      "loss": 2.2475,
+      "step": 60464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.557924958480814e-05,
+      "loss": 2.2645,
+      "step": 60472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.555934906467495e-05,
+      "loss": 2.2303,
+      "step": 60480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.553945500757391e-05,
+      "loss": 2.2486,
+      "step": 60488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.551956741550026e-05,
+      "loss": 2.2518,
+      "step": 60496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5499686290448644e-05,
+      "loss": 2.2583,
+      "step": 60504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.547981163441302e-05,
+      "loss": 2.2638,
+      "step": 60512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.545994344938667e-05,
+      "loss": 2.2542,
+      "step": 60520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.544008173736234e-05,
+      "loss": 2.2595,
+      "step": 60528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5420226500331934e-05,
+      "loss": 2.2627,
+      "step": 60536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.540037774028694e-05,
+      "loss": 2.2492,
+      "step": 60544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.538053545921799e-05,
+      "loss": 2.2411,
+      "step": 60552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5360699659115156e-05,
+      "loss": 2.2633,
+      "step": 60560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5340870341967925e-05,
+      "loss": 2.2557,
+      "step": 60568
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5321047509764964e-05,
+      "loss": 2.2639,
+      "step": 60576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.530123116449449e-05,
+      "loss": 2.2273,
+      "step": 60584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.528142130814394e-05,
+      "loss": 2.2538,
+      "step": 60592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.526161794270007e-05,
+      "loss": 2.2556,
+      "step": 60600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5241821070149136e-05,
+      "loss": 2.2631,
+      "step": 60608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5222030692476564e-05,
+      "loss": 2.257,
+      "step": 60616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.520224681166728e-05,
+      "loss": 2.2344,
+      "step": 60624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.518246942970552e-05,
+      "loss": 2.2627,
+      "step": 60632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.516269854857473e-05,
+      "loss": 2.2613,
+      "step": 60640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5142934170257916e-05,
+      "loss": 2.2417,
+      "step": 60648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5123176296737315e-05,
+      "loss": 2.2636,
+      "step": 60656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.510342492999452e-05,
+      "loss": 2.2659,
+      "step": 60664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.508368007201049e-05,
+      "loss": 2.2497,
+      "step": 60672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.506394172476547e-05,
+      "loss": 2.2507,
+      "step": 60680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5044209890239166e-05,
+      "loss": 2.2528,
+      "step": 60688
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.502448457041056e-05,
+      "loss": 2.2663,
+      "step": 60696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.500476576725798e-05,
+      "loss": 2.2527,
+      "step": 60704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.49850534827591e-05,
+      "loss": 2.2556,
+      "step": 60712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.496534771889097e-05,
+      "loss": 2.2288,
+      "step": 60720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.494564847762995e-05,
+      "loss": 2.2357,
+      "step": 60728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4925955760951774e-05,
+      "loss": 2.2741,
+      "step": 60736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4906269570831504e-05,
+      "loss": 2.2596,
+      "step": 60744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.488658990924356e-05,
+      "loss": 2.2427,
+      "step": 60752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4866916778161694e-05,
+      "loss": 2.2482,
+      "step": 60760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4847250179558994e-05,
+      "loss": 2.2484,
+      "step": 60768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4827590115407944e-05,
+      "loss": 2.2587,
+      "step": 60776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.480793658768031e-05,
+      "loss": 2.2521,
+      "step": 60784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.478828959834722e-05,
+      "loss": 2.2588,
+      "step": 60792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4768649149379225e-05,
+      "loss": 2.2583,
+      "step": 60800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.474901524274606e-05,
+      "loss": 2.2471,
+      "step": 60808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.472938788041694e-05,
+      "loss": 2.2439,
+      "step": 60816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4709767064360375e-05,
+      "loss": 2.2672,
+      "step": 60824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4690152796544196e-05,
+      "loss": 2.26,
+      "step": 60832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4670545078935666e-05,
+      "loss": 2.2484,
+      "step": 60840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.465094391350128e-05,
+      "loss": 2.2262,
+      "step": 60848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.463134930220688e-05,
+      "loss": 2.2514,
+      "step": 60856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.461176124701781e-05,
+      "loss": 2.2684,
+      "step": 60864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4592179749898535e-05,
+      "loss": 2.2466,
+      "step": 60872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.457260481281305e-05,
+      "loss": 2.2603,
+      "step": 60880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.455303643772452e-05,
+      "loss": 2.2565,
+      "step": 60888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4533474626595616e-05,
+      "loss": 2.2402,
+      "step": 60896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.451391938138829e-05,
+      "loss": 2.2803,
+      "step": 60904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.449437070406372e-05,
+      "loss": 2.2539,
+      "step": 60912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.447482859658263e-05,
+      "loss": 2.2601,
+      "step": 60920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4455293060904944e-05,
+      "loss": 2.2669,
+      "step": 60928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.443576409898996e-05,
+      "loss": 2.2602,
+      "step": 60936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.441624171279636e-05,
+      "loss": 2.2494,
+      "step": 60944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.439672590428205e-05,
+      "loss": 2.2534,
+      "step": 60952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.437721667540442e-05,
+      "loss": 2.2476,
+      "step": 60960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.435771402812013e-05,
+      "loss": 2.2561,
+      "step": 60968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.433821796438516e-05,
+      "loss": 2.2428,
+      "step": 60976
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.431872848615487e-05,
+      "loss": 2.2378,
+      "step": 60984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.429924559538396e-05,
+      "loss": 2.2493,
+      "step": 60992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.427976929402642e-05,
+      "loss": 2.2729,
+      "step": 61000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4260299584035625e-05,
+      "loss": 2.2669,
+      "step": 61008
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.424083646736429e-05,
+      "loss": 2.2665,
+      "step": 61016
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4221379945964436e-05,
+      "loss": 2.2701,
+      "step": 61024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.420193002178746e-05,
+      "loss": 2.2682,
+      "step": 61032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.418248669678406e-05,
+      "loss": 2.2474,
+      "step": 61040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.416304997290431e-05,
+      "loss": 2.2605,
+      "step": 61048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4143619852097595e-05,
+      "loss": 2.2618,
+      "step": 61056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.412419633631262e-05,
+      "loss": 2.2452,
+      "step": 61064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4104779427497536e-05,
+      "loss": 2.2538,
+      "step": 61072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4085369127599664e-05,
+      "loss": 2.251,
+      "step": 61080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.406596543856577e-05,
+      "loss": 2.256,
+      "step": 61088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4046568362341943e-05,
+      "loss": 2.2561,
+      "step": 61096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.402717790087357e-05,
+      "loss": 2.2566,
+      "step": 61104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.400779405610547e-05,
+      "loss": 2.2852,
+      "step": 61112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.398841682998169e-05,
+      "loss": 2.2717,
+      "step": 61120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.396904622444562e-05,
+      "loss": 2.2525,
+      "step": 61128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.3949682241440116e-05,
+      "loss": 2.2632,
+      "step": 61136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.393032488290718e-05,
+      "loss": 2.2408,
+      "step": 61144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.391097415078832e-05,
+      "loss": 2.2476,
+      "step": 61152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.3891630047024254e-05,
+      "loss": 2.2431,
+      "step": 61160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.387229257355509e-05,
+      "loss": 2.2537,
+      "step": 61168
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.385296173232032e-05,
+      "loss": 2.2551,
+      "step": 61176
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.383363752525863e-05,
+      "loss": 2.2322,
+      "step": 61184
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.3814319954308196e-05,
+      "loss": 2.259,
+      "step": 61192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3795009021406475e-05,
+      "loss": 2.2675,
+      "step": 61200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.377570472849016e-05,
+      "loss": 2.2385,
+      "step": 61208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3756407077495464e-05,
+      "loss": 2.2284,
+      "step": 61216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3737116070357717e-05,
+      "loss": 2.2472,
+      "step": 61224
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.37178317090118e-05,
+      "loss": 2.2594,
+      "step": 61232
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.369855399539179e-05,
+      "loss": 2.2747,
+      "step": 61240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.367928293143112e-05,
+      "loss": 2.2641,
+      "step": 61248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3660018519062584e-05,
+      "loss": 2.2514,
+      "step": 61256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.364076076021829e-05,
+      "loss": 2.2627,
+      "step": 61264
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.362150965682967e-05,
+      "loss": 2.2475,
+      "step": 61272
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.360226521082752e-05,
+      "loss": 2.274,
+      "step": 61280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.358302742414193e-05,
+      "loss": 2.2716,
+      "step": 61288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.356379629870236e-05,
+      "loss": 2.2662,
+      "step": 61296
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.354457183643757e-05,
+      "loss": 2.2446,
+      "step": 61304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.352535403927567e-05,
+      "loss": 2.2795,
+      "step": 61312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3506142909144086e-05,
+      "loss": 2.2645,
+      "step": 61320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.348693844796959e-05,
+      "loss": 2.2676,
+      "step": 61328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3467740657678305e-05,
+      "loss": 2.2703,
+      "step": 61336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.344854954019562e-05,
+      "loss": 2.2447,
+      "step": 61344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3429365097446315e-05,
+      "loss": 2.2623,
+      "step": 61352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3410187331354455e-05,
+      "loss": 2.2598,
+      "step": 61360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.339101624384355e-05,
+      "loss": 2.2485,
+      "step": 61368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3371851836836266e-05,
+      "loss": 2.2627,
+      "step": 61376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3352694112254694e-05,
+      "loss": 2.2402,
+      "step": 61384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3333543072020256e-05,
+      "loss": 2.2408,
+      "step": 61392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.331439871805367e-05,
+      "loss": 2.2396,
+      "step": 61400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.329526105227509e-05,
+      "loss": 2.2648,
+      "step": 61408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.32761300766038e-05,
+      "loss": 2.257,
+      "step": 61416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.325700579295862e-05,
+      "loss": 2.2448,
+      "step": 61424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.323788820325759e-05,
+      "loss": 2.2746,
+      "step": 61432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.321877730941802e-05,
+      "loss": 2.2679,
+      "step": 61440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.319967311335675e-05,
+      "loss": 2.2715,
+      "step": 61448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.318057561698969e-05,
+      "loss": 2.2654,
+      "step": 61456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.316148482223231e-05,
+      "loss": 2.2483,
+      "step": 61464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.31424007309993e-05,
+      "loss": 2.2534,
+      "step": 61472
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.31233233452046e-05,
+      "loss": 2.255,
+      "step": 61480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.310425266676166e-05,
+      "loss": 2.2609,
+      "step": 61488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.308518869758312e-05,
+      "loss": 2.2645,
+      "step": 61496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.306613143958099e-05,
+      "loss": 2.2465,
+      "step": 61504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3047080894666647e-05,
+      "loss": 2.2491,
+      "step": 61512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.302803706475065e-05,
+      "loss": 2.2558,
+      "step": 61520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.300899995174308e-05,
+      "loss": 2.2417,
+      "step": 61528
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.298996955755323e-05,
+      "loss": 2.2382,
+      "step": 61536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.297094588408973e-05,
+      "loss": 2.2476,
+      "step": 61544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.295192893326056e-05,
+      "loss": 2.2636,
+      "step": 61552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.293291870697298e-05,
+      "loss": 2.2615,
+      "step": 61560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.291391520713364e-05,
+      "loss": 2.2768,
+      "step": 61568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2894918435648477e-05,
+      "loss": 2.2409,
+      "step": 61576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.287592839442276e-05,
+      "loss": 2.2491,
+      "step": 61584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2856945085361076e-05,
+      "loss": 2.2573,
+      "step": 61592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.283796851036735e-05,
+      "loss": 2.2635,
+      "step": 61600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2818998671344815e-05,
+      "loss": 2.2631,
+      "step": 61608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.280003557019604e-05,
+      "loss": 2.2411,
+      "step": 61616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.278107920882293e-05,
+      "loss": 2.2463,
+      "step": 61624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.276212958912666e-05,
+      "loss": 2.2527,
+      "step": 61632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.274318671300787e-05,
+      "loss": 2.268,
+      "step": 61640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.272425058236631e-05,
+      "loss": 2.2371,
+      "step": 61648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.270532119910121e-05,
+      "loss": 2.2382,
+      "step": 61656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.268639856511109e-05,
+      "loss": 2.2368,
+      "step": 61664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.266748268229375e-05,
+      "loss": 2.2512,
+      "step": 61672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.264857355254641e-05,
+      "loss": 2.2751,
+      "step": 61680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2629671177765494e-05,
+      "loss": 2.2638,
+      "step": 61688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2610775559846786e-05,
+      "loss": 2.2448,
+      "step": 61696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.25918867006855e-05,
+      "loss": 2.2577,
+      "step": 61704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.257300460217596e-05,
+      "loss": 2.2378,
+      "step": 61712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.255412926621205e-05,
+      "loss": 2.2751,
+      "step": 61720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.253526069468679e-05,
+      "loss": 2.2744,
+      "step": 61728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2516398889492574e-05,
+      "loss": 2.265,
+      "step": 61736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2497543852521214e-05,
+      "loss": 2.2636,
+      "step": 61744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.247869558566366e-05,
+      "loss": 2.2524,
+      "step": 61752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.245985409081038e-05,
+      "loss": 2.2532,
+      "step": 61760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.244101936985103e-05,
+      "loss": 2.2553,
+      "step": 61768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.242219142467462e-05,
+      "loss": 2.2419,
+      "step": 61776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.240337025716952e-05,
+      "loss": 2.2575,
+      "step": 61784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.238455586922331e-05,
+      "loss": 2.2563,
+      "step": 61792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2365748262723035e-05,
+      "loss": 2.2603,
+      "step": 61800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.234694743955498e-05,
+      "loss": 2.2439,
+      "step": 61808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.232815340160475e-05,
+      "loss": 2.2588,
+      "step": 61816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2309366150757294e-05,
+      "loss": 2.2829,
+      "step": 61824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2290585688896846e-05,
+      "loss": 2.2771,
+      "step": 61832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2271812017907e-05,
+      "loss": 2.2627,
+      "step": 61840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2253045139670624e-05,
+      "loss": 2.251,
+      "step": 61848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.223428505606996e-05,
+      "loss": 2.2435,
+      "step": 61856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.221553176898652e-05,
+      "loss": 2.2613,
+      "step": 61864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.219678528030114e-05,
+      "loss": 2.2647,
+      "step": 61872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2178045591894025e-05,
+      "loss": 2.2575,
+      "step": 61880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.215931270564463e-05,
+      "loss": 2.2672,
+      "step": 61888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2140586623431756e-05,
+      "loss": 2.2533,
+      "step": 61896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.212186734713354e-05,
+      "loss": 2.2622,
+      "step": 61904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2103154878627406e-05,
+      "loss": 2.2585,
+      "step": 61912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.208444921979011e-05,
+      "loss": 2.2471,
+      "step": 61920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2065750372497724e-05,
+      "loss": 2.2636,
+      "step": 61928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.204705833862566e-05,
+      "loss": 2.2277,
+      "step": 61936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.202837312004856e-05,
+      "loss": 2.2741,
+      "step": 61944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2009694718640536e-05,
+      "loss": 2.2535,
+      "step": 61952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1991023136274856e-05,
+      "loss": 2.2434,
+      "step": 61960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.197235837482417e-05,
+      "loss": 2.2567,
+      "step": 61968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.195370043616052e-05,
+      "loss": 2.2599,
+      "step": 61976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1935049322155095e-05,
+      "loss": 2.2515,
+      "step": 61984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1916405034678604e-05,
+      "loss": 2.2617,
+      "step": 61992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1897767575600874e-05,
+      "loss": 2.2603,
+      "step": 62000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1879136946791143e-05,
+      "loss": 2.2503,
+      "step": 62008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1860513150118035e-05,
+      "loss": 2.2586,
+      "step": 62016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1841896187449305e-05,
+      "loss": 2.2535,
+      "step": 62024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.182328606065221e-05,
+      "loss": 2.2472,
+      "step": 62032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.180468277159323e-05,
+      "loss": 2.2621,
+      "step": 62040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.178608632213811e-05,
+      "loss": 2.2315,
+      "step": 62048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1767496714152064e-05,
+      "loss": 2.2545,
+      "step": 62056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1748913949499416e-05,
+      "loss": 2.2742,
+      "step": 62064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1730338030043994e-05,
+      "loss": 2.2555,
+      "step": 62072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1711768957648845e-05,
+      "loss": 2.2598,
+      "step": 62080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1693206734176296e-05,
+      "loss": 2.2478,
+      "step": 62088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.167465136148806e-05,
+      "loss": 2.2551,
+      "step": 62096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.165610284144517e-05,
+      "loss": 2.2649,
+      "step": 62104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.163756117590789e-05,
+      "loss": 2.2677,
+      "step": 62112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.161902636673586e-05,
+      "loss": 2.2593,
+      "step": 62120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1600498415788034e-05,
+      "loss": 2.2702,
+      "step": 62128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1581977324922624e-05,
+      "loss": 2.2607,
+      "step": 62136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.156346309599723e-05,
+      "loss": 2.2665,
+      "step": 62144
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1544955730868694e-05,
+      "loss": 2.2388,
+      "step": 62152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.152645523139321e-05,
+      "loss": 2.2649,
+      "step": 62160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.150796159942627e-05,
+      "loss": 2.263,
+      "step": 62168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.148947483682268e-05,
+      "loss": 2.2682,
+      "step": 62176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.147099494543658e-05,
+      "loss": 2.252,
+      "step": 62184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.145252192712138e-05,
+      "loss": 2.2785,
+      "step": 62192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.14340557837298e-05,
+      "loss": 2.2498,
+      "step": 62200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1415596517113925e-05,
+      "loss": 2.2553,
+      "step": 62208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.139714412912509e-05,
+      "loss": 2.2561,
+      "step": 62216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.137869862161398e-05,
+      "loss": 2.2609,
+      "step": 62224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1360259996430564e-05,
+      "loss": 2.245,
+      "step": 62232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.134182825542411e-05,
+      "loss": 2.2614,
+      "step": 62240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1323403400443296e-05,
+      "loss": 2.2846,
+      "step": 62248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.130498543333593e-05,
+      "loss": 2.2513,
+      "step": 62256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1286574355949314e-05,
+      "loss": 2.2461,
+      "step": 62264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.126817017012993e-05,
+      "loss": 2.2721,
+      "step": 62272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.124977287772359e-05,
+      "loss": 2.2653,
+      "step": 62280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1231382480575524e-05,
+      "loss": 2.2457,
+      "step": 62288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.121299898053008e-05,
+      "loss": 2.2585,
+      "step": 62296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.119462237943108e-05,
+      "loss": 2.2635,
+      "step": 62304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.117625267912163e-05,
+      "loss": 2.2699,
+      "step": 62312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1157889881444004e-05,
+      "loss": 2.2559,
+      "step": 62320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1139533988239985e-05,
+      "loss": 2.2532,
+      "step": 62328
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.112118500135047e-05,
+      "loss": 2.2463,
+      "step": 62336
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.110284292261585e-05,
+      "loss": 2.2604,
+      "step": 62344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1084507753875696e-05,
+      "loss": 2.2707,
+      "step": 62352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.106617949696888e-05,
+      "loss": 2.2383,
+      "step": 62360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1047858153733684e-05,
+      "loss": 2.2677,
+      "step": 62368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.102954372600761e-05,
+      "loss": 2.2536,
+      "step": 62376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.10112362156275e-05,
+      "loss": 2.2531,
+      "step": 62384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.09929356244295e-05,
+      "loss": 2.2628,
+      "step": 62392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.097464195424899e-05,
+      "loss": 2.2542,
+      "step": 62400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.09563552069208e-05,
+      "loss": 2.2359,
+      "step": 62408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.093807538427897e-05,
+      "loss": 2.2563,
+      "step": 62416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0919802488156844e-05,
+      "loss": 2.2642,
+      "step": 62424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.09015365203871e-05,
+      "loss": 2.2693,
+      "step": 62432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0883277482801705e-05,
+      "loss": 2.2307,
+      "step": 62440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.086502537723195e-05,
+      "loss": 2.2605,
+      "step": 62448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0846780205508396e-05,
+      "loss": 2.2482,
+      "step": 62456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0828541969460964e-05,
+      "loss": 2.2555,
+      "step": 62464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0810310670918784e-05,
+      "loss": 2.2676,
+      "step": 62472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.079208631171044e-05,
+      "loss": 2.2492,
+      "step": 62480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.077386889366367e-05,
+      "loss": 2.2502,
+      "step": 62488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0755658418605606e-05,
+      "loss": 2.2556,
+      "step": 62496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.073745488836263e-05,
+      "loss": 2.2733,
+      "step": 62504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.071925830476043e-05,
+      "loss": 2.2669,
+      "step": 62512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.070106866962411e-05,
+      "loss": 2.2576,
+      "step": 62520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.068288598477792e-05,
+      "loss": 2.2472,
+      "step": 62528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.066471025204546e-05,
+      "loss": 2.2684,
+      "step": 62536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0646541473249725e-05,
+      "loss": 2.2563,
+      "step": 62544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.062837965021287e-05,
+      "loss": 2.2587,
+      "step": 62552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.06102247847565e-05,
+      "loss": 2.2531,
+      "step": 62560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.059207687870137e-05,
+      "loss": 2.2544,
+      "step": 62568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.057393593386761e-05,
+      "loss": 2.2436,
+      "step": 62576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0555801952074755e-05,
+      "loss": 2.2714,
+      "step": 62584
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.053767493514142e-05,
+      "loss": 2.2368,
+      "step": 62592
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.05195548848857e-05,
+      "loss": 2.2374,
+      "step": 62600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.050144180312494e-05,
+      "loss": 2.2551,
+      "step": 62608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.048333569167577e-05,
+      "loss": 2.2436,
+      "step": 62616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.046523655235415e-05,
+      "loss": 2.2603,
+      "step": 62624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.044714438697525e-05,
+      "loss": 2.251,
+      "step": 62632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.042905919735367e-05,
+      "loss": 2.247,
+      "step": 62640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.041098098530326e-05,
+      "loss": 2.2539,
+      "step": 62648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.039290975263714e-05,
+      "loss": 2.2494,
+      "step": 62656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0374845501167746e-05,
+      "loss": 2.2513,
+      "step": 62664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0356788232706846e-05,
+      "loss": 2.249,
+      "step": 62672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0338737949065454e-05,
+      "loss": 2.246,
+      "step": 62680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.032069465205392e-05,
+      "loss": 2.2685,
+      "step": 62688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0302658343481904e-05,
+      "loss": 2.2632,
+      "step": 62696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.028462902515832e-05,
+      "loss": 2.2604,
+      "step": 62704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.026660669889141e-05,
+      "loss": 2.2596,
+      "step": 62712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.024859136648872e-05,
+      "loss": 2.243,
+      "step": 62720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.023058302975709e-05,
+      "loss": 2.2712,
+      "step": 62728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.021258169050264e-05,
+      "loss": 2.2811,
+      "step": 62736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.019458735053081e-05,
+      "loss": 2.2775,
+      "step": 62744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.017660001164633e-05,
+      "loss": 2.2655,
+      "step": 62752
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.015861967565323e-05,
+      "loss": 2.2531,
+      "step": 62760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0140646344354827e-05,
+      "loss": 2.2252,
+      "step": 62768
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.012268001955375e-05,
+      "loss": 2.2527,
+      "step": 62776
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.01047207030519e-05,
+      "loss": 2.2531,
+      "step": 62784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.008676839665058e-05,
+      "loss": 2.2658,
+      "step": 62792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.006882310215019e-05,
+      "loss": 2.2371,
+      "step": 62800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.005088482135058e-05,
+      "loss": 2.2492,
+      "step": 62808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.003295355605091e-05,
+      "loss": 2.2739,
+      "step": 62816
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.00150293080495e-05,
+      "loss": 2.2658,
+      "step": 62824
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.999711207914414e-05,
+      "loss": 2.2545,
+      "step": 62832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9979201871131754e-05,
+      "loss": 2.2576,
+      "step": 62840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.996129868580864e-05,
+      "loss": 2.2336,
+      "step": 62848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9943402524970455e-05,
+      "loss": 2.2617,
+      "step": 62856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9925513390411985e-05,
+      "loss": 2.2664,
+      "step": 62864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9907631283927486e-05,
+      "loss": 2.2651,
+      "step": 62872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9889756207310425e-05,
+      "loss": 2.2607,
+      "step": 62880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9871888162353496e-05,
+      "loss": 2.2457,
+      "step": 62888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.985402715084887e-05,
+      "loss": 2.2694,
+      "step": 62896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.983617317458781e-05,
+      "loss": 2.2461,
+      "step": 62904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.981832623536102e-05,
+      "loss": 2.2657,
+      "step": 62912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9800486334958475e-05,
+      "loss": 2.2515,
+      "step": 62920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.978265347516932e-05,
+      "loss": 2.2507,
+      "step": 62928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9764827657782175e-05,
+      "loss": 2.2752,
+      "step": 62936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.974700888458484e-05,
+      "loss": 2.2514,
+      "step": 62944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.972919715736445e-05,
+      "loss": 2.2614,
+      "step": 62952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9711392477907406e-05,
+      "loss": 2.2705,
+      "step": 62960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9693594847999404e-05,
+      "loss": 2.2544,
+      "step": 62968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.967580426942547e-05,
+      "loss": 2.2502,
+      "step": 62976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9658020743969895e-05,
+      "loss": 2.2852,
+      "step": 62984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.964024427341625e-05,
+      "loss": 2.2543,
+      "step": 62992
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.962247485954743e-05,
+      "loss": 2.2518,
+      "step": 63000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9604712504145604e-05,
+      "loss": 2.2444,
+      "step": 63008
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9586957208992215e-05,
+      "loss": 2.2482,
+      "step": 63016
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.956920897586804e-05,
+      "loss": 2.261,
+      "step": 63024
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9551467806553147e-05,
+      "loss": 2.2724,
+      "step": 63032
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.953373370282682e-05,
+      "loss": 2.261,
+      "step": 63040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.951600666646774e-05,
+      "loss": 2.2604,
+      "step": 63048
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.949828669925379e-05,
+      "loss": 2.2552,
+      "step": 63056
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9480573802962205e-05,
+      "loss": 2.257,
+      "step": 63064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9462867979369486e-05,
+      "loss": 2.2685,
+      "step": 63072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.94451692302514e-05,
+      "loss": 2.2647,
+      "step": 63080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.94274775573831e-05,
+      "loss": 2.248,
+      "step": 63088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.940979296253889e-05,
+      "loss": 2.2478,
+      "step": 63096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.939211544749246e-05,
+      "loss": 2.2597,
+      "step": 63104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9374445014016766e-05,
+      "loss": 2.2578,
+      "step": 63112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9356781663884024e-05,
+      "loss": 2.2628,
+      "step": 63120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.933912539886585e-05,
+      "loss": 2.2383,
+      "step": 63128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9321476220732944e-05,
+      "loss": 2.2429,
+      "step": 63136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.930383413125553e-05,
+      "loss": 2.2537,
+      "step": 63144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9286199132202985e-05,
+      "loss": 2.2699,
+      "step": 63152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.926857122534394e-05,
+      "loss": 2.2411,
+      "step": 63160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9250950412446455e-05,
+      "loss": 2.2642,
+      "step": 63168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.92333366952777e-05,
+      "loss": 2.2699,
+      "step": 63176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.921573007560433e-05,
+      "loss": 2.2478,
+      "step": 63184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.919813055519216e-05,
+      "loss": 2.2579,
+      "step": 63192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.918053813580627e-05,
+      "loss": 2.2592,
+      "step": 63200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.916295281921114e-05,
+      "loss": 2.232,
+      "step": 63208
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.914537460717045e-05,
+      "loss": 2.2552,
+      "step": 63216
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.91278035014472e-05,
+      "loss": 2.2382,
+      "step": 63224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9110239503803715e-05,
+      "loss": 2.2551,
+      "step": 63232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9092682616001475e-05,
+      "loss": 2.2595,
+      "step": 63240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9075132839801396e-05,
+      "loss": 2.2453,
+      "step": 63248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9057590176963624e-05,
+      "loss": 2.2366,
+      "step": 63256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.904005462924756e-05,
+      "loss": 2.2633,
+      "step": 63264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.902252619841196e-05,
+      "loss": 2.2609,
+      "step": 63272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.900500488621478e-05,
+      "loss": 2.2544,
+      "step": 63280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.898749069441333e-05,
+      "loss": 2.2483,
+      "step": 63288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8969983624764205e-05,
+      "loss": 2.2483,
+      "step": 63296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.895248367902322e-05,
+      "loss": 2.2599,
+      "step": 63304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8934990858945536e-05,
+      "loss": 2.236,
+      "step": 63312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8917505166285634e-05,
+      "loss": 2.2378,
+      "step": 63320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.890002660279717e-05,
+      "loss": 2.2526,
+      "step": 63328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.888255517023316e-05,
+      "loss": 2.241,
+      "step": 63336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.886509087034589e-05,
+      "loss": 2.2386,
+      "step": 63344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.884763370488691e-05,
+      "loss": 2.2412,
+      "step": 63352
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8830183675607146e-05,
+      "loss": 2.249,
+      "step": 63360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8812740784256664e-05,
+      "loss": 2.2372,
+      "step": 63368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8795305032584895e-05,
+      "loss": 2.2585,
+      "step": 63376
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8777876422340597e-05,
+      "loss": 2.2643,
+      "step": 63384
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8760454955271686e-05,
+      "loss": 2.2563,
+      "step": 63392
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.874304063312554e-05,
+      "loss": 2.2698,
+      "step": 63400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8725633457648606e-05,
+      "loss": 2.2651,
+      "step": 63408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.870823343058676e-05,
+      "loss": 2.2585,
+      "step": 63416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.869084055368519e-05,
+      "loss": 2.2734,
+      "step": 63424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.86734548286882e-05,
+      "loss": 2.2171,
+      "step": 63432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8656076257339565e-05,
+      "loss": 2.2314,
+      "step": 63440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8638704841382246e-05,
+      "loss": 2.2491,
+      "step": 63448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.862134058255842e-05,
+      "loss": 2.2397,
+      "step": 63456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.860398348260973e-05,
+      "loss": 2.2586,
+      "step": 63464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.85866335432769e-05,
+      "loss": 2.2448,
+      "step": 63472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.85692907663001e-05,
+      "loss": 2.2707,
+      "step": 63480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.855195515341868e-05,
+      "loss": 2.262,
+      "step": 63488
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8534626706371315e-05,
+      "loss": 2.2364,
+      "step": 63496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.851730542689596e-05,
+      "loss": 2.2687,
+      "step": 63504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8499991316729806e-05,
+      "loss": 2.255,
+      "step": 63512
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.848268437760939e-05,
+      "loss": 2.2539,
+      "step": 63520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.84653846112705e-05,
+      "loss": 2.2664,
+      "step": 63528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.844809201944818e-05,
+      "loss": 2.2718,
+      "step": 63536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.843080660387681e-05,
+      "loss": 2.2543,
+      "step": 63544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.841352836629e-05,
+      "loss": 2.2224,
+      "step": 63552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.839625730842066e-05,
+      "loss": 2.2453,
+      "step": 63560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.837899343200097e-05,
+      "loss": 2.2485,
+      "step": 63568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.836173673876243e-05,
+      "loss": 2.2686,
+      "step": 63576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.834448723043577e-05,
+      "loss": 2.2521,
+      "step": 63584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8327244908751015e-05,
+      "loss": 2.2661,
+      "step": 63592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.831000977543747e-05,
+      "loss": 2.2494,
+      "step": 63600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.829278183222374e-05,
+      "loss": 2.2559,
+      "step": 63608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.827556108083766e-05,
+      "loss": 2.2786,
+      "step": 63616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8258347523006397e-05,
+      "loss": 2.2652,
+      "step": 63624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.824114116045637e-05,
+      "loss": 2.2656,
+      "step": 63632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.822394199491327e-05,
+      "loss": 2.2472,
+      "step": 63640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8206750028102064e-05,
+      "loss": 2.2537,
+      "step": 63648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.818956526174708e-05,
+      "loss": 2.2385,
+      "step": 63656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8172387697571746e-05,
+      "loss": 2.2725,
+      "step": 63664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.815521733729898e-05,
+      "loss": 2.2616,
+      "step": 63672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.813805418265079e-05,
+      "loss": 2.2306,
+      "step": 63680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.812089823534854e-05,
+      "loss": 2.2621,
+      "step": 63688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.810374949711298e-05,
+      "loss": 2.2492,
+      "step": 63696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.80866079696639e-05,
+      "loss": 2.2443,
+      "step": 63704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.806947365472058e-05,
+      "loss": 2.2489,
+      "step": 63712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.805234655400149e-05,
+      "loss": 2.2356,
+      "step": 63720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.803522666922432e-05,
+      "loss": 2.2473,
+      "step": 63728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.801811400210619e-05,
+      "loss": 2.264,
+      "step": 63736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8001008554363304e-05,
+      "loss": 2.2453,
+      "step": 63744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7983910327711316e-05,
+      "loss": 2.2666,
+      "step": 63752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7966819323865076e-05,
+      "loss": 2.2603,
+      "step": 63760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.794973554453865e-05,
+      "loss": 2.2462,
+      "step": 63768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7932658991445514e-05,
+      "loss": 2.2478,
+      "step": 63776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7915589666298325e-05,
+      "loss": 2.276,
+      "step": 63784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.789852757080904e-05,
+      "loss": 2.2476,
+      "step": 63792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7881472706688915e-05,
+      "loss": 2.2345,
+      "step": 63800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.786442507564838e-05,
+      "loss": 2.2661,
+      "step": 63808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.784738467939729e-05,
+      "loss": 2.257,
+      "step": 63816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.783035151964469e-05,
+      "loss": 2.2543,
+      "step": 63824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.781332559809889e-05,
+      "loss": 2.2584,
+      "step": 63832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7796306916467496e-05,
+      "loss": 2.2477,
+      "step": 63840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.777929547645739e-05,
+      "loss": 2.2567,
+      "step": 63848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.776229127977472e-05,
+      "loss": 2.2557,
+      "step": 63856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.774529432812492e-05,
+      "loss": 2.2505,
+      "step": 63864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7728304623212684e-05,
+      "loss": 2.2724,
+      "step": 63872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.771132216674197e-05,
+      "loss": 2.2566,
+      "step": 63880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.769434696041604e-05,
+      "loss": 2.2562,
+      "step": 63888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.76773790059374e-05,
+      "loss": 2.2545,
+      "step": 63896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7660418305007835e-05,
+      "loss": 2.2454,
+      "step": 63904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7643464859328425e-05,
+      "loss": 2.2442,
+      "step": 63912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.762651867059947e-05,
+      "loss": 2.2221,
+      "step": 63920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.760957974052066e-05,
+      "loss": 2.2512,
+      "step": 63928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.759264807079077e-05,
+      "loss": 2.253,
+      "step": 63936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7575723663108006e-05,
+      "loss": 2.2698,
+      "step": 63944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.755880651916979e-05,
+      "loss": 2.2576,
+      "step": 63952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7541896640672766e-05,
+      "loss": 2.2655,
+      "step": 63960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7524994029313e-05,
+      "loss": 2.2589,
+      "step": 63968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.750809868678564e-05,
+      "loss": 2.2304,
+      "step": 63976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7491210614785185e-05,
+      "loss": 2.2762,
+      "step": 63984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.747432981500551e-05,
+      "loss": 2.2409,
+      "step": 63992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.745745628913954e-05,
+      "loss": 2.2408,
+      "step": 64000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7440590038879715e-05,
+      "loss": 2.2466,
+      "step": 64008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.742373106591751e-05,
+      "loss": 2.247,
+      "step": 64016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.740687937194387e-05,
+      "loss": 2.2467,
+      "step": 64024
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7390034958648904e-05,
+      "loss": 2.2718,
+      "step": 64032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.737319782772196e-05,
+      "loss": 2.2464,
+      "step": 64040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.735636798085178e-05,
+      "loss": 2.2592,
+      "step": 64048
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7339545419726266e-05,
+      "loss": 2.2535,
+      "step": 64056
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.732273014603262e-05,
+      "loss": 2.2612,
+      "step": 64064
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.730592216145736e-05,
+      "loss": 2.2463,
+      "step": 64072
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.728912146768616e-05,
+      "loss": 2.2653,
+      "step": 64080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7272328066404085e-05,
+      "loss": 2.2714,
+      "step": 64088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.72555419592954e-05,
+      "loss": 2.2461,
+      "step": 64096
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.723876314804368e-05,
+      "loss": 2.2623,
+      "step": 64104
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.722199163433173e-05,
+      "loss": 2.26,
+      "step": 64112
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7205227419841626e-05,
+      "loss": 2.2644,
+      "step": 64120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.718847050625475e-05,
+      "loss": 2.2504,
+      "step": 64128
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.717172089525169e-05,
+      "loss": 2.24,
+      "step": 64136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.715497858851236e-05,
+      "loss": 2.2622,
+      "step": 64144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.713824358771592e-05,
+      "loss": 2.2697,
+      "step": 64152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.712151589454079e-05,
+      "loss": 2.2482,
+      "step": 64160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7104795510664655e-05,
+      "loss": 2.247,
+      "step": 64168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7088082437764495e-05,
+      "loss": 2.2649,
+      "step": 64176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.707137667751652e-05,
+      "loss": 2.2513,
+      "step": 64184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.705467823159621e-05,
+      "loss": 2.2519,
+      "step": 64192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.703798710167838e-05,
+      "loss": 2.2748,
+      "step": 64200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7021303289437005e-05,
+      "loss": 2.2475,
+      "step": 64208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.700462679654536e-05,
+      "loss": 2.2333,
+      "step": 64216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.698795762467607e-05,
+      "loss": 2.2512,
+      "step": 64224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.697129577550089e-05,
+      "loss": 2.2531,
+      "step": 64232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.695464125069098e-05,
+      "loss": 2.2628,
+      "step": 64240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.693799405191664e-05,
+      "loss": 2.2456,
+      "step": 64248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6921354180847466e-05,
+      "loss": 2.2515,
+      "step": 64256
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.690472163915244e-05,
+      "loss": 2.2638,
+      "step": 64264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.688809642849961e-05,
+      "loss": 2.2597,
+      "step": 64272
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6871478550556434e-05,
+      "loss": 2.2344,
+      "step": 64280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.685486800698963e-05,
+      "loss": 2.2565,
+      "step": 64288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.683826479946506e-05,
+      "loss": 2.2629,
+      "step": 64296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6821668929648014e-05,
+      "loss": 2.2205,
+      "step": 64304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.680508039920288e-05,
+      "loss": 2.257,
+      "step": 64312
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6788499209793445e-05,
+      "loss": 2.2642,
+      "step": 64320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.677192536308274e-05,
+      "loss": 2.2347,
+      "step": 64328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6755358860732935e-05,
+      "loss": 2.2588,
+      "step": 64336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6738799704405635e-05,
+      "loss": 2.2537,
+      "step": 64344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6722247895761595e-05,
+      "loss": 2.2535,
+      "step": 64352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670570343646089e-05,
+      "loss": 2.2876,
+      "step": 64360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668916632816282e-05,
+      "loss": 2.2425,
+      "step": 64368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667263657252595e-05,
+      "loss": 2.2708,
+      "step": 64376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6656114171208155e-05,
+      "loss": 2.2679,
+      "step": 64384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6639599125866505e-05,
+      "loss": 2.2559,
+      "step": 64392
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6623091438157384e-05,
+      "loss": 2.2657,
+      "step": 64400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660659110973642e-05,
+      "loss": 2.2385,
+      "step": 64408
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6590098142258494e-05,
+      "loss": 2.2497,
+      "step": 64416
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6573612537377764e-05,
+      "loss": 2.2447,
+      "step": 64424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655713429674764e-05,
+      "loss": 2.2683,
+      "step": 64432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6540663422020796e-05,
+      "loss": 2.2629,
+      "step": 64440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652419991484916e-05,
+      "loss": 2.2558,
+      "step": 64448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650774377688394e-05,
+      "loss": 2.2449,
+      "step": 64456
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.649129500977558e-05,
+      "loss": 2.2561,
+      "step": 64464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647485361517383e-05,
+      "loss": 2.2625,
+      "step": 64472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.645841959472763e-05,
+      "loss": 2.2563,
+      "step": 64480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644199295008521e-05,
+      "loss": 2.2601,
+      "step": 64488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.642557368289414e-05,
+      "loss": 2.2591,
+      "step": 64496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6409161794801125e-05,
+      "loss": 2.2672,
+      "step": 64504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.639275728745218e-05,
+      "loss": 2.2461,
+      "step": 64512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.63763601624926e-05,
+      "loss": 2.2313,
+      "step": 64520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6359970421566904e-05,
+      "loss": 2.259,
+      "step": 64528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.634358806631895e-05,
+      "loss": 2.2748,
+      "step": 64536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.63272130983917e-05,
+      "loss": 2.2577,
+      "step": 64544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.631084551942755e-05,
+      "loss": 2.2318,
+      "step": 64552
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.629448533106808e-05,
+      "loss": 2.2644,
+      "step": 64560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.627813253495404e-05,
+      "loss": 2.2419,
+      "step": 64568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6261787132725614e-05,
+      "loss": 2.2657,
+      "step": 64576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.624544912602209e-05,
+      "loss": 2.2399,
+      "step": 64584
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6229118516482105e-05,
+      "loss": 2.2702,
+      "step": 64592
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6212795305743555e-05,
+      "loss": 2.2537,
+      "step": 64600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.619647949544349e-05,
+      "loss": 2.2365,
+      "step": 64608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.618017108721836e-05,
+      "loss": 2.2607,
+      "step": 64616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6163870082703774e-05,
+      "loss": 2.2725,
+      "step": 64624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6147576483534654e-05,
+      "loss": 2.2457,
+      "step": 64632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.613129029134516e-05,
+      "loss": 2.2436,
+      "step": 64640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.611501150776864e-05,
+      "loss": 2.2593,
+      "step": 64648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.609874013443783e-05,
+      "loss": 2.2728,
+      "step": 64656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.608247617298464e-05,
+      "loss": 2.2583,
+      "step": 64664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.606621962504025e-05,
+      "loss": 2.2427,
+      "step": 64672
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6049970492235105e-05,
+      "loss": 2.2705,
+      "step": 64680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6033728776198885e-05,
+      "loss": 2.2494,
+      "step": 64688
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.601749447856055e-05,
+      "loss": 2.251,
+      "step": 64696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.600126760094832e-05,
+      "loss": 2.2487,
+      "step": 64704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.598504814498965e-05,
+      "loss": 2.2637,
+      "step": 64712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.596883611231127e-05,
+      "loss": 2.2462,
+      "step": 64720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.595263150453912e-05,
+      "loss": 2.2461,
+      "step": 64728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.593643432329847e-05,
+      "loss": 2.2423,
+      "step": 64736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.592024457021379e-05,
+      "loss": 2.2682,
+      "step": 64744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.590406224690881e-05,
+      "loss": 2.2445,
+      "step": 64752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5887887355006535e-05,
+      "loss": 2.2356,
+      "step": 64760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.587171989612924e-05,
+      "loss": 2.2436,
+      "step": 64768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.585555987189839e-05,
+      "loss": 2.2796,
+      "step": 64776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5839407283934755e-05,
+      "loss": 2.2481,
+      "step": 64784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.582326213385836e-05,
+      "loss": 2.2455,
+      "step": 64792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.580712442328843e-05,
+      "loss": 2.2446,
+      "step": 64800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.579099415384358e-05,
+      "loss": 2.2534,
+      "step": 64808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.577487132714149e-05,
+      "loss": 2.2649,
+      "step": 64816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.57587559447992e-05,
+      "loss": 2.2531,
+      "step": 64824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.574264800843308e-05,
+      "loss": 2.2536,
+      "step": 64832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5726547519658534e-05,
+      "loss": 2.2644,
+      "step": 64840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5710454480090465e-05,
+      "loss": 2.2795,
+      "step": 64848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5694368891342854e-05,
+      "loss": 2.234,
+      "step": 64856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.567829075502897e-05,
+      "loss": 2.253,
+      "step": 64864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.566222007276143e-05,
+      "loss": 2.2643,
+      "step": 64872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.564615684615196e-05,
+      "loss": 2.226,
+      "step": 64880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.563010107681167e-05,
+      "loss": 2.2592,
+      "step": 64888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.561405276635085e-05,
+      "loss": 2.2527,
+      "step": 64896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5598011916379023e-05,
+      "loss": 2.2555,
+      "step": 64904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.558197852850505e-05,
+      "loss": 2.2412,
+      "step": 64912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5565952604336907e-05,
+      "loss": 2.2661,
+      "step": 64920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5549934145481964e-05,
+      "loss": 2.2588,
+      "step": 64928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.553392315354678e-05,
+      "loss": 2.2455,
+      "step": 64936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.551791963013715e-05,
+      "loss": 2.2435,
+      "step": 64944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5501923576858144e-05,
+      "loss": 2.27,
+      "step": 64952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5485934995314086e-05,
+      "loss": 2.2567,
+      "step": 64960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.546995388710852e-05,
+      "loss": 2.2376,
+      "step": 64968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5453980253844256e-05,
+      "loss": 2.2438,
+      "step": 64976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.543801409712339e-05,
+      "loss": 2.2538,
+      "step": 64984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.542205541854721e-05,
+      "loss": 2.2501,
+      "step": 64992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5406104219716285e-05,
+      "loss": 2.2693,
+      "step": 65000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.539016050223044e-05,
+      "loss": 2.2487,
+      "step": 65008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5374224267688734e-05,
+      "loss": 2.2502,
+      "step": 65016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.535829551768948e-05,
+      "loss": 2.2756,
+      "step": 65024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.534237425383023e-05,
+      "loss": 2.2657,
+      "step": 65032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.532646047770784e-05,
+      "loss": 2.2596,
+      "step": 65040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.531055419091832e-05,
+      "loss": 2.246,
+      "step": 65048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.529465539505698e-05,
+      "loss": 2.2376,
+      "step": 65056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.527876409171844e-05,
+      "loss": 2.2415,
+      "step": 65064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5262880282496434e-05,
+      "loss": 2.2532,
+      "step": 65072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5247003968984094e-05,
+      "loss": 2.254,
+      "step": 65080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5231135152773645e-05,
+      "loss": 2.2552,
+      "step": 65088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.521527383545667e-05,
+      "loss": 2.2534,
+      "step": 65096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.519942001862402e-05,
+      "loss": 2.2501,
+      "step": 65104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5183573703865656e-05,
+      "loss": 2.2675,
+      "step": 65112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.516773489277092e-05,
+      "loss": 2.2428,
+      "step": 65120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5151903586928394e-05,
+      "loss": 2.2215,
+      "step": 65128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.513607978792577e-05,
+      "loss": 2.2473,
+      "step": 65136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.512026349735019e-05,
+      "loss": 2.2516,
+      "step": 65144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.510445471678785e-05,
+      "loss": 2.25,
+      "step": 65152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.508865344782433e-05,
+      "loss": 2.2458,
+      "step": 65160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5072859692044426e-05,
+      "loss": 2.2739,
+      "step": 65168
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.505707345103209e-05,
+      "loss": 2.2773,
+      "step": 65176
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5041294726370654e-05,
+      "loss": 2.2545,
+      "step": 65184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.502552351964263e-05,
+      "loss": 2.2218,
+      "step": 65192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5009759832429766e-05,
+      "loss": 2.2608,
+      "step": 65200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.499400366631308e-05,
+      "loss": 2.2645,
+      "step": 65208
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4978255022872783e-05,
+      "loss": 2.2668,
+      "step": 65216
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.496251390368843e-05,
+      "loss": 2.2554,
+      "step": 65224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.494678031033876e-05,
+      "loss": 2.2669,
+      "step": 65232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.493105424440174e-05,
+      "loss": 2.2689,
+      "step": 65240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.491533570745462e-05,
+      "loss": 2.263,
+      "step": 65248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.489962470107387e-05,
+      "loss": 2.2335,
+      "step": 65256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.488392122683522e-05,
+      "loss": 2.2587,
+      "step": 65264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.486822528631364e-05,
+      "loss": 2.26,
+      "step": 65272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.485253688108335e-05,
+      "loss": 2.2398,
+      "step": 65280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.483685601271779e-05,
+      "loss": 2.2488,
+      "step": 65288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.482118268278968e-05,
+      "loss": 2.2417,
+      "step": 65296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.480551689287097e-05,
+      "loss": 2.2521,
+      "step": 65304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4789858644532823e-05,
+      "loss": 2.2512,
+      "step": 65312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4774207939345704e-05,
+      "loss": 2.2698,
+      "step": 65320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4758564778879244e-05,
+      "loss": 2.2655,
+      "step": 65328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.474292916470245e-05,
+      "loss": 2.2489,
+      "step": 65336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.47273010983834e-05,
+      "loss": 2.264,
+      "step": 65344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.471168058148953e-05,
+      "loss": 2.2698,
+      "step": 65352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4696067615587494e-05,
+      "loss": 2.2392,
+      "step": 65360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.468046220224316e-05,
+      "loss": 2.2554,
+      "step": 65368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.466486434302172e-05,
+      "loss": 2.2679,
+      "step": 65376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.46492740394875e-05,
+      "loss": 2.2286,
+      "step": 65384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.463369129320411e-05,
+      "loss": 2.2442,
+      "step": 65392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.461811610573447e-05,
+      "loss": 2.2533,
+      "step": 65400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4602548478640616e-05,
+      "loss": 2.2518,
+      "step": 65408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4586988413483965e-05,
+      "loss": 2.2594,
+      "step": 65416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.457143591182501e-05,
+      "loss": 2.2496,
+      "step": 65424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.455589097522366e-05,
+      "loss": 2.2656,
+      "step": 65432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.454035360523897e-05,
+      "loss": 2.2726,
+      "step": 65440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4524823803429204e-05,
+      "loss": 2.2488,
+      "step": 65448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.450930157135196e-05,
+      "loss": 2.246,
+      "step": 65456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4493786910564017e-05,
+      "loss": 2.2247,
+      "step": 65464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4478279822621404e-05,
+      "loss": 2.2657,
+      "step": 65472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.446278030907941e-05,
+      "loss": 2.26,
+      "step": 65480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.44472883714925e-05,
+      "loss": 2.2608,
+      "step": 65488
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4431804011414476e-05,
+      "loss": 2.2454,
+      "step": 65496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.441632723039831e-05,
+      "loss": 2.2658,
+      "step": 65504
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4400858029996265e-05,
+      "loss": 2.2642,
+      "step": 65512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.438539641175977e-05,
+      "loss": 2.2388,
+      "step": 65520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4369942377239566e-05,
+      "loss": 2.2422,
+      "step": 65528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4354495927985605e-05,
+      "loss": 2.2397,
+      "step": 65536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.433905706554706e-05,
+      "loss": 2.2595,
+      "step": 65544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.432362579147237e-05,
+      "loss": 2.2565,
+      "step": 65552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.430820210730922e-05,
+      "loss": 2.2411,
+      "step": 65560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.429278601460451e-05,
+      "loss": 2.2461,
+      "step": 65568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4277377514904365e-05,
+      "loss": 2.2609,
+      "step": 65576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.426197660975419e-05,
+      "loss": 2.2484,
+      "step": 65584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.424658330069861e-05,
+      "loss": 2.2505,
+      "step": 65592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.423119758928146e-05,
+      "loss": 2.2635,
+      "step": 65600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.421581947704592e-05,
+      "loss": 2.2438,
+      "step": 65608
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.420044896553423e-05,
+      "loss": 2.2455,
+      "step": 65616
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4185086056288e-05,
+      "loss": 2.2584,
+      "step": 65624
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.416973075084806e-05,
+      "loss": 2.2632,
+      "step": 65632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.415438305075442e-05,
+      "loss": 2.2379,
+      "step": 65640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.413904295754644e-05,
+      "loss": 2.2747,
+      "step": 65648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4123710472762584e-05,
+      "loss": 2.2407,
+      "step": 65656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4108385597940594e-05,
+      "loss": 2.2542,
+      "step": 65664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.409306833461756e-05,
+      "loss": 2.2605,
+      "step": 65672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4077758684329606e-05,
+      "loss": 2.2492,
+      "step": 65680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.406245664861231e-05,
+      "loss": 2.2658,
+      "step": 65688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4047162229000295e-05,
+      "loss": 2.2581,
+      "step": 65696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.403187542702753e-05,
+      "loss": 2.2244,
+      "step": 65704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.401659624422722e-05,
+      "loss": 2.2619,
+      "step": 65712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.400132468213173e-05,
+      "loss": 2.2446,
+      "step": 65720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.398606074227276e-05,
+      "loss": 2.2523,
+      "step": 65728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.397080442618121e-05,
+      "loss": 2.2301,
+      "step": 65736
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.395555573538712e-05,
+      "loss": 2.2557,
+      "step": 65744
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.394031467141995e-05,
+      "loss": 2.254,
+      "step": 65752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.392508123580819e-05,
+      "loss": 2.2449,
+      "step": 65760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.390985543007974e-05,
+      "loss": 2.2228,
+      "step": 65768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.389463725576164e-05,
+      "loss": 2.2688,
+      "step": 65776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.387942671438018e-05,
+      "loss": 2.2444,
+      "step": 65784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3864223807460906e-05,
+      "loss": 2.2576,
+      "step": 65792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.384902853652857e-05,
+      "loss": 2.2607,
+      "step": 65800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.383384090310718e-05,
+      "loss": 2.2629,
+      "step": 65808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.381866090871997e-05,
+      "loss": 2.238,
+      "step": 65816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3803488554889396e-05,
+      "loss": 2.2541,
+      "step": 65824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.378832384313717e-05,
+      "loss": 2.255,
+      "step": 65832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.377316677498423e-05,
+      "loss": 2.2358,
+      "step": 65840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.375801735195072e-05,
+      "loss": 2.2507,
+      "step": 65848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.374287557555608e-05,
+      "loss": 2.2538,
+      "step": 65856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3727741447318904e-05,
+      "loss": 2.2484,
+      "step": 65864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3712614968757096e-05,
+      "loss": 2.2637,
+      "step": 65872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.369749614138773e-05,
+      "loss": 2.2474,
+      "step": 65880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.368238496672714e-05,
+      "loss": 2.2615,
+      "step": 65888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.366728144629088e-05,
+      "loss": 2.2559,
+      "step": 65896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3652185581593816e-05,
+      "loss": 2.2581,
+      "step": 65904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3637097374149905e-05,
+      "loss": 2.2425,
+      "step": 65912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.362201682547243e-05,
+      "loss": 2.275,
+      "step": 65920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.360694393707389e-05,
+      "loss": 2.2554,
+      "step": 65928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3591878710465973e-05,
+      "loss": 2.2453,
+      "step": 65936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.357682114715972e-05,
+      "loss": 2.2292,
+      "step": 65944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.356177124866523e-05,
+      "loss": 2.2631,
+      "step": 65952
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.354672901649198e-05,
+      "loss": 2.2275,
+      "step": 65960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3531694452148606e-05,
+      "loss": 2.2601,
+      "step": 65968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3516667557142955e-05,
+      "loss": 2.2627,
+      "step": 65976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3501648332982213e-05,
+      "loss": 2.2453,
+      "step": 65984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.348663678117264e-05,
+      "loss": 2.2382,
+      "step": 65992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.347163290321987e-05,
+      "loss": 2.2523,
+      "step": 66000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.345663670062871e-05,
+      "loss": 2.2535,
+      "step": 66008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3441648174903115e-05,
+      "loss": 2.2637,
+      "step": 66016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3426667327546434e-05,
+      "loss": 2.2445,
+      "step": 66024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3411694160061134e-05,
+      "loss": 2.2555,
+      "step": 66032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.339672867394893e-05,
+      "loss": 2.2585,
+      "step": 66040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.338177087071081e-05,
+      "loss": 2.244,
+      "step": 66048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.336682075184688e-05,
+      "loss": 2.2268,
+      "step": 66056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.335187831885662e-05,
+      "loss": 2.2532,
+      "step": 66064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.333694357323867e-05,
+      "loss": 2.2307,
+      "step": 66072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.332201651649086e-05,
+      "loss": 2.2695,
+      "step": 66080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.330709715011032e-05,
+      "loss": 2.2407,
+      "step": 66088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.329218547559337e-05,
+      "loss": 2.2579,
+      "step": 66096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3277281494435576e-05,
+      "loss": 2.2573,
+      "step": 66104
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.326238520813171e-05,
+      "loss": 2.2455,
+      "step": 66112
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.324749661817578e-05,
+      "loss": 2.2528,
+      "step": 66120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.323261572606104e-05,
+      "loss": 2.2437,
+      "step": 66128
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.321774253327995e-05,
+      "loss": 2.2724,
+      "step": 66136
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.320287704132423e-05,
+      "loss": 2.2582,
+      "step": 66144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.318801925168477e-05,
+      "loss": 2.2469,
+      "step": 66152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.317316916585174e-05,
+      "loss": 2.2373,
+      "step": 66160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.315832678531449e-05,
+      "loss": 2.2654,
+      "step": 66168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.31434921115617e-05,
+      "loss": 2.2523,
+      "step": 66176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.312866514608115e-05,
+      "loss": 2.2623,
+      "step": 66184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.311384589035991e-05,
+      "loss": 2.256,
+      "step": 66192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.309903434588424e-05,
+      "loss": 2.2419,
+      "step": 66200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.308423051413968e-05,
+      "loss": 2.2336,
+      "step": 66208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.306943439661101e-05,
+      "loss": 2.2455,
+      "step": 66216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.305464599478214e-05,
+      "loss": 2.2613,
+      "step": 66224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.303986531013625e-05,
+      "loss": 2.2576,
+      "step": 66232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.302509234415583e-05,
+      "loss": 2.2667,
+      "step": 66240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3010327098322445e-05,
+      "loss": 2.2555,
+      "step": 66248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2995569574117056e-05,
+      "loss": 2.2547,
+      "step": 66256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.298081977301967e-05,
+      "loss": 2.2374,
+      "step": 66264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.296607769650965e-05,
+      "loss": 2.2201,
+      "step": 66272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2951343346065565e-05,
+      "loss": 2.2791,
+      "step": 66280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2936616723165115e-05,
+      "loss": 2.247,
+      "step": 66288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.292189782928536e-05,
+      "loss": 2.2488,
+      "step": 66296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2907186665902525e-05,
+      "loss": 2.2235,
+      "step": 66304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.289248323449202e-05,
+      "loss": 2.2651,
+      "step": 66312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.287778753652855e-05,
+      "loss": 2.2738,
+      "step": 66320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.286309957348596e-05,
+      "loss": 2.2466,
+      "step": 66328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2848419346837414e-05,
+      "loss": 2.2497,
+      "step": 66336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.283374685805524e-05,
+      "loss": 2.2476,
+      "step": 66344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.281908210861101e-05,
+      "loss": 2.2549,
+      "step": 66352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2804425099975525e-05,
+      "loss": 2.2299,
+      "step": 66360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.278977583361877e-05,
+      "loss": 2.2515,
+      "step": 66368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.277513431101002e-05,
+      "loss": 2.2469,
+      "step": 66376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.276050053361771e-05,
+      "loss": 2.2558,
+      "step": 66384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2745874502909526e-05,
+      "loss": 2.2558,
+      "step": 66392
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.273125622035239e-05,
+      "loss": 2.2312,
+      "step": 66400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.271664568741242e-05,
+      "loss": 2.2433,
+      "step": 66408
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.270204290555498e-05,
+      "loss": 2.2262,
+      "step": 66416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.268744787624465e-05,
+      "loss": 2.2407,
+      "step": 66424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.267286060094521e-05,
+      "loss": 2.2339,
+      "step": 66432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.265828108111969e-05,
+      "loss": 2.2455,
+      "step": 66440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.264370931823036e-05,
+      "loss": 2.2514,
+      "step": 66448
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.262914531373864e-05,
+      "loss": 2.2593,
+      "step": 66456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.261458906910524e-05,
+      "loss": 2.2513,
+      "step": 66464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.260004058579006e-05,
+      "loss": 2.2717,
+      "step": 66472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2585499865252225e-05,
+      "loss": 2.2531,
+      "step": 66480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.257096690895014e-05,
+      "loss": 2.2519,
+      "step": 66488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2556441718341325e-05,
+      "loss": 2.2317,
+      "step": 66496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2541924294882554e-05,
+      "loss": 2.2616,
+      "step": 66504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.252741464002991e-05,
+      "loss": 2.2614,
+      "step": 66512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2512912755238555e-05,
+      "loss": 2.2541,
+      "step": 66520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.249841864196303e-05,
+      "loss": 2.2485,
+      "step": 66528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.248393230165695e-05,
+      "loss": 2.258,
+      "step": 66536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.246945373577321e-05,
+      "loss": 2.265,
+      "step": 66544
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.245498294576398e-05,
+      "loss": 2.2617,
+      "step": 66552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.244051993308054e-05,
+      "loss": 2.217,
+      "step": 66560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.242606469917348e-05,
+      "loss": 2.2494,
+      "step": 66568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.241161724549259e-05,
+      "loss": 2.2586,
+      "step": 66576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.239717757348683e-05,
+      "loss": 2.2329,
+      "step": 66584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.238274568460446e-05,
+      "loss": 2.2256,
+      "step": 66592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.236832158029286e-05,
+      "loss": 2.2528,
+      "step": 66600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.235390526199874e-05,
+      "loss": 2.2336,
+      "step": 66608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2339496731167954e-05,
+      "loss": 2.2355,
+      "step": 66616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2325095989245604e-05,
+      "loss": 2.2555,
+      "step": 66624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.231070303767599e-05,
+      "loss": 2.2709,
+      "step": 66632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.229631787790264e-05,
+      "loss": 2.2583,
+      "step": 66640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.228194051136833e-05,
+      "loss": 2.2511,
+      "step": 66648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.226757093951501e-05,
+      "loss": 2.2591,
+      "step": 66656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.225320916378386e-05,
+      "loss": 2.2554,
+      "step": 66664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2238855185615294e-05,
+      "loss": 2.2696,
+      "step": 66672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.222450900644894e-05,
+      "loss": 2.2704,
+      "step": 66680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.221017062772363e-05,
+      "loss": 2.2281,
+      "step": 66688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.219584005087742e-05,
+      "loss": 2.227,
+      "step": 66696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.218151727734759e-05,
+      "loss": 2.2286,
+      "step": 66704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2167202308570645e-05,
+      "loss": 2.2672,
+      "step": 66712
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2152895145982274e-05,
+      "loss": 2.2748,
+      "step": 66720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2138595791017416e-05,
+      "loss": 2.2477,
+      "step": 66728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.212430424511021e-05,
+      "loss": 2.2185,
+      "step": 66736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.211002050969403e-05,
+      "loss": 2.2905,
+      "step": 66744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2095744586201435e-05,
+      "loss": 2.2839,
+      "step": 66752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2081476476064225e-05,
+      "loss": 2.2444,
+      "step": 66760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.206721618071342e-05,
+      "loss": 2.2445,
+      "step": 66768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.205296370157922e-05,
+      "loss": 2.2229,
+      "step": 66776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.203871904009113e-05,
+      "loss": 2.2662,
+      "step": 66784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2024482197677726e-05,
+      "loss": 2.2498,
+      "step": 66792
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2010253175766966e-05,
+      "loss": 2.2227,
+      "step": 66800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.199603197578587e-05,
+      "loss": 2.249,
+      "step": 66808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1981818599160765e-05,
+      "loss": 2.2586,
+      "step": 66816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.19676130473172e-05,
+      "loss": 2.2392,
+      "step": 66824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.195341532167987e-05,
+      "loss": 2.2687,
+      "step": 66832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.193922542367276e-05,
+      "loss": 2.2246,
+      "step": 66840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.192504335471904e-05,
+      "loss": 2.2723,
+      "step": 66848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.191086911624104e-05,
+      "loss": 2.261,
+      "step": 66856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1896702709660434e-05,
+      "loss": 2.2474,
+      "step": 66864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.188254413639794e-05,
+      "loss": 2.2484,
+      "step": 66872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1868393397873644e-05,
+      "loss": 2.2538,
+      "step": 66880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.18542504955068e-05,
+      "loss": 2.2563,
+      "step": 66888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.18401154307158e-05,
+      "loss": 2.2508,
+      "step": 66896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1825988204918355e-05,
+      "loss": 2.2435,
+      "step": 66904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1811868819531345e-05,
+      "loss": 2.2412,
+      "step": 66912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.179775727597084e-05,
+      "loss": 2.2502,
+      "step": 66920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.178365357565219e-05,
+      "loss": 2.2375,
+      "step": 66928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.176955771998985e-05,
+      "loss": 2.27,
+      "step": 66936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.17554697103976e-05,
+      "loss": 2.2658,
+      "step": 66944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1741389548288395e-05,
+      "loss": 2.2309,
+      "step": 66952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1727317235074374e-05,
+      "loss": 2.2407,
+      "step": 66960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.171325277216692e-05,
+      "loss": 2.2498,
+      "step": 66968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1699196160976605e-05,
+      "loss": 2.2548,
+      "step": 66976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.168514740291325e-05,
+      "loss": 2.2494,
+      "step": 66984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.167110649938586e-05,
+      "loss": 2.2344,
+      "step": 66992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1657073451802636e-05,
+      "loss": 2.271,
+      "step": 67000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.164304826157102e-05,
+      "loss": 2.2628,
+      "step": 67008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.16290309300977e-05,
+      "loss": 2.2333,
+      "step": 67016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.161502145878849e-05,
+      "loss": 2.2653,
+      "step": 67024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.160101984904847e-05,
+      "loss": 2.2383,
+      "step": 67032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.158702610228193e-05,
+      "loss": 2.2354,
+      "step": 67040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1573040219892335e-05,
+      "loss": 2.2784,
+      "step": 67048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.155906220328245e-05,
+      "loss": 2.2363,
+      "step": 67056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.154509205385415e-05,
+      "loss": 2.2547,
+      "step": 67064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1531129773008535e-05,
+      "loss": 2.2536,
+      "step": 67072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.151717536214602e-05,
+      "loss": 2.239,
+      "step": 67080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.150322882266606e-05,
+      "loss": 2.2725,
+      "step": 67088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1489290155967513e-05,
+      "loss": 2.2463,
+      "step": 67096
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.147535936344827e-05,
+      "loss": 2.2363,
+      "step": 67104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.146143644650553e-05,
+      "loss": 2.2504,
+      "step": 67112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1447521406535724e-05,
+      "loss": 2.2286,
+      "step": 67120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1433614244934364e-05,
+      "loss": 2.2464,
+      "step": 67128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.141971496309635e-05,
+      "loss": 2.236,
+      "step": 67136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.140582356241567e-05,
+      "loss": 2.2415,
+      "step": 67144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.139194004428552e-05,
+      "loss": 2.2619,
+      "step": 67152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1378064410098406e-05,
+      "loss": 2.2692,
+      "step": 67160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.136419666124589e-05,
+      "loss": 2.2488,
+      "step": 67168
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1350336799118904e-05,
+      "loss": 2.2698,
+      "step": 67176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.133648482510747e-05,
+      "loss": 2.2305,
+      "step": 67184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1322640740600886e-05,
+      "loss": 2.2447,
+      "step": 67192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.130880454698761e-05,
+      "loss": 2.2675,
+      "step": 67200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.129497624565537e-05,
+      "loss": 2.2594,
+      "step": 67208
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1281155837991025e-05,
+      "loss": 2.2471,
+      "step": 67216
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.126734332538072e-05,
+      "loss": 2.2416,
+      "step": 67224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.125353870920975e-05,
+      "loss": 2.2586,
+      "step": 67232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.123974199086264e-05,
+      "loss": 2.2703,
+      "step": 67240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.122595317172313e-05,
+      "loss": 2.2194,
+      "step": 67248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.121217225317416e-05,
+      "loss": 2.261,
+      "step": 67256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1198399236597884e-05,
+      "loss": 2.2569,
+      "step": 67264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.118463412337564e-05,
+      "loss": 2.2531,
+      "step": 67272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1170876914888e-05,
+      "loss": 2.2586,
+      "step": 67280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.115712761251473e-05,
+      "loss": 2.262,
+      "step": 67288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1143386217634835e-05,
+      "loss": 2.2306,
+      "step": 67296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.112965273162646e-05,
+      "loss": 2.2577,
+      "step": 67304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.111592715586702e-05,
+      "loss": 2.2533,
+      "step": 67312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.110220949173308e-05,
+      "loss": 2.2613,
+      "step": 67320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.10884997406005e-05,
+      "loss": 2.259,
+      "step": 67328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1074797903844255e-05,
+      "loss": 2.2376,
+      "step": 67336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1061103982838534e-05,
+      "loss": 2.2503,
+      "step": 67344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.104741797895684e-05,
+      "loss": 2.2349,
+      "step": 67352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.10337398935717e-05,
+      "loss": 2.2482,
+      "step": 67360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.102006972805505e-05,
+      "loss": 2.2456,
+      "step": 67368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1006407483777854e-05,
+      "loss": 2.226,
+      "step": 67376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.099275316211038e-05,
+      "loss": 2.2482,
+      "step": 67384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.097910676442211e-05,
+      "loss": 2.2504,
+      "step": 67392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.096546829208163e-05,
+      "loss": 2.25,
+      "step": 67400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.095183774645689e-05,
+      "loss": 2.2508,
+      "step": 67408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.093821512891491e-05,
+      "loss": 2.2391,
+      "step": 67416
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0924600440821925e-05,
+      "loss": 2.2487,
+      "step": 67424
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.09109936835435e-05,
+      "loss": 2.2459,
+      "step": 67432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.089739485844422e-05,
+      "loss": 2.2512,
+      "step": 67440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.088380396688804e-05,
+      "loss": 2.2524,
+      "step": 67448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0870221010238045e-05,
+      "loss": 2.2506,
+      "step": 67456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0856645989856467e-05,
+      "loss": 2.2434,
+      "step": 67464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0843078907104865e-05,
+      "loss": 2.2879,
+      "step": 67472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0829519763343916e-05,
+      "loss": 2.2371,
+      "step": 67480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.081596855993354e-05,
+      "loss": 2.234,
+      "step": 67488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.080242529823283e-05,
+      "loss": 2.2686,
+      "step": 67496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.07888899796001e-05,
+      "loss": 2.2543,
+      "step": 67504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.077536260539287e-05,
+      "loss": 2.2563,
+      "step": 67512
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.076184317696786e-05,
+      "loss": 2.2502,
+      "step": 67520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0748331695681004e-05,
+      "loss": 2.2457,
+      "step": 67528
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0734828162887396e-05,
+      "loss": 2.2475,
+      "step": 67536
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.07213325799414e-05,
+      "loss": 2.2522,
+      "step": 67544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.070784494819651e-05,
+      "loss": 2.2518,
+      "step": 67552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.069436526900549e-05,
+      "loss": 2.2386,
+      "step": 67560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.068089354372026e-05,
+      "loss": 2.2469,
+      "step": 67568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0667429773691966e-05,
+      "loss": 2.2568,
+      "step": 67576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.065397396027093e-05,
+      "loss": 2.2659,
+      "step": 67584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0640526104806724e-05,
+      "loss": 2.2448,
+      "step": 67592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.062708620864806e-05,
+      "loss": 2.224,
+      "step": 67600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.06136542731429e-05,
+      "loss": 2.2597,
+      "step": 67608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0600230299638374e-05,
+      "loss": 2.2474,
+      "step": 67616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.058681428948088e-05,
+      "loss": 2.2282,
+      "step": 67624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.057340624401591e-05,
+      "loss": 2.2672,
+      "step": 67632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.056000616458825e-05,
+      "loss": 2.2257,
+      "step": 67640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.054661405254182e-05,
+      "loss": 2.25,
+      "step": 67648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0533229909219796e-05,
+      "loss": 2.2532,
+      "step": 67656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0519853735964556e-05,
+      "loss": 2.2584,
+      "step": 67664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0506485534117586e-05,
+      "loss": 2.2387,
+      "step": 67672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0493125305019698e-05,
+      "loss": 2.2381,
+      "step": 67680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0479773050010863e-05,
+      "loss": 2.2468,
+      "step": 67688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0466428770430154e-05,
+      "loss": 2.244,
+      "step": 67696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0453092467616024e-05,
+      "loss": 2.26,
+      "step": 67704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0439764142905935e-05,
+      "loss": 2.2349,
+      "step": 67712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.042644379763672e-05,
+      "loss": 2.2544,
+      "step": 67720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0413131433144314e-05,
+      "loss": 2.2526,
+      "step": 67728
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.039982705076383e-05,
+      "loss": 2.2455,
+      "step": 67736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0386530651829664e-05,
+      "loss": 2.2511,
+      "step": 67744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.037324223767536e-05,
+      "loss": 2.2395,
+      "step": 67752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.035996180963368e-05,
+      "loss": 2.2336,
+      "step": 67760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0346689369036573e-05,
+      "loss": 2.2904,
+      "step": 67768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.033342491721515e-05,
+      "loss": 2.2546,
+      "step": 67776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0320168455499805e-05,
+      "loss": 2.2334,
+      "step": 67784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.030691998522008e-05,
+      "loss": 2.2363,
+      "step": 67792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0293679507704715e-05,
+      "loss": 2.2421,
+      "step": 67800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0280447024281644e-05,
+      "loss": 2.2608,
+      "step": 67808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0267222536278034e-05,
+      "loss": 2.2559,
+      "step": 67816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.02540060450202e-05,
+      "loss": 2.2584,
+      "step": 67824
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0240797551833704e-05,
+      "loss": 2.26,
+      "step": 67832
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0227597058043272e-05,
+      "loss": 2.2339,
+      "step": 67840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0214404564972813e-05,
+      "loss": 2.259,
+      "step": 67848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0201220073945535e-05,
+      "loss": 2.2416,
+      "step": 67856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.018804358628368e-05,
+      "loss": 2.2531,
+      "step": 67864
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0174875103308837e-05,
+      "loss": 2.2414,
+      "step": 67872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0161714626341696e-05,
+      "loss": 2.2552,
+      "step": 67880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0148562156702163e-05,
+      "loss": 2.2273,
+      "step": 67888
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0135417695709433e-05,
+      "loss": 2.2479,
+      "step": 67896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.012228124468175e-05,
+      "loss": 2.256,
+      "step": 67904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0109152804936624e-05,
+      "loss": 2.2643,
+      "step": 67912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.009603237779083e-05,
+      "loss": 2.2358,
+      "step": 67920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0082919964560173e-05,
+      "loss": 2.2535,
+      "step": 67928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0069815566559866e-05,
+      "loss": 2.2345,
+      "step": 67936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0056719185104114e-05,
+      "loss": 2.2498,
+      "step": 67944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.004363082150644e-05,
+      "loss": 2.2664,
+      "step": 67952
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0030550477079568e-05,
+      "loss": 2.2182,
+      "step": 67960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.00174781531353e-05,
+      "loss": 2.2253,
+      "step": 67968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0004413850984802e-05,
+      "loss": 2.2422,
+      "step": 67976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9991357571938328e-05,
+      "loss": 2.2167,
+      "step": 67984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9978309317305303e-05,
+      "loss": 2.2355,
+      "step": 67992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9965269088394455e-05,
+      "loss": 2.2652,
+      "step": 68000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9952236886513586e-05,
+      "loss": 2.2358,
+      "step": 68008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9939212712969793e-05,
+      "loss": 2.2619,
+      "step": 68016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9926196569069325e-05,
+      "loss": 2.2639,
+      "step": 68024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9913188456117607e-05,
+      "loss": 2.2468,
+      "step": 68032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9900188375419296e-05,
+      "loss": 2.233,
+      "step": 68040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9887196328278223e-05,
+      "loss": 2.2416,
+      "step": 68048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9874212315997414e-05,
+      "loss": 2.2612,
+      "step": 68056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9861236339879085e-05,
+      "loss": 2.2507,
+      "step": 68064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9848268401224666e-05,
+      "loss": 2.2446,
+      "step": 68072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9835308501334753e-05,
+      "loss": 2.2554,
+      "step": 68080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9822356641509165e-05,
+      "loss": 2.2703,
+      "step": 68088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.98094128230469e-05,
+      "loss": 2.2513,
+      "step": 68096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9796477047246135e-05,
+      "loss": 2.2228,
+      "step": 68104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.978354931540427e-05,
+      "loss": 2.2552,
+      "step": 68112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9770629628817868e-05,
+      "loss": 2.2525,
+      "step": 68120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9757717988782723e-05,
+      "loss": 2.2336,
+      "step": 68128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9744814396593777e-05,
+      "loss": 2.2668,
+      "step": 68136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9731918853545206e-05,
+      "loss": 2.2461,
+      "step": 68144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9719031360930337e-05,
+      "loss": 2.2479,
+      "step": 68152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9706151920041735e-05,
+      "loss": 2.2499,
+      "step": 68160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.969328053217112e-05,
+      "loss": 2.2617,
+      "step": 68168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9680417198609417e-05,
+      "loss": 2.2192,
+      "step": 68176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9667561920646748e-05,
+      "loss": 2.267,
+      "step": 68184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.965471469957246e-05,
+      "loss": 2.2546,
+      "step": 68192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9641875536674996e-05,
+      "loss": 2.2279,
+      "step": 68200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.962904443324211e-05,
+      "loss": 2.2424,
+      "step": 68208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9616221390560645e-05,
+      "loss": 2.2197,
+      "step": 68216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9603406409916684e-05,
+      "loss": 2.2274,
+      "step": 68224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9590599492595532e-05,
+      "loss": 2.2605,
+      "step": 68232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9577800639881597e-05,
+      "loss": 2.2386,
+      "step": 68240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9565009853058574e-05,
+      "loss": 2.2233,
+      "step": 68248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.955222713340932e-05,
+      "loss": 2.2557,
+      "step": 68256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9539452482215802e-05,
+      "loss": 2.2606,
+      "step": 68264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.952668590075932e-05,
+      "loss": 2.2267,
+      "step": 68272
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.951392739032021e-05,
+      "loss": 2.2501,
+      "step": 68280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9501176952178155e-05,
+      "loss": 2.2334,
+      "step": 68288
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9488434587611936e-05,
+      "loss": 2.2446,
+      "step": 68296
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9475700297899485e-05,
+      "loss": 2.2251,
+      "step": 68304
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9462974084318035e-05,
+      "loss": 2.2391,
+      "step": 68312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.945025594814394e-05,
+      "loss": 2.2621,
+      "step": 68320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9437545890652746e-05,
+      "loss": 2.2282,
+      "step": 68328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9424843913119238e-05,
+      "loss": 2.2689,
+      "step": 68336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.941215001681728e-05,
+      "loss": 2.2365,
+      "step": 68344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.939946420302005e-05,
+      "loss": 2.2343,
+      "step": 68352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9386786472999867e-05,
+      "loss": 2.2386,
+      "step": 68360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9374116828028202e-05,
+      "loss": 2.2578,
+      "step": 68368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9361455269375786e-05,
+      "loss": 2.254,
+      "step": 68376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9348801798312475e-05,
+      "loss": 2.2579,
+      "step": 68384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9336156416107353e-05,
+      "loss": 2.2638,
+      "step": 68392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9323519124028676e-05,
+      "loss": 2.2563,
+      "step": 68400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.931088992334388e-05,
+      "loss": 2.2296,
+      "step": 68408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.929826881531963e-05,
+      "loss": 2.2448,
+      "step": 68416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.928565580122173e-05,
+      "loss": 2.2711,
+      "step": 68424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.92730508823152e-05,
+      "loss": 2.2415,
+      "step": 68432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9260454059864246e-05,
+      "loss": 2.2437,
+      "step": 68440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.924786533513225e-05,
+      "loss": 2.2541,
+      "step": 68448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9235284709381773e-05,
+      "loss": 2.2438,
+      "step": 68456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.922271218387464e-05,
+      "loss": 2.2524,
+      "step": 68464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9210147759871755e-05,
+      "loss": 2.2331,
+      "step": 68472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.919759143863326e-05,
+      "loss": 2.2392,
+      "step": 68480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9185043221418495e-05,
+      "loss": 2.236,
+      "step": 68488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.917250310948596e-05,
+      "loss": 2.2359,
+      "step": 68496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9159971104093403e-05,
+      "loss": 2.2461,
+      "step": 68504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9147447206497657e-05,
+      "loss": 2.2473,
+      "step": 68512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9134931417954804e-05,
+      "loss": 2.2482,
+      "step": 68520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.912242373972016e-05,
+      "loss": 2.2413,
+      "step": 68528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9109924173048102e-05,
+      "loss": 2.2653,
+      "step": 68536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9097432719192336e-05,
+      "loss": 2.2558,
+      "step": 68544
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9084949379405614e-05,
+      "loss": 2.2246,
+      "step": 68552
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9072474154939994e-05,
+      "loss": 2.2421,
+      "step": 68560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.906000704704668e-05,
+      "loss": 2.2576,
+      "step": 68568
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.904754805697598e-05,
+      "loss": 2.2211,
+      "step": 68576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9035097185977533e-05,
+      "loss": 2.2582,
+      "step": 68584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9022654435300067e-05,
+      "loss": 2.2431,
+      "step": 68592
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9010219806191523e-05,
+      "loss": 2.243,
+      "step": 68600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8997793299899035e-05,
+      "loss": 2.2516,
+      "step": 68608
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8985374917668852e-05,
+      "loss": 2.2395,
+      "step": 68616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8972964660746533e-05,
+      "loss": 2.2494,
+      "step": 68624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.896056253037673e-05,
+      "loss": 2.2506,
+      "step": 68632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.894816852780331e-05,
+      "loss": 2.2253,
+      "step": 68640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.893578265426933e-05,
+      "loss": 2.2683,
+      "step": 68648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.892340491101701e-05,
+      "loss": 2.2526,
+      "step": 68656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.891103529928778e-05,
+      "loss": 2.256,
+      "step": 68664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.889867382032222e-05,
+      "loss": 2.2201,
+      "step": 68672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.888632047536014e-05,
+      "loss": 2.2721,
+      "step": 68680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8873975265640496e-05,
+      "loss": 2.2602,
+      "step": 68688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8861638192401446e-05,
+      "loss": 2.2409,
+      "step": 68696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8849309256880336e-05,
+      "loss": 2.2454,
+      "step": 68704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8836988460313674e-05,
+      "loss": 2.2517,
+      "step": 68712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.882467580393719e-05,
+      "loss": 2.2412,
+      "step": 68720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8812371288985723e-05,
+      "loss": 2.2379,
+      "step": 68728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8800074916693433e-05,
+      "loss": 2.2637,
+      "step": 68736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8787786688293495e-05,
+      "loss": 2.2438,
+      "step": 68744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.877550660501836e-05,
+      "loss": 2.2476,
+      "step": 68752
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.876323466809971e-05,
+      "loss": 2.2562,
+      "step": 68760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8750970878768274e-05,
+      "loss": 2.2407,
+      "step": 68768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8738715238254106e-05,
+      "loss": 2.254,
+      "step": 68776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.872646774778634e-05,
+      "loss": 2.2541,
+      "step": 68784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8714228408593312e-05,
+      "loss": 2.2586,
+      "step": 68792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8701997221902628e-05,
+      "loss": 2.2567,
+      "step": 68800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8689774188940914e-05,
+      "loss": 2.2603,
+      "step": 68808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.867755931093415e-05,
+      "loss": 2.2403,
+      "step": 68816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8665352589107397e-05,
+      "loss": 2.2362,
+      "step": 68824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.865315402468488e-05,
+      "loss": 2.2518,
+      "step": 68832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.864096361889012e-05,
+      "loss": 2.2633,
+      "step": 68840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8628781372945656e-05,
+      "loss": 2.2291,
+      "step": 68848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.861660728807337e-05,
+      "loss": 2.2435,
+      "step": 68856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.860444136549425e-05,
+      "loss": 2.2648,
+      "step": 68864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8592283606428405e-05,
+      "loss": 2.2367,
+      "step": 68872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8580134012095262e-05,
+      "loss": 2.2402,
+      "step": 68880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8567992583713314e-05,
+      "loss": 2.2539,
+      "step": 68888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.85558593225003e-05,
+      "loss": 2.2376,
+      "step": 68896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.85437342296731e-05,
+      "loss": 2.2521,
+      "step": 68904
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8531617306447806e-05,
+      "loss": 2.2481,
+      "step": 68912
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8519508554039665e-05,
+      "loss": 2.2253,
+      "step": 68920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.850740797366313e-05,
+      "loss": 2.2525,
+      "step": 68928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8495315566531816e-05,
+      "loss": 2.2203,
+      "step": 68936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.848323133385851e-05,
+      "loss": 2.233,
+      "step": 68944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8471155276855205e-05,
+      "loss": 2.2382,
+      "step": 68952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8459087396733052e-05,
+      "loss": 2.244,
+      "step": 68960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8447027694702404e-05,
+      "loss": 2.2551,
+      "step": 68968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8434976171972776e-05,
+      "loss": 2.2337,
+      "step": 68976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.842293282975286e-05,
+      "loss": 2.2619,
+      "step": 68984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8410897669250548e-05,
+      "loss": 2.2249,
+      "step": 68992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8398870691672873e-05,
+      "loss": 2.2537,
+      "step": 69000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8386851898226108e-05,
+      "loss": 2.2287,
+      "step": 69008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.837484129011564e-05,
+      "loss": 2.2336,
+      "step": 69016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8362838868546066e-05,
+      "loss": 2.2558,
+      "step": 69024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.83508446347212e-05,
+      "loss": 2.2727,
+      "step": 69032
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8338858589843965e-05,
+      "loss": 2.2575,
+      "step": 69040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.832688073511648e-05,
+      "loss": 2.2432,
+      "step": 69048
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8314911071740074e-05,
+      "loss": 2.2761,
+      "step": 69056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8302949600915223e-05,
+      "loss": 2.2327,
+      "step": 69064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8290996323841634e-05,
+      "loss": 2.2703,
+      "step": 69072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.82790512417181e-05,
+      "loss": 2.2523,
+      "step": 69080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.826711435574268e-05,
+      "loss": 2.2392,
+      "step": 69088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8255185667112594e-05,
+      "loss": 2.2437,
+      "step": 69096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.824326517702415e-05,
+      "loss": 2.2344,
+      "step": 69104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.823135288667298e-05,
+      "loss": 2.2321,
+      "step": 69112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.821944879725376e-05,
+      "loss": 2.2482,
+      "step": 69120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.820755290996045e-05,
+      "loss": 2.245,
+      "step": 69128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.819566522598613e-05,
+      "loss": 2.2393,
+      "step": 69136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8183785746523027e-05,
+      "loss": 2.2322,
+      "step": 69144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.817191447276265e-05,
+      "loss": 2.2385,
+      "step": 69152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.816005140589557e-05,
+      "loss": 2.2588,
+      "step": 69160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.814819654711161e-05,
+      "loss": 2.248,
+      "step": 69168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.813634989759976e-05,
+      "loss": 2.2468,
+      "step": 69176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8124511458548108e-05,
+      "loss": 2.2611,
+      "step": 69184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8112681231144043e-05,
+      "loss": 2.2353,
+      "step": 69192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.810085921657406e-05,
+      "loss": 2.229,
+      "step": 69200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8089045416023818e-05,
+      "loss": 2.2605,
+      "step": 69208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8077239830678197e-05,
+      "loss": 2.2413,
+      "step": 69216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8065442461721226e-05,
+      "loss": 2.2428,
+      "step": 69224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.805365331033609e-05,
+      "loss": 2.2346,
+      "step": 69232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.804187237770521e-05,
+      "loss": 2.2485,
+      "step": 69240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8030099665010122e-05,
+      "loss": 2.2453,
+      "step": 69248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.801833517343158e-05,
+      "loss": 2.2248,
+      "step": 69256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8006578904149484e-05,
+      "loss": 2.2478,
+      "step": 69264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7994830858342914e-05,
+      "loss": 2.2438,
+      "step": 69272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7983091037190152e-05,
+      "loss": 2.2694,
+      "step": 69280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.797135944186862e-05,
+      "loss": 2.2484,
+      "step": 69288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7959636073554922e-05,
+      "loss": 2.2387,
+      "step": 69296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7947920933424885e-05,
+      "loss": 2.2504,
+      "step": 69304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7936214022653446e-05,
+      "loss": 2.2324,
+      "step": 69312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7924515342414727e-05,
+      "loss": 2.2343,
+      "step": 69320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7912824893882062e-05,
+      "loss": 2.26,
+      "step": 69328
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7901142678227908e-05,
+      "loss": 2.2506,
+      "step": 69336
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7889468696623976e-05,
+      "loss": 2.2456,
+      "step": 69344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7877802950241054e-05,
+      "loss": 2.2553,
+      "step": 69352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7866145440249146e-05,
+      "loss": 2.2607,
+      "step": 69360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.78544961678175e-05,
+      "loss": 2.258,
+      "step": 69368
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7842855134114387e-05,
+      "loss": 2.2442,
+      "step": 69376
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7831222340307412e-05,
+      "loss": 2.2306,
+      "step": 69384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7819597787563217e-05,
+      "loss": 2.2399,
+      "step": 69392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7807981477047696e-05,
+      "loss": 2.227,
+      "step": 69400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7796373409925935e-05,
+      "loss": 2.2312,
+      "step": 69408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7784773587362097e-05,
+      "loss": 2.2609,
+      "step": 69416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7773182010519632e-05,
+      "loss": 2.2126,
+      "step": 69424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7761598680561086e-05,
+      "loss": 2.2677,
+      "step": 69432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.77500235986482e-05,
+      "loss": 2.2418,
+      "step": 69440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7738456765941918e-05,
+      "loss": 2.2484,
+      "step": 69448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7726898183602264e-05,
+      "loss": 2.2539,
+      "step": 69456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7715347852788547e-05,
+      "loss": 2.2307,
+      "step": 69464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7703805774659204e-05,
+      "loss": 2.2262,
+      "step": 69472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7692271950371817e-05,
+      "loss": 2.2384,
+      "step": 69480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.768074638108318e-05,
+      "loss": 2.2409,
+      "step": 69488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7669229067949222e-05,
+      "loss": 2.2278,
+      "step": 69496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.765772001212509e-05,
+      "loss": 2.258,
+      "step": 69504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7646219214765056e-05,
+      "loss": 2.2603,
+      "step": 69512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7634726677022592e-05,
+      "loss": 2.2429,
+      "step": 69520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7623242400050335e-05,
+      "loss": 2.244,
+      "step": 69528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7611766385000092e-05,
+      "loss": 2.2346,
+      "step": 69536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.760029863302284e-05,
+      "loss": 2.2324,
+      "step": 69544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.758883914526875e-05,
+      "loss": 2.2654,
+      "step": 69552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7577387922887105e-05,
+      "loss": 2.2516,
+      "step": 69560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.756594496702643e-05,
+      "loss": 2.2474,
+      "step": 69568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.755451027883438e-05,
+      "loss": 2.234,
+      "step": 69576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7543083859457773e-05,
+      "loss": 2.2472,
+      "step": 69584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.753166571004263e-05,
+      "loss": 2.2352,
+      "step": 69592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.752025583173414e-05,
+      "loss": 2.2379,
+      "step": 69600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7508854225676622e-05,
+      "loss": 2.2512,
+      "step": 69608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7497460893013622e-05,
+      "loss": 2.2183,
+      "step": 69616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.748607583488779e-05,
+      "loss": 2.2475,
+      "step": 69624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7474699052441e-05,
+      "loss": 2.2596,
+      "step": 69632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7463330546814303e-05,
+      "loss": 2.258,
+      "step": 69640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7451970319147847e-05,
+      "loss": 2.23,
+      "step": 69648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7440618370581043e-05,
+      "loss": 2.2531,
+      "step": 69656
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7429274702252415e-05,
+      "loss": 2.2537,
+      "step": 69664
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7417939315299646e-05,
+      "loss": 2.2326,
+      "step": 69672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.740661221085965e-05,
+      "loss": 2.2372,
+      "step": 69680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7395293390068416e-05,
+      "loss": 2.2364,
+      "step": 69688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7383982854061208e-05,
+      "loss": 2.2454,
+      "step": 69696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7372680603972406e-05,
+      "loss": 2.2332,
+      "step": 69704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.736138664093552e-05,
+      "loss": 2.2303,
+      "step": 69712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7350100966083307e-05,
+      "loss": 2.2418,
+      "step": 69720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7338823580547657e-05,
+      "loss": 2.2383,
+      "step": 69728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.732755448545961e-05,
+      "loss": 2.2502,
+      "step": 69736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7316293681949416e-05,
+      "loss": 2.2296,
+      "step": 69744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7305041171146428e-05,
+      "loss": 2.2378,
+      "step": 69752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7293796954179253e-05,
+      "loss": 2.2512,
+      "step": 69760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7282561032175592e-05,
+      "loss": 2.2472,
+      "step": 69768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7271333406262363e-05,
+      "loss": 2.2466,
+      "step": 69776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7260114077565624e-05,
+      "loss": 2.2453,
+      "step": 69784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7248903047210622e-05,
+      "loss": 2.2563,
+      "step": 69792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7237700316321745e-05,
+      "loss": 2.232,
+      "step": 69800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7226505886022573e-05,
+      "loss": 2.2515,
+      "step": 69808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7215319757435837e-05,
+      "loss": 2.2466,
+      "step": 69816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7204141931683443e-05,
+      "loss": 2.2623,
+      "step": 69824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.719297240988647e-05,
+      "loss": 2.2274,
+      "step": 69832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7181811193165147e-05,
+      "loss": 2.2367,
+      "step": 69840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7170658282638895e-05,
+      "loss": 2.254,
+      "step": 69848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7159513679426274e-05,
+      "loss": 2.2372,
+      "step": 69856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7148377384645023e-05,
+      "loss": 2.2248,
+      "step": 69864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.713724939941209e-05,
+      "loss": 2.2674,
+      "step": 69872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.712612972484349e-05,
+      "loss": 2.2531,
+      "step": 69880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.711501836205449e-05,
+      "loss": 2.2285,
+      "step": 69888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.71039153121595e-05,
+      "loss": 2.2471,
+      "step": 69896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7092820576272083e-05,
+      "loss": 2.2504,
+      "step": 69904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7081734155505e-05,
+      "loss": 2.2558,
+      "step": 69912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7070656050970134e-05,
+      "loss": 2.2171,
+      "step": 69920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7059586263778534e-05,
+      "loss": 2.2526,
+      "step": 69928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.704852479504052e-05,
+      "loss": 2.261,
+      "step": 69936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.703747164586539e-05,
+      "loss": 2.2546,
+      "step": 69944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7026426817361804e-05,
+      "loss": 2.2314,
+      "step": 69952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.701539031063742e-05,
+      "loss": 2.2506,
+      "step": 69960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7004362126799183e-05,
+      "loss": 2.2254,
+      "step": 69968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.699334226695316e-05,
+      "loss": 2.2268,
+      "step": 69976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.698233073220453e-05,
+      "loss": 2.2369,
+      "step": 69984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.697132752365775e-05,
+      "loss": 2.2345,
+      "step": 69992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6960332642416348e-05,
+      "loss": 2.2376,
+      "step": 70000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6949346089583054e-05,
+      "loss": 2.2457,
+      "step": 70008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6938367866259774e-05,
+      "loss": 2.2338,
+      "step": 70016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.692739797354752e-05,
+      "loss": 2.2507,
+      "step": 70024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6916436412546537e-05,
+      "loss": 2.2379,
+      "step": 70032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.690548318435621e-05,
+      "loss": 2.231,
+      "step": 70040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6894538290075078e-05,
+      "loss": 2.2284,
+      "step": 70048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6883601730800844e-05,
+      "loss": 2.2324,
+      "step": 70056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6872673507630414e-05,
+      "loss": 2.25,
+      "step": 70064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.686175362165979e-05,
+      "loss": 2.2615,
+      "step": 70072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6850842073984196e-05,
+      "loss": 2.2395,
+      "step": 70080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6839938865698e-05,
+      "loss": 2.2541,
+      "step": 70088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6829043997894714e-05,
+      "loss": 2.2388,
+      "step": 70096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6818157471667036e-05,
+      "loss": 2.249,
+      "step": 70104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6807279288106845e-05,
+      "loss": 2.2459,
+      "step": 70112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6796409448305145e-05,
+      "loss": 2.2409,
+      "step": 70120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6785547953352108e-05,
+      "loss": 2.2253,
+      "step": 70128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6774694804337093e-05,
+      "loss": 2.2496,
+      "step": 70136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6763850002348622e-05,
+      "loss": 2.2649,
+      "step": 70144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6753013548474347e-05,
+      "loss": 2.2599,
+      "step": 70152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.674218544380111e-05,
+      "loss": 2.2362,
+      "step": 70160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6731365689414905e-05,
+      "loss": 2.2124,
+      "step": 70168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6720554286400878e-05,
+      "loss": 2.2433,
+      "step": 70176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6709751235843404e-05,
+      "loss": 2.2515,
+      "step": 70184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6698956538825914e-05,
+      "loss": 2.2439,
+      "step": 70192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.668817019643106e-05,
+      "loss": 2.2473,
+      "step": 70200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6677392209740704e-05,
+      "loss": 2.2494,
+      "step": 70208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6666622579835733e-05,
+      "loss": 2.2322,
+      "step": 70216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.665586130779636e-05,
+      "loss": 2.2667,
+      "step": 70224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6645108394701834e-05,
+      "loss": 2.2428,
+      "step": 70232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6634363841630604e-05,
+      "loss": 2.231,
+      "step": 70240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6623627649660322e-05,
+      "loss": 2.2454,
+      "step": 70248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.661289981986773e-05,
+      "loss": 2.2518,
+      "step": 70256
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.660218035332881e-05,
+      "loss": 2.231,
+      "step": 70264
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6591469251118628e-05,
+      "loss": 2.2453,
+      "step": 70272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.658076651431148e-05,
+      "loss": 2.2341,
+      "step": 70280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6570072143980766e-05,
+      "loss": 2.2548,
+      "step": 70288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.655938614119906e-05,
+      "loss": 2.2648,
+      "step": 70296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6548708507038134e-05,
+      "loss": 2.2509,
+      "step": 70304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6538039242568887e-05,
+      "loss": 2.2511,
+      "step": 70312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6527378348861374e-05,
+      "loss": 2.2556,
+      "step": 70320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.651672582698484e-05,
+      "loss": 2.2587,
+      "step": 70328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6506081678007656e-05,
+      "loss": 2.2378,
+      "step": 70336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6495445902997383e-05,
+      "loss": 2.2418,
+      "step": 70344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6484818503020714e-05,
+      "loss": 2.2381,
+      "step": 70352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6474199479143535e-05,
+      "loss": 2.2583,
+      "step": 70360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6463588832430864e-05,
+      "loss": 2.2469,
+      "step": 70368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.645298656394689e-05,
+      "loss": 2.2353,
+      "step": 70376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.644239267475495e-05,
+      "loss": 2.2416,
+      "step": 70384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6431807165917572e-05,
+      "loss": 2.2253,
+      "step": 70392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6421230038496413e-05,
+      "loss": 2.2248,
+      "step": 70400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6410661293552296e-05,
+      "loss": 2.2555,
+      "step": 70408
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6400100932145205e-05,
+      "loss": 2.2366,
+      "step": 70416
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6389548955334282e-05,
+      "loss": 2.257,
+      "step": 70424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6379005364177828e-05,
+      "loss": 2.243,
+      "step": 70432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.636847015973335e-05,
+      "loss": 2.2432,
+      "step": 70440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6357943343057388e-05,
+      "loss": 2.2539,
+      "step": 70448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6347424915205808e-05,
+      "loss": 2.2385,
+      "step": 70456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6336914877233487e-05,
+      "loss": 2.2434,
+      "step": 70464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.632641323019453e-05,
+      "loss": 2.2285,
+      "step": 70472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6315919975142234e-05,
+      "loss": 2.2441,
+      "step": 70480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.630543511312895e-05,
+      "loss": 2.2404,
+      "step": 70488
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6294958645206307e-05,
+      "loss": 2.2343,
+      "step": 70496
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6284490572425023e-05,
+      "loss": 2.2448,
+      "step": 70504
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.627403089583495e-05,
+      "loss": 2.2272,
+      "step": 70512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6263579616485185e-05,
+      "loss": 2.2442,
+      "step": 70520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6253136735423884e-05,
+      "loss": 2.2354,
+      "step": 70528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6242702253698444e-05,
+      "loss": 2.2375,
+      "step": 70536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6232276172355386e-05,
+      "loss": 2.253,
+      "step": 70544
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.622185849244035e-05,
+      "loss": 2.2367,
+      "step": 70552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.62114492149982e-05,
+      "loss": 2.229,
+      "step": 70560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.620104834107293e-05,
+      "loss": 2.2321,
+      "step": 70568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.619065587170768e-05,
+      "loss": 2.231,
+      "step": 70576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6180271807944774e-05,
+      "loss": 2.2498,
+      "step": 70584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6169896150825618e-05,
+      "loss": 2.2471,
+      "step": 70592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.615952890139089e-05,
+      "loss": 2.2352,
+      "step": 70600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6149170060680344e-05,
+      "loss": 2.2423,
+      "step": 70608
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.613881962973292e-05,
+      "loss": 2.2388,
+      "step": 70616
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6128477609586698e-05,
+      "loss": 2.2416,
+      "step": 70624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.611814400127894e-05,
+      "loss": 2.2464,
+      "step": 70632
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6107818805846028e-05,
+      "loss": 2.2523,
+      "step": 70640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6097502024323538e-05,
+      "loss": 2.2315,
+      "step": 70648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6087193657746173e-05,
+      "loss": 2.231,
+      "step": 70656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6076893707147802e-05,
+      "loss": 2.2401,
+      "step": 70664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6066602173561467e-05,
+      "loss": 2.247,
+      "step": 70672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.605631905801934e-05,
+      "loss": 2.245,
+      "step": 70680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6046044361552752e-05,
+      "loss": 2.2391,
+      "step": 70688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6035778085192218e-05,
+      "loss": 2.2277,
+      "step": 70696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.602552022996735e-05,
+      "loss": 2.2423,
+      "step": 70704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.601527079690701e-05,
+      "loss": 2.2483,
+      "step": 70712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.600502978703911e-05,
+      "loss": 2.2175,
+      "step": 70720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.599479720139078e-05,
+      "loss": 2.2641,
+      "step": 70728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.59845730409883e-05,
+      "loss": 2.2322,
+      "step": 70736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.597435730685706e-05,
+      "loss": 2.2425,
+      "step": 70744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.59641500000217e-05,
+      "loss": 2.2098,
+      "step": 70752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.595395112150591e-05,
+      "loss": 2.2403,
+      "step": 70760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.594376067233257e-05,
+      "loss": 2.2539,
+      "step": 70768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5933578653523776e-05,
+      "loss": 2.2437,
+      "step": 70776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.592340506610067e-05,
+      "loss": 2.2542,
+      "step": 70784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.591323991108366e-05,
+      "loss": 2.2404,
+      "step": 70792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.59030831894922e-05,
+      "loss": 2.2505,
+      "step": 70800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5892934902344985e-05,
+      "loss": 2.2269,
+      "step": 70808
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5882795050659836e-05,
+      "loss": 2.2446,
+      "step": 70816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.587266363545368e-05,
+      "loss": 2.2335,
+      "step": 70824
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.586254065774269e-05,
+      "loss": 2.2512,
+      "step": 70832
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5852426118542123e-05,
+      "loss": 2.2235,
+      "step": 70840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.584232001886641e-05,
+      "loss": 2.2458,
+      "step": 70848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.583222235972914e-05,
+      "loss": 2.221,
+      "step": 70856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5822133142143023e-05,
+      "loss": 2.2457,
+      "step": 70864
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.581205236712e-05,
+      "loss": 2.2718,
+      "step": 70872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5801980035671075e-05,
+      "loss": 2.2216,
+      "step": 70880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5791916148806467e-05,
+      "loss": 2.2622,
+      "step": 70888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.578186070753552e-05,
+      "loss": 2.2647,
+      "step": 70896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5771813712866728e-05,
+      "loss": 2.2284,
+      "step": 70904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5761775165807766e-05,
+      "loss": 2.2709,
+      "step": 70912
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5751745067365434e-05,
+      "loss": 2.2099,
+      "step": 70920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.574172341854569e-05,
+      "loss": 2.2521,
+      "step": 70928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5731710220353667e-05,
+      "loss": 2.2364,
+      "step": 70936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5721705473793607e-05,
+      "loss": 2.2263,
+      "step": 70944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5711709179868938e-05,
+      "loss": 2.2395,
+      "step": 70952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5701721339582236e-05,
+      "loss": 2.2513,
+      "step": 70960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5691741953935226e-05,
+      "loss": 2.2285,
+      "step": 70968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5681771023928775e-05,
+      "loss": 2.2477,
+      "step": 70976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5671808550562938e-05,
+      "loss": 2.2382,
+      "step": 70984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.566185453483685e-05,
+      "loss": 2.2506,
+      "step": 70992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5651908977748884e-05,
+      "loss": 2.2585,
+      "step": 71000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.564197188029649e-05,
+      "loss": 2.26,
+      "step": 71008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5632043243476306e-05,
+      "loss": 2.2573,
+      "step": 71016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5622123068284164e-05,
+      "loss": 2.2284,
+      "step": 71024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5612211355714947e-05,
+      "loss": 2.2082,
+      "step": 71032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5602308106762756e-05,
+      "loss": 2.2336,
+      "step": 71040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5592413322420867e-05,
+      "loss": 2.2227,
+      "step": 71048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5582527003681607e-05,
+      "loss": 2.2466,
+      "step": 71056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5572649151536583e-05,
+      "loss": 2.2434,
+      "step": 71064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5562779766976436e-05,
+      "loss": 2.2441,
+      "step": 71072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5552918850991027e-05,
+      "loss": 2.2316,
+      "step": 71080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5543066404569376e-05,
+      "loss": 2.2619,
+      "step": 71088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.553322242869959e-05,
+      "loss": 2.2562,
+      "step": 71096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5523386924368982e-05,
+      "loss": 2.2597,
+      "step": 71104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.551355989256401e-05,
+      "loss": 2.2234,
+      "step": 71112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5503741334270226e-05,
+      "loss": 2.2562,
+      "step": 71120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5493931250472425e-05,
+      "loss": 2.2738,
+      "step": 71128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.548412964215446e-05,
+      "loss": 2.2414,
+      "step": 71136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5474336510299417e-05,
+      "loss": 2.2605,
+      "step": 71144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5464551855889472e-05,
+      "loss": 2.2252,
+      "step": 71152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5454775679905968e-05,
+      "loss": 2.2447,
+      "step": 71160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.54450079833294e-05,
+      "loss": 2.2306,
+      "step": 71168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5435248767139425e-05,
+      "loss": 2.2321,
+      "step": 71176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5425498032314832e-05,
+      "loss": 2.2267,
+      "step": 71184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5415755779833543e-05,
+      "loss": 2.2515,
+      "step": 71192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.540602201067268e-05,
+      "loss": 2.2403,
+      "step": 71200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5396296725808477e-05,
+      "loss": 2.2613,
+      "step": 71208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5386579926216312e-05,
+      "loss": 2.2538,
+      "step": 71216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5376871612870738e-05,
+      "loss": 2.2423,
+      "step": 71224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5367171786745442e-05,
+      "loss": 2.2674,
+      "step": 71232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.535748044881326e-05,
+      "loss": 2.2433,
+      "step": 71240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.534779760004618e-05,
+      "loss": 2.2525,
+      "step": 71248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5338123241415334e-05,
+      "loss": 2.2384,
+      "step": 71256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5328457373890997e-05,
+      "loss": 2.2438,
+      "step": 71264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.531879999844262e-05,
+      "loss": 2.2372,
+      "step": 71272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.530915111603877e-05,
+      "loss": 2.2423,
+      "step": 71280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.529951072764718e-05,
+      "loss": 2.241,
+      "step": 71288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5289878834234726e-05,
+      "loss": 2.2451,
+      "step": 71296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.528025543676743e-05,
+      "loss": 2.232,
+      "step": 71304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.527064053621045e-05,
+      "loss": 2.2387,
+      "step": 71312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5261034133528138e-05,
+      "loss": 2.2429,
+      "step": 71320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.525143622968393e-05,
+      "loss": 2.2372,
+      "step": 71328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5241846825640485e-05,
+      "loss": 2.2613,
+      "step": 71336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5232265922359516e-05,
+      "loss": 2.2501,
+      "step": 71344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5222693520801947e-05,
+      "loss": 2.2455,
+      "step": 71352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.521312962192786e-05,
+      "loss": 2.2312,
+      "step": 71360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.520357422669643e-05,
+      "loss": 2.2335,
+      "step": 71368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5194027336066028e-05,
+      "loss": 2.2451,
+      "step": 71376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5184488950994163e-05,
+      "loss": 2.207,
+      "step": 71384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.517495907243743e-05,
+      "loss": 2.2572,
+      "step": 71392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5165437701351702e-05,
+      "loss": 2.2407,
+      "step": 71400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.515592483869183e-05,
+      "loss": 2.2363,
+      "step": 71408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5146420485411958e-05,
+      "loss": 2.2416,
+      "step": 71416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.513692464246533e-05,
+      "loss": 2.2361,
+      "step": 71424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5127437310804263e-05,
+      "loss": 2.253,
+      "step": 71432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5117958491380337e-05,
+      "loss": 2.259,
+      "step": 71440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5108488185144203e-05,
+      "loss": 2.2423,
+      "step": 71448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5099026393045682e-05,
+      "loss": 2.2262,
+      "step": 71456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.508957311603376e-05,
+      "loss": 2.2344,
+      "step": 71464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5080128355056483e-05,
+      "loss": 2.2377,
+      "step": 71472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.507069211106118e-05,
+      "loss": 2.247,
+      "step": 71480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.506126438499421e-05,
+      "loss": 2.2473,
+      "step": 71488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5051845177801136e-05,
+      "loss": 2.2285,
+      "step": 71496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5042434490426648e-05,
+      "loss": 2.2405,
+      "step": 71504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5033032323814587e-05,
+      "loss": 2.2332,
+      "step": 71512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5023638678907925e-05,
+      "loss": 2.2419,
+      "step": 71520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5014253556648813e-05,
+      "loss": 2.2503,
+      "step": 71528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5004876957978508e-05,
+      "loss": 2.2474,
+      "step": 71536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.499550888383742e-05,
+      "loss": 2.2355,
+      "step": 71544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4986149335165166e-05,
+      "loss": 2.2421,
+      "step": 71552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4976798312900394e-05,
+      "loss": 2.2343,
+      "step": 71560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.496745581798099e-05,
+      "loss": 2.2387,
+      "step": 71568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4958121851343943e-05,
+      "loss": 2.2423,
+      "step": 71576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4948796413925384e-05,
+      "loss": 2.2533,
+      "step": 71584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.493947950666065e-05,
+      "loss": 2.237,
+      "step": 71592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.493017113048413e-05,
+      "loss": 2.2426,
+      "step": 71600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4920871286329396e-05,
+      "loss": 2.2578,
+      "step": 71608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.491157997512921e-05,
+      "loss": 2.2512,
+      "step": 71616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.49022971978154e-05,
+      "loss": 2.2414,
+      "step": 71624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.489302295531901e-05,
+      "loss": 2.2216,
+      "step": 71632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4883757248570156e-05,
+      "loss": 2.238,
+      "step": 71640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.487450007849815e-05,
+      "loss": 2.252,
+      "step": 71648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4865251446031466e-05,
+      "loss": 2.2622,
+      "step": 71656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4856011352097634e-05,
+      "loss": 2.2203,
+      "step": 71664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4846779797623418e-05,
+      "loss": 2.2486,
+      "step": 71672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4837556783534682e-05,
+      "loss": 2.263,
+      "step": 71680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.482834231075646e-05,
+      "loss": 2.2584,
+      "step": 71688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.481913638021289e-05,
+      "loss": 2.2488,
+      "step": 71696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4809938992827265e-05,
+      "loss": 2.2176,
+      "step": 71704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4800750149522052e-05,
+      "loss": 2.2406,
+      "step": 71712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.479156985121884e-05,
+      "loss": 2.241,
+      "step": 71720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.478239809883834e-05,
+      "loss": 2.2553,
+      "step": 71728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4773234893300457e-05,
+      "loss": 2.2085,
+      "step": 71736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.476408023552418e-05,
+      "loss": 2.2558,
+      "step": 71744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4754934126427693e-05,
+      "loss": 2.2164,
+      "step": 71752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.474579656692829e-05,
+      "loss": 2.2482,
+      "step": 71760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4736667557942408e-05,
+      "loss": 2.2404,
+      "step": 71768
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.472754710038564e-05,
+      "loss": 2.2312,
+      "step": 71776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4718435195172714e-05,
+      "loss": 2.2645,
+      "step": 71784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4709331843217513e-05,
+      "loss": 2.2436,
+      "step": 71792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.470023704543304e-05,
+      "loss": 2.2385,
+      "step": 71800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4691150802731462e-05,
+      "loss": 2.2283,
+      "step": 71808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4682073116024074e-05,
+      "loss": 2.2222,
+      "step": 71816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4673003986221305e-05,
+      "loss": 2.2497,
+      "step": 71824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.466394341423275e-05,
+      "loss": 2.2376,
+      "step": 71832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4654891400967137e-05,
+      "loss": 2.2467,
+      "step": 71840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4645847947332316e-05,
+      "loss": 2.2292,
+      "step": 71848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4636813054235286e-05,
+      "loss": 2.262,
+      "step": 71856
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.462778672258225e-05,
+      "loss": 2.2515,
+      "step": 71864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4618768953278443e-05,
+      "loss": 2.2694,
+      "step": 71872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.460975974722829e-05,
+      "loss": 2.2376,
+      "step": 71880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4600759105335423e-05,
+      "loss": 2.2451,
+      "step": 71888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4591767028502496e-05,
+      "loss": 2.2445,
+      "step": 71896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4582783517631407e-05,
+      "loss": 2.2358,
+      "step": 71904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.457380857362311e-05,
+      "loss": 2.2382,
+      "step": 71912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4564842197377763e-05,
+      "loss": 2.2368,
+      "step": 71920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4555884389794665e-05,
+      "loss": 2.2267,
+      "step": 71928
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4546935151772176e-05,
+      "loss": 2.2466,
+      "step": 71936
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.45379944842079e-05,
+      "loss": 2.2288,
+      "step": 71944
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4529062387998544e-05,
+      "loss": 2.2312,
+      "step": 71952
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4520138864039895e-05,
+      "loss": 2.2384,
+      "step": 71960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.451122391322699e-05,
+      "loss": 2.2516,
+      "step": 71968
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4502317536453888e-05,
+      "loss": 2.2426,
+      "step": 71976
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4493419734613895e-05,
+      "loss": 2.2497,
+      "step": 71984
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.448453050859942e-05,
+      "loss": 2.2436,
+      "step": 71992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4475649859301942e-05,
+      "loss": 2.2314,
+      "step": 72000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4466777787612188e-05,
+      "loss": 2.2255,
+      "step": 72008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.445791429441997e-05,
+      "loss": 2.2361,
+      "step": 72016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4449059380614235e-05,
+      "loss": 2.2421,
+      "step": 72024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4440213047083088e-05,
+      "loss": 2.2444,
+      "step": 72032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.443137529471376e-05,
+      "loss": 2.2237,
+      "step": 72040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4422546124392634e-05,
+      "loss": 2.2472,
+      "step": 72048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4413725537005222e-05,
+      "loss": 2.2488,
+      "step": 72056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4404913533436186e-05,
+      "loss": 2.2232,
+      "step": 72064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4396110114569306e-05,
+      "loss": 2.2529,
+      "step": 72072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4387315281287533e-05,
+      "loss": 2.2511,
+      "step": 72080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4378529034472916e-05,
+      "loss": 2.2327,
+      "step": 72088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.436975137500669e-05,
+      "loss": 2.2384,
+      "step": 72096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4360982303769183e-05,
+      "loss": 2.2277,
+      "step": 72104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4352221821639904e-05,
+      "loss": 2.2349,
+      "step": 72112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4343469929497463e-05,
+      "loss": 2.2447,
+      "step": 72120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.433472662821963e-05,
+      "loss": 2.2593,
+      "step": 72128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4325991918683318e-05,
+      "loss": 2.2425,
+      "step": 72136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4317265801764555e-05,
+      "loss": 2.2288,
+      "step": 72144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4308548278338517e-05,
+      "loss": 2.2612,
+      "step": 72152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4299839349279556e-05,
+      "loss": 2.2252,
+      "step": 72160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.429113901546109e-05,
+      "loss": 2.2297,
+      "step": 72168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.428244727775573e-05,
+      "loss": 2.2373,
+      "step": 72176
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4273764137035205e-05,
+      "loss": 2.226,
+      "step": 72184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.426508959417037e-05,
+      "loss": 2.2365,
+      "step": 72192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4256423650031275e-05,
+      "loss": 2.2273,
+      "step": 72200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4247766305487012e-05,
+      "loss": 2.2509,
+      "step": 72208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4239117561405905e-05,
+      "loss": 2.2645,
+      "step": 72216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4230477418655377e-05,
+      "loss": 2.2291,
+      "step": 72224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4221845878101928e-05,
+      "loss": 2.2258,
+      "step": 72232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.421322294061133e-05,
+      "loss": 2.2296,
+      "step": 72240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4204608607048344e-05,
+      "loss": 2.2507,
+      "step": 72248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.419600287827698e-05,
+      "loss": 2.2416,
+      "step": 72256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4187405755160354e-05,
+      "loss": 2.2535,
+      "step": 72264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.417881723856066e-05,
+      "loss": 2.2291,
+      "step": 72272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4170237329339324e-05,
+      "loss": 2.2499,
+      "step": 72280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.416166602835684e-05,
+      "loss": 2.2451,
+      "step": 72288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.415310333647287e-05,
+      "loss": 2.2463,
+      "step": 72296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4144549254546218e-05,
+      "loss": 2.2284,
+      "step": 72304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4136003783434765e-05,
+      "loss": 2.2449,
+      "step": 72312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.412746692399561e-05,
+      "loss": 2.2328,
+      "step": 72320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4118938677084942e-05,
+      "loss": 2.2147,
+      "step": 72328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.411041904355809e-05,
+      "loss": 2.2534,
+      "step": 72336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4101908024269546e-05,
+      "loss": 2.2433,
+      "step": 72344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4093405620072887e-05,
+      "loss": 2.2412,
+      "step": 72352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4084911831820876e-05,
+      "loss": 2.2381,
+      "step": 72360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4076426660365385e-05,
+      "loss": 2.2363,
+      "step": 72368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4067950106557436e-05,
+      "loss": 2.2731,
+      "step": 72376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4059482171247154e-05,
+      "loss": 2.2359,
+      "step": 72384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4051022855283866e-05,
+      "loss": 2.2324,
+      "step": 72392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4042572159515968e-05,
+      "loss": 2.2496,
+      "step": 72400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.403413008479102e-05,
+      "loss": 2.244,
+      "step": 72408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4025696631955712e-05,
+      "loss": 2.2523,
+      "step": 72416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.401727180185586e-05,
+      "loss": 2.2378,
+      "step": 72424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.400885559533648e-05,
+      "loss": 2.2199,
+      "step": 72432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.40004480132416e-05,
+      "loss": 2.2172,
+      "step": 72440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3992049056414485e-05,
+      "loss": 2.2259,
+      "step": 72448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3983658725697518e-05,
+      "loss": 2.2466,
+      "step": 72456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3975277021932168e-05,
+      "loss": 2.2305,
+      "step": 72464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.396690394595911e-05,
+      "loss": 2.2357,
+      "step": 72472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3958539498618082e-05,
+      "loss": 2.2139,
+      "step": 72480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3950183680747993e-05,
+      "loss": 2.233,
+      "step": 72488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.394183649318692e-05,
+      "loss": 2.2435,
+      "step": 72496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3933497936771994e-05,
+      "loss": 2.2604,
+      "step": 72504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3925168012339557e-05,
+      "loss": 2.2299,
+      "step": 72512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.391684672072505e-05,
+      "loss": 2.2552,
+      "step": 72520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.390853406276302e-05,
+      "loss": 2.2435,
+      "step": 72528
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3900230039287222e-05,
+      "loss": 2.2323,
+      "step": 72536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3891934651130466e-05,
+      "loss": 2.2479,
+      "step": 72544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.388364789912476e-05,
+      "loss": 2.2519,
+      "step": 72552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.387536978410121e-05,
+      "loss": 2.2426,
+      "step": 72560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.386710030689005e-05,
+      "loss": 2.2562,
+      "step": 72568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3858839468320684e-05,
+      "loss": 2.2445,
+      "step": 72576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3850587269221615e-05,
+      "loss": 2.2157,
+      "step": 72584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3842343710420487e-05,
+      "loss": 2.234,
+      "step": 72592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3834108792744093e-05,
+      "loss": 2.2532,
+      "step": 72600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3825882517018343e-05,
+      "loss": 2.2421,
+      "step": 72608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3817664884068293e-05,
+      "loss": 2.2458,
+      "step": 72616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3809455894718114e-05,
+      "loss": 2.2349,
+      "step": 72624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.380125554979113e-05,
+      "loss": 2.2359,
+      "step": 72632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.379306385010978e-05,
+      "loss": 2.2359,
+      "step": 72640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.378488079649565e-05,
+      "loss": 2.2504,
+      "step": 72648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3776706389769457e-05,
+      "loss": 2.2283,
+      "step": 72656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3768540630751054e-05,
+      "loss": 2.2394,
+      "step": 72664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3760383520259394e-05,
+      "loss": 2.2448,
+      "step": 72672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3752235059112615e-05,
+      "loss": 2.2354,
+      "step": 72680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3744095248127953e-05,
+      "loss": 2.2217,
+      "step": 72688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3735964088121787e-05,
+      "loss": 2.2614,
+      "step": 72696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3727841579909617e-05,
+      "loss": 2.2403,
+      "step": 72704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3719727724306103e-05,
+      "loss": 2.2361,
+      "step": 72712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3711622522124992e-05,
+      "loss": 2.246,
+      "step": 72720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3703525974179223e-05,
+      "loss": 2.2425,
+      "step": 72728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.369543808128079e-05,
+      "loss": 2.2479,
+      "step": 72736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3687358844240918e-05,
+      "loss": 2.2534,
+      "step": 72744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3679288263869868e-05,
+      "loss": 2.2172,
+      "step": 72752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.367122634097707e-05,
+      "loss": 2.2632,
+      "step": 72760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3663173076371133e-05,
+      "loss": 2.2315,
+      "step": 72768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3655128470859696e-05,
+      "loss": 2.2397,
+      "step": 72776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3647092525249638e-05,
+      "loss": 2.2361,
+      "step": 72784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3639065240346896e-05,
+      "loss": 2.2346,
+      "step": 72792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3631046616956543e-05,
+      "loss": 2.2343,
+      "step": 72800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3623036655882845e-05,
+      "loss": 2.263,
+      "step": 72808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.361503535792911e-05,
+      "loss": 2.2355,
+      "step": 72816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.360704272389785e-05,
+      "loss": 2.2362,
+      "step": 72824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3599058754590693e-05,
+      "loss": 2.2481,
+      "step": 72832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3591083450808348e-05,
+      "loss": 2.2473,
+      "step": 72840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3583116813350718e-05,
+      "loss": 2.2281,
+      "step": 72848
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.357515884301681e-05,
+      "loss": 2.2411,
+      "step": 72856
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3567209540604757e-05,
+      "loss": 2.2526,
+      "step": 72864
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3559268906911843e-05,
+      "loss": 2.2414,
+      "step": 72872
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3551336942734437e-05,
+      "loss": 2.2606,
+      "step": 72880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.354341364886811e-05,
+      "loss": 2.2315,
+      "step": 72888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3535499026107486e-05,
+      "loss": 2.2453,
+      "step": 72896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.352759307524639e-05,
+      "loss": 2.2432,
+      "step": 72904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3519695797077716e-05,
+      "loss": 2.2373,
+      "step": 72912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3511807192393536e-05,
+      "loss": 2.2215,
+      "step": 72920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3503927261985035e-05,
+      "loss": 2.2476,
+      "step": 72928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3496056006642507e-05,
+      "loss": 2.235,
+      "step": 72936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3488193427155402e-05,
+      "loss": 2.2443,
+      "step": 72944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3480339524312296e-05,
+      "loss": 2.2372,
+      "step": 72952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3472494298900888e-05,
+      "loss": 2.2146,
+      "step": 72960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3464657751708003e-05,
+      "loss": 2.2418,
+      "step": 72968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3456829883519615e-05,
+      "loss": 2.2437,
+      "step": 72976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3449010695120803e-05,
+      "loss": 2.2618,
+      "step": 72984
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.344120018729579e-05,
+      "loss": 2.218,
+      "step": 72992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3433398360827954e-05,
+      "loss": 2.2361,
+      "step": 73000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3425605216499736e-05,
+      "loss": 2.2446,
+      "step": 73008
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3417820755092755e-05,
+      "loss": 2.2199,
+      "step": 73016
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3410044977387745e-05,
+      "loss": 2.2439,
+      "step": 73024
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3402277884164576e-05,
+      "loss": 2.2437,
+      "step": 73032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3394519476202258e-05,
+      "loss": 2.2289,
+      "step": 73040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3386769754278902e-05,
+      "loss": 2.244,
+      "step": 73048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3379028719171746e-05,
+      "loss": 2.2458,
+      "step": 73056
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.337129637165721e-05,
+      "loss": 2.234,
+      "step": 73064
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3363572712510757e-05,
+      "loss": 2.2549,
+      "step": 73072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3355857742507083e-05,
+      "loss": 2.2332,
+      "step": 73080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.334815146241989e-05,
+      "loss": 2.2226,
+      "step": 73088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3340453873022126e-05,
+      "loss": 2.2375,
+      "step": 73096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3332764975085812e-05,
+      "loss": 2.2267,
+      "step": 73104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3325084769382067e-05,
+      "loss": 2.2389,
+      "step": 73112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3317413256681205e-05,
+      "loss": 2.2213,
+      "step": 73120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3309750437752616e-05,
+      "loss": 2.2467,
+      "step": 73128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3302096313364847e-05,
+      "loss": 2.2128,
+      "step": 73136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3294450884285567e-05,
+      "loss": 2.2412,
+      "step": 73144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.328681415128155e-05,
+      "loss": 2.2498,
+      "step": 73152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3279186115118727e-05,
+      "loss": 2.2464,
+      "step": 73160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.327156677656216e-05,
+      "loss": 2.2272,
+      "step": 73168
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.326395613637602e-05,
+      "loss": 2.2558,
+      "step": 73176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.325635419532359e-05,
+      "loss": 2.2422,
+      "step": 73184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3248760954167325e-05,
+      "loss": 2.2183,
+      "step": 73192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3241176413668785e-05,
+      "loss": 2.2405,
+      "step": 73200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.323360057458864e-05,
+      "loss": 2.2438,
+      "step": 73208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3226033437686713e-05,
+      "loss": 2.2333,
+      "step": 73216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3218475003721937e-05,
+      "loss": 2.2523,
+      "step": 73224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3210925273452393e-05,
+      "loss": 2.2477,
+      "step": 73232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3203384247635274e-05,
+      "loss": 2.2445,
+      "step": 73240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3195851927026893e-05,
+      "loss": 2.2506,
+      "step": 73248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3188328312382704e-05,
+      "loss": 2.2442,
+      "step": 73256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.318081340445728e-05,
+      "loss": 2.2323,
+      "step": 73264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.317330720400434e-05,
+      "loss": 2.2358,
+      "step": 73272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3165809711776692e-05,
+      "loss": 2.2431,
+      "step": 73280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.315832092852629e-05,
+      "loss": 2.2532,
+      "step": 73288
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.315084085500425e-05,
+      "loss": 2.2188,
+      "step": 73296
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.314336949196074e-05,
+      "loss": 2.225,
+      "step": 73304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3135906840145122e-05,
+      "loss": 2.2402,
+      "step": 73312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.312845290030585e-05,
+      "loss": 2.2209,
+      "step": 73320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3121007673190487e-05,
+      "loss": 2.241,
+      "step": 73328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3113571159545804e-05,
+      "loss": 2.2372,
+      "step": 73336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3106143360117582e-05,
+      "loss": 2.2319,
+      "step": 73344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.309872427565083e-05,
+      "loss": 2.2489,
+      "step": 73352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.309131390688963e-05,
+      "loss": 2.2549,
+      "step": 73360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3083912254577166e-05,
+      "loss": 2.2468,
+      "step": 73368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.307651931945584e-05,
+      "loss": 2.2375,
+      "step": 73376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3069135102267067e-05,
+      "loss": 2.2609,
+      "step": 73384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.306175960375147e-05,
+      "loss": 2.251,
+      "step": 73392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3054392824648786e-05,
+      "loss": 2.2297,
+      "step": 73400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3047034765697812e-05,
+      "loss": 2.2424,
+      "step": 73408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3039685427636555e-05,
+      "loss": 2.2385,
+      "step": 73416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3032344811202112e-05,
+      "loss": 2.2333,
+      "step": 73424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.30250129171307e-05,
+      "loss": 2.2453,
+      "step": 73432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.301768974615766e-05,
+      "loss": 2.2435,
+      "step": 73440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.301037529901748e-05,
+      "loss": 2.2157,
+      "step": 73448
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3003069576443737e-05,
+      "loss": 2.2493,
+      "step": 73456
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2995772579169167e-05,
+      "loss": 2.2574,
+      "step": 73464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2988484307925616e-05,
+      "loss": 2.2287,
+      "step": 73472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2981204763444047e-05,
+      "loss": 2.2468,
+      "step": 73480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.297393394645457e-05,
+      "loss": 2.2125,
+      "step": 73488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2966671857686405e-05,
+      "loss": 2.2367,
+      "step": 73496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2959418497867882e-05,
+      "loss": 2.2361,
+      "step": 73504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2952173867726504e-05,
+      "loss": 2.2535,
+      "step": 73512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2944937967988834e-05,
+      "loss": 2.2436,
+      "step": 73520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.29377107993806e-05,
+      "loss": 2.2437,
+      "step": 73528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.293049236262666e-05,
+      "loss": 2.2429,
+      "step": 73536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2923282658450974e-05,
+      "loss": 2.2349,
+      "step": 73544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2916081687576624e-05,
+      "loss": 2.2485,
+      "step": 73552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2908889450725828e-05,
+      "loss": 2.2416,
+      "step": 73560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2901705948619965e-05,
+      "loss": 2.2445,
+      "step": 73568
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.289453118197945e-05,
+      "loss": 2.2475,
+      "step": 73576
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.28873651515239e-05,
+      "loss": 2.2672,
+      "step": 73584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.288020785797201e-05,
+      "loss": 2.2272,
+      "step": 73592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2873059302041616e-05,
+      "loss": 2.2342,
+      "step": 73600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2865919484449696e-05,
+      "loss": 2.2638,
+      "step": 73608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2858788405912316e-05,
+      "loss": 2.2553,
+      "step": 73616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2851666067144695e-05,
+      "loss": 2.2227,
+      "step": 73624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.284455246886117e-05,
+      "loss": 2.2435,
+      "step": 73632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2837447611775158e-05,
+      "loss": 2.2444,
+      "step": 73640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2830351496599283e-05,
+      "loss": 2.2393,
+      "step": 73648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2823264124045197e-05,
+      "loss": 2.2453,
+      "step": 73656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2816185494823765e-05,
+      "loss": 2.2238,
+      "step": 73664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2809115609644926e-05,
+      "loss": 2.2321,
+      "step": 73672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.280205446921772e-05,
+      "loss": 2.2292,
+      "step": 73680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.279500207425037e-05,
+      "loss": 2.2216,
+      "step": 73688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.278795842545019e-05,
+      "loss": 2.2499,
+      "step": 73696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2780923523523594e-05,
+      "loss": 2.2472,
+      "step": 73704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.277389736917618e-05,
+      "loss": 2.2353,
+      "step": 73712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2766879963112585e-05,
+      "loss": 2.2495,
+      "step": 73720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2759871306036642e-05,
+      "loss": 2.2578,
+      "step": 73728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.275287139865128e-05,
+      "loss": 2.2326,
+      "step": 73736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.274588024165855e-05,
+      "loss": 2.2488,
+      "step": 73744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2738897835759632e-05,
+      "loss": 2.2316,
+      "step": 73752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2731924181654802e-05,
+      "loss": 2.2325,
+      "step": 73760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2724959280043493e-05,
+      "loss": 2.2398,
+      "step": 73768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2718003131624244e-05,
+      "loss": 2.2366,
+      "step": 73776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2711055737094705e-05,
+      "loss": 2.2219,
+      "step": 73784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.270411709715167e-05,
+      "loss": 2.2491,
+      "step": 73792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2697187212491044e-05,
+      "loss": 2.2384,
+      "step": 73800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2690266083807858e-05,
+      "loss": 2.2363,
+      "step": 73808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2683353711796265e-05,
+      "loss": 2.2424,
+      "step": 73816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2676450097149528e-05,
+      "loss": 2.2354,
+      "step": 73824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.266955524056003e-05,
+      "loss": 2.2519,
+      "step": 73832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.266266914271932e-05,
+      "loss": 2.253,
+      "step": 73840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2655791804318004e-05,
+      "loss": 2.2375,
+      "step": 73848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2648923226045855e-05,
+      "loss": 2.2244,
+      "step": 73856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.264206340859174e-05,
+      "loss": 2.2306,
+      "step": 73864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2635212352643668e-05,
+      "loss": 2.2539,
+      "step": 73872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2628370058888776e-05,
+      "loss": 2.2418,
+      "step": 73880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2621536528013284e-05,
+      "loss": 2.2527,
+      "step": 73888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2614711760702555e-05,
+      "loss": 2.2277,
+      "step": 73896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2607895757641095e-05,
+      "loss": 2.2313,
+      "step": 73904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2601088519512496e-05,
+      "loss": 2.2304,
+      "step": 73912
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2594290046999498e-05,
+      "loss": 2.2391,
+      "step": 73920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.258750034078393e-05,
+      "loss": 2.244,
+      "step": 73928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2580719401546764e-05,
+      "loss": 2.2447,
+      "step": 73936
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2573947229968116e-05,
+      "loss": 2.2323,
+      "step": 73944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2567183826727155e-05,
+      "loss": 2.2453,
+      "step": 73952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2560429192502245e-05,
+      "loss": 2.2421,
+      "step": 73960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.255368332797083e-05,
+      "loss": 2.2383,
+      "step": 73968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.254694623380948e-05,
+      "loss": 2.2425,
+      "step": 73976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2540217910693895e-05,
+      "loss": 2.2526,
+      "step": 73984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.253349835929886e-05,
+      "loss": 2.2678,
+      "step": 73992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2526787580298332e-05,
+      "loss": 2.2509,
+      "step": 74000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2520085574365368e-05,
+      "loss": 2.2405,
+      "step": 74008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.251339234217213e-05,
+      "loss": 2.214,
+      "step": 74016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2506707884389907e-05,
+      "loss": 2.2212,
+      "step": 74024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2500032201689118e-05,
+      "loss": 2.2477,
+      "step": 74032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2493365294739305e-05,
+      "loss": 2.2135,
+      "step": 74040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2486707164209112e-05,
+      "loss": 2.2225,
+      "step": 74048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2480057810766304e-05,
+      "loss": 2.2488,
+      "step": 74056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2473417235077793e-05,
+      "loss": 2.2312,
+      "step": 74064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.246678543780958e-05,
+      "loss": 2.2213,
+      "step": 74072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.246016241962679e-05,
+      "loss": 2.2471,
+      "step": 74080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2453548181193686e-05,
+      "loss": 2.2164,
+      "step": 74088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2446942723173632e-05,
+      "loss": 2.2524,
+      "step": 74096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2440346046229122e-05,
+      "loss": 2.2369,
+      "step": 74104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2433758151021757e-05,
+      "loss": 2.2337,
+      "step": 74112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2427179038212282e-05,
+      "loss": 2.2543,
+      "step": 74120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2420608708460517e-05,
+      "loss": 2.2668,
+      "step": 74128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2414047162425464e-05,
+      "loss": 2.2326,
+      "step": 74136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.240749440076518e-05,
+      "loss": 2.2321,
+      "step": 74144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.24009504241369e-05,
+      "loss": 2.2145,
+      "step": 74152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2394415233196913e-05,
+      "loss": 2.2554,
+      "step": 74160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2387888828600674e-05,
+      "loss": 2.225,
+      "step": 74168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.238137121100276e-05,
+      "loss": 2.2486,
+      "step": 74176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.237486238105682e-05,
+      "loss": 2.2354,
+      "step": 74184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2368362339415687e-05,
+      "loss": 2.2372,
+      "step": 74192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.236187108673127e-05,
+      "loss": 2.2425,
+      "step": 74200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.235538862365458e-05,
+      "loss": 2.2495,
+      "step": 74208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2348914950835802e-05,
+      "loss": 2.2478,
+      "step": 74216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2342450068924187e-05,
+      "loss": 2.2495,
+      "step": 74224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2335993978568147e-05,
+      "loss": 2.253,
+      "step": 74232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.232954668041518e-05,
+      "loss": 2.2379,
+      "step": 74240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.232310817511189e-05,
+      "loss": 2.2293,
+      "step": 74248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2316678463304063e-05,
+      "loss": 2.2253,
+      "step": 74256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2310257545636545e-05,
+      "loss": 2.2323,
+      "step": 74264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2303845422753315e-05,
+      "loss": 2.2411,
+      "step": 74272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2297442095297484e-05,
+      "loss": 2.2255,
+      "step": 74280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2291047563911245e-05,
+      "loss": 2.2447,
+      "step": 74288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2284661829235952e-05,
+      "loss": 2.2515,
+      "step": 74296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.227828489191206e-05,
+      "loss": 2.2521,
+      "step": 74304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.227191675257913e-05,
+      "loss": 2.2384,
+      "step": 74312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.226555741187586e-05,
+      "loss": 2.2222,
+      "step": 74320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2259206870440054e-05,
+      "loss": 2.2283,
+      "step": 74328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.225286512890863e-05,
+      "loss": 2.2227,
+      "step": 74336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2246532187917622e-05,
+      "loss": 2.2353,
+      "step": 74344
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2240208048102204e-05,
+      "loss": 2.2332,
+      "step": 74352
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.223389271009665e-05,
+      "loss": 2.2411,
+      "step": 74360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2227586174534346e-05,
+      "loss": 2.2463,
+      "step": 74368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.222128844204779e-05,
+      "loss": 2.2389,
+      "step": 74376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.221499951326863e-05,
+      "loss": 2.2362,
+      "step": 74384
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2208719388827606e-05,
+      "loss": 2.2432,
+      "step": 74392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.220244806935456e-05,
+      "loss": 2.2397,
+      "step": 74400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2196185555478503e-05,
+      "loss": 2.2443,
+      "step": 74408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2189931847827505e-05,
+      "loss": 2.2294,
+      "step": 74416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2183686947028775e-05,
+      "loss": 2.2347,
+      "step": 74424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2177450853708653e-05,
+      "loss": 2.2404,
+      "step": 74432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2171223568492575e-05,
+      "loss": 2.2496,
+      "step": 74440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2165005092005124e-05,
+      "loss": 2.2322,
+      "step": 74448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2158795424869953e-05,
+      "loss": 2.2258,
+      "step": 74456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.215259456770986e-05,
+      "loss": 2.2242,
+      "step": 74464
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2146402521146776e-05,
+      "loss": 2.2418,
+      "step": 74472
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.214021928580169e-05,
+      "loss": 2.2343,
+      "step": 74480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2134044862294793e-05,
+      "loss": 2.2341,
+      "step": 74488
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2127879251245298e-05,
+      "loss": 2.2423,
+      "step": 74496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2121722453271615e-05,
+      "loss": 2.2222,
+      "step": 74504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2115574468991234e-05,
+      "loss": 2.2346,
+      "step": 74512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2109435299020738e-05,
+      "loss": 2.2403,
+      "step": 74520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.210330494397587e-05,
+      "loss": 2.2356,
+      "step": 74528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2097183404471472e-05,
+      "loss": 2.2358,
+      "step": 74536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2091070681121487e-05,
+      "loss": 2.221,
+      "step": 74544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2084966774539002e-05,
+      "loss": 2.2475,
+      "step": 74552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2078871685336177e-05,
+      "loss": 2.2547,
+      "step": 74560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2072785414124344e-05,
+      "loss": 2.2278,
+      "step": 74568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.206670796151391e-05,
+      "loss": 2.2455,
+      "step": 74576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2060639328114418e-05,
+      "loss": 2.2492,
+      "step": 74584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2054579514534497e-05,
+      "loss": 2.2276,
+      "step": 74592
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.204852852138194e-05,
+      "loss": 2.2211,
+      "step": 74600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.204248634926359e-05,
+      "loss": 2.2473,
+      "step": 74608
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2036452998785483e-05,
+      "loss": 2.2257,
+      "step": 74616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.203042847055271e-05,
+      "loss": 2.2306,
+      "step": 74624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2024412765169485e-05,
+      "loss": 2.2487,
+      "step": 74632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.201840588323917e-05,
+      "loss": 2.2304,
+      "step": 74640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2012407825364224e-05,
+      "loss": 2.2428,
+      "step": 74648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.20064185921462e-05,
+      "loss": 2.2495,
+      "step": 74656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.200043818418579e-05,
+      "loss": 2.2406,
+      "step": 74664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1994466602082787e-05,
+      "loss": 2.2218,
+      "step": 74672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.198850384643614e-05,
+      "loss": 2.2199,
+      "step": 74680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.198254991784384e-05,
+      "loss": 2.2365,
+      "step": 74688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1976604816903057e-05,
+      "loss": 2.2437,
+      "step": 74696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.197066854421004e-05,
+      "loss": 2.2447,
+      "step": 74704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.196474110036015e-05,
+      "loss": 2.232,
+      "step": 74712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.195882248594792e-05,
+      "loss": 2.2239,
+      "step": 74720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1952912701566915e-05,
+      "loss": 2.2467,
+      "step": 74728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1947011747809853e-05,
+      "loss": 2.2503,
+      "step": 74736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.194111962526859e-05,
+      "loss": 2.2403,
+      "step": 74744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.193523633453405e-05,
+      "loss": 2.237,
+      "step": 74752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.192936187619632e-05,
+      "loss": 2.2312,
+      "step": 74760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1923496250844545e-05,
+      "loss": 2.2346,
+      "step": 74768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1917639459067017e-05,
+      "loss": 2.2353,
+      "step": 74776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1911791501451173e-05,
+      "loss": 2.2112,
+      "step": 74784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.190595237858349e-05,
+      "loss": 2.2431,
+      "step": 74792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.190012209104962e-05,
+      "loss": 2.2456,
+      "step": 74800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1894300639434298e-05,
+      "loss": 2.2187,
+      "step": 74808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1888488024321395e-05,
+      "loss": 2.2246,
+      "step": 74816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1882684246293877e-05,
+      "loss": 2.2386,
+      "step": 74824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1876889305933813e-05,
+      "loss": 2.2439,
+      "step": 74832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.187110320382244e-05,
+      "loss": 2.2476,
+      "step": 74840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.186532594054004e-05,
+      "loss": 2.2203,
+      "step": 74848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1859557516666044e-05,
+      "loss": 2.2419,
+      "step": 74856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1853797932779003e-05,
+      "loss": 2.2227,
+      "step": 74864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.184804718945657e-05,
+      "loss": 2.2541,
+      "step": 74872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.18423052872755e-05,
+      "loss": 2.2235,
+      "step": 74880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1836572226811686e-05,
+      "loss": 2.2348,
+      "step": 74888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1830848008640108e-05,
+      "loss": 2.2277,
+      "step": 74896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1825132633334893e-05,
+      "loss": 2.2353,
+      "step": 74904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1819426101469235e-05,
+      "loss": 2.2291,
+      "step": 74912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1813728413615496e-05,
+      "loss": 2.204,
+      "step": 74920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1808039570345097e-05,
+      "loss": 2.2422,
+      "step": 74928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.18023595722286e-05,
+      "loss": 2.2405,
+      "step": 74936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1796688419835697e-05,
+      "loss": 2.2357,
+      "step": 74944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.179102611373514e-05,
+      "loss": 2.2395,
+      "step": 74952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1785372654494857e-05,
+      "loss": 2.2342,
+      "step": 74960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.177972804268184e-05,
+      "loss": 2.2323,
+      "step": 74968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1774092278862223e-05,
+      "loss": 2.231,
+      "step": 74976
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1768465363601216e-05,
+      "loss": 2.2238,
+      "step": 74984
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1762847297463215e-05,
+      "loss": 2.2265,
+      "step": 74992
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1757238081011637e-05,
+      "loss": 2.2464,
+      "step": 75000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1751637714809063e-05,
+      "loss": 2.2364,
+      "step": 75008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.174604619941719e-05,
+      "loss": 2.2308,
+      "step": 75016
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1740463535396795e-05,
+      "loss": 2.2516,
+      "step": 75024
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1734889723307818e-05,
+      "loss": 2.2361,
+      "step": 75032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1729324763709264e-05,
+      "loss": 2.2388,
+      "step": 75040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.172376865715926e-05,
+      "loss": 2.2584,
+      "step": 75048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1718221404215065e-05,
+      "loss": 2.2253,
+      "step": 75056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1712683005433018e-05,
+      "loss": 2.2283,
+      "step": 75064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1707153461368617e-05,
+      "loss": 2.2354,
+      "step": 75072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.170163277257643e-05,
+      "loss": 2.2322,
+      "step": 75080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1696120939610137e-05,
+      "loss": 2.2576,
+      "step": 75088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1690617963022554e-05,
+      "loss": 2.2361,
+      "step": 75096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1685123843365613e-05,
+      "loss": 2.2428,
+      "step": 75104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1679638581190324e-05,
+      "loss": 2.2447,
+      "step": 75112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.167416217704685e-05,
+      "loss": 2.2353,
+      "step": 75120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.166869463148441e-05,
+      "loss": 2.2542,
+      "step": 75128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1663235945051392e-05,
+      "loss": 2.2343,
+      "step": 75136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1657786118295256e-05,
+      "loss": 2.2461,
+      "step": 75144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1652345151762614e-05,
+      "loss": 2.2224,
+      "step": 75152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.164691304599914e-05,
+      "loss": 2.2271,
+      "step": 75160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.164148980154965e-05,
+      "loss": 2.2227,
+      "step": 75168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1636075418958067e-05,
+      "loss": 2.2508,
+      "step": 75176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1630669898767417e-05,
+      "loss": 2.2357,
+      "step": 75184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1625273241519855e-05,
+      "loss": 2.2426,
+      "step": 75192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.161988544775662e-05,
+      "loss": 2.2377,
+      "step": 75200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1614506518018094e-05,
+      "loss": 2.2481,
+      "step": 75208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1609136452843744e-05,
+      "loss": 2.2258,
+      "step": 75216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.160377525277216e-05,
+      "loss": 2.2032,
+      "step": 75224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1598422918341026e-05,
+      "loss": 2.2316,
+      "step": 75232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1593079450087166e-05,
+      "loss": 2.2341,
+      "step": 75240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.158774484854651e-05,
+      "loss": 2.2407,
+      "step": 75248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1582419114254056e-05,
+      "loss": 2.2417,
+      "step": 75256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1577102247743973e-05,
+      "loss": 2.2406,
+      "step": 75264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1571794249549502e-05,
+      "loss": 2.2199,
+      "step": 75272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1566495120202995e-05,
+      "loss": 2.2576,
+      "step": 75280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1561204860235952e-05,
+      "loss": 2.2155,
+      "step": 75288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1555923470178934e-05,
+      "loss": 2.2335,
+      "step": 75296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.155065095056163e-05,
+      "loss": 2.2478,
+      "step": 75304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1545387301912873e-05,
+      "loss": 2.2459,
+      "step": 75312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1540132524760534e-05,
+      "loss": 2.26,
+      "step": 75320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1534886619631683e-05,
+      "loss": 2.2326,
+      "step": 75328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1529649587052415e-05,
+      "loss": 2.2193,
+      "step": 75336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1524421427548002e-05,
+      "loss": 2.2322,
+      "step": 75344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1519202141642805e-05,
+      "loss": 2.253,
+      "step": 75352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1513991729860243e-05,
+      "loss": 2.2394,
+      "step": 75360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1508790192722942e-05,
+      "loss": 2.228,
+      "step": 75368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1503597530752563e-05,
+      "loss": 2.2567,
+      "step": 75376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1498413744469898e-05,
+      "loss": 2.2294,
+      "step": 75384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.149323883439487e-05,
+      "loss": 2.2359,
+      "step": 75392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.148807280104646e-05,
+      "loss": 2.2376,
+      "step": 75400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1482915644942824e-05,
+      "loss": 2.2386,
+      "step": 75408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.147776736660119e-05,
+      "loss": 2.2475,
+      "step": 75416
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1472627966537884e-05,
+      "loss": 2.238,
+      "step": 75424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1467497445268367e-05,
+      "loss": 2.2305,
+      "step": 75432
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.146237580330721e-05,
+      "loss": 2.2387,
+      "step": 75440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1457263041168072e-05,
+      "loss": 2.2314,
+      "step": 75448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1452159159363742e-05,
+      "loss": 2.2354,
+      "step": 75456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1447064158406117e-05,
+      "loss": 2.2385,
+      "step": 75464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1441978038806173e-05,
+      "loss": 2.2464,
+      "step": 75472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.143690080107405e-05,
+      "loss": 2.2307,
+      "step": 75480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.143183244571894e-05,
+      "loss": 2.2457,
+      "step": 75488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.142677297324919e-05,
+      "loss": 2.2368,
+      "step": 75496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.142172238417222e-05,
+      "loss": 2.231,
+      "step": 75504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1416680678994573e-05,
+      "loss": 2.2311,
+      "step": 75512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1411647858221933e-05,
+      "loss": 2.2324,
+      "step": 75520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1406623922359037e-05,
+      "loss": 2.2232,
+      "step": 75528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.140160887190976e-05,
+      "loss": 2.2303,
+      "step": 75536
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1396602707377092e-05,
+      "loss": 2.2486,
+      "step": 75544
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1391605429263105e-05,
+      "loss": 2.2087,
+      "step": 75552
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1386617038069028e-05,
+      "loss": 2.2148,
+      "step": 75560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1381637534295142e-05,
+      "loss": 2.2314,
+      "step": 75568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1376666918440866e-05,
+      "loss": 2.2323,
+      "step": 75576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.137170519100474e-05,
+      "loss": 2.2413,
+      "step": 75584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.136675235248438e-05,
+      "loss": 2.2466,
+      "step": 75592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1361808403376545e-05,
+      "loss": 2.2355,
+      "step": 75600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.135687334417708e-05,
+      "loss": 2.2621,
+      "step": 75608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.135194717538092e-05,
+      "loss": 2.2427,
+      "step": 75616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1347029897482162e-05,
+      "loss": 2.2395,
+      "step": 75624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1342121510973965e-05,
+      "loss": 2.2454,
+      "step": 75632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1337222016348623e-05,
+      "loss": 2.2385,
+      "step": 75640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1332331414097532e-05,
+      "loss": 2.251,
+      "step": 75648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1327449704711154e-05,
+      "loss": 2.2387,
+      "step": 75656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1322576888679147e-05,
+      "loss": 2.2252,
+      "step": 75664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1317712966490194e-05,
+      "loss": 2.254,
+      "step": 75672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.131285793863213e-05,
+      "loss": 2.2415,
+      "step": 75680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.130801180559188e-05,
+      "loss": 2.2133,
+      "step": 75688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1303174567855504e-05,
+      "loss": 2.2506,
+      "step": 75696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1298346225908124e-05,
+      "loss": 2.2317,
+      "step": 75704
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1293526780234015e-05,
+      "loss": 2.2606,
+      "step": 75712
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.128871623131652e-05,
+      "loss": 2.2449,
+      "step": 75720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1283914579638127e-05,
+      "loss": 2.2413,
+      "step": 75728
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.127912182568041e-05,
+      "loss": 2.2343,
+      "step": 75736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1274337969924053e-05,
+      "loss": 2.2433,
+      "step": 75744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1269563012848852e-05,
+      "loss": 2.2403,
+      "step": 75752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1264796954933705e-05,
+      "loss": 2.2141,
+      "step": 75760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.126003979665662e-05,
+      "loss": 2.249,
+      "step": 75768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1255291538494716e-05,
+      "loss": 2.246,
+      "step": 75776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.125055218092422e-05,
+      "loss": 2.2447,
+      "step": 75784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1245821724420452e-05,
+      "loss": 2.2255,
+      "step": 75792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1241100169457858e-05,
+      "loss": 2.2512,
+      "step": 75800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1236387516509975e-05,
+      "loss": 2.2507,
+      "step": 75808
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1231683766049474e-05,
+      "loss": 2.2261,
+      "step": 75816
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.12269889185481e-05,
+      "loss": 2.2401,
+      "step": 75824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1222302974476722e-05,
+      "loss": 2.2361,
+      "step": 75832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1217625934305306e-05,
+      "loss": 2.251,
+      "step": 75840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1212957798502937e-05,
+      "loss": 2.2386,
+      "step": 75848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1208298567537817e-05,
+      "loss": 2.2223,
+      "step": 75856
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1203648241877217e-05,
+      "loss": 2.2267,
+      "step": 75864
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.119900682198757e-05,
+      "loss": 2.238,
+      "step": 75872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.119437430833435e-05,
+      "loss": 2.2564,
+      "step": 75880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1189750701382184e-05,
+      "loss": 2.2322,
+      "step": 75888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1185136001594807e-05,
+      "loss": 2.2453,
+      "step": 75896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1180530209435026e-05,
+      "loss": 2.2442,
+      "step": 75904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1175933325364788e-05,
+      "loss": 2.2405,
+      "step": 75912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.117134534984514e-05,
+      "loss": 2.2277,
+      "step": 75920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1166766283336213e-05,
+      "loss": 2.2376,
+      "step": 75928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1162196126297286e-05,
+      "loss": 2.2292,
+      "step": 75936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1157634879186693e-05,
+      "loss": 2.2351,
+      "step": 75944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.115308254246191e-05,
+      "loss": 2.217,
+      "step": 75952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1148539116579526e-05,
+      "loss": 2.2187,
+      "step": 75960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1144004601995186e-05,
+      "loss": 2.243,
+      "step": 75968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1139478999163723e-05,
+      "loss": 2.2268,
+      "step": 75976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.113496230853899e-05,
+      "loss": 2.2519,
+      "step": 75984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.113045453057401e-05,
+      "loss": 2.234,
+      "step": 75992
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1125955665720877e-05,
+      "loss": 2.2305,
+      "step": 76000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1121465714430792e-05,
+      "loss": 2.2575,
+      "step": 76008
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1116984677154094e-05,
+      "loss": 2.2239,
+      "step": 76016
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.111251255434018e-05,
+      "loss": 2.2308,
+      "step": 76024
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1108049346437613e-05,
+      "loss": 2.269,
+      "step": 76032
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1103595053893995e-05,
+      "loss": 2.2363,
+      "step": 76040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.109914967715608e-05,
+      "loss": 2.232,
+      "step": 76048
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.109471321666971e-05,
+      "loss": 2.2469,
+      "step": 76056
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.109028567287984e-05,
+      "loss": 2.2459,
+      "step": 76064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1085867046230527e-05,
+      "loss": 2.2436,
+      "step": 76072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1081457337164928e-05,
+      "loss": 2.2493,
+      "step": 76080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1077056546125335e-05,
+      "loss": 2.2415,
+      "step": 76088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.107266467355309e-05,
+      "loss": 2.2267,
+      "step": 76096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.106828171988869e-05,
+      "loss": 2.2126,
+      "step": 76104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1063907685571726e-05,
+      "loss": 2.2479,
+      "step": 76112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.105954257104087e-05,
+      "loss": 2.2442,
+      "step": 76120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1055186376733946e-05,
+      "loss": 2.2278,
+      "step": 76128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.105083910308783e-05,
+      "loss": 2.254,
+      "step": 76136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1046500750538533e-05,
+      "loss": 2.2425,
+      "step": 76144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1042171319521184e-05,
+      "loss": 2.2385,
+      "step": 76152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1037850810469978e-05,
+      "loss": 2.258,
+      "step": 76160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1033539223818252e-05,
+      "loss": 2.2431,
+      "step": 76168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.102923655999844e-05,
+      "loss": 2.2272,
+      "step": 76176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1024942819442046e-05,
+      "loss": 2.2313,
+      "step": 76184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1020658002579742e-05,
+      "loss": 2.2429,
+      "step": 76192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1016382109841242e-05,
+      "loss": 2.229,
+      "step": 76200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1012115141655418e-05,
+      "loss": 2.2388,
+      "step": 76208
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1007857098450207e-05,
+      "loss": 2.2401,
+      "step": 76216
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1003607980652682e-05,
+      "loss": 2.2248,
+      "step": 76224
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0999367788688986e-05,
+      "loss": 2.2369,
+      "step": 76232
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0995136522984393e-05,
+      "loss": 2.2496,
+      "step": 76240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0990914183963285e-05,
+      "loss": 2.2263,
+      "step": 76248
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0986700772049125e-05,
+      "loss": 2.2265,
+      "step": 76256
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0982496287664505e-05,
+      "loss": 2.2445,
+      "step": 76264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.09783007312311e-05,
+      "loss": 2.2595,
+      "step": 76272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0974114103169712e-05,
+      "loss": 2.2307,
+      "step": 76280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0969936403900237e-05,
+      "loss": 2.2428,
+      "step": 76288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.096576763384166e-05,
+      "loss": 2.2354,
+      "step": 76296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.096160779341209e-05,
+      "loss": 2.2173,
+      "step": 76304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0957456883028748e-05,
+      "loss": 2.2393,
+      "step": 76312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0953314903107927e-05,
+      "loss": 2.2408,
+      "step": 76320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.094918185406507e-05,
+      "loss": 2.2343,
+      "step": 76328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0945057736314678e-05,
+      "loss": 2.2339,
+      "step": 76336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.094094255027038e-05,
+      "loss": 2.2546,
+      "step": 76344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0936836296344903e-05,
+      "loss": 2.2234,
+      "step": 76352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0932738974950095e-05,
+      "loss": 2.228,
+      "step": 76360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0928650586496873e-05,
+      "loss": 2.2398,
+      "step": 76368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.09245711313953e-05,
+      "loss": 2.2478,
+      "step": 76376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0920500610054515e-05,
+      "loss": 2.2262,
+      "step": 76384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0916439022882756e-05,
+      "loss": 2.2402,
+      "step": 76392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0912386370287397e-05,
+      "loss": 2.215,
+      "step": 76400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.090834265267489e-05,
+      "loss": 2.244,
+      "step": 76408
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.090430787045078e-05,
+      "loss": 2.2251,
+      "step": 76416
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.090028202401976e-05,
+      "loss": 2.2402,
+      "step": 76424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.089626511378557e-05,
+      "loss": 2.25,
+      "step": 76432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0892257140151124e-05,
+      "loss": 2.2311,
+      "step": 76440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0888258103518353e-05,
+      "loss": 2.22,
+      "step": 76448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.088426800428836e-05,
+      "loss": 2.2566,
+      "step": 76456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0880286842861346e-05,
+      "loss": 2.229,
+      "step": 76464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0876314619636568e-05,
+      "loss": 2.2548,
+      "step": 76472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0872351335012433e-05,
+      "loss": 2.2492,
+      "step": 76480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0868396989386437e-05,
+      "loss": 2.2348,
+      "step": 76488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0864451583155166e-05,
+      "loss": 2.2561,
+      "step": 76496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0860515116714342e-05,
+      "loss": 2.2596,
+      "step": 76504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0856587590458753e-05,
+      "loss": 2.2306,
+      "step": 76512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.085266900478232e-05,
+      "loss": 2.238,
+      "step": 76520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.084875936007806e-05,
+      "loss": 2.2385,
+      "step": 76528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0844858656738057e-05,
+      "loss": 2.2453,
+      "step": 76536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0840966895153553e-05,
+      "loss": 2.2062,
+      "step": 76544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.083708407571487e-05,
+      "loss": 2.2469,
+      "step": 76552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0833210198811433e-05,
+      "loss": 2.2279,
+      "step": 76560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0829345264831763e-05,
+      "loss": 2.2523,
+      "step": 76568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0825489274163497e-05,
+      "loss": 2.2186,
+      "step": 76576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.082164222719336e-05,
+      "loss": 2.2257,
+      "step": 76584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08178041243072e-05,
+      "loss": 2.2485,
+      "step": 76592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0813974965889955e-05,
+      "loss": 2.2378,
+      "step": 76600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0810154752325655e-05,
+      "loss": 2.2468,
+      "step": 76608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.080634348399746e-05,
+      "loss": 2.2371,
+      "step": 76616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.080254116128762e-05,
+      "loss": 2.2318,
+      "step": 76624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0798747784577474e-05,
+      "loss": 2.2211,
+      "step": 76632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0794963354247494e-05,
+      "loss": 2.2234,
+      "step": 76640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0791187870677214e-05,
+      "loss": 2.2508,
+      "step": 76648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0787421334245304e-05,
+      "loss": 2.2229,
+      "step": 76656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0783663745329536e-05,
+      "loss": 2.2505,
+      "step": 76664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0779915104306773e-05,
+      "loss": 2.2243,
+      "step": 76672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.077617541155296e-05,
+      "loss": 2.2237,
+      "step": 76680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.077244466744319e-05,
+      "loss": 2.2423,
+      "step": 76688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0768722872351633e-05,
+      "loss": 2.2419,
+      "step": 76696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.076501002665156e-05,
+      "loss": 2.2397,
+      "step": 76704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0761306130715338e-05,
+      "loss": 2.2437,
+      "step": 76712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0757611184914464e-05,
+      "loss": 2.2341,
+      "step": 76720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0753925189619506e-05,
+      "loss": 2.216,
+      "step": 76728
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0750248145200168e-05,
+      "loss": 2.2277,
+      "step": 76736
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0746580052025205e-05,
+      "loss": 2.2487,
+      "step": 76744
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0742920910462544e-05,
+      "loss": 2.2416,
+      "step": 76752
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.073927072087915e-05,
+      "loss": 2.2349,
+      "step": 76760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0735629483641123e-05,
+      "loss": 2.2419,
+      "step": 76768
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.073199719911366e-05,
+      "loss": 2.2411,
+      "step": 76776
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0728373867661055e-05,
+      "loss": 2.2221,
+      "step": 76784
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.072475948964671e-05,
+      "loss": 2.2256,
+      "step": 76792
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.072115406543314e-05,
+      "loss": 2.2426,
+      "step": 76800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.071755759538192e-05,
+      "loss": 2.2584,
+      "step": 76808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0713970079853773e-05,
+      "loss": 2.225,
+      "step": 76816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.07103915192085e-05,
+      "loss": 2.2333,
+      "step": 76824
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0706821913805026e-05,
+      "loss": 2.2245,
+      "step": 76832
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0703261264001354e-05,
+      "loss": 2.2356,
+      "step": 76840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.069970957015458e-05,
+      "loss": 2.2199,
+      "step": 76848
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0696166832620943e-05,
+      "loss": 2.2511,
+      "step": 76856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0692633051755745e-05,
+      "loss": 2.2168,
+      "step": 76864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.068910822791341e-05,
+      "loss": 2.2543,
+      "step": 76872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0685592361447457e-05,
+      "loss": 2.2399,
+      "step": 76880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.06820854527105e-05,
+      "loss": 2.2422,
+      "step": 76888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0678587502054273e-05,
+      "loss": 2.2219,
+      "step": 76896
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.06750985098296e-05,
+      "loss": 2.2463,
+      "step": 76904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0671618476386406e-05,
+      "loss": 2.2289,
+      "step": 76912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.066814740207371e-05,
+      "loss": 2.227,
+      "step": 76920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0664685287239648e-05,
+      "loss": 2.2387,
+      "step": 76928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0661232132231454e-05,
+      "loss": 2.226,
+      "step": 76936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065778793739545e-05,
+      "loss": 2.229,
+      "step": 76944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0654352703077074e-05,
+      "loss": 2.2434,
+      "step": 76952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065092642962086e-05,
+      "loss": 2.2277,
+      "step": 76960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0647509117370455e-05,
+      "loss": 2.2219,
+      "step": 76968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0644100766668583e-05,
+      "loss": 2.2315,
+      "step": 76976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.064070137785708e-05,
+      "loss": 2.2423,
+      "step": 76984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0637310951276906e-05,
+      "loss": 2.2514,
+      "step": 76992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0633929487268075e-05,
+      "loss": 2.2376,
+      "step": 77000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.063055698616974e-05,
+      "loss": 2.2254,
+      "step": 77008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0627193448320145e-05,
+      "loss": 2.2406,
+      "step": 77016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0623838874056637e-05,
+      "loss": 2.24,
+      "step": 77024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.062049326371566e-05,
+      "loss": 2.2303,
+      "step": 77032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.061715661763274e-05,
+      "loss": 2.2417,
+      "step": 77040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0613828936142547e-05,
+      "loss": 2.2405,
+      "step": 77048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0610510219578837e-05,
+      "loss": 2.229,
+      "step": 77056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0607200468274425e-05,
+      "loss": 2.2318,
+      "step": 77064
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0603899682561285e-05,
+      "loss": 2.2188,
+      "step": 77072
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0600607862770453e-05,
+      "loss": 2.2497,
+      "step": 77080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0597325009232105e-05,
+      "loss": 2.2568,
+      "step": 77088
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.059405112227546e-05,
+      "loss": 2.2505,
+      "step": 77096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.059078620222889e-05,
+      "loss": 2.235,
+      "step": 77104
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0587530249419843e-05,
+      "loss": 2.2255,
+      "step": 77112
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0584283264174877e-05,
+      "loss": 2.2332,
+      "step": 77120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0581045246819642e-05,
+      "loss": 2.216,
+      "step": 77128
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.057781619767889e-05,
+      "loss": 2.2365,
+      "step": 77136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0574596117076477e-05,
+      "loss": 2.2283,
+      "step": 77144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.057138500533536e-05,
+      "loss": 2.236,
+      "step": 77152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0568182862777607e-05,
+      "loss": 2.2369,
+      "step": 77160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0564989689724354e-05,
+      "loss": 2.2327,
+      "step": 77168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.056180548649587e-05,
+      "loss": 2.2527,
+      "step": 77176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.055863025341151e-05,
+      "loss": 2.2318,
+      "step": 77184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.055546399078974e-05,
+      "loss": 2.2305,
+      "step": 77192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0552306698948108e-05,
+      "loss": 2.2203,
+      "step": 77200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0549158378203272e-05,
+      "loss": 2.2119,
+      "step": 77208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054601902887099e-05,
+      "loss": 2.2455,
+      "step": 77216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054288865126614e-05,
+      "loss": 2.2329,
+      "step": 77224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0539767245702658e-05,
+      "loss": 2.2334,
+      "step": 77232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0536654812493607e-05,
+      "loss": 2.2265,
+      "step": 77240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.053355135195116e-05,
+      "loss": 2.2399,
+      "step": 77248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.053045686438656e-05,
+      "loss": 2.2381,
+      "step": 77256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.052737135011018e-05,
+      "loss": 2.2565,
+      "step": 77264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0524294809431467e-05,
+      "loss": 2.219,
+      "step": 77272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0521227242658988e-05,
+      "loss": 2.2331,
+      "step": 77280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0518168650100415e-05,
+      "loss": 2.237,
+      "step": 77288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0515119032062475e-05,
+      "loss": 2.2303,
+      "step": 77296
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0512078388851057e-05,
+      "loss": 2.2413,
+      "step": 77304
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.050904672077111e-05,
+      "loss": 2.2469,
+      "step": 77312
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.050602402812669e-05,
+      "loss": 2.2276,
+      "step": 77320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0503010311220966e-05,
+      "loss": 2.2487,
+      "step": 77328
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0500005570356188e-05,
+      "loss": 2.2489,
+      "step": 77336
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0497009805833717e-05,
+      "loss": 2.2434,
+      "step": 77344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0494023017954002e-05,
+      "loss": 2.2272,
+      "step": 77352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.049104520701661e-05,
+      "loss": 2.227,
+      "step": 77360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.048807637332021e-05,
+      "loss": 2.2296,
+      "step": 77368
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.048511651716254e-05,
+      "loss": 2.2368,
+      "step": 77376
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0482165638840468e-05,
+      "loss": 2.2245,
+      "step": 77384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.047922373864994e-05,
+      "loss": 2.2451,
+      "step": 77392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0476290816886016e-05,
+      "loss": 2.2392,
+      "step": 77400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0473366873842868e-05,
+      "loss": 2.24,
+      "step": 77408
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0470451909813724e-05,
+      "loss": 2.2324,
+      "step": 77416
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0467545925090947e-05,
+      "loss": 2.2356,
+      "step": 77424
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0464648919966006e-05,
+      "loss": 2.2299,
+      "step": 77432
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.046176089472944e-05,
+      "loss": 2.2367,
+      "step": 77440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0458881849670902e-05,
+      "loss": 2.2399,
+      "step": 77448
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0456011785079154e-05,
+      "loss": 2.2354,
+      "step": 77456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.045315070124203e-05,
+      "loss": 2.2395,
+      "step": 77464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.04502985984465e-05,
+      "loss": 2.2284,
+      "step": 77472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.04474554769786e-05,
+      "loss": 2.2539,
+      "step": 77480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0444621337123475e-05,
+      "loss": 2.27,
+      "step": 77488
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.044179617916539e-05,
+      "loss": 2.2458,
+      "step": 77496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0438980003387685e-05,
+      "loss": 2.2241,
+      "step": 77504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0436172810072803e-05,
+      "loss": 2.2251,
+      "step": 77512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.043337459950229e-05,
+      "loss": 2.2332,
+      "step": 77520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0430585371956778e-05,
+      "loss": 2.247,
+      "step": 77528
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.042780512771605e-05,
+      "loss": 2.2257,
+      "step": 77536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0425033867058905e-05,
+      "loss": 2.2296,
+      "step": 77544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.042227159026332e-05,
+      "loss": 2.2403,
+      "step": 77552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0419518297606302e-05,
+      "loss": 2.2169,
+      "step": 77560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0416773989364017e-05,
+      "loss": 2.2491,
+      "step": 77568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.04140386658117e-05,
+      "loss": 2.2308,
+      "step": 77576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.041131232722367e-05,
+      "loss": 2.2296,
+      "step": 77584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0408594973873377e-05,
+      "loss": 2.2345,
+      "step": 77592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0405886606033367e-05,
+      "loss": 2.2345,
+      "step": 77600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.040318722397525e-05,
+      "loss": 2.2302,
+      "step": 77608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0400496827969778e-05,
+      "loss": 2.2143,
+      "step": 77616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0397815418286778e-05,
+      "loss": 2.2484,
+      "step": 77624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0395142995195167e-05,
+      "loss": 2.2574,
+      "step": 77632
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0392479558963003e-05,
+      "loss": 2.2431,
+      "step": 77640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0389825109857372e-05,
+      "loss": 2.2536,
+      "step": 77648
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.038717964814454e-05,
+      "loss": 2.2308,
+      "step": 77656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0384543174089806e-05,
+      "loss": 2.2174,
+      "step": 77664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.038191568795761e-05,
+      "loss": 2.2079,
+      "step": 77672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0379297190011457e-05,
+      "loss": 2.2404,
+      "step": 77680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037668768051398e-05,
+      "loss": 2.233,
+      "step": 77688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.03740871597269e-05,
+      "loss": 2.2352,
+      "step": 77696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037149562791103e-05,
+      "loss": 2.2245,
+      "step": 77704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0368913085326274e-05,
+      "loss": 2.2479,
+      "step": 77712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036633953223167e-05,
+      "loss": 2.2419,
+      "step": 77720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0363774968885317e-05,
+      "loss": 2.2211,
+      "step": 77728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036121939554442e-05,
+      "loss": 2.2511,
+      "step": 77736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.03586728124653e-05,
+      "loss": 2.2308,
+      "step": 77744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0356135219903362e-05,
+      "loss": 2.2328,
+      "step": 77752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.035360661811312e-05,
+      "loss": 2.2325,
+      "step": 77760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.035108700734816e-05,
+      "loss": 2.2477,
+      "step": 77768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0348576387861192e-05,
+      "loss": 2.2618,
+      "step": 77776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034607475990403e-05,
+      "loss": 2.2364,
+      "step": 77784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0343582123727553e-05,
+      "loss": 2.2406,
+      "step": 77792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0341098479581766e-05,
+      "loss": 2.23,
+      "step": 77800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0338623827715773e-05,
+      "loss": 2.2564,
+      "step": 77808
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.033615816837776e-05,
+      "loss": 2.2391,
+      "step": 77816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.033370150181501e-05,
+      "loss": 2.2595,
+      "step": 77824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0331253828273933e-05,
+      "loss": 2.2398,
+      "step": 77832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0328815148000002e-05,
+      "loss": 2.2326,
+      "step": 77840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0326385461237807e-05,
+      "loss": 2.2433,
+      "step": 77848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0323964768231035e-05,
+      "loss": 2.2549,
+      "step": 77856
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0321553069222457e-05,
+      "loss": 2.2342,
+      "step": 77864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.031915036445397e-05,
+      "loss": 2.2278,
+      "step": 77872
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0316756654166528e-05,
+      "loss": 2.2382,
+      "step": 77880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0314371938600228e-05,
+      "loss": 2.2377,
+      "step": 77888
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.031199621799424e-05,
+      "loss": 2.2243,
+      "step": 77896
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0309629492586822e-05,
+      "loss": 2.235,
+      "step": 77904
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0307271762615365e-05,
+      "loss": 2.2197,
+      "step": 77912
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0304923028316314e-05,
+      "loss": 2.2204,
+      "step": 77920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0302583289925244e-05,
+      "loss": 2.2251,
+      "step": 77928
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.030025254767682e-05,
+      "loss": 2.2559,
+      "step": 77936
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0297930801804793e-05,
+      "loss": 2.2527,
+      "step": 77944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.029561805254203e-05,
+      "loss": 2.2118,
+      "step": 77952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0293314300120483e-05,
+      "loss": 2.2519,
+      "step": 77960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.02910195447712e-05,
+      "loss": 2.2457,
+      "step": 77968
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.028873378672434e-05,
+      "loss": 2.2243,
+      "step": 77976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0286457026209153e-05,
+      "loss": 2.2467,
+      "step": 77984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0284189263453984e-05,
+      "loss": 2.2174,
+      "step": 77992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.028193049868628e-05,
+      "loss": 2.2481,
+      "step": 78000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0279680732132556e-05,
+      "loss": 2.251,
+      "step": 78008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0277439964018484e-05,
+      "loss": 2.2378,
+      "step": 78016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.027520819456879e-05,
+      "loss": 2.2365,
+      "step": 78024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0272985424007298e-05,
+      "loss": 2.2334,
+      "step": 78032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0270771652556944e-05,
+      "loss": 2.221,
+      "step": 78040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.026856688043977e-05,
+      "loss": 2.2326,
+      "step": 78048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0266371107876887e-05,
+      "loss": 2.2215,
+      "step": 78056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0264184335088525e-05,
+      "loss": 2.2252,
+      "step": 78064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0262006562293996e-05,
+      "loss": 2.2394,
+      "step": 78072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0259837789711737e-05,
+      "loss": 2.2565,
+      "step": 78080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0257678017559244e-05,
+      "loss": 2.239,
+      "step": 78088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.025552724605313e-05,
+      "loss": 2.2255,
+      "step": 78096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0253385475409124e-05,
+      "loss": 2.237,
+      "step": 78104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.025125270584202e-05,
+      "loss": 2.2484,
+      "step": 78112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0249128937565728e-05,
+      "loss": 2.2143,
+      "step": 78120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.024701417079324e-05,
+      "loss": 2.2338,
+      "step": 78128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0244908405736664e-05,
+      "loss": 2.2482,
+      "step": 78136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0242811642607202e-05,
+      "loss": 2.2371,
+      "step": 78144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0240723881615138e-05,
+      "loss": 2.2308,
+      "step": 78152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0238645122969857e-05,
+      "loss": 2.2317,
+      "step": 78160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.023657536687986e-05,
+      "loss": 2.235,
+      "step": 78168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0234514613552726e-05,
+      "loss": 2.2461,
+      "step": 78176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0232462863195135e-05,
+      "loss": 2.2293,
+      "step": 78184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0230420116012868e-05,
+      "loss": 2.2316,
+      "step": 78192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0228386372210803e-05,
+      "loss": 2.2238,
+      "step": 78200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0226361631992915e-05,
+      "loss": 2.2248,
+      "step": 78208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0224345895562262e-05,
+      "loss": 2.238,
+      "step": 78216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0222339163121017e-05,
+      "loss": 2.2434,
+      "step": 78224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.022034143487045e-05,
+      "loss": 2.2432,
+      "step": 78232
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021835271101092e-05,
+      "loss": 2.2306,
+      "step": 78240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021637299174188e-05,
+      "loss": 2.2382,
+      "step": 78248
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0214402277261886e-05,
+      "loss": 2.2451,
+      "step": 78256
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021244056776859e-05,
+      "loss": 2.2471,
+      "step": 78264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0210487863458742e-05,
+      "loss": 2.223,
+      "step": 78272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0208544164528183e-05,
+      "loss": 2.2515,
+      "step": 78280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0206609471171864e-05,
+      "loss": 2.2605,
+      "step": 78288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0204683783583813e-05,
+      "loss": 2.2305,
+      "step": 78296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0202767101957172e-05,
+      "loss": 2.2419,
+      "step": 78304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020085942648417e-05,
+      "loss": 2.2399,
+      "step": 78312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019896075735614e-05,
+      "loss": 2.2321,
+      "step": 78320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01970710947635e-05,
+      "loss": 2.2241,
+      "step": 78328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0195190438895785e-05,
+      "loss": 2.2436,
+      "step": 78336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01933187899416e-05,
+      "loss": 2.27,
+      "step": 78344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0191456148088675e-05,
+      "loss": 2.2251,
+      "step": 78352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0189602513523814e-05,
+      "loss": 2.2259,
+      "step": 78360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018775788643292e-05,
+      "loss": 2.2393,
+      "step": 78368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0185922267001014e-05,
+      "loss": 2.2333,
+      "step": 78376
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018409565541218e-05,
+      "loss": 2.2339,
+      "step": 78384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018227805184963e-05,
+      "loss": 2.2417,
+      "step": 78392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018046945649566e-05,
+      "loss": 2.224,
+      "step": 78400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0178669869531652e-05,
+      "loss": 2.2355,
+      "step": 78408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0176879291138118e-05,
+      "loss": 2.2329,
+      "step": 78416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0175097721494605e-05,
+      "loss": 2.2112,
+      "step": 78424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0173325160779817e-05,
+      "loss": 2.2422,
+      "step": 78432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0171561609171542e-05,
+      "loss": 2.2275,
+      "step": 78440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016980706684663e-05,
+      "loss": 2.2327,
+      "step": 78448
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0168061533981074e-05,
+      "loss": 2.2377,
+      "step": 78456
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0166325010749924e-05,
+      "loss": 2.2249,
+      "step": 78464
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0164597497327343e-05,
+      "loss": 2.2303,
+      "step": 78472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016287899388661e-05,
+      "loss": 2.2343,
+      "step": 78480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016116950060005e-05,
+      "loss": 2.2415,
+      "step": 78488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015946901763915e-05,
+      "loss": 2.2523,
+      "step": 78496
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0157777545174446e-05,
+      "loss": 2.2186,
+      "step": 78504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015609508337557e-05,
+      "loss": 2.2368,
+      "step": 78512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0154421632411277e-05,
+      "loss": 2.2377,
+      "step": 78520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01527571924494e-05,
+      "loss": 2.2248,
+      "step": 78528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0151101763656882e-05,
+      "loss": 2.2348,
+      "step": 78536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0149455346199738e-05,
+      "loss": 2.2116,
+      "step": 78544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0147817940243103e-05,
+      "loss": 2.2268,
+      "step": 78552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0146189545951193e-05,
+      "loss": 2.2386,
+      "step": 78560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014457016348734e-05,
+      "loss": 2.2375,
+      "step": 78568
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0142959793013947e-05,
+      "loss": 2.2757,
+      "step": 78576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0141358434692523e-05,
+      "loss": 2.2283,
+      "step": 78584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0139766088683692e-05,
+      "loss": 2.2348,
+      "step": 78592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0138182755147142e-05,
+      "loss": 2.2494,
+      "step": 78600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013660843424167e-05,
+      "loss": 2.2272,
+      "step": 78608
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013504312612519e-05,
+      "loss": 2.2405,
+      "step": 78616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0133486830954674e-05,
+      "loss": 2.2454,
+      "step": 78624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0131939548886213e-05,
+      "loss": 2.2536,
+      "step": 78632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0130401280075004e-05,
+      "loss": 2.2421,
+      "step": 78640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.012887202467531e-05,
+      "loss": 2.2443,
+      "step": 78648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.012735178284051e-05,
+      "loss": 2.2466,
+      "step": 78656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0125840554723096e-05,
+      "loss": 2.237,
+      "step": 78664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0124338340474614e-05,
+      "loss": 2.2343,
+      "step": 78672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0122845140245718e-05,
+      "loss": 2.2003,
+      "step": 78680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0121360954186197e-05,
+      "loss": 2.2468,
+      "step": 78688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0119885782444885e-05,
+      "loss": 2.2313,
+      "step": 78696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.011841962516975e-05,
+      "loss": 2.2486,
+      "step": 78704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.011696248250782e-05,
+      "loss": 2.2223,
+      "step": 78712
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0115514354605253e-05,
+      "loss": 2.2464,
+      "step": 78720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0114075241607274e-05,
+      "loss": 2.2285,
+      "step": 78728
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0112645143658232e-05,
+      "loss": 2.2374,
+      "step": 78736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.011122406090155e-05,
+      "loss": 2.2452,
+      "step": 78744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0109811993479772e-05,
+      "loss": 2.233,
+      "step": 78752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0108408941534486e-05,
+      "loss": 2.2334,
+      "step": 78760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0107014905206432e-05,
+      "loss": 2.2396,
+      "step": 78768
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.010562988463543e-05,
+      "loss": 2.2313,
+      "step": 78776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.010425387996038e-05,
+      "loss": 2.2481,
+      "step": 78784
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0102886891319288e-05,
+      "loss": 2.2329,
+      "step": 78792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0101528918849243e-05,
+      "loss": 2.2414,
+      "step": 78800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010017996268647e-05,
+      "loss": 2.2341,
+      "step": 78808
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0098840022966242e-05,
+      "loss": 2.2519,
+      "step": 78816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0097509099822948e-05,
+      "loss": 2.2226,
+      "step": 78824
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009618719339008e-05,
+      "loss": 2.2493,
+      "step": 78832
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0094874303800207e-05,
+      "loss": 2.2458,
+      "step": 78840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009357043118502e-05,
+      "loss": 2.2471,
+      "step": 78848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0092275575675272e-05,
+      "loss": 2.2533,
+      "step": 78856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009098973740084e-05,
+      "loss": 2.2366,
+      "step": 78864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0089712916490687e-05,
+      "loss": 2.2184,
+      "step": 78872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0088445113072866e-05,
+      "loss": 2.2289,
+      "step": 78880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0087186327274542e-05,
+      "loss": 2.231,
+      "step": 78888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0085936559221945e-05,
+      "loss": 2.239,
+      "step": 78896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0084695809040437e-05,
+      "loss": 2.2492,
+      "step": 78904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0083464076854454e-05,
+      "loss": 2.2466,
+      "step": 78912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0082241362787516e-05,
+      "loss": 2.2354,
+      "step": 78920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0081027666962274e-05,
+      "loss": 2.2462,
+      "step": 78928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007982298950045e-05,
+      "loss": 2.2307,
+      "step": 78936
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0078627330522863e-05,
+      "loss": 2.2384,
+      "step": 78944
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007744069014944e-05,
+      "loss": 2.2423,
+      "step": 78952
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0076263068499173e-05,
+      "loss": 2.2155,
+      "step": 78960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0075094465690194e-05,
+      "loss": 2.2197,
+      "step": 78968
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0073934881839693e-05,
+      "loss": 2.2277,
+      "step": 78976
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007278431706397e-05,
+      "loss": 2.2463,
+      "step": 78984
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007164277147843e-05,
+      "loss": 2.2526,
+      "step": 78992
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0070510245197554e-05,
+      "loss": 2.2248,
+      "step": 79000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006938673833494e-05,
+      "loss": 2.229,
+      "step": 79008
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006827225100325e-05,
+      "loss": 2.2295,
+      "step": 79016
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006716678331428e-05,
+      "loss": 2.2329,
+      "step": 79024
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0066070335378896e-05,
+      "loss": 2.2316,
+      "step": 79032
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0064982907307055e-05,
+      "loss": 2.2535,
+      "step": 79040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0063904499207836e-05,
+      "loss": 2.2423,
+      "step": 79048
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006283511118939e-05,
+      "loss": 2.2355,
+      "step": 79056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0061774743358958e-05,
+      "loss": 2.2312,
+      "step": 79064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006072339582292e-05,
+      "loss": 2.2442,
+      "step": 79072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0059681068686686e-05,
+      "loss": 2.2277,
+      "step": 79080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0058647762054817e-05,
+      "loss": 2.2309,
+      "step": 79088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005762347603094e-05,
+      "loss": 2.2382,
+      "step": 79096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0056608210717788e-05,
+      "loss": 2.2268,
+      "step": 79104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0055601966217186e-05,
+      "loss": 2.2441,
+      "step": 79112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005460474263005e-05,
+      "loss": 2.2518,
+      "step": 79120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0053616540056407e-05,
+      "loss": 2.2292,
+      "step": 79128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0052637358595355e-05,
+      "loss": 2.2464,
+      "step": 79136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0051667198345107e-05,
+      "loss": 2.2307,
+      "step": 79144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0050706059402963e-05,
+      "loss": 2.2525,
+      "step": 79152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0049753941865323e-05,
+      "loss": 2.2297,
+      "step": 79160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0048810845827668e-05,
+      "loss": 2.238,
+      "step": 79168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0047876771384603e-05,
+      "loss": 2.2423,
+      "step": 79176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00469517186298e-05,
+      "loss": 2.2368,
+      "step": 79184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0046035687656034e-05,
+      "loss": 2.2312,
+      "step": 79192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004512867855519e-05,
+      "loss": 2.2288,
+      "step": 79200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0044230691418218e-05,
+      "loss": 2.2322,
+      "step": 79208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0043341726335195e-05,
+      "loss": 2.2314,
+      "step": 79216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004246178339527e-05,
+      "loss": 2.2284,
+      "step": 79224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00415908626867e-05,
+      "loss": 2.2363,
+      "step": 79232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004072896429684e-05,
+      "loss": 2.2534,
+      "step": 79240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0039876088312115e-05,
+      "loss": 2.2355,
+      "step": 79248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0039032234818092e-05,
+      "loss": 2.236,
+      "step": 79256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003819740389938e-05,
+      "loss": 2.2601,
+      "step": 79264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0037371595639714e-05,
+      "loss": 2.2406,
+      "step": 79272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0036554810121933e-05,
+      "loss": 2.214,
+      "step": 79280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003574704742793e-05,
+      "loss": 2.2247,
+      "step": 79288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003494830763874e-05,
+      "loss": 2.2331,
+      "step": 79296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003415859083446e-05,
+      "loss": 2.2412,
+      "step": 79304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0033377897094306e-05,
+      "loss": 2.22,
+      "step": 79312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0032606226496564e-05,
+      "loss": 2.231,
+      "step": 79320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0031843579118633e-05,
+      "loss": 2.2411,
+      "step": 79328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0031089955037012e-05,
+      "loss": 2.214,
+      "step": 79336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0030345354327265e-05,
+      "loss": 2.2529,
+      "step": 79344
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0029609777064096e-05,
+      "loss": 2.2182,
+      "step": 79352
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0028883223321254e-05,
+      "loss": 2.2367,
+      "step": 79360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0028165693171628e-05,
+      "loss": 2.2448,
+      "step": 79368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002745718668717e-05,
+      "loss": 2.2383,
+      "step": 79376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0026757703938943e-05,
+      "loss": 2.2237,
+      "step": 79384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0026067244997105e-05,
+      "loss": 2.2053,
+      "step": 79392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00253858099309e-05,
+      "loss": 2.2349,
+      "step": 79400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0024713398808675e-05,
+      "loss": 2.2406,
+      "step": 79408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0024050011697874e-05,
+      "loss": 2.2394,
+      "step": 79416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0023395648665012e-05,
+      "loss": 2.2366,
+      "step": 79424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0022750309775738e-05,
+      "loss": 2.2293,
+      "step": 79432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0022113995094772e-05,
+      "loss": 2.2284,
+      "step": 79440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0021486704685923e-05,
+      "loss": 2.2304,
+      "step": 79448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0020868438612118e-05,
+      "loss": 2.2389,
+      "step": 79456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0020259196935346e-05,
+      "loss": 2.2463,
+      "step": 79464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001965897971674e-05,
+      "loss": 2.238,
+      "step": 79472
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0019067787016464e-05,
+      "loss": 2.2463,
+      "step": 79480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001848561889385e-05,
+      "loss": 2.2288,
+      "step": 79488
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0017912475407247e-05,
+      "loss": 2.221,
+      "step": 79496
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001734835661417e-05,
+      "loss": 2.2409,
+      "step": 79504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001679326257117e-05,
+      "loss": 2.2415,
+      "step": 79512
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001624719333394e-05,
+      "loss": 2.246,
+      "step": 79520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001571014895724e-05,
+      "loss": 2.2236,
+      "step": 79528
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0015182129494924e-05,
+      "loss": 2.242,
+      "step": 79536
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0014663134999974e-05,
+      "loss": 2.2388,
+      "step": 79544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0014153165524418e-05,
+      "loss": 2.2419,
+      "step": 79552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001365222111941e-05,
+      "loss": 2.2431,
+      "step": 79560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0013160301835194e-05,
+      "loss": 2.2318,
+      "step": 79568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0012677407721108e-05,
+      "loss": 2.2461,
+      "step": 79576
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0012203538825586e-05,
+      "loss": 2.2449,
+      "step": 79584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0011738695196146e-05,
+      "loss": 2.2485,
+      "step": 79592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0011282876879414e-05,
+      "loss": 2.2389,
+      "step": 79600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0010836083921104e-05,
+      "loss": 2.244,
+      "step": 79608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001039831636603e-05,
+      "loss": 2.2274,
+      "step": 79616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000996957425809e-05,
+      "loss": 2.2348,
+      "step": 79624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.00095498576403e-05,
+      "loss": 2.2384,
+      "step": 79632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000913916655473e-05,
+      "loss": 2.2064,
+      "step": 79640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0008737501042602e-05,
+      "loss": 2.2263,
+      "step": 79648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0008344861144177e-05,
+      "loss": 2.2422,
+      "step": 79656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007961246898837e-05,
+      "loss": 2.2516,
+      "step": 79664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007586658345072e-05,
+      "loss": 2.2487,
+      "step": 79672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000722109552043e-05,
+      "loss": 2.2108,
+      "step": 79680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0006864558461583e-05,
+      "loss": 2.2362,
+      "step": 79688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0006517047204305e-05,
+      "loss": 2.242,
+      "step": 79696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000617856178342e-05,
+      "loss": 2.2201,
+      "step": 79704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005849102232897e-05,
+      "loss": 2.236,
+      "step": 79712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005528668585782e-05,
+      "loss": 2.2307,
+      "step": 79720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005217260874207e-05,
+      "loss": 2.2414,
+      "step": 79728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0004914879129394e-05,
+      "loss": 2.2384,
+      "step": 79736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0004621523381677e-05,
+      "loss": 2.2263,
+      "step": 79744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0004337193660482e-05,
+      "loss": 2.2403,
+      "step": 79752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000406188999432e-05,
+      "loss": 2.2434,
+      "step": 79760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003795612410807e-05,
+      "loss": 2.2473,
+      "step": 79768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003538360936653e-05,
+      "loss": 2.2207,
+      "step": 79776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000329013559765e-05,
+      "loss": 2.2414,
+      "step": 79784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003050936418692e-05,
+      "loss": 2.2256,
+      "step": 79792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000282076342378e-05,
+      "loss": 2.2544,
+      "step": 79800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000259961663599e-05,
+      "loss": 2.2385,
+      "step": 79808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002387496077508e-05,
+      "loss": 2.2498,
+      "step": 79816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.00021844017696e-05,
+      "loss": 2.216,
+      "step": 79824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001990333732647e-05,
+      "loss": 2.2405,
+      "step": 79832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001805291986103e-05,
+      "loss": 2.2351,
+      "step": 79840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000162927654854e-05,
+      "loss": 2.2393,
+      "step": 79848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000146228743759e-05,
+      "loss": 2.2303,
+      "step": 79856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000130432467001e-05,
+      "loss": 2.2186,
+      "step": 79864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001155388261656e-05,
+      "loss": 2.2078,
+      "step": 79872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001015478227448e-05,
+      "loss": 2.244,
+      "step": 79880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000884594581425e-05,
+      "loss": 2.2441,
+      "step": 79888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000076273733672e-05,
+      "loss": 2.2301,
+      "step": 79896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000649906505543e-05,
+      "loss": 2.2391,
+      "step": 79904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000546102099226e-05,
+      "loss": 2.2307,
+      "step": 79912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000045132412816e-05,
+      "loss": 2.2252,
+      "step": 79920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000365572601862e-05,
+      "loss": 2.2481,
+      "step": 79928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000028884752893e-05,
+      "loss": 2.235,
+      "step": 79936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000221148917062e-05,
+      "loss": 2.2301,
+      "step": 79944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000162476773048e-05,
+      "loss": 2.2623,
+      "step": 79952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000112831102773e-05,
+      "loss": 2.2336,
+      "step": 79960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000072211911203e-05,
+      "loss": 2.2275,
+      "step": 79968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000040619202434e-05,
+      "loss": 2.2352,
+      "step": 79976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000018052979615e-05,
+      "loss": 2.2677,
+      "step": 79984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000004513245018e-05,
+      "loss": 2.2489,
+      "step": 79992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 2.2439,
+      "step": 80000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.38147229679313e-05,
+      "loss": 2.2966,
+      "step": 80008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.380880403687231e-05,
+      "loss": 2.3107,
+      "step": 80016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.380288476252493e-05,
+      "loss": 2.286,
+      "step": 80024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.379696514503625e-05,
+      "loss": 2.2986,
+      "step": 80032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.379104518455349e-05,
+      "loss": 2.2907,
+      "step": 80040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.378512488122382e-05,
+      "loss": 2.3108,
+      "step": 80048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.377920423519435e-05,
+      "loss": 2.2866,
+      "step": 80056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.377328324661236e-05,
+      "loss": 2.3207,
+      "step": 80064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.376736191562496e-05,
+      "loss": 2.2933,
+      "step": 80072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.376144024237946e-05,
+      "loss": 2.3103,
+      "step": 80080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.375551822702295e-05,
+      "loss": 2.2943,
+      "step": 80088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.374959586970274e-05,
+      "loss": 2.2651,
+      "step": 80096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.374367317056606e-05,
+      "loss": 2.2916,
+      "step": 80104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.373775012976008e-05,
+      "loss": 2.2736,
+      "step": 80112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.373182674743209e-05,
+      "loss": 2.2699,
+      "step": 80120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.372590302372936e-05,
+      "loss": 2.2665,
+      "step": 80128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.371997895879916e-05,
+      "loss": 2.2664,
+      "step": 80136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.371405455278872e-05,
+      "loss": 2.2644,
+      "step": 80144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.370812980584536e-05,
+      "loss": 2.2634,
+      "step": 80152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.370220471811635e-05,
+      "loss": 2.2723,
+      "step": 80160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.369627928974898e-05,
+      "loss": 2.2508,
+      "step": 80168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.36903535208906e-05,
+      "loss": 2.2603,
+      "step": 80176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.368442741168846e-05,
+      "loss": 2.2474,
+      "step": 80184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.367850096228994e-05,
+      "loss": 2.2315,
+      "step": 80192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.367257417284235e-05,
+      "loss": 2.2511,
+      "step": 80200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.366664704349303e-05,
+      "loss": 2.2528,
+      "step": 80208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.366071957438931e-05,
+      "loss": 2.247,
+      "step": 80216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.36547917656786e-05,
+      "loss": 2.2485,
+      "step": 80224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.364886361750822e-05,
+      "loss": 2.2404,
+      "step": 80232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.364293513002556e-05,
+      "loss": 2.2477,
+      "step": 80240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.363700630337803e-05,
+      "loss": 2.2476,
+      "step": 80248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.363107713771295e-05,
+      "loss": 2.25,
+      "step": 80256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.362514763317778e-05,
+      "loss": 2.2357,
+      "step": 80264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.361921778991992e-05,
+      "loss": 2.2522,
+      "step": 80272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.361328760808676e-05,
+      "loss": 2.2628,
+      "step": 80280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.360735708782575e-05,
+      "loss": 2.2491,
+      "step": 80288
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.360142622928432e-05,
+      "loss": 2.2538,
+      "step": 80296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.359549503260992e-05,
+      "loss": 2.2652,
+      "step": 80304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.358956349794996e-05,
+      "loss": 2.2593,
+      "step": 80312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.358363162545194e-05,
+      "loss": 2.2539,
+      "step": 80320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.357769941526334e-05,
+      "loss": 2.2555,
+      "step": 80328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.357176686753156e-05,
+      "loss": 2.2586,
+      "step": 80336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.356583398240415e-05,
+      "loss": 2.2395,
+      "step": 80344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.355990076002861e-05,
+      "loss": 2.2513,
+      "step": 80352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.355396720055241e-05,
+      "loss": 2.2422,
+      "step": 80360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.354803330412303e-05,
+      "loss": 2.2432,
+      "step": 80368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.354209907088804e-05,
+      "loss": 2.2373,
+      "step": 80376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.353616450099496e-05,
+      "loss": 2.2376,
+      "step": 80384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.35302295945913e-05,
+      "loss": 2.2508,
+      "step": 80392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.352429435182462e-05,
+      "loss": 2.2441,
+      "step": 80400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.351835877284246e-05,
+      "loss": 2.2479,
+      "step": 80408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.351242285779241e-05,
+      "loss": 2.2488,
+      "step": 80416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.350648660682197e-05,
+      "loss": 2.243,
+      "step": 80424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.350055002007878e-05,
+      "loss": 2.2452,
+      "step": 80432
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.34946130977104e-05,
+      "loss": 2.2454,
+      "step": 80440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.348867583986443e-05,
+      "loss": 2.2646,
+      "step": 80448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.348273824668846e-05,
+      "loss": 2.2679,
+      "step": 80456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.347680031833008e-05,
+      "loss": 2.255,
+      "step": 80464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.347086205493696e-05,
+      "loss": 2.2689,
+      "step": 80472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.34649234566567e-05,
+      "loss": 2.2683,
+      "step": 80480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.345898452363693e-05,
+      "loss": 2.2572,
+      "step": 80488
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.345304525602528e-05,
+      "loss": 2.2435,
+      "step": 80496
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.344710565396942e-05,
+      "loss": 2.2718,
+      "step": 80504
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.344116571761701e-05,
+      "loss": 2.252,
+      "step": 80512
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.343522544711572e-05,
+      "loss": 2.2653,
+      "step": 80520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.34292848426132e-05,
+      "loss": 2.2687,
+      "step": 80528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.342334390425715e-05,
+      "loss": 2.2641,
+      "step": 80536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.34174026321953e-05,
+      "loss": 2.2609,
+      "step": 80544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.341146102657528e-05,
+      "loss": 2.2604,
+      "step": 80552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.340551908754485e-05,
+      "loss": 2.2573,
+      "step": 80560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.339957681525171e-05,
+      "loss": 2.2714,
+      "step": 80568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.33936342098436e-05,
+      "loss": 2.2757,
+      "step": 80576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.33876912714682e-05,
+      "loss": 2.2602,
+      "step": 80584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.338174800027333e-05,
+      "loss": 2.263,
+      "step": 80592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.337580439640668e-05,
+      "loss": 2.2645,
+      "step": 80600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.336986046001603e-05,
+      "loss": 2.2812,
+      "step": 80608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.336391619124918e-05,
+      "loss": 2.2673,
+      "step": 80616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.335797159025386e-05,
+      "loss": 2.2672,
+      "step": 80624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.335202665717786e-05,
+      "loss": 2.2697,
+      "step": 80632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.334608139216899e-05,
+      "loss": 2.2499,
+      "step": 80640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.334013579537503e-05,
+      "loss": 2.2636,
+      "step": 80648
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.333418986694381e-05,
+      "loss": 2.2493,
+      "step": 80656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.33282436070231e-05,
+      "loss": 2.2479,
+      "step": 80664
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.332229701576078e-05,
+      "loss": 2.2426,
+      "step": 80672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.331635009330466e-05,
+      "loss": 2.259,
+      "step": 80680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.33104028398026e-05,
+      "loss": 2.2529,
+      "step": 80688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.330445525540242e-05,
+      "loss": 2.2591,
+      "step": 80696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.329850734025199e-05,
+      "loss": 2.2717,
+      "step": 80704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.329255909449917e-05,
+      "loss": 2.2394,
+      "step": 80712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.328661051829182e-05,
+      "loss": 2.2555,
+      "step": 80720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.328066161177789e-05,
+      "loss": 2.2517,
+      "step": 80728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.327471237510517e-05,
+      "loss": 2.2377,
+      "step": 80736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.326876280842166e-05,
+      "loss": 2.2513,
+      "step": 80744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.32628129118752e-05,
+      "loss": 2.2567,
+      "step": 80752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.325686268561372e-05,
+      "loss": 2.2512,
+      "step": 80760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.325091212978512e-05,
+      "loss": 2.2509,
+      "step": 80768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.32449612445374e-05,
+      "loss": 2.2414,
+      "step": 80776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.323901003001842e-05,
+      "loss": 2.2325,
+      "step": 80784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.32330584863762e-05,
+      "loss": 2.2383,
+      "step": 80792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.322710661375864e-05,
+      "loss": 2.2356,
+      "step": 80800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.322115441231374e-05,
+      "loss": 2.2269,
+      "step": 80808
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.321520188218944e-05,
+      "loss": 2.2349,
+      "step": 80816
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.320924902353374e-05,
+      "loss": 2.231,
+      "step": 80824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.320329583649466e-05,
+      "loss": 2.248,
+      "step": 80832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.319734232122011e-05,
+      "loss": 2.2198,
+      "step": 80840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.319138847785818e-05,
+      "loss": 2.2308,
+      "step": 80848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.318543430655686e-05,
+      "loss": 2.2546,
+      "step": 80856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.317947980746416e-05,
+      "loss": 2.2496,
+      "step": 80864
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.317352498072809e-05,
+      "loss": 2.2337,
+      "step": 80872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.316756982649675e-05,
+      "loss": 2.2613,
+      "step": 80880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.316161434491813e-05,
+      "loss": 2.2481,
+      "step": 80888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.315565853614028e-05,
+      "loss": 2.2504,
+      "step": 80896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.314970240031131e-05,
+      "loss": 2.2487,
+      "step": 80904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.314374593757924e-05,
+      "loss": 2.2561,
+      "step": 80912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.31377891480922e-05,
+      "loss": 2.2646,
+      "step": 80920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.313183203199823e-05,
+      "loss": 2.2603,
+      "step": 80928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.312587458944542e-05,
+      "loss": 2.2585,
+      "step": 80936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.311991682058193e-05,
+      "loss": 2.241,
+      "step": 80944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.311395872555582e-05,
+      "loss": 2.2645,
+      "step": 80952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.310800030451523e-05,
+      "loss": 2.2588,
+      "step": 80960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.310204155760829e-05,
+      "loss": 2.2659,
+      "step": 80968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.309608248498312e-05,
+      "loss": 2.2607,
+      "step": 80976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.309012308678787e-05,
+      "loss": 2.2587,
+      "step": 80984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.308416336317068e-05,
+      "loss": 2.255,
+      "step": 80992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.307820331427973e-05,
+      "loss": 2.2608,
+      "step": 81000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.307224294026318e-05,
+      "loss": 2.2463,
+      "step": 81008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.306628224126921e-05,
+      "loss": 2.2428,
+      "step": 81016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.306032121744597e-05,
+      "loss": 2.2431,
+      "step": 81024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.305435986894169e-05,
+      "loss": 2.2466,
+      "step": 81032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.304839819590459e-05,
+      "loss": 2.2468,
+      "step": 81040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.304243619848281e-05,
+      "loss": 2.2519,
+      "step": 81048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.303647387682462e-05,
+      "loss": 2.2492,
+      "step": 81056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.303051123107822e-05,
+      "loss": 2.2464,
+      "step": 81064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.302454826139185e-05,
+      "loss": 2.2334,
+      "step": 81072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.301858496791376e-05,
+      "loss": 2.2378,
+      "step": 81080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.301262135079216e-05,
+      "loss": 2.2409,
+      "step": 81088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.300665741017535e-05,
+      "loss": 2.2441,
+      "step": 81096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.300069314621161e-05,
+      "loss": 2.2521,
+      "step": 81104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.299472855904914e-05,
+      "loss": 2.258,
+      "step": 81112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.298876364883626e-05,
+      "loss": 2.2478,
+      "step": 81120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.298279841572126e-05,
+      "loss": 2.2501,
+      "step": 81128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.297683285985246e-05,
+      "loss": 2.2488,
+      "step": 81136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.297086698137812e-05,
+      "loss": 2.2596,
+      "step": 81144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.296490078044656e-05,
+      "loss": 2.2385,
+      "step": 81152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.295893425720615e-05,
+      "loss": 2.2482,
+      "step": 81160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.295296741180513e-05,
+      "loss": 2.2539,
+      "step": 81168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.294700024439191e-05,
+      "loss": 2.2507,
+      "step": 81176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.294103275511484e-05,
+      "loss": 2.2616,
+      "step": 81184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.293506494412222e-05,
+      "loss": 2.2651,
+      "step": 81192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.292909681156241e-05,
+      "loss": 2.2529,
+      "step": 81200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.292312835758382e-05,
+      "loss": 2.2598,
+      "step": 81208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.291715958233482e-05,
+      "loss": 2.2445,
+      "step": 81216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.291119048596377e-05,
+      "loss": 2.2491,
+      "step": 81224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.29052210686191e-05,
+      "loss": 2.258,
+      "step": 81232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.289925133044915e-05,
+      "loss": 2.2497,
+      "step": 81240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.28932812716024e-05,
+      "loss": 2.2516,
+      "step": 81248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.28873108922272e-05,
+      "loss": 2.2635,
+      "step": 81256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.288134019247203e-05,
+      "loss": 2.2599,
+      "step": 81264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.287536917248528e-05,
+      "loss": 2.2758,
+      "step": 81272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.286939783241542e-05,
+      "loss": 2.2603,
+      "step": 81280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.286342617241089e-05,
+      "loss": 2.2543,
+      "step": 81288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.285745419262012e-05,
+      "loss": 2.2551,
+      "step": 81296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.28514818931916e-05,
+      "loss": 2.2465,
+      "step": 81304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.284550927427383e-05,
+      "loss": 2.2359,
+      "step": 81312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.283953633601523e-05,
+      "loss": 2.2736,
+      "step": 81320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.283356307856432e-05,
+      "loss": 2.248,
+      "step": 81328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.282758950206958e-05,
+      "loss": 2.2566,
+      "step": 81336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.282161560667955e-05,
+      "loss": 2.2479,
+      "step": 81344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.281564139254271e-05,
+      "loss": 2.2379,
+      "step": 81352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.28096668598076e-05,
+      "loss": 2.2586,
+      "step": 81360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.280369200862269e-05,
+      "loss": 2.2411,
+      "step": 81368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.279771683913661e-05,
+      "loss": 2.2503,
+      "step": 81376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.279174135149782e-05,
+      "loss": 2.2509,
+      "step": 81384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.278576554585493e-05,
+      "loss": 2.2578,
+      "step": 81392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.277978942235644e-05,
+      "loss": 2.2582,
+      "step": 81400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.277381298115098e-05,
+      "loss": 2.2685,
+      "step": 81408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.27678362223871e-05,
+      "loss": 2.2655,
+      "step": 81416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.276185914621334e-05,
+      "loss": 2.2534,
+      "step": 81424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.275588175277837e-05,
+      "loss": 2.2563,
+      "step": 81432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.274990404223074e-05,
+      "loss": 2.2573,
+      "step": 81440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.274392601471907e-05,
+      "loss": 2.2627,
+      "step": 81448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.273794767039195e-05,
+      "loss": 2.2621,
+      "step": 81456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.273196900939802e-05,
+      "loss": 2.2709,
+      "step": 81464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.272599003188593e-05,
+      "loss": 2.255,
+      "step": 81472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.27200107380043e-05,
+      "loss": 2.2664,
+      "step": 81480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.271403112790177e-05,
+      "loss": 2.2685,
+      "step": 81488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.270805120172702e-05,
+      "loss": 2.2815,
+      "step": 81496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.270207095962864e-05,
+      "loss": 2.2348,
+      "step": 81504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.26960904017554e-05,
+      "loss": 2.2584,
+      "step": 81512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.26901095282559e-05,
+      "loss": 2.2523,
+      "step": 81520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.268412833927887e-05,
+      "loss": 2.2553,
+      "step": 81528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.267814683497298e-05,
+      "loss": 2.2599,
+      "step": 81536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.267216501548692e-05,
+      "loss": 2.261,
+      "step": 81544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.266618288096944e-05,
+      "loss": 2.2645,
+      "step": 81552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.266020043156922e-05,
+      "loss": 2.277,
+      "step": 81560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.2654217667435e-05,
+      "loss": 2.2653,
+      "step": 81568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.264823458871552e-05,
+      "loss": 2.2611,
+      "step": 81576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.26422511955595e-05,
+      "loss": 2.2761,
+      "step": 81584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.26362674881157e-05,
+      "loss": 2.2728,
+      "step": 81592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.263028346653287e-05,
+      "loss": 2.2616,
+      "step": 81600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.262429913095976e-05,
+      "loss": 2.2612,
+      "step": 81608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.261831448154516e-05,
+      "loss": 2.2662,
+      "step": 81616
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.261232951843789e-05,
+      "loss": 2.2677,
+      "step": 81624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.260634424178665e-05,
+      "loss": 2.2633,
+      "step": 81632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.260035865174028e-05,
+      "loss": 2.2668,
+      "step": 81640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.25943727484476e-05,
+      "loss": 2.2859,
+      "step": 81648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.258838653205737e-05,
+      "loss": 2.2493,
+      "step": 81656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.258240000271848e-05,
+      "loss": 2.2704,
+      "step": 81664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.257641316057969e-05,
+      "loss": 2.2799,
+      "step": 81672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.257042600578985e-05,
+      "loss": 2.2558,
+      "step": 81680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.256443853849784e-05,
+      "loss": 2.2631,
+      "step": 81688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.255845075885244e-05,
+      "loss": 2.2643,
+      "step": 81696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.255246266700258e-05,
+      "loss": 2.2586,
+      "step": 81704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.254647426309706e-05,
+      "loss": 2.2551,
+      "step": 81712
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.254048554728479e-05,
+      "loss": 2.2583,
+      "step": 81720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.253449651971467e-05,
+      "loss": 2.2445,
+      "step": 81728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.252850718053552e-05,
+      "loss": 2.253,
+      "step": 81736
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.252251752989629e-05,
+      "loss": 2.2464,
+      "step": 81744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.251652756794587e-05,
+      "loss": 2.2358,
+      "step": 81752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.251053729483317e-05,
+      "loss": 2.2483,
+      "step": 81760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.250454671070712e-05,
+      "loss": 2.2392,
+      "step": 81768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.249855581571662e-05,
+      "loss": 2.2474,
+      "step": 81776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.249256461001064e-05,
+      "loss": 2.2435,
+      "step": 81784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.248657309373807e-05,
+      "loss": 2.2508,
+      "step": 81792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.24805812670479e-05,
+      "loss": 2.2451,
+      "step": 81800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.247458913008911e-05,
+      "loss": 2.2468,
+      "step": 81808
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.246859668301063e-05,
+      "loss": 2.2458,
+      "step": 81816
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.24626039259614e-05,
+      "loss": 2.2378,
+      "step": 81824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.245661085909048e-05,
+      "loss": 2.2381,
+      "step": 81832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.24506174825468e-05,
+      "loss": 2.2402,
+      "step": 81840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.244462379647936e-05,
+      "loss": 2.2334,
+      "step": 81848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.24386298010372e-05,
+      "loss": 2.2462,
+      "step": 81856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.243263549636927e-05,
+      "loss": 2.239,
+      "step": 81864
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.242664088262465e-05,
+      "loss": 2.2508,
+      "step": 81872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.242064595995234e-05,
+      "loss": 2.2407,
+      "step": 81880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.24146507285014e-05,
+      "loss": 2.2584,
+      "step": 81888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.240865518842081e-05,
+      "loss": 2.2453,
+      "step": 81896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.24026593398597e-05,
+      "loss": 2.2648,
+      "step": 81904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.239666318296707e-05,
+      "loss": 2.2606,
+      "step": 81912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.239066671789198e-05,
+      "loss": 2.243,
+      "step": 81920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.238466994478357e-05,
+      "loss": 2.2477,
+      "step": 81928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.237867286379085e-05,
+      "loss": 2.2656,
+      "step": 81936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.237267547506295e-05,
+      "loss": 2.2684,
+      "step": 81944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.23666777787489e-05,
+      "loss": 2.2658,
+      "step": 81952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.23606797749979e-05,
+      "loss": 2.2537,
+      "step": 81960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.235468146395901e-05,
+      "loss": 2.2549,
+      "step": 81968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.234868284578136e-05,
+      "loss": 2.2607,
+      "step": 81976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.234268392061402e-05,
+      "loss": 2.2431,
+      "step": 81984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.233668468860622e-05,
+      "loss": 2.2528,
+      "step": 81992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.233068514990704e-05,
+      "loss": 2.2399,
+      "step": 82000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.232468530466566e-05,
+      "loss": 2.2587,
+      "step": 82008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.23186851530312e-05,
+      "loss": 2.2555,
+      "step": 82016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.231268469515284e-05,
+      "loss": 2.253,
+      "step": 82024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.230668393117979e-05,
+      "loss": 2.2544,
+      "step": 82032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.230068286126117e-05,
+      "loss": 2.258,
+      "step": 82040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.229468148554619e-05,
+      "loss": 2.2482,
+      "step": 82048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.228867980418405e-05,
+      "loss": 2.2572,
+      "step": 82056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.228267781732397e-05,
+      "loss": 2.2559,
+      "step": 82064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.227667552511512e-05,
+      "loss": 2.2448,
+      "step": 82072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.227067292770675e-05,
+      "loss": 2.2496,
+      "step": 82080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.226467002524806e-05,
+      "loss": 2.2634,
+      "step": 82088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.225866681788833e-05,
+      "loss": 2.2274,
+      "step": 82096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.225266330577673e-05,
+      "loss": 2.2452,
+      "step": 82104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.224665948906258e-05,
+      "loss": 2.2395,
+      "step": 82112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.224065536789508e-05,
+      "loss": 2.2495,
+      "step": 82120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.22346509424235e-05,
+      "loss": 2.2621,
+      "step": 82128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.222864621279714e-05,
+      "loss": 2.2444,
+      "step": 82136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.222264117916529e-05,
+      "loss": 2.2538,
+      "step": 82144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.221663584167718e-05,
+      "loss": 2.246,
+      "step": 82152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.221063020048213e-05,
+      "loss": 2.2546,
+      "step": 82160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.220462425572947e-05,
+      "loss": 2.2229,
+      "step": 82168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.219861800756844e-05,
+      "loss": 2.2587,
+      "step": 82176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.219261145614841e-05,
+      "loss": 2.2364,
+      "step": 82184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.218660460161872e-05,
+      "loss": 2.2613,
+      "step": 82192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.218059744412864e-05,
+      "loss": 2.2516,
+      "step": 82200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.217458998382756e-05,
+      "loss": 2.2416,
+      "step": 82208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.21685822208648e-05,
+      "loss": 2.254,
+      "step": 82216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.216257415538971e-05,
+      "loss": 2.26,
+      "step": 82224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.21565657875517e-05,
+      "loss": 2.2307,
+      "step": 82232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.215055711750007e-05,
+      "loss": 2.2419,
+      "step": 82240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.214454814538423e-05,
+      "loss": 2.2425,
+      "step": 82248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.213853887135355e-05,
+      "loss": 2.2364,
+      "step": 82256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.213252929555745e-05,
+      "loss": 2.2425,
+      "step": 82264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.21265194181453e-05,
+      "loss": 2.2477,
+      "step": 82272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.212050923926653e-05,
+      "loss": 2.253,
+      "step": 82280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.211449875907052e-05,
+      "loss": 2.2556,
+      "step": 82288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.210848797770673e-05,
+      "loss": 2.2464,
+      "step": 82296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.210247689532456e-05,
+      "loss": 2.2655,
+      "step": 82304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.209646551207348e-05,
+      "loss": 2.254,
+      "step": 82312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.209045382810286e-05,
+      "loss": 2.2426,
+      "step": 82320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.208444184356225e-05,
+      "loss": 2.2469,
+      "step": 82328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.207842955860105e-05,
+      "loss": 2.2529,
+      "step": 82336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.207241697336871e-05,
+      "loss": 2.2771,
+      "step": 82344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.206640408801476e-05,
+      "loss": 2.2681,
+      "step": 82352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.206039090268861e-05,
+      "loss": 2.2641,
+      "step": 82360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.205437741753985e-05,
+      "loss": 2.2525,
+      "step": 82368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.204836363271788e-05,
+      "loss": 2.252,
+      "step": 82376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.204234954837223e-05,
+      "loss": 2.2666,
+      "step": 82384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.203633516465242e-05,
+      "loss": 2.2679,
+      "step": 82392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.203032048170797e-05,
+      "loss": 2.2674,
+      "step": 82400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.202430549968841e-05,
+      "loss": 2.2706,
+      "step": 82408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.201829021874327e-05,
+      "loss": 2.2649,
+      "step": 82416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.201227463902207e-05,
+      "loss": 2.2745,
+      "step": 82424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.200625876067439e-05,
+      "loss": 2.281,
+      "step": 82432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.200024258384977e-05,
+      "loss": 2.2768,
+      "step": 82440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.199422610869777e-05,
+      "loss": 2.2875,
+      "step": 82448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.198820933536796e-05,
+      "loss": 2.2823,
+      "step": 82456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.198219226400993e-05,
+      "loss": 2.2704,
+      "step": 82464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.197617489477327e-05,
+      "loss": 2.2702,
+      "step": 82472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.19701572278075e-05,
+      "loss": 2.2734,
+      "step": 82480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.196413926326233e-05,
+      "loss": 2.2815,
+      "step": 82488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.19581210012873e-05,
+      "loss": 2.2658,
+      "step": 82496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.195210244203205e-05,
+      "loss": 2.271,
+      "step": 82504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.194608358564618e-05,
+      "loss": 2.2777,
+      "step": 82512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.194006443227934e-05,
+      "loss": 2.2748,
+      "step": 82520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.193404498208116e-05,
+      "loss": 2.2643,
+      "step": 82528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.192802523520128e-05,
+      "loss": 2.2604,
+      "step": 82536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.192200519178933e-05,
+      "loss": 2.2815,
+      "step": 82544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.191598485199505e-05,
+      "loss": 2.2765,
+      "step": 82552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.190996421596798e-05,
+      "loss": 2.2724,
+      "step": 82560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.190394328385792e-05,
+      "loss": 2.2564,
+      "step": 82568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.189792205581445e-05,
+      "loss": 2.2486,
+      "step": 82576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.189190053198734e-05,
+      "loss": 2.2341,
+      "step": 82584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.188587871252624e-05,
+      "loss": 2.2757,
+      "step": 82592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.187985659758085e-05,
+      "loss": 2.2657,
+      "step": 82600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.18738341873009e-05,
+      "loss": 2.2527,
+      "step": 82608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.18678114818361e-05,
+      "loss": 2.262,
+      "step": 82616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.186178848133618e-05,
+      "loss": 2.2693,
+      "step": 82624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.185576518595085e-05,
+      "loss": 2.2448,
+      "step": 82632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.184974159582987e-05,
+      "loss": 2.2549,
+      "step": 82640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.184371771112299e-05,
+      "loss": 2.2588,
+      "step": 82648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.183769353197998e-05,
+      "loss": 2.2652,
+      "step": 82656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.183166905855054e-05,
+      "loss": 2.2532,
+      "step": 82664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.182564429098451e-05,
+      "loss": 2.2644,
+      "step": 82672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.181961922943162e-05,
+      "loss": 2.2583,
+      "step": 82680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.18135938740417e-05,
+      "loss": 2.2663,
+      "step": 82688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.180756822496449e-05,
+      "loss": 2.2569,
+      "step": 82696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.180154228234979e-05,
+      "loss": 2.2576,
+      "step": 82704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.179551604634742e-05,
+      "loss": 2.2575,
+      "step": 82712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.178948951710724e-05,
+      "loss": 2.2486,
+      "step": 82720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.1783462694779e-05,
+      "loss": 2.239,
+      "step": 82728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.177743557951255e-05,
+      "loss": 2.2637,
+      "step": 82736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.177140817145775e-05,
+      "loss": 2.2522,
+      "step": 82744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.176538047076442e-05,
+      "loss": 2.2419,
+      "step": 82752
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.175935247758242e-05,
+      "loss": 2.252,
+      "step": 82760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.175332419206157e-05,
+      "loss": 2.2455,
+      "step": 82768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.174729561435178e-05,
+      "loss": 2.2605,
+      "step": 82776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.17412667446029e-05,
+      "loss": 2.2584,
+      "step": 82784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.173523758296481e-05,
+      "loss": 2.2553,
+      "step": 82792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.17292081295874e-05,
+      "loss": 2.25,
+      "step": 82800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.172317838462054e-05,
+      "loss": 2.2651,
+      "step": 82808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.171714834821417e-05,
+      "loss": 2.2611,
+      "step": 82816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.171111802051818e-05,
+      "loss": 2.2599,
+      "step": 82824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.170508740168247e-05,
+      "loss": 2.2501,
+      "step": 82832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.1699056491857e-05,
+      "loss": 2.2524,
+      "step": 82840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.169302529119165e-05,
+      "loss": 2.2522,
+      "step": 82848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.168699379983636e-05,
+      "loss": 2.2737,
+      "step": 82856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.168096201794112e-05,
+      "loss": 2.2611,
+      "step": 82864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.167492994565586e-05,
+      "loss": 2.2644,
+      "step": 82872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.166889758313053e-05,
+      "loss": 2.2417,
+      "step": 82880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.166286493051509e-05,
+      "loss": 2.2413,
+      "step": 82888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.165683198795952e-05,
+      "loss": 2.2536,
+      "step": 82896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.16507987556138e-05,
+      "loss": 2.2655,
+      "step": 82904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.164476523362791e-05,
+      "loss": 2.2566,
+      "step": 82912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.163873142215186e-05,
+      "loss": 2.258,
+      "step": 82920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.163269732133562e-05,
+      "loss": 2.2585,
+      "step": 82928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.162666293132922e-05,
+      "loss": 2.2676,
+      "step": 82936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.162062825228268e-05,
+      "loss": 2.2548,
+      "step": 82944
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.161459328434603e-05,
+      "loss": 2.2558,
+      "step": 82952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.160855802766928e-05,
+      "loss": 2.2521,
+      "step": 82960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.160252248240249e-05,
+      "loss": 2.2601,
+      "step": 82968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.159648664869568e-05,
+      "loss": 2.2397,
+      "step": 82976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.15904505266989e-05,
+      "loss": 2.2479,
+      "step": 82984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.158441411656222e-05,
+      "loss": 2.2351,
+      "step": 82992
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.157837741843574e-05,
+      "loss": 2.2545,
+      "step": 83000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.157234043246946e-05,
+      "loss": 2.2335,
+      "step": 83008
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.15663031588135e-05,
+      "loss": 2.2452,
+      "step": 83016
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.156026559761795e-05,
+      "loss": 2.2375,
+      "step": 83024
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.15542277490329e-05,
+      "loss": 2.249,
+      "step": 83032
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.154818961320849e-05,
+      "loss": 2.2578,
+      "step": 83040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.154215119029475e-05,
+      "loss": 2.2398,
+      "step": 83048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.153611248044183e-05,
+      "loss": 2.2569,
+      "step": 83056
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.153007348379987e-05,
+      "loss": 2.2575,
+      "step": 83064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.152403420051899e-05,
+      "loss": 2.2509,
+      "step": 83072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.151799463074936e-05,
+      "loss": 2.2492,
+      "step": 83080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.151195477464104e-05,
+      "loss": 2.2289,
+      "step": 83088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.150591463234426e-05,
+      "loss": 2.2448,
+      "step": 83096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.149987420400914e-05,
+      "loss": 2.2501,
+      "step": 83104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.149383348978587e-05,
+      "loss": 2.2479,
+      "step": 83112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.14877924898246e-05,
+      "loss": 2.2367,
+      "step": 83120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.148175120427552e-05,
+      "loss": 2.2548,
+      "step": 83128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.147570963328882e-05,
+      "loss": 2.2445,
+      "step": 83136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.146966777701469e-05,
+      "loss": 2.2582,
+      "step": 83144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.146362563560333e-05,
+      "loss": 2.262,
+      "step": 83152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.145758320920495e-05,
+      "loss": 2.2427,
+      "step": 83160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.145154049796976e-05,
+      "loss": 2.2685,
+      "step": 83168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.144549750204801e-05,
+      "loss": 2.256,
+      "step": 83176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.143945422158988e-05,
+      "loss": 2.2506,
+      "step": 83184
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.143341065674567e-05,
+      "loss": 2.264,
+      "step": 83192
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.142736680766555e-05,
+      "loss": 2.2491,
+      "step": 83200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.142132267449984e-05,
+      "loss": 2.2572,
+      "step": 83208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.141527825739875e-05,
+      "loss": 2.2794,
+      "step": 83216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.140923355651257e-05,
+      "loss": 2.2724,
+      "step": 83224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.140318857199155e-05,
+      "loss": 2.263,
+      "step": 83232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.1397143303986e-05,
+      "loss": 2.262,
+      "step": 83240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.139109775264617e-05,
+      "loss": 2.2444,
+      "step": 83248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.13850519181224e-05,
+      "loss": 2.2507,
+      "step": 83256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.137900580056494e-05,
+      "loss": 2.2522,
+      "step": 83264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.137295940012412e-05,
+      "loss": 2.2545,
+      "step": 83272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.136691271695024e-05,
+      "loss": 2.2658,
+      "step": 83280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.136086575119366e-05,
+      "loss": 2.2469,
+      "step": 83288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.135481850300467e-05,
+      "loss": 2.2468,
+      "step": 83296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.134877097253363e-05,
+      "loss": 2.269,
+      "step": 83304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.134272315993087e-05,
+      "loss": 2.2697,
+      "step": 83312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.133667506534672e-05,
+      "loss": 2.2593,
+      "step": 83320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.133062668893155e-05,
+      "loss": 2.2559,
+      "step": 83328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.132457803083576e-05,
+      "loss": 2.2667,
+      "step": 83336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.131852909120968e-05,
+      "loss": 2.269,
+      "step": 83344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.131247987020368e-05,
+      "loss": 2.2502,
+      "step": 83352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.130643036796818e-05,
+      "loss": 2.2697,
+      "step": 83360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.130038058465355e-05,
+      "loss": 2.2432,
+      "step": 83368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.12943305204102e-05,
+      "loss": 2.257,
+      "step": 83376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.128828017538849e-05,
+      "loss": 2.254,
+      "step": 83384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.12822295497389e-05,
+      "loss": 2.2448,
+      "step": 83392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.127617864361183e-05,
+      "loss": 2.2589,
+      "step": 83400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.127012745715766e-05,
+      "loss": 2.2354,
+      "step": 83408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.126407599052687e-05,
+      "loss": 2.2514,
+      "step": 83416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.125802424386989e-05,
+      "loss": 2.2562,
+      "step": 83424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.125197221733717e-05,
+      "loss": 2.2517,
+      "step": 83432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.124591991107914e-05,
+      "loss": 2.26,
+      "step": 83440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.123986732524628e-05,
+      "loss": 2.2482,
+      "step": 83448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.123381445998906e-05,
+      "loss": 2.2492,
+      "step": 83456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.122776131545797e-05,
+      "loss": 2.2308,
+      "step": 83464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.122170789180347e-05,
+      "loss": 2.2531,
+      "step": 83472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.121565418917604e-05,
+      "loss": 2.2589,
+      "step": 83480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.12096002077262e-05,
+      "loss": 2.2524,
+      "step": 83488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.120354594760443e-05,
+      "loss": 2.2568,
+      "step": 83496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.119749140896124e-05,
+      "loss": 2.2542,
+      "step": 83504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.119143659194718e-05,
+      "loss": 2.2605,
+      "step": 83512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.118538149671275e-05,
+      "loss": 2.2524,
+      "step": 83520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.117932612340847e-05,
+      "loss": 2.2628,
+      "step": 83528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.11732704721849e-05,
+      "loss": 2.2435,
+      "step": 83536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.116721454319256e-05,
+      "loss": 2.2373,
+      "step": 83544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.116115833658201e-05,
+      "loss": 2.2498,
+      "step": 83552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.115510185250383e-05,
+      "loss": 2.2364,
+      "step": 83560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.114904509110856e-05,
+      "loss": 2.2681,
+      "step": 83568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.114298805254675e-05,
+      "loss": 2.2619,
+      "step": 83576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.113693073696905e-05,
+      "loss": 2.2579,
+      "step": 83584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.113087314452598e-05,
+      "loss": 2.2558,
+      "step": 83592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.112481527536816e-05,
+      "loss": 2.248,
+      "step": 83600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.11187571296462e-05,
+      "loss": 2.263,
+      "step": 83608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.111269870751065e-05,
+      "loss": 2.2614,
+      "step": 83616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.11066400091122e-05,
+      "loss": 2.2429,
+      "step": 83624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.11005810346014e-05,
+      "loss": 2.2571,
+      "step": 83632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.109452178412895e-05,
+      "loss": 2.2657,
+      "step": 83640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.108846225784543e-05,
+      "loss": 2.2684,
+      "step": 83648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.108240245590147e-05,
+      "loss": 2.2455,
+      "step": 83656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.107634237844776e-05,
+      "loss": 2.2543,
+      "step": 83664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.107028202563492e-05,
+      "loss": 2.2591,
+      "step": 83672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.106422139761366e-05,
+      "loss": 2.2517,
+      "step": 83680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.10581604945346e-05,
+      "loss": 2.2556,
+      "step": 83688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.105209931654842e-05,
+      "loss": 2.2418,
+      "step": 83696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.104603786380581e-05,
+      "loss": 2.2615,
+      "step": 83704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.103997613645748e-05,
+      "loss": 2.2365,
+      "step": 83712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.103391413465408e-05,
+      "loss": 2.2385,
+      "step": 83720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.102785185854634e-05,
+      "loss": 2.258,
+      "step": 83728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.102178930828497e-05,
+      "loss": 2.2485,
+      "step": 83736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.101572648402071e-05,
+      "loss": 2.2472,
+      "step": 83744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.10096633859042e-05,
+      "loss": 2.2491,
+      "step": 83752
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.100360001408625e-05,
+      "loss": 2.2468,
+      "step": 83760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.099753636871757e-05,
+      "loss": 2.2475,
+      "step": 83768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.09914724499489e-05,
+      "loss": 2.2511,
+      "step": 83776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.098540825793101e-05,
+      "loss": 2.2716,
+      "step": 83784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.097934379281461e-05,
+      "loss": 2.2544,
+      "step": 83792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.097327905475051e-05,
+      "loss": 2.2428,
+      "step": 83800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.096721404388944e-05,
+      "loss": 2.2555,
+      "step": 83808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.096114876038223e-05,
+      "loss": 2.2446,
+      "step": 83816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.09550832043796e-05,
+      "loss": 2.2555,
+      "step": 83824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.094901737603239e-05,
+      "loss": 2.2544,
+      "step": 83832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.09429512754914e-05,
+      "loss": 2.2544,
+      "step": 83840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.093688490290738e-05,
+      "loss": 2.255,
+      "step": 83848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.093081825843117e-05,
+      "loss": 2.2317,
+      "step": 83856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.092475134221361e-05,
+      "loss": 2.2404,
+      "step": 83864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.09186841544055e-05,
+      "loss": 2.2733,
+      "step": 83872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.091261669515767e-05,
+      "loss": 2.2654,
+      "step": 83880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.090654896462098e-05,
+      "loss": 2.2284,
+      "step": 83888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.090048096294626e-05,
+      "loss": 2.2625,
+      "step": 83896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.089441269028433e-05,
+      "loss": 2.2347,
+      "step": 83904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.088834414678612e-05,
+      "loss": 2.2376,
+      "step": 83912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.088227533260244e-05,
+      "loss": 2.2321,
+      "step": 83920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.087620624788416e-05,
+      "loss": 2.254,
+      "step": 83928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.087013689278219e-05,
+      "loss": 2.2453,
+      "step": 83936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.086406726744737e-05,
+      "loss": 2.2339,
+      "step": 83944
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.085799737203064e-05,
+      "loss": 2.2417,
+      "step": 83952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.085192720668288e-05,
+      "loss": 2.2462,
+      "step": 83960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.084585677155498e-05,
+      "loss": 2.2567,
+      "step": 83968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.083978606679787e-05,
+      "loss": 2.2476,
+      "step": 83976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.083371509256247e-05,
+      "loss": 2.2514,
+      "step": 83984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.082764384899969e-05,
+      "loss": 2.2555,
+      "step": 83992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.082157233626046e-05,
+      "loss": 2.2408,
+      "step": 84000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.081550055449573e-05,
+      "loss": 2.2528,
+      "step": 84008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.080942850385645e-05,
+      "loss": 2.2587,
+      "step": 84016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.080335618449357e-05,
+      "loss": 2.2437,
+      "step": 84024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.079728359655804e-05,
+      "loss": 2.2539,
+      "step": 84032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.079121074020081e-05,
+      "loss": 2.2458,
+      "step": 84040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.078513761557289e-05,
+      "loss": 2.2379,
+      "step": 84048
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.077906422282523e-05,
+      "loss": 2.2446,
+      "step": 84056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.077299056210884e-05,
+      "loss": 2.2356,
+      "step": 84064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.076691663357469e-05,
+      "loss": 2.2438,
+      "step": 84072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.076084243737376e-05,
+      "loss": 2.2507,
+      "step": 84080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.07547679736571e-05,
+      "loss": 2.2628,
+      "step": 84088
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.07486932425757e-05,
+      "loss": 2.2429,
+      "step": 84096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.074261824428057e-05,
+      "loss": 2.2451,
+      "step": 84104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.073654297892276e-05,
+      "loss": 2.2418,
+      "step": 84112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.073046744665328e-05,
+      "loss": 2.2519,
+      "step": 84120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.072439164762317e-05,
+      "loss": 2.2551,
+      "step": 84128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.07183155819835e-05,
+      "loss": 2.2412,
+      "step": 84136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.071223924988528e-05,
+      "loss": 2.2539,
+      "step": 84144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.07061626514796e-05,
+      "loss": 2.2493,
+      "step": 84152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.070008578691751e-05,
+      "loss": 2.2277,
+      "step": 84160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.06940086563501e-05,
+      "loss": 2.2482,
+      "step": 84168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.068793125992843e-05,
+      "loss": 2.259,
+      "step": 84176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.068185359780358e-05,
+      "loss": 2.2385,
+      "step": 84184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.067577567012664e-05,
+      "loss": 2.2575,
+      "step": 84192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.066969747704874e-05,
+      "loss": 2.2397,
+      "step": 84200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.066361901872098e-05,
+      "loss": 2.2625,
+      "step": 84208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.065754029529444e-05,
+      "loss": 2.2415,
+      "step": 84216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.065146130692022e-05,
+      "loss": 2.2456,
+      "step": 84224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.064538205374953e-05,
+      "loss": 2.2505,
+      "step": 84232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.063930253593341e-05,
+      "loss": 2.2584,
+      "step": 84240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.063322275362306e-05,
+      "loss": 2.2596,
+      "step": 84248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.062714270696957e-05,
+      "loss": 2.2395,
+      "step": 84256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.062106239612414e-05,
+      "loss": 2.2464,
+      "step": 84264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.061498182123792e-05,
+      "loss": 2.248,
+      "step": 84272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.060890098246202e-05,
+      "loss": 2.2727,
+      "step": 84280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.060281987994768e-05,
+      "loss": 2.243,
+      "step": 84288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.059673851384602e-05,
+      "loss": 2.2448,
+      "step": 84296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.05906568843083e-05,
+      "loss": 2.2532,
+      "step": 84304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.058457499148563e-05,
+      "loss": 2.2555,
+      "step": 84312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.057849283552924e-05,
+      "loss": 2.2436,
+      "step": 84320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.057241041659035e-05,
+      "loss": 2.251,
+      "step": 84328
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.056632773482012e-05,
+      "loss": 2.255,
+      "step": 84336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.056024479036982e-05,
+      "loss": 2.2623,
+      "step": 84344
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.055416158339066e-05,
+      "loss": 2.2664,
+      "step": 84352
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.054807811403384e-05,
+      "loss": 2.2647,
+      "step": 84360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.054199438245062e-05,
+      "loss": 2.2485,
+      "step": 84368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.053591038879221e-05,
+      "loss": 2.2654,
+      "step": 84376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.052982613320993e-05,
+      "loss": 2.2678,
+      "step": 84384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.052374161585495e-05,
+      "loss": 2.2547,
+      "step": 84392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.05176568368786e-05,
+      "loss": 2.2511,
+      "step": 84400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.051157179643213e-05,
+      "loss": 2.2734,
+      "step": 84408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.050548649466677e-05,
+      "loss": 2.2698,
+      "step": 84416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.049940093173389e-05,
+      "loss": 2.2446,
+      "step": 84424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.049331510778468e-05,
+      "loss": 2.2431,
+      "step": 84432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.048722902297049e-05,
+      "loss": 2.2504,
+      "step": 84440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.048114267744262e-05,
+      "loss": 2.2524,
+      "step": 84448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.047505607135235e-05,
+      "loss": 2.2543,
+      "step": 84456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.0468969204851e-05,
+      "loss": 2.26,
+      "step": 84464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.046288207808991e-05,
+      "loss": 2.2542,
+      "step": 84472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.045679469122041e-05,
+      "loss": 2.2705,
+      "step": 84480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.045070704439385e-05,
+      "loss": 2.2524,
+      "step": 84488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.044461913776148e-05,
+      "loss": 2.2632,
+      "step": 84496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.043853097147475e-05,
+      "loss": 2.256,
+      "step": 84504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.043244254568495e-05,
+      "loss": 2.2507,
+      "step": 84512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.042635386054345e-05,
+      "loss": 2.2633,
+      "step": 84520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.042026491620164e-05,
+      "loss": 2.2486,
+      "step": 84528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.041417571281089e-05,
+      "loss": 2.2547,
+      "step": 84536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.040808625052253e-05,
+      "loss": 2.2653,
+      "step": 84544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.0401996529488e-05,
+      "loss": 2.2391,
+      "step": 84552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.039590654985867e-05,
+      "loss": 2.2689,
+      "step": 84560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.03898163117859e-05,
+      "loss": 2.2586,
+      "step": 84568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.038372581542116e-05,
+      "loss": 2.2485,
+      "step": 84576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.037763506091584e-05,
+      "loss": 2.2616,
+      "step": 84584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.037154404842133e-05,
+      "loss": 2.2692,
+      "step": 84592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.036545277808908e-05,
+      "loss": 2.2605,
+      "step": 84600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.03593612500705e-05,
+      "loss": 2.2522,
+      "step": 84608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.035326946451706e-05,
+      "loss": 2.2671,
+      "step": 84616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.034717742158015e-05,
+      "loss": 2.2599,
+      "step": 84624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.034108512141127e-05,
+      "loss": 2.2649,
+      "step": 84632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.033499256416185e-05,
+      "loss": 2.253,
+      "step": 84640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.032889974998336e-05,
+      "loss": 2.2673,
+      "step": 84648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.032280667902724e-05,
+      "loss": 2.2588,
+      "step": 84656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.031671335144501e-05,
+      "loss": 2.2557,
+      "step": 84664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.031061976738814e-05,
+      "loss": 2.2702,
+      "step": 84672
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.030452592700806e-05,
+      "loss": 2.2631,
+      "step": 84680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.029843183045635e-05,
+      "loss": 2.2538,
+      "step": 84688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.029233747788446e-05,
+      "loss": 2.2537,
+      "step": 84696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.02862428694439e-05,
+      "loss": 2.2692,
+      "step": 84704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.028014800528619e-05,
+      "loss": 2.2571,
+      "step": 84712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.027405288556286e-05,
+      "loss": 2.26,
+      "step": 84720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.026795751042541e-05,
+      "loss": 2.2572,
+      "step": 84728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.026186188002537e-05,
+      "loss": 2.2563,
+      "step": 84736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.025576599451432e-05,
+      "loss": 2.27,
+      "step": 84744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.024966985404376e-05,
+      "loss": 2.2609,
+      "step": 84752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.024357345876527e-05,
+      "loss": 2.2706,
+      "step": 84760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.023747680883037e-05,
+      "loss": 2.2646,
+      "step": 84768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.023137990439066e-05,
+      "loss": 2.265,
+      "step": 84776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.02252827455977e-05,
+      "loss": 2.272,
+      "step": 84784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.021918533260308e-05,
+      "loss": 2.2684,
+      "step": 84792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.021308766555834e-05,
+      "loss": 2.2523,
+      "step": 84800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.020698974461511e-05,
+      "loss": 2.2656,
+      "step": 84808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.020089156992497e-05,
+      "loss": 2.2766,
+      "step": 84816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.019479314163951e-05,
+      "loss": 2.264,
+      "step": 84824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.018869445991037e-05,
+      "loss": 2.2779,
+      "step": 84832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.01825955248891e-05,
+      "loss": 2.2599,
+      "step": 84840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.01764963367274e-05,
+      "loss": 2.2503,
+      "step": 84848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.017039689557686e-05,
+      "loss": 2.2519,
+      "step": 84856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.01642972015891e-05,
+      "loss": 2.2594,
+      "step": 84864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.015819725491577e-05,
+      "loss": 2.2694,
+      "step": 84872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.015209705570851e-05,
+      "loss": 2.2895,
+      "step": 84880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.0145996604119e-05,
+      "loss": 2.2595,
+      "step": 84888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.013989590029885e-05,
+      "loss": 2.267,
+      "step": 84896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.013379494439978e-05,
+      "loss": 2.2637,
+      "step": 84904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.012769373657342e-05,
+      "loss": 2.2559,
+      "step": 84912
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.012159227697143e-05,
+      "loss": 2.2916,
+      "step": 84920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.011549056574555e-05,
+      "loss": 2.2583,
+      "step": 84928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.010938860304743e-05,
+      "loss": 2.2785,
+      "step": 84936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.010328638902876e-05,
+      "loss": 2.2617,
+      "step": 84944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.00971839238413e-05,
+      "loss": 2.2486,
+      "step": 84952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.009108120763667e-05,
+      "loss": 2.2561,
+      "step": 84960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.008497824056665e-05,
+      "loss": 2.2801,
+      "step": 84968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.007887502278297e-05,
+      "loss": 2.2679,
+      "step": 84976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.007277155443727e-05,
+      "loss": 2.2672,
+      "step": 84984
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.006666783568137e-05,
+      "loss": 2.2623,
+      "step": 84992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.006056386666694e-05,
+      "loss": 2.2533,
+      "step": 85000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.005445964754582e-05,
+      "loss": 2.2587,
+      "step": 85008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.004835517846969e-05,
+      "loss": 2.2697,
+      "step": 85016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.00422504595903e-05,
+      "loss": 2.2482,
+      "step": 85024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.003614549105946e-05,
+      "loss": 2.2414,
+      "step": 85032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.003004027302888e-05,
+      "loss": 2.2542,
+      "step": 85040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.002393480565041e-05,
+      "loss": 2.266,
+      "step": 85048
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.00178290890758e-05,
+      "loss": 2.2651,
+      "step": 85056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.00117231234568e-05,
+      "loss": 2.2467,
+      "step": 85064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.000561690894527e-05,
+      "loss": 2.2784,
+      "step": 85072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.999951044569298e-05,
+      "loss": 2.2382,
+      "step": 85080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.999340373385172e-05,
+      "loss": 2.2536,
+      "step": 85088
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.998729677357332e-05,
+      "loss": 2.2561,
+      "step": 85096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.998118956500962e-05,
+      "loss": 2.2468,
+      "step": 85104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.997508210831242e-05,
+      "loss": 2.259,
+      "step": 85112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.996897440363356e-05,
+      "loss": 2.2539,
+      "step": 85120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.996286645112488e-05,
+      "loss": 2.2511,
+      "step": 85128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.995675825093821e-05,
+      "loss": 2.2517,
+      "step": 85136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.995064980322543e-05,
+      "loss": 2.258,
+      "step": 85144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.994454110813838e-05,
+      "loss": 2.2453,
+      "step": 85152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.993843216582891e-05,
+      "loss": 2.2408,
+      "step": 85160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.993232297644891e-05,
+      "loss": 2.2653,
+      "step": 85168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.992621354015024e-05,
+      "loss": 2.2579,
+      "step": 85176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.992010385708478e-05,
+      "loss": 2.2491,
+      "step": 85184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.991399392740446e-05,
+      "loss": 2.251,
+      "step": 85192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.990788375126109e-05,
+      "loss": 2.2546,
+      "step": 85200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.990177332880665e-05,
+      "loss": 2.2506,
+      "step": 85208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.9895662660193e-05,
+      "loss": 2.2566,
+      "step": 85216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.988955174557208e-05,
+      "loss": 2.2691,
+      "step": 85224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.98834405850958e-05,
+      "loss": 2.2405,
+      "step": 85232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.987732917891605e-05,
+      "loss": 2.2535,
+      "step": 85240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.987121752718479e-05,
+      "loss": 2.2466,
+      "step": 85248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.986510563005398e-05,
+      "loss": 2.2283,
+      "step": 85256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.985899348767553e-05,
+      "loss": 2.2585,
+      "step": 85264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.985288110020138e-05,
+      "loss": 2.2503,
+      "step": 85272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.98467684677835e-05,
+      "loss": 2.2437,
+      "step": 85280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.984065559057389e-05,
+      "loss": 2.2509,
+      "step": 85288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.983454246872444e-05,
+      "loss": 2.2397,
+      "step": 85296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.982842910238718e-05,
+      "loss": 2.2499,
+      "step": 85304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.982231549171405e-05,
+      "loss": 2.2451,
+      "step": 85312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.981620163685707e-05,
+      "loss": 2.2509,
+      "step": 85320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.98100875379682e-05,
+      "loss": 2.2524,
+      "step": 85328
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.980397319519947e-05,
+      "loss": 2.25,
+      "step": 85336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.979785860870288e-05,
+      "loss": 2.2464,
+      "step": 85344
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.979174377863038e-05,
+      "loss": 2.2609,
+      "step": 85352
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.978562870513408e-05,
+      "loss": 2.2505,
+      "step": 85360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.977951338836591e-05,
+      "loss": 2.2403,
+      "step": 85368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.977339782847796e-05,
+      "loss": 2.2497,
+      "step": 85376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.976728202562225e-05,
+      "loss": 2.2443,
+      "step": 85384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.976116597995081e-05,
+      "loss": 2.2443,
+      "step": 85392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.975504969161568e-05,
+      "loss": 2.2356,
+      "step": 85400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.974893316076891e-05,
+      "loss": 2.2571,
+      "step": 85408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.97428163875626e-05,
+      "loss": 2.2249,
+      "step": 85416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.973669937214877e-05,
+      "loss": 2.2403,
+      "step": 85424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.973058211467948e-05,
+      "loss": 2.2618,
+      "step": 85432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.972446461530685e-05,
+      "loss": 2.2554,
+      "step": 85440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.971834687418293e-05,
+      "loss": 2.2349,
+      "step": 85448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.971222889145984e-05,
+      "loss": 2.2491,
+      "step": 85456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.970611066728961e-05,
+      "loss": 2.2639,
+      "step": 85464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.969999220182439e-05,
+      "loss": 2.2479,
+      "step": 85472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.96938734952163e-05,
+      "loss": 2.2498,
+      "step": 85480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.96877545476174e-05,
+      "loss": 2.2486,
+      "step": 85488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.968163535917985e-05,
+      "loss": 2.2561,
+      "step": 85496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.967551593005574e-05,
+      "loss": 2.2497,
+      "step": 85504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.966939626039723e-05,
+      "loss": 2.2605,
+      "step": 85512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.966327635035646e-05,
+      "loss": 2.2325,
+      "step": 85520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.965715620008553e-05,
+      "loss": 2.2499,
+      "step": 85528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.965103580973665e-05,
+      "loss": 2.2493,
+      "step": 85536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.96449151794619e-05,
+      "loss": 2.2485,
+      "step": 85544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.963879430941349e-05,
+      "loss": 2.2778,
+      "step": 85552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.963267319974356e-05,
+      "loss": 2.2639,
+      "step": 85560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.96265518506043e-05,
+      "loss": 2.2475,
+      "step": 85568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.962043026214789e-05,
+      "loss": 2.2642,
+      "step": 85576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.96143084345265e-05,
+      "loss": 2.2556,
+      "step": 85584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.960818636789233e-05,
+      "loss": 2.2386,
+      "step": 85592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.960206406239756e-05,
+      "loss": 2.2641,
+      "step": 85600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.95959415181944e-05,
+      "loss": 2.2567,
+      "step": 85608
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.958981873543507e-05,
+      "loss": 2.268,
+      "step": 85616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.958369571427174e-05,
+      "loss": 2.2621,
+      "step": 85624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.957757245485669e-05,
+      "loss": 2.2506,
+      "step": 85632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.957144895734212e-05,
+      "loss": 2.2544,
+      "step": 85640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.956532522188022e-05,
+      "loss": 2.2599,
+      "step": 85648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.955920124862328e-05,
+      "loss": 2.2657,
+      "step": 85656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.955307703772352e-05,
+      "loss": 2.2414,
+      "step": 85664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.954695258933321e-05,
+      "loss": 2.2801,
+      "step": 85672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.954082790360456e-05,
+      "loss": 2.2534,
+      "step": 85680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.953470298068988e-05,
+      "loss": 2.2643,
+      "step": 85688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.95285778207414e-05,
+      "loss": 2.253,
+      "step": 85696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.952245242391142e-05,
+      "loss": 2.251,
+      "step": 85704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.95163267903522e-05,
+      "loss": 2.2628,
+      "step": 85712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.951020092021603e-05,
+      "loss": 2.2504,
+      "step": 85720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.95040748136552e-05,
+      "loss": 2.2648,
+      "step": 85728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.9497948470822e-05,
+      "loss": 2.2517,
+      "step": 85736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.949182189186871e-05,
+      "loss": 2.2465,
+      "step": 85744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.948569507694769e-05,
+      "loss": 2.2647,
+      "step": 85752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.947956802621122e-05,
+      "loss": 2.245,
+      "step": 85760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.947344073981163e-05,
+      "loss": 2.2408,
+      "step": 85768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.946731321790123e-05,
+      "loss": 2.2398,
+      "step": 85776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.946118546063237e-05,
+      "loss": 2.2501,
+      "step": 85784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.94550574681574e-05,
+      "loss": 2.2558,
+      "step": 85792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.944892924062861e-05,
+      "loss": 2.2527,
+      "step": 85800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.944280077819838e-05,
+      "loss": 2.2496,
+      "step": 85808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.943667208101907e-05,
+      "loss": 2.2476,
+      "step": 85816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.943054314924303e-05,
+      "loss": 2.2456,
+      "step": 85824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.942441398302264e-05,
+      "loss": 2.2518,
+      "step": 85832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.941828458251023e-05,
+      "loss": 2.2498,
+      "step": 85840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.941215494785825e-05,
+      "loss": 2.2535,
+      "step": 85848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.940602507921902e-05,
+      "loss": 2.2658,
+      "step": 85856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.939989497674495e-05,
+      "loss": 2.2591,
+      "step": 85864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.939376464058844e-05,
+      "loss": 2.2579,
+      "step": 85872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.938763407090188e-05,
+      "loss": 2.2475,
+      "step": 85880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.93815032678377e-05,
+      "loss": 2.2563,
+      "step": 85888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.937537223154828e-05,
+      "loss": 2.2421,
+      "step": 85896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.936924096218607e-05,
+      "loss": 2.253,
+      "step": 85904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.936310945990347e-05,
+      "loss": 2.2453,
+      "step": 85912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.935697772485292e-05,
+      "loss": 2.2574,
+      "step": 85920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.935084575718685e-05,
+      "loss": 2.2359,
+      "step": 85928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.934471355705771e-05,
+      "loss": 2.2653,
+      "step": 85936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.933858112461794e-05,
+      "loss": 2.2693,
+      "step": 85944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.933244846002001e-05,
+      "loss": 2.2528,
+      "step": 85952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.932631556341635e-05,
+      "loss": 2.2348,
+      "step": 85960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.932018243495944e-05,
+      "loss": 2.2594,
+      "step": 85968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.931404907480173e-05,
+      "loss": 2.2535,
+      "step": 85976
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.930791548309574e-05,
+      "loss": 2.2578,
+      "step": 85984
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.930178165999393e-05,
+      "loss": 2.2678,
+      "step": 85992
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.929564760564875e-05,
+      "loss": 2.249,
+      "step": 86000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.928951332021276e-05,
+      "loss": 2.2441,
+      "step": 86008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.928337880383842e-05,
+      "loss": 2.2651,
+      "step": 86016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.927724405667824e-05,
+      "loss": 2.2483,
+      "step": 86024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.927110907888473e-05,
+      "loss": 2.2611,
+      "step": 86032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.926497387061042e-05,
+      "loss": 2.2646,
+      "step": 86040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.92588384320078e-05,
+      "loss": 2.2505,
+      "step": 86048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.925270276322941e-05,
+      "loss": 2.2559,
+      "step": 86056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.924656686442783e-05,
+      "loss": 2.2602,
+      "step": 86064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.924043073575554e-05,
+      "loss": 2.2747,
+      "step": 86072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.92342943773651e-05,
+      "loss": 2.2624,
+      "step": 86080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.922815778940907e-05,
+      "loss": 2.2588,
+      "step": 86088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.922202097204e-05,
+      "loss": 2.2466,
+      "step": 86096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.921588392541047e-05,
+      "loss": 2.2552,
+      "step": 86104
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.920974664967302e-05,
+      "loss": 2.2554,
+      "step": 86112
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.92036091449802e-05,
+      "loss": 2.262,
+      "step": 86120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.919747141148467e-05,
+      "loss": 2.2657,
+      "step": 86128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.919133344933896e-05,
+      "loss": 2.263,
+      "step": 86136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.918519525869567e-05,
+      "loss": 2.2634,
+      "step": 86144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.917905683970738e-05,
+      "loss": 2.2628,
+      "step": 86152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.917291819252671e-05,
+      "loss": 2.2493,
+      "step": 86160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.916677931730628e-05,
+      "loss": 2.2679,
+      "step": 86168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.916064021419868e-05,
+      "loss": 2.2683,
+      "step": 86176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.915450088335653e-05,
+      "loss": 2.2533,
+      "step": 86184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.914836132493247e-05,
+      "loss": 2.2612,
+      "step": 86192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.914222153907912e-05,
+      "loss": 2.2513,
+      "step": 86200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.913608152594911e-05,
+      "loss": 2.2554,
+      "step": 86208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.912994128569508e-05,
+      "loss": 2.2486,
+      "step": 86216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.912380081846972e-05,
+      "loss": 2.2574,
+      "step": 86224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.911766012442562e-05,
+      "loss": 2.2671,
+      "step": 86232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.911151920371547e-05,
+      "loss": 2.2621,
+      "step": 86240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.910537805649194e-05,
+      "loss": 2.2553,
+      "step": 86248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.909923668290767e-05,
+      "loss": 2.2658,
+      "step": 86256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.909309508311538e-05,
+      "loss": 2.2468,
+      "step": 86264
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.90869532572677e-05,
+      "loss": 2.2458,
+      "step": 86272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.908081120551737e-05,
+      "loss": 2.2633,
+      "step": 86280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.907466892801703e-05,
+      "loss": 2.2664,
+      "step": 86288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.906852642491941e-05,
+      "loss": 2.2457,
+      "step": 86296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.906238369637721e-05,
+      "loss": 2.2417,
+      "step": 86304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.905624074254313e-05,
+      "loss": 2.2494,
+      "step": 86312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.90500975635699e-05,
+      "loss": 2.2431,
+      "step": 86320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.904395415961022e-05,
+      "loss": 2.2505,
+      "step": 86328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.903781053081682e-05,
+      "loss": 2.2415,
+      "step": 86336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.903166667734244e-05,
+      "loss": 2.2583,
+      "step": 86344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.902552259933983e-05,
+      "loss": 2.2459,
+      "step": 86352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.90193782969617e-05,
+      "loss": 2.2451,
+      "step": 86360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.90132337703608e-05,
+      "loss": 2.2529,
+      "step": 86368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.90070890196899e-05,
+      "loss": 2.2494,
+      "step": 86376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.900094404510179e-05,
+      "loss": 2.2391,
+      "step": 86384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.899479884674918e-05,
+      "loss": 2.2596,
+      "step": 86392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.898865342478484e-05,
+      "loss": 2.2585,
+      "step": 86400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.898250777936159e-05,
+      "loss": 2.2482,
+      "step": 86408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.89763619106322e-05,
+      "loss": 2.2564,
+      "step": 86416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.897021581874943e-05,
+      "loss": 2.2446,
+      "step": 86424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.89640695038661e-05,
+      "loss": 2.2573,
+      "step": 86432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.895792296613498e-05,
+      "loss": 2.2605,
+      "step": 86440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.895177620570889e-05,
+      "loss": 2.2488,
+      "step": 86448
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.894562922274065e-05,
+      "loss": 2.2668,
+      "step": 86456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.893948201738305e-05,
+      "loss": 2.2546,
+      "step": 86464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.893333458978894e-05,
+      "loss": 2.2544,
+      "step": 86472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.892718694011114e-05,
+      "loss": 2.2479,
+      "step": 86480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.892103906850245e-05,
+      "loss": 2.2491,
+      "step": 86488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.891489097511572e-05,
+      "loss": 2.2567,
+      "step": 86496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.890874266010381e-05,
+      "loss": 2.2584,
+      "step": 86504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.890259412361956e-05,
+      "loss": 2.2493,
+      "step": 86512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.889644536581583e-05,
+      "loss": 2.2556,
+      "step": 86520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.889029638684546e-05,
+      "loss": 2.2456,
+      "step": 86528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.888414718686133e-05,
+      "loss": 2.26,
+      "step": 86536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.88779977660163e-05,
+      "loss": 2.2661,
+      "step": 86544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.887184812446325e-05,
+      "loss": 2.265,
+      "step": 86552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.886569826235507e-05,
+      "loss": 2.2602,
+      "step": 86560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.885954817984462e-05,
+      "loss": 2.2508,
+      "step": 86568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.885339787708482e-05,
+      "loss": 2.2718,
+      "step": 86576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.884724735422854e-05,
+      "loss": 2.2637,
+      "step": 86584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.884109661142871e-05,
+      "loss": 2.2613,
+      "step": 86592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.883494564883823e-05,
+      "loss": 2.2637,
+      "step": 86600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.882879446661001e-05,
+      "loss": 2.2509,
+      "step": 86608
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.882264306489696e-05,
+      "loss": 2.2473,
+      "step": 86616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.881649144385202e-05,
+      "loss": 2.2441,
+      "step": 86624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.881033960362811e-05,
+      "loss": 2.2658,
+      "step": 86632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.880418754437817e-05,
+      "loss": 2.25,
+      "step": 86640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.879803526625514e-05,
+      "loss": 2.2593,
+      "step": 86648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.879188276941197e-05,
+      "loss": 2.2412,
+      "step": 86656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.87857300540016e-05,
+      "loss": 2.2632,
+      "step": 86664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.877957712017699e-05,
+      "loss": 2.2218,
+      "step": 86672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.877342396809112e-05,
+      "loss": 2.2558,
+      "step": 86680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.876727059789694e-05,
+      "loss": 2.2527,
+      "step": 86688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.876111700974743e-05,
+      "loss": 2.2569,
+      "step": 86696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.875496320379557e-05,
+      "loss": 2.2607,
+      "step": 86704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.874880918019435e-05,
+      "loss": 2.226,
+      "step": 86712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.874265493909674e-05,
+      "loss": 2.2544,
+      "step": 86720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.873650048065577e-05,
+      "loss": 2.2478,
+      "step": 86728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.873034580502437e-05,
+      "loss": 2.2394,
+      "step": 86736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.872419091235564e-05,
+      "loss": 2.2506,
+      "step": 86744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.871803580280252e-05,
+      "loss": 2.2536,
+      "step": 86752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.871188047651807e-05,
+      "loss": 2.2457,
+      "step": 86760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.870572493365527e-05,
+      "loss": 2.253,
+      "step": 86768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.86995691743672e-05,
+      "loss": 2.255,
+      "step": 86776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.869341319880685e-05,
+      "loss": 2.2341,
+      "step": 86784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.868725700712727e-05,
+      "loss": 2.2402,
+      "step": 86792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.86811005994815e-05,
+      "loss": 2.2486,
+      "step": 86800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.867494397602261e-05,
+      "loss": 2.2389,
+      "step": 86808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.866878713690363e-05,
+      "loss": 2.2519,
+      "step": 86816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.866263008227764e-05,
+      "loss": 2.2613,
+      "step": 86824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.865647281229768e-05,
+      "loss": 2.2289,
+      "step": 86832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.865031532711684e-05,
+      "loss": 2.2466,
+      "step": 86840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.864415762688819e-05,
+      "loss": 2.2364,
+      "step": 86848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.863799971176481e-05,
+      "loss": 2.2486,
+      "step": 86856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.863184158189979e-05,
+      "loss": 2.2349,
+      "step": 86864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.862568323744622e-05,
+      "loss": 2.2382,
+      "step": 86872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.861952467855721e-05,
+      "loss": 2.2398,
+      "step": 86880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.861336590538583e-05,
+      "loss": 2.2427,
+      "step": 86888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.860720691808521e-05,
+      "loss": 2.2481,
+      "step": 86896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.860104771680847e-05,
+      "loss": 2.2425,
+      "step": 86904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.859488830170872e-05,
+      "loss": 2.2498,
+      "step": 86912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.858872867293907e-05,
+      "loss": 2.2347,
+      "step": 86920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.858256883065267e-05,
+      "loss": 2.2499,
+      "step": 86928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.857640877500265e-05,
+      "loss": 2.2391,
+      "step": 86936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.857024850614213e-05,
+      "loss": 2.2527,
+      "step": 86944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.856408802422429e-05,
+      "loss": 2.2606,
+      "step": 86952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.855792732940223e-05,
+      "loss": 2.2492,
+      "step": 86960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.855176642182916e-05,
+      "loss": 2.2404,
+      "step": 86968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.854560530165822e-05,
+      "loss": 2.2579,
+      "step": 86976
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.853944396904258e-05,
+      "loss": 2.252,
+      "step": 86984
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.853328242413536e-05,
+      "loss": 2.2406,
+      "step": 86992
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.852712066708984e-05,
+      "loss": 2.2518,
+      "step": 87000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.852095869805912e-05,
+      "loss": 2.2404,
+      "step": 87008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.85147965171964e-05,
+      "loss": 2.2545,
+      "step": 87016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.850863412465489e-05,
+      "loss": 2.2682,
+      "step": 87024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.850247152058777e-05,
+      "loss": 2.2581,
+      "step": 87032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.849630870514828e-05,
+      "loss": 2.2619,
+      "step": 87040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.849014567848958e-05,
+      "loss": 2.2613,
+      "step": 87048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.848398244076492e-05,
+      "loss": 2.2589,
+      "step": 87056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.847781899212751e-05,
+      "loss": 2.257,
+      "step": 87064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.847165533273055e-05,
+      "loss": 2.2504,
+      "step": 87072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.846549146272729e-05,
+      "loss": 2.2604,
+      "step": 87080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.845932738227097e-05,
+      "loss": 2.2576,
+      "step": 87088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.845316309151481e-05,
+      "loss": 2.27,
+      "step": 87096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.84469985906121e-05,
+      "loss": 2.263,
+      "step": 87104
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.844083387971603e-05,
+      "loss": 2.2453,
+      "step": 87112
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.843466895897987e-05,
+      "loss": 2.2737,
+      "step": 87120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.842850382855692e-05,
+      "loss": 2.2543,
+      "step": 87128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.842233848860042e-05,
+      "loss": 2.2581,
+      "step": 87136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.841617293926362e-05,
+      "loss": 2.2553,
+      "step": 87144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.841000718069982e-05,
+      "loss": 2.259,
+      "step": 87152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.840384121306232e-05,
+      "loss": 2.2735,
+      "step": 87160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.839767503650436e-05,
+      "loss": 2.2684,
+      "step": 87168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.839150865117927e-05,
+      "loss": 2.2514,
+      "step": 87176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.838534205724034e-05,
+      "loss": 2.2442,
+      "step": 87184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.837917525484085e-05,
+      "loss": 2.2626,
+      "step": 87192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.837300824413413e-05,
+      "loss": 2.2447,
+      "step": 87200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.836684102527346e-05,
+      "loss": 2.2562,
+      "step": 87208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.836067359841223e-05,
+      "loss": 2.2434,
+      "step": 87216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.835450596370369e-05,
+      "loss": 2.2603,
+      "step": 87224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.83483381213012e-05,
+      "loss": 2.2493,
+      "step": 87232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.834217007135807e-05,
+      "loss": 2.2486,
+      "step": 87240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.833600181402767e-05,
+      "loss": 2.2475,
+      "step": 87248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.832983334946335e-05,
+      "loss": 2.2502,
+      "step": 87256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.83236646778184e-05,
+      "loss": 2.2601,
+      "step": 87264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.831749579924623e-05,
+      "loss": 2.2715,
+      "step": 87272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.831132671390019e-05,
+      "loss": 2.2622,
+      "step": 87280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.830515742193362e-05,
+      "loss": 2.246,
+      "step": 87288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.829898792349993e-05,
+      "loss": 2.2582,
+      "step": 87296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.829281821875244e-05,
+      "loss": 2.2445,
+      "step": 87304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.828664830784457e-05,
+      "loss": 2.2518,
+      "step": 87312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.82804781909297e-05,
+      "loss": 2.2444,
+      "step": 87320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.82743078681612e-05,
+      "loss": 2.2591,
+      "step": 87328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.826813733969248e-05,
+      "loss": 2.2639,
+      "step": 87336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.826196660567694e-05,
+      "loss": 2.2773,
+      "step": 87344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.825579566626798e-05,
+      "loss": 2.2602,
+      "step": 87352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.824962452161902e-05,
+      "loss": 2.2586,
+      "step": 87360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.824345317188347e-05,
+      "loss": 2.2518,
+      "step": 87368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.823728161721477e-05,
+      "loss": 2.258,
+      "step": 87376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.823110985776628e-05,
+      "loss": 2.2805,
+      "step": 87384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.822493789369149e-05,
+      "loss": 2.259,
+      "step": 87392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.821876572514386e-05,
+      "loss": 2.2629,
+      "step": 87400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.821259335227674e-05,
+      "loss": 2.2583,
+      "step": 87408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.820642077524366e-05,
+      "loss": 2.2437,
+      "step": 87416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.820024799419802e-05,
+      "loss": 2.2739,
+      "step": 87424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.81940750092933e-05,
+      "loss": 2.2644,
+      "step": 87432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.818790182068295e-05,
+      "loss": 2.2552,
+      "step": 87440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.818172842852047e-05,
+      "loss": 2.2479,
+      "step": 87448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.817555483295927e-05,
+      "loss": 2.2538,
+      "step": 87456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.816938103415287e-05,
+      "loss": 2.2618,
+      "step": 87464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.816320703225475e-05,
+      "loss": 2.2665,
+      "step": 87472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.815703282741839e-05,
+      "loss": 2.2471,
+      "step": 87480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.815085841979724e-05,
+      "loss": 2.247,
+      "step": 87488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.814468380954488e-05,
+      "loss": 2.2584,
+      "step": 87496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.813850899681475e-05,
+      "loss": 2.2551,
+      "step": 87504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.813233398176038e-05,
+      "loss": 2.26,
+      "step": 87512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.812615876453527e-05,
+      "loss": 2.2602,
+      "step": 87520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.811998334529295e-05,
+      "loss": 2.2588,
+      "step": 87528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.811380772418692e-05,
+      "loss": 2.2545,
+      "step": 87536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.810763190137072e-05,
+      "loss": 2.2586,
+      "step": 87544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.810145587699787e-05,
+      "loss": 2.2626,
+      "step": 87552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.809527965122194e-05,
+      "loss": 2.2654,
+      "step": 87560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.808910322419646e-05,
+      "loss": 2.2233,
+      "step": 87568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.808292659607494e-05,
+      "loss": 2.2642,
+      "step": 87576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.807674976701096e-05,
+      "loss": 2.2767,
+      "step": 87584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.807057273715809e-05,
+      "loss": 2.2532,
+      "step": 87592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.806439550666986e-05,
+      "loss": 2.2507,
+      "step": 87600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.805821807569986e-05,
+      "loss": 2.2574,
+      "step": 87608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.805204044440166e-05,
+      "loss": 2.2561,
+      "step": 87616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.804586261292882e-05,
+      "loss": 2.2508,
+      "step": 87624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.803968458143495e-05,
+      "loss": 2.2578,
+      "step": 87632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.80335063500736e-05,
+      "loss": 2.2566,
+      "step": 87640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.80273279189984e-05,
+      "loss": 2.2482,
+      "step": 87648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.802114928836291e-05,
+      "loss": 2.2651,
+      "step": 87656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.801497045832075e-05,
+      "loss": 2.2618,
+      "step": 87664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.800879142902552e-05,
+      "loss": 2.2705,
+      "step": 87672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.800261220063085e-05,
+      "loss": 2.2571,
+      "step": 87680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.799643277329035e-05,
+      "loss": 2.2548,
+      "step": 87688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.79902531471576e-05,
+      "loss": 2.2441,
+      "step": 87696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.79840733223863e-05,
+      "loss": 2.2634,
+      "step": 87704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.797789329913002e-05,
+      "loss": 2.2756,
+      "step": 87712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.797171307754242e-05,
+      "loss": 2.2519,
+      "step": 87720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.796553265777717e-05,
+      "loss": 2.2561,
+      "step": 87728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.795935203998782e-05,
+      "loss": 2.2539,
+      "step": 87736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.795317122432814e-05,
+      "loss": 2.2528,
+      "step": 87744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.794699021095172e-05,
+      "loss": 2.2647,
+      "step": 87752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.794080900001222e-05,
+      "loss": 2.2424,
+      "step": 87760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.793462759166331e-05,
+      "loss": 2.2518,
+      "step": 87768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.792844598605868e-05,
+      "loss": 2.2465,
+      "step": 87776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.7922264183352e-05,
+      "loss": 2.2465,
+      "step": 87784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.791608218369693e-05,
+      "loss": 2.2642,
+      "step": 87792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.790989998724717e-05,
+      "loss": 2.2429,
+      "step": 87800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.790371759415642e-05,
+      "loss": 2.2572,
+      "step": 87808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.789753500457835e-05,
+      "loss": 2.247,
+      "step": 87816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.78913522186667e-05,
+      "loss": 2.2456,
+      "step": 87824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.78851692365751e-05,
+      "loss": 2.2589,
+      "step": 87832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.787898605845736e-05,
+      "loss": 2.2388,
+      "step": 87840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.787280268446712e-05,
+      "loss": 2.2504,
+      "step": 87848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.786661911475811e-05,
+      "loss": 2.2574,
+      "step": 87856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.786043534948407e-05,
+      "loss": 2.2541,
+      "step": 87864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.785425138879875e-05,
+      "loss": 2.2582,
+      "step": 87872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.784806723285584e-05,
+      "loss": 2.2506,
+      "step": 87880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.784188288180909e-05,
+      "loss": 2.2689,
+      "step": 87888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.783569833581228e-05,
+      "loss": 2.2762,
+      "step": 87896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.78295135950191e-05,
+      "loss": 2.2427,
+      "step": 87904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.782332865958334e-05,
+      "loss": 2.2473,
+      "step": 87912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.781714352965876e-05,
+      "loss": 2.241,
+      "step": 87920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.781095820539912e-05,
+      "loss": 2.268,
+      "step": 87928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.780477268695818e-05,
+      "loss": 2.2401,
+      "step": 87936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.779858697448972e-05,
+      "loss": 2.2542,
+      "step": 87944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.779240106814749e-05,
+      "loss": 2.2429,
+      "step": 87952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.77862149680853e-05,
+      "loss": 2.2573,
+      "step": 87960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.778002867445695e-05,
+      "loss": 2.2703,
+      "step": 87968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.77738421874162e-05,
+      "loss": 2.2616,
+      "step": 87976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.776765550711687e-05,
+      "loss": 2.2723,
+      "step": 87984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.776146863371275e-05,
+      "loss": 2.2499,
+      "step": 87992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.775528156735764e-05,
+      "loss": 2.2685,
+      "step": 88000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.774909430820537e-05,
+      "loss": 2.2504,
+      "step": 88008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.774290685640974e-05,
+      "loss": 2.2546,
+      "step": 88016
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.773671921212459e-05,
+      "loss": 2.2662,
+      "step": 88024
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.773053137550371e-05,
+      "loss": 2.2532,
+      "step": 88032
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.772434334670098e-05,
+      "loss": 2.2503,
+      "step": 88040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.771815512587019e-05,
+      "loss": 2.2742,
+      "step": 88048
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.77119667131652e-05,
+      "loss": 2.2516,
+      "step": 88056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.770577810873987e-05,
+      "loss": 2.2686,
+      "step": 88064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.769958931274799e-05,
+      "loss": 2.2557,
+      "step": 88072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.769340032534348e-05,
+      "loss": 2.2564,
+      "step": 88080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.768721114668017e-05,
+      "loss": 2.2466,
+      "step": 88088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.768102177691194e-05,
+      "loss": 2.262,
+      "step": 88096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.767483221619264e-05,
+      "loss": 2.2595,
+      "step": 88104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.766864246467615e-05,
+      "loss": 2.2466,
+      "step": 88112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.766245252251636e-05,
+      "loss": 2.248,
+      "step": 88120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.765626238986712e-05,
+      "loss": 2.2568,
+      "step": 88128
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.765007206688236e-05,
+      "loss": 2.2595,
+      "step": 88136
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.764388155371592e-05,
+      "loss": 2.2573,
+      "step": 88144
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.763769085052174e-05,
+      "loss": 2.2502,
+      "step": 88152
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.763149995745372e-05,
+      "loss": 2.2472,
+      "step": 88160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.762530887466576e-05,
+      "loss": 2.2331,
+      "step": 88168
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.761911760231175e-05,
+      "loss": 2.2428,
+      "step": 88176
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.761292614054562e-05,
+      "loss": 2.2535,
+      "step": 88184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.76067344895213e-05,
+      "loss": 2.2409,
+      "step": 88192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.76005426493927e-05,
+      "loss": 2.2414,
+      "step": 88200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.759435062031375e-05,
+      "loss": 2.2657,
+      "step": 88208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.758815840243841e-05,
+      "loss": 2.2649,
+      "step": 88216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.75819659959206e-05,
+      "loss": 2.2678,
+      "step": 88224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.757577340091425e-05,
+      "loss": 2.2342,
+      "step": 88232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.756958061757332e-05,
+      "loss": 2.2536,
+      "step": 88240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.756338764605179e-05,
+      "loss": 2.2427,
+      "step": 88248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.755719448650359e-05,
+      "loss": 2.2434,
+      "step": 88256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.755100113908266e-05,
+      "loss": 2.2487,
+      "step": 88264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.754480760394303e-05,
+      "loss": 2.2548,
+      "step": 88272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.753861388123862e-05,
+      "loss": 2.255,
+      "step": 88280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.753241997112342e-05,
+      "loss": 2.2528,
+      "step": 88288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.752622587375141e-05,
+      "loss": 2.2342,
+      "step": 88296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.75200315892766e-05,
+      "loss": 2.2529,
+      "step": 88304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.751383711785296e-05,
+      "loss": 2.2503,
+      "step": 88312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.750764245963447e-05,
+      "loss": 2.2319,
+      "step": 88320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.750144761477512e-05,
+      "loss": 2.2568,
+      "step": 88328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.749525258342899e-05,
+      "loss": 2.2403,
+      "step": 88336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.748905736575003e-05,
+      "loss": 2.2599,
+      "step": 88344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.748286196189226e-05,
+      "loss": 2.2518,
+      "step": 88352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.747666637200967e-05,
+      "loss": 2.2575,
+      "step": 88360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.747047059625636e-05,
+      "loss": 2.2722,
+      "step": 88368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.746427463478631e-05,
+      "loss": 2.2647,
+      "step": 88376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.745807848775353e-05,
+      "loss": 2.2583,
+      "step": 88384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.745188215531208e-05,
+      "loss": 2.2584,
+      "step": 88392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.744568563761602e-05,
+      "loss": 2.2584,
+      "step": 88400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.743948893481938e-05,
+      "loss": 2.2499,
+      "step": 88408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.74332920470762e-05,
+      "loss": 2.2693,
+      "step": 88416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.742709497454055e-05,
+      "loss": 2.2669,
+      "step": 88424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.742089771736651e-05,
+      "loss": 2.259,
+      "step": 88432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.741470027570808e-05,
+      "loss": 2.2397,
+      "step": 88440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.740850264971939e-05,
+      "loss": 2.2514,
+      "step": 88448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.74023048395545e-05,
+      "loss": 2.2539,
+      "step": 88456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.739610684536746e-05,
+      "loss": 2.2582,
+      "step": 88464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.73899086673124e-05,
+      "loss": 2.2543,
+      "step": 88472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.738371030554336e-05,
+      "loss": 2.2621,
+      "step": 88480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.737751176021445e-05,
+      "loss": 2.2626,
+      "step": 88488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.737131303147978e-05,
+      "loss": 2.2467,
+      "step": 88496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.736511411949344e-05,
+      "loss": 2.2503,
+      "step": 88504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.735891502440954e-05,
+      "loss": 2.2451,
+      "step": 88512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.735271574638218e-05,
+      "loss": 2.2456,
+      "step": 88520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.73465162855655e-05,
+      "loss": 2.2509,
+      "step": 88528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.734031664211357e-05,
+      "loss": 2.2661,
+      "step": 88536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.733411681618057e-05,
+      "loss": 2.2548,
+      "step": 88544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.732791680792058e-05,
+      "loss": 2.2581,
+      "step": 88552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.732171661748779e-05,
+      "loss": 2.2503,
+      "step": 88560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.731551624503626e-05,
+      "loss": 2.2569,
+      "step": 88568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.73093156907202e-05,
+      "loss": 2.2507,
+      "step": 88576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.730311495469374e-05,
+      "loss": 2.2311,
+      "step": 88584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.729691403711103e-05,
+      "loss": 2.2592,
+      "step": 88592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.729071293812619e-05,
+      "loss": 2.2531,
+      "step": 88600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.728451165789342e-05,
+      "loss": 2.2434,
+      "step": 88608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.727831019656688e-05,
+      "loss": 2.2555,
+      "step": 88616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.727210855430072e-05,
+      "loss": 2.2447,
+      "step": 88624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.726590673124913e-05,
+      "loss": 2.2487,
+      "step": 88632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.725970472756629e-05,
+      "loss": 2.2635,
+      "step": 88640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.725350254340637e-05,
+      "loss": 2.2497,
+      "step": 88648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.724730017892355e-05,
+      "loss": 2.2646,
+      "step": 88656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.724109763427206e-05,
+      "loss": 2.2573,
+      "step": 88664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.723489490960606e-05,
+      "loss": 2.2319,
+      "step": 88672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.722869200507976e-05,
+      "loss": 2.2479,
+      "step": 88680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.722248892084736e-05,
+      "loss": 2.2436,
+      "step": 88688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.721628565706308e-05,
+      "loss": 2.2584,
+      "step": 88696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.721008221388112e-05,
+      "loss": 2.2599,
+      "step": 88704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.720387859145572e-05,
+      "loss": 2.2438,
+      "step": 88712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.719767478994109e-05,
+      "loss": 2.2467,
+      "step": 88720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.719147080949144e-05,
+      "loss": 2.2556,
+      "step": 88728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.718526665026104e-05,
+      "loss": 2.2573,
+      "step": 88736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.71790623124041e-05,
+      "loss": 2.259,
+      "step": 88744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.717285779607485e-05,
+      "loss": 2.237,
+      "step": 88752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.716665310142755e-05,
+      "loss": 2.2583,
+      "step": 88760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.716044822861648e-05,
+      "loss": 2.2594,
+      "step": 88768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.715424317779585e-05,
+      "loss": 2.2587,
+      "step": 88776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.714803794911992e-05,
+      "loss": 2.2373,
+      "step": 88784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.714183254274297e-05,
+      "loss": 2.2413,
+      "step": 88792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.713562695881928e-05,
+      "loss": 2.2512,
+      "step": 88800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.712942119750308e-05,
+      "loss": 2.2581,
+      "step": 88808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.712321525894867e-05,
+      "loss": 2.2469,
+      "step": 88816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.711700914331034e-05,
+      "loss": 2.2575,
+      "step": 88824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.711080285074236e-05,
+      "loss": 2.2496,
+      "step": 88832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.710459638139903e-05,
+      "loss": 2.2676,
+      "step": 88840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.709838973543463e-05,
+      "loss": 2.25,
+      "step": 88848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.709218291300346e-05,
+      "loss": 2.2408,
+      "step": 88856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.708597591425983e-05,
+      "loss": 2.2462,
+      "step": 88864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.707976873935804e-05,
+      "loss": 2.2594,
+      "step": 88872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.707356138845242e-05,
+      "loss": 2.2512,
+      "step": 88880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.706735386169724e-05,
+      "loss": 2.2675,
+      "step": 88888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.706114615924688e-05,
+      "loss": 2.2537,
+      "step": 88896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.705493828125564e-05,
+      "loss": 2.2515,
+      "step": 88904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.704873022787783e-05,
+      "loss": 2.2612,
+      "step": 88912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.704252199926778e-05,
+      "loss": 2.2404,
+      "step": 88920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.703631359557984e-05,
+      "loss": 2.2505,
+      "step": 88928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.703010501696839e-05,
+      "loss": 2.2547,
+      "step": 88936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.702389626358772e-05,
+      "loss": 2.2377,
+      "step": 88944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.70176873355922e-05,
+      "loss": 2.2476,
+      "step": 88952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.70114782331362e-05,
+      "loss": 2.2466,
+      "step": 88960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.700526895637404e-05,
+      "loss": 2.2431,
+      "step": 88968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.699905950546012e-05,
+      "loss": 2.2633,
+      "step": 88976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.699284988054879e-05,
+      "loss": 2.2349,
+      "step": 88984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.698664008179445e-05,
+      "loss": 2.2335,
+      "step": 88992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.698043010935143e-05,
+      "loss": 2.2598,
+      "step": 89000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.697421996337413e-05,
+      "loss": 2.2434,
+      "step": 89008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.696800964401695e-05,
+      "loss": 2.2536,
+      "step": 89016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.696179915143428e-05,
+      "loss": 2.2394,
+      "step": 89024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.695558848578049e-05,
+      "loss": 2.2605,
+      "step": 89032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.694937764720998e-05,
+      "loss": 2.2528,
+      "step": 89040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.694316663587717e-05,
+      "loss": 2.2357,
+      "step": 89048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.693695545193648e-05,
+      "loss": 2.2562,
+      "step": 89056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.693074409554227e-05,
+      "loss": 2.2192,
+      "step": 89064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.692453256684899e-05,
+      "loss": 2.2508,
+      "step": 89072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.691832086601106e-05,
+      "loss": 2.2353,
+      "step": 89080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.69121089931829e-05,
+      "loss": 2.24,
+      "step": 89088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.690589694851893e-05,
+      "loss": 2.2371,
+      "step": 89096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.689968473217356e-05,
+      "loss": 2.2313,
+      "step": 89104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.68934723443013e-05,
+      "loss": 2.2446,
+      "step": 89112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.688725978505652e-05,
+      "loss": 2.246,
+      "step": 89120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.68810470545937e-05,
+      "loss": 2.2629,
+      "step": 89128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.687483415306727e-05,
+      "loss": 2.2416,
+      "step": 89136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.686862108063168e-05,
+      "loss": 2.2382,
+      "step": 89144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.686240783744142e-05,
+      "loss": 2.2433,
+      "step": 89152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.685619442365092e-05,
+      "loss": 2.2367,
+      "step": 89160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.684998083941466e-05,
+      "loss": 2.2364,
+      "step": 89168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.684376708488712e-05,
+      "loss": 2.2636,
+      "step": 89176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.683755316022274e-05,
+      "loss": 2.2455,
+      "step": 89184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.683133906557604e-05,
+      "loss": 2.2599,
+      "step": 89192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.682512480110149e-05,
+      "loss": 2.2509,
+      "step": 89200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.681891036695353e-05,
+      "loss": 2.2545,
+      "step": 89208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.681269576328674e-05,
+      "loss": 2.2553,
+      "step": 89216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.680648099025554e-05,
+      "loss": 2.2571,
+      "step": 89224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.680026604801445e-05,
+      "loss": 2.2478,
+      "step": 89232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.679405093671802e-05,
+      "loss": 2.2516,
+      "step": 89240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.678783565652072e-05,
+      "loss": 2.2462,
+      "step": 89248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.678162020757704e-05,
+      "loss": 2.2486,
+      "step": 89256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.677540459004152e-05,
+      "loss": 2.2668,
+      "step": 89264
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.676918880406869e-05,
+      "loss": 2.2289,
+      "step": 89272
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.676297284981306e-05,
+      "loss": 2.25,
+      "step": 89280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.675675672742918e-05,
+      "loss": 2.2582,
+      "step": 89288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.675054043707157e-05,
+      "loss": 2.2513,
+      "step": 89296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.674432397889475e-05,
+      "loss": 2.2748,
+      "step": 89304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.67381073530533e-05,
+      "loss": 2.2593,
+      "step": 89312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.673189055970173e-05,
+      "loss": 2.2586,
+      "step": 89320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.672567359899461e-05,
+      "loss": 2.2564,
+      "step": 89328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.671945647108648e-05,
+      "loss": 2.2375,
+      "step": 89336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.671323917613193e-05,
+      "loss": 2.2498,
+      "step": 89344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.670702171428549e-05,
+      "loss": 2.252,
+      "step": 89352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.670080408570172e-05,
+      "loss": 2.2511,
+      "step": 89360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.669458629053524e-05,
+      "loss": 2.2599,
+      "step": 89368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.668836832894059e-05,
+      "loss": 2.2618,
+      "step": 89376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.668215020107232e-05,
+      "loss": 2.2562,
+      "step": 89384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.667593190708508e-05,
+      "loss": 2.2645,
+      "step": 89392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.66697134471334e-05,
+      "loss": 2.2515,
+      "step": 89400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.666349482137191e-05,
+      "loss": 2.258,
+      "step": 89408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.665727602995518e-05,
+      "loss": 2.2627,
+      "step": 89416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.665105707303783e-05,
+      "loss": 2.2532,
+      "step": 89424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.664483795077448e-05,
+      "loss": 2.2698,
+      "step": 89432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.663861866331968e-05,
+      "loss": 2.2669,
+      "step": 89440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.663239921082807e-05,
+      "loss": 2.2682,
+      "step": 89448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.662617959345428e-05,
+      "loss": 2.2668,
+      "step": 89456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.661995981135293e-05,
+      "loss": 2.2603,
+      "step": 89464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.661373986467863e-05,
+      "loss": 2.2788,
+      "step": 89472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.660751975358602e-05,
+      "loss": 2.2575,
+      "step": 89480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.660129947822973e-05,
+      "loss": 2.2544,
+      "step": 89488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.659507903876438e-05,
+      "loss": 2.2669,
+      "step": 89496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.658885843534465e-05,
+      "loss": 2.2611,
+      "step": 89504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.658263766812514e-05,
+      "loss": 2.2624,
+      "step": 89512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.657641673726051e-05,
+      "loss": 2.2461,
+      "step": 89520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.657019564290545e-05,
+      "loss": 2.2479,
+      "step": 89528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.656397438521458e-05,
+      "loss": 2.2432,
+      "step": 89536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.655775296434255e-05,
+      "loss": 2.2706,
+      "step": 89544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.655153138044408e-05,
+      "loss": 2.2487,
+      "step": 89552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.654530963367377e-05,
+      "loss": 2.2424,
+      "step": 89560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.653908772418633e-05,
+      "loss": 2.2381,
+      "step": 89568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.653286565213646e-05,
+      "loss": 2.2856,
+      "step": 89576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.65266434176788e-05,
+      "loss": 2.2515,
+      "step": 89584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.652042102096806e-05,
+      "loss": 2.2476,
+      "step": 89592
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.651419846215892e-05,
+      "loss": 2.2554,
+      "step": 89600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.650797574140608e-05,
+      "loss": 2.259,
+      "step": 89608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.650175285886422e-05,
+      "loss": 2.2593,
+      "step": 89616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.649552981468807e-05,
+      "loss": 2.2597,
+      "step": 89624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.648930660903229e-05,
+      "loss": 2.2667,
+      "step": 89632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.648308324205166e-05,
+      "loss": 2.2625,
+      "step": 89640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.647685971390082e-05,
+      "loss": 2.2627,
+      "step": 89648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.647063602473455e-05,
+      "loss": 2.2546,
+      "step": 89656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.646441217470752e-05,
+      "loss": 2.2482,
+      "step": 89664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.64581881639745e-05,
+      "loss": 2.244,
+      "step": 89672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.645196399269018e-05,
+      "loss": 2.2681,
+      "step": 89680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.644573966100933e-05,
+      "loss": 2.246,
+      "step": 89688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.643951516908664e-05,
+      "loss": 2.2553,
+      "step": 89696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.64332905170769e-05,
+      "loss": 2.2591,
+      "step": 89704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.642706570513482e-05,
+      "loss": 2.2474,
+      "step": 89712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.64208407334152e-05,
+      "loss": 2.2395,
+      "step": 89720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.641461560207271e-05,
+      "loss": 2.2561,
+      "step": 89728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.640839031126218e-05,
+      "loss": 2.2521,
+      "step": 89736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.640216486113835e-05,
+      "loss": 2.2485,
+      "step": 89744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.639593925185598e-05,
+      "loss": 2.2581,
+      "step": 89752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.638971348356986e-05,
+      "loss": 2.2745,
+      "step": 89760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.638348755643471e-05,
+      "loss": 2.2711,
+      "step": 89768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.637726147060539e-05,
+      "loss": 2.2642,
+      "step": 89776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.63710352262366e-05,
+      "loss": 2.2739,
+      "step": 89784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.636480882348317e-05,
+      "loss": 2.2619,
+      "step": 89792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.635858226249987e-05,
+      "loss": 2.2484,
+      "step": 89800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.635235554344151e-05,
+      "loss": 2.2572,
+      "step": 89808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.634612866646286e-05,
+      "loss": 2.2527,
+      "step": 89816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.633990163171877e-05,
+      "loss": 2.2672,
+      "step": 89824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.6333674439364e-05,
+      "loss": 2.2598,
+      "step": 89832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.632744708955337e-05,
+      "loss": 2.2577,
+      "step": 89840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.632121958244171e-05,
+      "loss": 2.2706,
+      "step": 89848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.63149919181838e-05,
+      "loss": 2.2663,
+      "step": 89856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.630876409693449e-05,
+      "loss": 2.2628,
+      "step": 89864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.63025361188486e-05,
+      "loss": 2.273,
+      "step": 89872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.629630798408095e-05,
+      "loss": 2.259,
+      "step": 89880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.629007969278638e-05,
+      "loss": 2.2721,
+      "step": 89888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.628385124511974e-05,
+      "loss": 2.255,
+      "step": 89896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.627762264123584e-05,
+      "loss": 2.2551,
+      "step": 89904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.627139388128953e-05,
+      "loss": 2.2521,
+      "step": 89912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.626516496543567e-05,
+      "loss": 2.2647,
+      "step": 89920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.62589358938291e-05,
+      "loss": 2.2563,
+      "step": 89928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.625270666662469e-05,
+      "loss": 2.2548,
+      "step": 89936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.624647728397728e-05,
+      "loss": 2.2639,
+      "step": 89944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.624024774604174e-05,
+      "loss": 2.2726,
+      "step": 89952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.623401805297295e-05,
+      "loss": 2.2561,
+      "step": 89960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.62277882049258e-05,
+      "loss": 2.2591,
+      "step": 89968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.622155820205508e-05,
+      "loss": 2.2468,
+      "step": 89976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.621532804451576e-05,
+      "loss": 2.2632,
+      "step": 89984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.620909773246268e-05,
+      "loss": 2.2732,
+      "step": 89992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.620286726605074e-05,
+      "loss": 2.2456,
+      "step": 90000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.619663664543483e-05,
+      "loss": 2.2404,
+      "step": 90008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.619040587076983e-05,
+      "loss": 2.2602,
+      "step": 90016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.618417494221064e-05,
+      "loss": 2.2589,
+      "step": 90024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.617794385991217e-05,
+      "loss": 2.2518,
+      "step": 90032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.617171262402933e-05,
+      "loss": 2.2503,
+      "step": 90040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.616548123471701e-05,
+      "loss": 2.2544,
+      "step": 90048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.615924969213015e-05,
+      "loss": 2.2615,
+      "step": 90056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.615301799642362e-05,
+      "loss": 2.2545,
+      "step": 90064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.61467861477524e-05,
+      "loss": 2.2421,
+      "step": 90072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.614055414627138e-05,
+      "loss": 2.2549,
+      "step": 90080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.61343219921355e-05,
+      "loss": 2.2602,
+      "step": 90088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.612808968549967e-05,
+      "loss": 2.2572,
+      "step": 90096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.612185722651885e-05,
+      "loss": 2.2263,
+      "step": 90104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.6115624615348e-05,
+      "loss": 2.2541,
+      "step": 90112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.6109391852142e-05,
+      "loss": 2.2563,
+      "step": 90120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.610315893705583e-05,
+      "loss": 2.2301,
+      "step": 90128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.609692587024447e-05,
+      "loss": 2.2537,
+      "step": 90136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.609069265186284e-05,
+      "loss": 2.2483,
+      "step": 90144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.608445928206592e-05,
+      "loss": 2.2467,
+      "step": 90152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.607822576100864e-05,
+      "loss": 2.2478,
+      "step": 90160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.607199208884598e-05,
+      "loss": 2.2527,
+      "step": 90168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.606575826573293e-05,
+      "loss": 2.2425,
+      "step": 90176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.605952429182445e-05,
+      "loss": 2.2639,
+      "step": 90184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.60532901672755e-05,
+      "loss": 2.2525,
+      "step": 90192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.60470558922411e-05,
+      "loss": 2.266,
+      "step": 90200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.60408214668762e-05,
+      "loss": 2.2418,
+      "step": 90208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.603458689133582e-05,
+      "loss": 2.2467,
+      "step": 90216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.60283521657749e-05,
+      "loss": 2.2491,
+      "step": 90224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.602211729034848e-05,
+      "loss": 2.2532,
+      "step": 90232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.601588226521156e-05,
+      "loss": 2.2487,
+      "step": 90240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.600964709051912e-05,
+      "loss": 2.2591,
+      "step": 90248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.60034117664262e-05,
+      "loss": 2.2445,
+      "step": 90256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.599717629308775e-05,
+      "loss": 2.2421,
+      "step": 90264
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.599094067065885e-05,
+      "loss": 2.2624,
+      "step": 90272
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.598470489929449e-05,
+      "loss": 2.2557,
+      "step": 90280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.59784689791497e-05,
+      "loss": 2.2502,
+      "step": 90288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.597223291037948e-05,
+      "loss": 2.2724,
+      "step": 90296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.596599669313889e-05,
+      "loss": 2.2573,
+      "step": 90304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.595976032758298e-05,
+      "loss": 2.2646,
+      "step": 90312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.595352381386672e-05,
+      "loss": 2.2539,
+      "step": 90320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.594728715214523e-05,
+      "loss": 2.2608,
+      "step": 90328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.594105034257349e-05,
+      "loss": 2.2624,
+      "step": 90336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.593481338530656e-05,
+      "loss": 2.2394,
+      "step": 90344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.592857628049953e-05,
+      "loss": 2.2452,
+      "step": 90352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.59223390283074e-05,
+      "loss": 2.2507,
+      "step": 90360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.591610162888529e-05,
+      "loss": 2.2566,
+      "step": 90368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.59098640823882e-05,
+      "loss": 2.2519,
+      "step": 90376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.590362638897125e-05,
+      "loss": 2.2633,
+      "step": 90384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.589738854878948e-05,
+      "loss": 2.2472,
+      "step": 90392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.589115056199795e-05,
+      "loss": 2.2508,
+      "step": 90400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.588491242875177e-05,
+      "loss": 2.2558,
+      "step": 90408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.5878674149206e-05,
+      "loss": 2.2418,
+      "step": 90416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.587243572351574e-05,
+      "loss": 2.2662,
+      "step": 90424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.586619715183606e-05,
+      "loss": 2.2575,
+      "step": 90432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.585995843432207e-05,
+      "loss": 2.2446,
+      "step": 90440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.585371957112886e-05,
+      "loss": 2.2582,
+      "step": 90448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.584748056241151e-05,
+      "loss": 2.2298,
+      "step": 90456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.584124140832516e-05,
+      "loss": 2.2617,
+      "step": 90464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.583500210902485e-05,
+      "loss": 2.2464,
+      "step": 90472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.582876266466578e-05,
+      "loss": 2.264,
+      "step": 90480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.5822523075403e-05,
+      "loss": 2.2327,
+      "step": 90488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.581628334139163e-05,
+      "loss": 2.2457,
+      "step": 90496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.58100434627868e-05,
+      "loss": 2.2387,
+      "step": 90504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.580380343974366e-05,
+      "loss": 2.2358,
+      "step": 90512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.579756327241731e-05,
+      "loss": 2.2348,
+      "step": 90520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.579132296096289e-05,
+      "loss": 2.2501,
+      "step": 90528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.57850825055355e-05,
+      "loss": 2.2623,
+      "step": 90536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.577884190629036e-05,
+      "loss": 2.2646,
+      "step": 90544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.577260116338254e-05,
+      "loss": 2.2576,
+      "step": 90552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.576636027696719e-05,
+      "loss": 2.2555,
+      "step": 90560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.57601192471995e-05,
+      "loss": 2.2457,
+      "step": 90568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.575387807423457e-05,
+      "loss": 2.2519,
+      "step": 90576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.574763675822761e-05,
+      "loss": 2.2541,
+      "step": 90584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.574139529933375e-05,
+      "loss": 2.2502,
+      "step": 90592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.573515369770815e-05,
+      "loss": 2.2622,
+      "step": 90600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.572891195350602e-05,
+      "loss": 2.262,
+      "step": 90608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.572267006688245e-05,
+      "loss": 2.2409,
+      "step": 90616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.571642803799268e-05,
+      "loss": 2.2585,
+      "step": 90624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.571018586699187e-05,
+      "loss": 2.2489,
+      "step": 90632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.57039435540352e-05,
+      "loss": 2.2711,
+      "step": 90640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.569770109927784e-05,
+      "loss": 2.2616,
+      "step": 90648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.569145850287501e-05,
+      "loss": 2.2562,
+      "step": 90656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.568521576498188e-05,
+      "loss": 2.2584,
+      "step": 90664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.567897288575362e-05,
+      "loss": 2.2681,
+      "step": 90672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.567272986534549e-05,
+      "loss": 2.2672,
+      "step": 90680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.566648670391265e-05,
+      "loss": 2.2489,
+      "step": 90688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.566024340161031e-05,
+      "loss": 2.2534,
+      "step": 90696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.565399995859369e-05,
+      "loss": 2.2539,
+      "step": 90704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.564775637501799e-05,
+      "loss": 2.2631,
+      "step": 90712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.564151265103845e-05,
+      "loss": 2.2616,
+      "step": 90720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.563526878681024e-05,
+      "loss": 2.2662,
+      "step": 90728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.562902478248865e-05,
+      "loss": 2.263,
+      "step": 90736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.562278063822885e-05,
+      "loss": 2.2616,
+      "step": 90744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.56165363541861e-05,
+      "loss": 2.2581,
+      "step": 90752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.561029193051563e-05,
+      "loss": 2.2538,
+      "step": 90760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.560404736737267e-05,
+      "loss": 2.2433,
+      "step": 90768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.559780266491247e-05,
+      "loss": 2.2618,
+      "step": 90776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.559155782329025e-05,
+      "loss": 2.2618,
+      "step": 90784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.558531284266127e-05,
+      "loss": 2.2606,
+      "step": 90792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.557906772318083e-05,
+      "loss": 2.2572,
+      "step": 90800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.557282246500409e-05,
+      "loss": 2.262,
+      "step": 90808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.556657706828639e-05,
+      "loss": 2.2443,
+      "step": 90816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.556033153318294e-05,
+      "loss": 2.2463,
+      "step": 90824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.555408585984903e-05,
+      "loss": 2.2585,
+      "step": 90832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.554784004843992e-05,
+      "loss": 2.2485,
+      "step": 90840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.554159409911089e-05,
+      "loss": 2.2586,
+      "step": 90848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.553534801201722e-05,
+      "loss": 2.2453,
+      "step": 90856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.552910178731418e-05,
+      "loss": 2.2657,
+      "step": 90864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.552285542515703e-05,
+      "loss": 2.2582,
+      "step": 90872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.551660892570107e-05,
+      "loss": 2.2617,
+      "step": 90880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.55103622891016e-05,
+      "loss": 2.2413,
+      "step": 90888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.550411551551392e-05,
+      "loss": 2.2596,
+      "step": 90896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.54978686050933e-05,
+      "loss": 2.2666,
+      "step": 90904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.549162155799504e-05,
+      "loss": 2.2418,
+      "step": 90912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.548537437437448e-05,
+      "loss": 2.2597,
+      "step": 90920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.547912705438687e-05,
+      "loss": 2.251,
+      "step": 90928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.547287959818755e-05,
+      "loss": 2.243,
+      "step": 90936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.546663200593184e-05,
+      "loss": 2.2495,
+      "step": 90944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.546038427777503e-05,
+      "loss": 2.2585,
+      "step": 90952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.545413641387248e-05,
+      "loss": 2.2422,
+      "step": 90960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.544788841437946e-05,
+      "loss": 2.2587,
+      "step": 90968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.544164027945134e-05,
+      "loss": 2.237,
+      "step": 90976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.543539200924343e-05,
+      "loss": 2.2498,
+      "step": 90984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.542914360391104e-05,
+      "loss": 2.237,
+      "step": 90992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.542289506360954e-05,
+      "loss": 2.2562,
+      "step": 91000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.54166463884943e-05,
+      "loss": 2.2598,
+      "step": 91008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.541039757872058e-05,
+      "loss": 2.2562,
+      "step": 91016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.54041486344438e-05,
+      "loss": 2.2584,
+      "step": 91024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.539789955581924e-05,
+      "loss": 2.2605,
+      "step": 91032
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.539165034300232e-05,
+      "loss": 2.2469,
+      "step": 91040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.538540099614837e-05,
+      "loss": 2.2469,
+      "step": 91048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.537915151541274e-05,
+      "loss": 2.2518,
+      "step": 91056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.537290190095078e-05,
+      "loss": 2.247,
+      "step": 91064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.53666521529179e-05,
+      "loss": 2.2748,
+      "step": 91072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.536040227146943e-05,
+      "loss": 2.2673,
+      "step": 91080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.535415225676079e-05,
+      "loss": 2.2672,
+      "step": 91088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.534790210894729e-05,
+      "loss": 2.2534,
+      "step": 91096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.534165182818435e-05,
+      "loss": 2.2687,
+      "step": 91104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.533540141462735e-05,
+      "loss": 2.2412,
+      "step": 91112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.532915086843168e-05,
+      "loss": 2.2615,
+      "step": 91120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.532290018975272e-05,
+      "loss": 2.2617,
+      "step": 91128
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.531664937874587e-05,
+      "loss": 2.2615,
+      "step": 91136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.531039843556649e-05,
+      "loss": 2.2505,
+      "step": 91144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.530414736037002e-05,
+      "loss": 2.2545,
+      "step": 91152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.529789615331187e-05,
+      "loss": 2.2488,
+      "step": 91160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.529164481454741e-05,
+      "loss": 2.2429,
+      "step": 91168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.528539334423207e-05,
+      "loss": 2.2661,
+      "step": 91176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.527914174252125e-05,
+      "loss": 2.2543,
+      "step": 91184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.52728900095704e-05,
+      "loss": 2.2649,
+      "step": 91192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.526663814553489e-05,
+      "loss": 2.2458,
+      "step": 91200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.52603861505702e-05,
+      "loss": 2.246,
+      "step": 91208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.525413402483167e-05,
+      "loss": 2.249,
+      "step": 91216
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.52478817684748e-05,
+      "loss": 2.2421,
+      "step": 91224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.524162938165501e-05,
+      "loss": 2.2598,
+      "step": 91232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.523537686452772e-05,
+      "loss": 2.2533,
+      "step": 91240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.522912421724837e-05,
+      "loss": 2.2544,
+      "step": 91248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.52228714399724e-05,
+      "loss": 2.2399,
+      "step": 91256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.521661853285527e-05,
+      "loss": 2.2487,
+      "step": 91264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.521036549605242e-05,
+      "loss": 2.2577,
+      "step": 91272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.520411232971928e-05,
+      "loss": 2.2741,
+      "step": 91280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.519785903401134e-05,
+      "loss": 2.2483,
+      "step": 91288
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.519160560908403e-05,
+      "loss": 2.2651,
+      "step": 91296
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.518535205509284e-05,
+      "loss": 2.2401,
+      "step": 91304
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.517909837219321e-05,
+      "loss": 2.2394,
+      "step": 91312
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.517284456054061e-05,
+      "loss": 2.244,
+      "step": 91320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.516659062029051e-05,
+      "loss": 2.2506,
+      "step": 91328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.51603365515984e-05,
+      "loss": 2.2694,
+      "step": 91336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.515408235461973e-05,
+      "loss": 2.2443,
+      "step": 91344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.514782802951001e-05,
+      "loss": 2.2475,
+      "step": 91352
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.51415735764247e-05,
+      "loss": 2.2534,
+      "step": 91360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.513531899551928e-05,
+      "loss": 2.2592,
+      "step": 91368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.512906428694926e-05,
+      "loss": 2.2477,
+      "step": 91376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.512280945087015e-05,
+      "loss": 2.2652,
+      "step": 91384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.511655448743739e-05,
+      "loss": 2.2405,
+      "step": 91392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.511029939680653e-05,
+      "loss": 2.244,
+      "step": 91400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.510404417913305e-05,
+      "loss": 2.2566,
+      "step": 91408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.509778883457245e-05,
+      "loss": 2.2619,
+      "step": 91416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.509153336328026e-05,
+      "loss": 2.2429,
+      "step": 91424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.508527776541198e-05,
+      "loss": 2.2532,
+      "step": 91432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.507902204112311e-05,
+      "loss": 2.2421,
+      "step": 91440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.507276619056919e-05,
+      "loss": 2.2536,
+      "step": 91448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.506651021390574e-05,
+      "loss": 2.2626,
+      "step": 91456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.506025411128826e-05,
+      "loss": 2.262,
+      "step": 91464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.505399788287232e-05,
+      "loss": 2.2518,
+      "step": 91472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.504774152881341e-05,
+      "loss": 2.2454,
+      "step": 91480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.504148504926708e-05,
+      "loss": 2.2568,
+      "step": 91488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.503522844438887e-05,
+      "loss": 2.2374,
+      "step": 91496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.502897171433431e-05,
+      "loss": 2.2515,
+      "step": 91504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.502271485925896e-05,
+      "loss": 2.2617,
+      "step": 91512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.501645787931835e-05,
+      "loss": 2.256,
+      "step": 91520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.501020077466801e-05,
+      "loss": 2.2555,
+      "step": 91528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.500394354546355e-05,
+      "loss": 2.2718,
+      "step": 91536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.499768619186048e-05,
+      "loss": 2.2511,
+      "step": 91544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.499142871401438e-05,
+      "loss": 2.2437,
+      "step": 91552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.49851711120808e-05,
+      "loss": 2.2555,
+      "step": 91560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.49789133862153e-05,
+      "loss": 2.2329,
+      "step": 91568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.497265553657346e-05,
+      "loss": 2.2479,
+      "step": 91576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.496639756331085e-05,
+      "loss": 2.2549,
+      "step": 91584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.4960139466583e-05,
+      "loss": 2.255,
+      "step": 91592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.495388124654557e-05,
+      "loss": 2.241,
+      "step": 91600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.494762290335407e-05,
+      "loss": 2.2673,
+      "step": 91608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.494136443716415e-05,
+      "loss": 2.253,
+      "step": 91616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.49351058481313e-05,
+      "loss": 2.2477,
+      "step": 91624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.49288471364112e-05,
+      "loss": 2.2416,
+      "step": 91632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.49225883021594e-05,
+      "loss": 2.2253,
+      "step": 91640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.49163293455315e-05,
+      "loss": 2.2519,
+      "step": 91648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.49100702666831e-05,
+      "loss": 2.2663,
+      "step": 91656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.490381106576981e-05,
+      "loss": 2.2511,
+      "step": 91664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.489755174294723e-05,
+      "loss": 2.2608,
+      "step": 91672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.489129229837095e-05,
+      "loss": 2.2571,
+      "step": 91680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.488503273219659e-05,
+      "loss": 2.2785,
+      "step": 91688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.487877304457979e-05,
+      "loss": 2.2416,
+      "step": 91696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.487251323567614e-05,
+      "loss": 2.2478,
+      "step": 91704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.486625330564125e-05,
+      "loss": 2.2487,
+      "step": 91712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.485999325463078e-05,
+      "loss": 2.2738,
+      "step": 91720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.48537330828003e-05,
+      "loss": 2.2456,
+      "step": 91728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.484747279030549e-05,
+      "loss": 2.2492,
+      "step": 91736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.484121237730197e-05,
+      "loss": 2.2612,
+      "step": 91744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.483495184394535e-05,
+      "loss": 2.2635,
+      "step": 91752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.48286911903913e-05,
+      "loss": 2.2776,
+      "step": 91760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.482243041679544e-05,
+      "loss": 2.2642,
+      "step": 91768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.481616952331342e-05,
+      "loss": 2.2595,
+      "step": 91776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.480990851010087e-05,
+      "loss": 2.2689,
+      "step": 91784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.480364737731348e-05,
+      "loss": 2.2689,
+      "step": 91792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.479738612510686e-05,
+      "loss": 2.2621,
+      "step": 91800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.479112475363668e-05,
+      "loss": 2.2716,
+      "step": 91808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.478486326305862e-05,
+      "loss": 2.2646,
+      "step": 91816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.477860165352829e-05,
+      "loss": 2.2612,
+      "step": 91824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.477233992520141e-05,
+      "loss": 2.2553,
+      "step": 91832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.476607807823363e-05,
+      "loss": 2.2425,
+      "step": 91840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.47598161127806e-05,
+      "loss": 2.2754,
+      "step": 91848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.475355402899803e-05,
+      "loss": 2.2653,
+      "step": 91856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.474729182704154e-05,
+      "loss": 2.2531,
+      "step": 91864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.474102950706687e-05,
+      "loss": 2.2523,
+      "step": 91872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.473476706922965e-05,
+      "loss": 2.2529,
+      "step": 91880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.472850451368561e-05,
+      "loss": 2.2487,
+      "step": 91888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.472224184059041e-05,
+      "loss": 2.261,
+      "step": 91896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.471597905009973e-05,
+      "loss": 2.2708,
+      "step": 91904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.470971614236932e-05,
+      "loss": 2.2671,
+      "step": 91912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.470345311755481e-05,
+      "loss": 2.2677,
+      "step": 91920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.469718997581194e-05,
+      "loss": 2.265,
+      "step": 91928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.46909267172964e-05,
+      "loss": 2.253,
+      "step": 91936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.468466334216389e-05,
+      "loss": 2.2556,
+      "step": 91944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.467839985057014e-05,
+      "loss": 2.2702,
+      "step": 91952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.467213624267081e-05,
+      "loss": 2.2499,
+      "step": 91960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.466587251862166e-05,
+      "loss": 2.2737,
+      "step": 91968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.465960867857842e-05,
+      "loss": 2.2476,
+      "step": 91976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.465334472269678e-05,
+      "loss": 2.2559,
+      "step": 91984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.464708065113244e-05,
+      "loss": 2.2573,
+      "step": 91992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.464081646404117e-05,
+      "loss": 2.2727,
+      "step": 92000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.463455216157869e-05,
+      "loss": 2.2535,
+      "step": 92008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.462828774390072e-05,
+      "loss": 2.2541,
+      "step": 92016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.462202321116299e-05,
+      "loss": 2.2557,
+      "step": 92024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.461575856352125e-05,
+      "loss": 2.2548,
+      "step": 92032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.460949380113124e-05,
+      "loss": 2.2563,
+      "step": 92040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.460322892414869e-05,
+      "loss": 2.2596,
+      "step": 92048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.459696393272935e-05,
+      "loss": 2.2529,
+      "step": 92056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.459069882702898e-05,
+      "loss": 2.2592,
+      "step": 92064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.45844336072033e-05,
+      "loss": 2.2658,
+      "step": 92072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.45781682734081e-05,
+      "loss": 2.26,
+      "step": 92080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.457190282579912e-05,
+      "loss": 2.262,
+      "step": 92088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.456563726453214e-05,
+      "loss": 2.2519,
+      "step": 92096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.455937158976288e-05,
+      "loss": 2.263,
+      "step": 92104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.455310580164715e-05,
+      "loss": 2.2657,
+      "step": 92112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.45468399003407e-05,
+      "loss": 2.2758,
+      "step": 92120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.454057388599928e-05,
+      "loss": 2.258,
+      "step": 92128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.45343077587787e-05,
+      "loss": 2.2445,
+      "step": 92136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.452804151883472e-05,
+      "loss": 2.273,
+      "step": 92144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.452177516632311e-05,
+      "loss": 2.2591,
+      "step": 92152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.451550870139966e-05,
+      "loss": 2.251,
+      "step": 92160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.450924212422018e-05,
+      "loss": 2.2539,
+      "step": 92168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.450297543494042e-05,
+      "loss": 2.2845,
+      "step": 92176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.449670863371619e-05,
+      "loss": 2.2644,
+      "step": 92184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.449044172070327e-05,
+      "loss": 2.2863,
+      "step": 92192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.448417469605748e-05,
+      "loss": 2.2655,
+      "step": 92200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.447790755993458e-05,
+      "loss": 2.2621,
+      "step": 92208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.447164031249041e-05,
+      "loss": 2.2693,
+      "step": 92216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.446537295388076e-05,
+      "loss": 2.2672,
+      "step": 92224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.445910548426144e-05,
+      "loss": 2.2695,
+      "step": 92232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.445283790378826e-05,
+      "loss": 2.2714,
+      "step": 92240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.444657021261702e-05,
+      "loss": 2.2527,
+      "step": 92248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.444030241090354e-05,
+      "loss": 2.2721,
+      "step": 92256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.443403449880366e-05,
+      "loss": 2.2471,
+      "step": 92264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.442776647647318e-05,
+      "loss": 2.2683,
+      "step": 92272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.442149834406791e-05,
+      "loss": 2.2676,
+      "step": 92280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.441523010174373e-05,
+      "loss": 2.259,
+      "step": 92288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.44089617496564e-05,
+      "loss": 2.2781,
+      "step": 92296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.440269328796183e-05,
+      "loss": 2.2472,
+      "step": 92304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.439642471681576e-05,
+      "loss": 2.2544,
+      "step": 92312
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.43901560363741e-05,
+      "loss": 2.2435,
+      "step": 92320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.43838872467927e-05,
+      "loss": 2.2627,
+      "step": 92328
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.437761834822731e-05,
+      "loss": 2.259,
+      "step": 92336
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.437134934083389e-05,
+      "loss": 2.267,
+      "step": 92344
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.436508022476819e-05,
+      "loss": 2.2525,
+      "step": 92352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.435881100018615e-05,
+      "loss": 2.2419,
+      "step": 92360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.435254166724354e-05,
+      "loss": 2.2604,
+      "step": 92368
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.434627222609628e-05,
+      "loss": 2.2699,
+      "step": 92376
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.434000267690022e-05,
+      "loss": 2.2584,
+      "step": 92384
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.433373301981119e-05,
+      "loss": 2.2637,
+      "step": 92392
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.432746325498504e-05,
+      "loss": 2.2513,
+      "step": 92400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.432119338257769e-05,
+      "loss": 2.2522,
+      "step": 92408
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.431492340274501e-05,
+      "loss": 2.2542,
+      "step": 92416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.430865331564283e-05,
+      "loss": 2.2665,
+      "step": 92424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.430238312142703e-05,
+      "loss": 2.2506,
+      "step": 92432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.429611282025354e-05,
+      "loss": 2.2626,
+      "step": 92440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.428984241227818e-05,
+      "loss": 2.2513,
+      "step": 92448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.428357189765686e-05,
+      "loss": 2.2664,
+      "step": 92456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.427730127654545e-05,
+      "loss": 2.2445,
+      "step": 92464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.427103054909987e-05,
+      "loss": 2.2516,
+      "step": 92472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.426475971547599e-05,
+      "loss": 2.2583,
+      "step": 92480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.42584887758297e-05,
+      "loss": 2.2459,
+      "step": 92488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.425221773031688e-05,
+      "loss": 2.2578,
+      "step": 92496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.424594657909349e-05,
+      "loss": 2.2714,
+      "step": 92504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.423967532231537e-05,
+      "loss": 2.2458,
+      "step": 92512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.423340396013844e-05,
+      "loss": 2.2482,
+      "step": 92520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.422713249271861e-05,
+      "loss": 2.2586,
+      "step": 92528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.42208609202118e-05,
+      "loss": 2.254,
+      "step": 92536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.421458924277392e-05,
+      "loss": 2.2448,
+      "step": 92544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.420831746056087e-05,
+      "loss": 2.2492,
+      "step": 92552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.42020455737286e-05,
+      "loss": 2.2511,
+      "step": 92560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.419577358243297e-05,
+      "loss": 2.2636,
+      "step": 92568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.418950148682996e-05,
+      "loss": 2.2476,
+      "step": 92576
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.418322928707547e-05,
+      "loss": 2.2542,
+      "step": 92584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.417695698332542e-05,
+      "loss": 2.2439,
+      "step": 92592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.417068457573576e-05,
+      "loss": 2.2426,
+      "step": 92600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.416441206446242e-05,
+      "loss": 2.2564,
+      "step": 92608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.415813944966131e-05,
+      "loss": 2.2429,
+      "step": 92616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.41518667314884e-05,
+      "loss": 2.2496,
+      "step": 92624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.414559391009962e-05,
+      "loss": 2.2456,
+      "step": 92632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.41393209856509e-05,
+      "loss": 2.2406,
+      "step": 92640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.41330479582982e-05,
+      "loss": 2.2536,
+      "step": 92648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.412677482819747e-05,
+      "loss": 2.2678,
+      "step": 92656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.412050159550465e-05,
+      "loss": 2.2784,
+      "step": 92664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.411422826037569e-05,
+      "loss": 2.262,
+      "step": 92672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.410795482296657e-05,
+      "loss": 2.2598,
+      "step": 92680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.410168128343321e-05,
+      "loss": 2.2632,
+      "step": 92688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.409540764193162e-05,
+      "loss": 2.2733,
+      "step": 92696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.408913389861773e-05,
+      "loss": 2.2506,
+      "step": 92704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.408286005364748e-05,
+      "loss": 2.2571,
+      "step": 92712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.407658610717691e-05,
+      "loss": 2.2654,
+      "step": 92720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.407031205936194e-05,
+      "loss": 2.2551,
+      "step": 92728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.406403791035854e-05,
+      "loss": 2.2564,
+      "step": 92736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.405776366032272e-05,
+      "loss": 2.2469,
+      "step": 92744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.405148930941045e-05,
+      "loss": 2.2706,
+      "step": 92752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.404521485777768e-05,
+      "loss": 2.2557,
+      "step": 92760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.403894030558041e-05,
+      "loss": 2.2528,
+      "step": 92768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.403266565297465e-05,
+      "loss": 2.2461,
+      "step": 92776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.402639090011635e-05,
+      "loss": 2.2598,
+      "step": 92784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.402011604716155e-05,
+      "loss": 2.2505,
+      "step": 92792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.401384109426617e-05,
+      "loss": 2.2397,
+      "step": 92800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.40075660415863e-05,
+      "loss": 2.2677,
+      "step": 92808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.400129088927786e-05,
+      "loss": 2.2451,
+      "step": 92816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.399501563749689e-05,
+      "loss": 2.2598,
+      "step": 92824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.398874028639937e-05,
+      "loss": 2.2551,
+      "step": 92832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.398246483614132e-05,
+      "loss": 2.2553,
+      "step": 92840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.397618928687876e-05,
+      "loss": 2.2662,
+      "step": 92848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.396991363876769e-05,
+      "loss": 2.2481,
+      "step": 92856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.39636378919641e-05,
+      "loss": 2.2483,
+      "step": 92864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.395736204662406e-05,
+      "loss": 2.2477,
+      "step": 92872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.395108610290352e-05,
+      "loss": 2.2438,
+      "step": 92880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.394481006095856e-05,
+      "loss": 2.2362,
+      "step": 92888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.393853392094516e-05,
+      "loss": 2.2562,
+      "step": 92896
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.393225768301938e-05,
+      "loss": 2.2567,
+      "step": 92904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.392598134733724e-05,
+      "loss": 2.246,
+      "step": 92912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.391970491405475e-05,
+      "loss": 2.2557,
+      "step": 92920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.391342838332795e-05,
+      "loss": 2.2399,
+      "step": 92928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.390715175531289e-05,
+      "loss": 2.268,
+      "step": 92936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.39008750301656e-05,
+      "loss": 2.2422,
+      "step": 92944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.389459820804214e-05,
+      "loss": 2.2472,
+      "step": 92952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.388832128909848e-05,
+      "loss": 2.2588,
+      "step": 92960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.388204427349075e-05,
+      "loss": 2.2472,
+      "step": 92968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.387576716137496e-05,
+      "loss": 2.2585,
+      "step": 92976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.386948995290718e-05,
+      "loss": 2.2456,
+      "step": 92984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.386321264824342e-05,
+      "loss": 2.2468,
+      "step": 92992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.385693524753978e-05,
+      "loss": 2.2618,
+      "step": 93000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.385065775095229e-05,
+      "loss": 2.2501,
+      "step": 93008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.384438015863701e-05,
+      "loss": 2.2313,
+      "step": 93016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.383810247074999e-05,
+      "loss": 2.2444,
+      "step": 93024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.383182468744735e-05,
+      "loss": 2.2472,
+      "step": 93032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.38255468088851e-05,
+      "loss": 2.2561,
+      "step": 93040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.381926883521931e-05,
+      "loss": 2.2559,
+      "step": 93048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.381299076660609e-05,
+      "loss": 2.2648,
+      "step": 93056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.38067126032015e-05,
+      "loss": 2.2468,
+      "step": 93064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.380043434516158e-05,
+      "loss": 2.2597,
+      "step": 93072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.379415599264247e-05,
+      "loss": 2.2305,
+      "step": 93080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.37878775458002e-05,
+      "loss": 2.2346,
+      "step": 93088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.378159900479086e-05,
+      "loss": 2.2535,
+      "step": 93096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.377532036977055e-05,
+      "loss": 2.2288,
+      "step": 93104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.376904164089534e-05,
+      "loss": 2.2459,
+      "step": 93112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.376276281832136e-05,
+      "loss": 2.2645,
+      "step": 93120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.375648390220466e-05,
+      "loss": 2.2575,
+      "step": 93128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.375020489270133e-05,
+      "loss": 2.2434,
+      "step": 93136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.374392578996752e-05,
+      "loss": 2.2589,
+      "step": 93144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.373764659415927e-05,
+      "loss": 2.2565,
+      "step": 93152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.373136730543272e-05,
+      "loss": 2.2468,
+      "step": 93160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.372508792394395e-05,
+      "loss": 2.2593,
+      "step": 93168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.371880844984907e-05,
+      "loss": 2.2508,
+      "step": 93176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.371252888330422e-05,
+      "loss": 2.2377,
+      "step": 93184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.370624922446548e-05,
+      "loss": 2.2505,
+      "step": 93192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.369996947348896e-05,
+      "loss": 2.2394,
+      "step": 93200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.369368963053077e-05,
+      "loss": 2.2491,
+      "step": 93208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.368740969574707e-05,
+      "loss": 2.2526,
+      "step": 93216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.368112966929394e-05,
+      "loss": 2.2539,
+      "step": 93224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.36748495513275e-05,
+      "loss": 2.2551,
+      "step": 93232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.36685693420039e-05,
+      "loss": 2.2473,
+      "step": 93240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.366228904147924e-05,
+      "loss": 2.262,
+      "step": 93248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.365600864990969e-05,
+      "loss": 2.2509,
+      "step": 93256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.364972816745134e-05,
+      "loss": 2.25,
+      "step": 93264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.364344759426033e-05,
+      "loss": 2.2445,
+      "step": 93272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.363716693049281e-05,
+      "loss": 2.2487,
+      "step": 93280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.363088617630492e-05,
+      "loss": 2.2505,
+      "step": 93288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.362460533185278e-05,
+      "loss": 2.2415,
+      "step": 93296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.361832439729256e-05,
+      "loss": 2.2565,
+      "step": 93304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.361204337278037e-05,
+      "loss": 2.2428,
+      "step": 93312
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.36057622584724e-05,
+      "loss": 2.241,
+      "step": 93320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.359948105452476e-05,
+      "loss": 2.2369,
+      "step": 93328
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.359319976109361e-05,
+      "loss": 2.255,
+      "step": 93336
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.358691837833513e-05,
+      "loss": 2.2408,
+      "step": 93344
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.358063690640546e-05,
+      "loss": 2.2535,
+      "step": 93352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.357435534546072e-05,
+      "loss": 2.2567,
+      "step": 93360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.356807369565713e-05,
+      "loss": 2.2423,
+      "step": 93368
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.356179195715083e-05,
+      "loss": 2.259,
+      "step": 93376
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.355551013009798e-05,
+      "loss": 2.235,
+      "step": 93384
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.354922821465475e-05,
+      "loss": 2.2474,
+      "step": 93392
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.35429462109773e-05,
+      "loss": 2.2509,
+      "step": 93400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.353666411922181e-05,
+      "loss": 2.2557,
+      "step": 93408
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.353038193954445e-05,
+      "loss": 2.2517,
+      "step": 93416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.352409967210142e-05,
+      "loss": 2.2283,
+      "step": 93424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.351781731704886e-05,
+      "loss": 2.2407,
+      "step": 93432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.351153487454298e-05,
+      "loss": 2.2543,
+      "step": 93440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.350525234473993e-05,
+      "loss": 2.2546,
+      "step": 93448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.349896972779591e-05,
+      "loss": 2.2438,
+      "step": 93456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.349268702386713e-05,
+      "loss": 2.2495,
+      "step": 93464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.348640423310976e-05,
+      "loss": 2.2473,
+      "step": 93472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.348012135567999e-05,
+      "loss": 2.2201,
+      "step": 93480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.3473838391734e-05,
+      "loss": 2.228,
+      "step": 93488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.3467555341428e-05,
+      "loss": 2.2517,
+      "step": 93496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.346127220491817e-05,
+      "loss": 2.2491,
+      "step": 93504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.345498898236076e-05,
+      "loss": 2.2398,
+      "step": 93512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.344870567391193e-05,
+      "loss": 2.2496,
+      "step": 93520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.344242227972786e-05,
+      "loss": 2.2526,
+      "step": 93528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.343613879996482e-05,
+      "loss": 2.262,
+      "step": 93536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.342985523477896e-05,
+      "loss": 2.2483,
+      "step": 93544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.342357158432652e-05,
+      "loss": 2.2409,
+      "step": 93552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.341728784876371e-05,
+      "loss": 2.2528,
+      "step": 93560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.341100402824674e-05,
+      "loss": 2.2512,
+      "step": 93568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.340472012293182e-05,
+      "loss": 2.2382,
+      "step": 93576
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.339843613297519e-05,
+      "loss": 2.248,
+      "step": 93584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.339215205853303e-05,
+      "loss": 2.2456,
+      "step": 93592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.338586789976161e-05,
+      "loss": 2.2446,
+      "step": 93600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.337958365681715e-05,
+      "loss": 2.2424,
+      "step": 93608
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.337329932985583e-05,
+      "loss": 2.2385,
+      "step": 93616
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.336701491903392e-05,
+      "loss": 2.2684,
+      "step": 93624
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.336073042450765e-05,
+      "loss": 2.2276,
+      "step": 93632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.335444584643324e-05,
+      "loss": 2.2486,
+      "step": 93640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.334816118496691e-05,
+      "loss": 2.2474,
+      "step": 93648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.334187644026494e-05,
+      "loss": 2.2491,
+      "step": 93656
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.333559161248355e-05,
+      "loss": 2.2667,
+      "step": 93664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.332930670177897e-05,
+      "loss": 2.2543,
+      "step": 93672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.332302170830745e-05,
+      "loss": 2.2524,
+      "step": 93680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.331673663222523e-05,
+      "loss": 2.2562,
+      "step": 93688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.33104514736886e-05,
+      "loss": 2.2582,
+      "step": 93696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.330416623285374e-05,
+      "loss": 2.2364,
+      "step": 93704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.329788090987695e-05,
+      "loss": 2.2747,
+      "step": 93712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.329159550491446e-05,
+      "loss": 2.2521,
+      "step": 93720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.328531001812254e-05,
+      "loss": 2.2511,
+      "step": 93728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.327902444965746e-05,
+      "loss": 2.2419,
+      "step": 93736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.327273879967545e-05,
+      "loss": 2.2684,
+      "step": 93744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.326645306833278e-05,
+      "loss": 2.254,
+      "step": 93752
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.326016725578575e-05,
+      "loss": 2.243,
+      "step": 93760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.325388136219055e-05,
+      "loss": 2.236,
+      "step": 93768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.324759538770353e-05,
+      "loss": 2.248,
+      "step": 93776
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.324130933248091e-05,
+      "loss": 2.2535,
+      "step": 93784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.3235023196679e-05,
+      "loss": 2.2475,
+      "step": 93792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.322873698045401e-05,
+      "loss": 2.2689,
+      "step": 93800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.322245068396229e-05,
+      "loss": 2.2506,
+      "step": 93808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.321616430736006e-05,
+      "loss": 2.2612,
+      "step": 93816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.320987785080364e-05,
+      "loss": 2.2363,
+      "step": 93824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.320359131444928e-05,
+      "loss": 2.264,
+      "step": 93832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.31973046984533e-05,
+      "loss": 2.258,
+      "step": 93840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.319101800297197e-05,
+      "loss": 2.2696,
+      "step": 93848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.318473122816155e-05,
+      "loss": 2.2517,
+      "step": 93856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.317844437417837e-05,
+      "loss": 2.2577,
+      "step": 93864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.31721574411787e-05,
+      "loss": 2.2537,
+      "step": 93872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.316587042931886e-05,
+      "loss": 2.256,
+      "step": 93880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.31595833387551e-05,
+      "loss": 2.2648,
+      "step": 93888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.315329616964377e-05,
+      "loss": 2.2683,
+      "step": 93896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.314700892214111e-05,
+      "loss": 2.2627,
+      "step": 93904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.314072159640347e-05,
+      "loss": 2.2493,
+      "step": 93912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.313443419258714e-05,
+      "loss": 2.2511,
+      "step": 93920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.312814671084844e-05,
+      "loss": 2.2645,
+      "step": 93928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.312185915134364e-05,
+      "loss": 2.2522,
+      "step": 93936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.311557151422908e-05,
+      "loss": 2.2553,
+      "step": 93944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.310928379966105e-05,
+      "loss": 2.253,
+      "step": 93952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.31029960077959e-05,
+      "loss": 2.2386,
+      "step": 93960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.309670813878992e-05,
+      "loss": 2.2608,
+      "step": 93968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.30904201927994e-05,
+      "loss": 2.2523,
+      "step": 93976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.30841321699807e-05,
+      "loss": 2.253,
+      "step": 93984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.307784407049013e-05,
+      "loss": 2.256,
+      "step": 93992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.307155589448403e-05,
+      "loss": 2.2509,
+      "step": 94000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.306526764211868e-05,
+      "loss": 2.2384,
+      "step": 94008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.305897931355042e-05,
+      "loss": 2.2635,
+      "step": 94016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.305269090893565e-05,
+      "loss": 2.2454,
+      "step": 94024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.304640242843058e-05,
+      "loss": 2.264,
+      "step": 94032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.304011387219164e-05,
+      "loss": 2.27,
+      "step": 94040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.30338252403751e-05,
+      "loss": 2.2672,
+      "step": 94048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.302753653313735e-05,
+      "loss": 2.2671,
+      "step": 94056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.30212477506347e-05,
+      "loss": 2.2574,
+      "step": 94064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.301495889302347e-05,
+      "loss": 2.2688,
+      "step": 94072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.300866996046004e-05,
+      "loss": 2.2559,
+      "step": 94080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.300238095310073e-05,
+      "loss": 2.2616,
+      "step": 94088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.299609187110192e-05,
+      "loss": 2.2536,
+      "step": 94096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.298980271461989e-05,
+      "loss": 2.2449,
+      "step": 94104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.298351348381104e-05,
+      "loss": 2.2375,
+      "step": 94112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.297722417883171e-05,
+      "loss": 2.2453,
+      "step": 94120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.297093479983827e-05,
+      "loss": 2.2383,
+      "step": 94128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.296464534698705e-05,
+      "loss": 2.2312,
+      "step": 94136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.295835582043441e-05,
+      "loss": 2.261,
+      "step": 94144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.295206622033671e-05,
+      "loss": 2.2603,
+      "step": 94152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.294577654685032e-05,
+      "loss": 2.2583,
+      "step": 94160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.293948680013159e-05,
+      "loss": 2.261,
+      "step": 94168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.293319698033688e-05,
+      "loss": 2.2491,
+      "step": 94176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.292690708762258e-05,
+      "loss": 2.2589,
+      "step": 94184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.292061712214504e-05,
+      "loss": 2.2706,
+      "step": 94192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.291432708406062e-05,
+      "loss": 2.2603,
+      "step": 94200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.29080369735257e-05,
+      "loss": 2.2655,
+      "step": 94208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.290174679069666e-05,
+      "loss": 2.258,
+      "step": 94216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.289545653572987e-05,
+      "loss": 2.2664,
+      "step": 94224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.288916620878171e-05,
+      "loss": 2.2674,
+      "step": 94232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.288287581000855e-05,
+      "loss": 2.2552,
+      "step": 94240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.287658533956677e-05,
+      "loss": 2.2594,
+      "step": 94248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.287029479761277e-05,
+      "loss": 2.2585,
+      "step": 94256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.286400418430291e-05,
+      "loss": 2.2599,
+      "step": 94264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.285771349979359e-05,
+      "loss": 2.2539,
+      "step": 94272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.285142274424118e-05,
+      "loss": 2.2608,
+      "step": 94280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.284513191780208e-05,
+      "loss": 2.276,
+      "step": 94288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.283884102063269e-05,
+      "loss": 2.2591,
+      "step": 94296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.283255005288938e-05,
+      "loss": 2.2593,
+      "step": 94304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.282625901472858e-05,
+      "loss": 2.2524,
+      "step": 94312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.281996790630667e-05,
+      "loss": 2.251,
+      "step": 94320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.281367672778e-05,
+      "loss": 2.2612,
+      "step": 94328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.280738547930505e-05,
+      "loss": 2.2544,
+      "step": 94336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.280109416103815e-05,
+      "loss": 2.2661,
+      "step": 94344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.279480277313576e-05,
+      "loss": 2.2536,
+      "step": 94352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.278851131575425e-05,
+      "loss": 2.2579,
+      "step": 94360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.278221978905003e-05,
+      "loss": 2.2485,
+      "step": 94368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.277592819317952e-05,
+      "loss": 2.2549,
+      "step": 94376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.276963652829914e-05,
+      "loss": 2.2625,
+      "step": 94384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.276334479456526e-05,
+      "loss": 2.2309,
+      "step": 94392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.275705299213433e-05,
+      "loss": 2.2486,
+      "step": 94400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.275076112116277e-05,
+      "loss": 2.2616,
+      "step": 94408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.274446918180694e-05,
+      "loss": 2.2363,
+      "step": 94416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.273817717422333e-05,
+      "loss": 2.2449,
+      "step": 94424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.273188509856831e-05,
+      "loss": 2.2449,
+      "step": 94432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.272559295499834e-05,
+      "loss": 2.2576,
+      "step": 94440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.27193007436698e-05,
+      "loss": 2.251,
+      "step": 94448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.271300846473916e-05,
+      "loss": 2.2468,
+      "step": 94456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.27067161183628e-05,
+      "loss": 2.2487,
+      "step": 94464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.27004237046972e-05,
+      "loss": 2.2437,
+      "step": 94472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.269413122389875e-05,
+      "loss": 2.2554,
+      "step": 94480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.268783867612389e-05,
+      "loss": 2.2517,
+      "step": 94488
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.268154606152908e-05,
+      "loss": 2.2466,
+      "step": 94496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.267525338027071e-05,
+      "loss": 2.2625,
+      "step": 94504
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.266896063250524e-05,
+      "loss": 2.2394,
+      "step": 94512
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.266266781838912e-05,
+      "loss": 2.2443,
+      "step": 94520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.265637493807878e-05,
+      "loss": 2.2616,
+      "step": 94528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.265008199173068e-05,
+      "loss": 2.2368,
+      "step": 94536
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.26437889795012e-05,
+      "loss": 2.2418,
+      "step": 94544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.263749590154687e-05,
+      "loss": 2.2532,
+      "step": 94552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.263120275802408e-05,
+      "loss": 2.249,
+      "step": 94560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.26249095490893e-05,
+      "loss": 2.2461,
+      "step": 94568
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.261861627489897e-05,
+      "loss": 2.2368,
+      "step": 94576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.261232293560953e-05,
+      "loss": 2.2414,
+      "step": 94584
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.260602953137747e-05,
+      "loss": 2.249,
+      "step": 94592
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.259973606235923e-05,
+      "loss": 2.2408,
+      "step": 94600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.259344252871123e-05,
+      "loss": 2.247,
+      "step": 94608
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.258714893058999e-05,
+      "loss": 2.2515,
+      "step": 94616
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.25808552681519e-05,
+      "loss": 2.2538,
+      "step": 94624
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.25745615415535e-05,
+      "loss": 2.2557,
+      "step": 94632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.256826775095117e-05,
+      "loss": 2.2493,
+      "step": 94640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.256197389650145e-05,
+      "loss": 2.2646,
+      "step": 94648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.255567997836075e-05,
+      "loss": 2.2464,
+      "step": 94656
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.254938599668557e-05,
+      "loss": 2.2553,
+      "step": 94664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.254309195163237e-05,
+      "loss": 2.2657,
+      "step": 94672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.253679784335762e-05,
+      "loss": 2.2669,
+      "step": 94680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.253050367201778e-05,
+      "loss": 2.2379,
+      "step": 94688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.252420943776933e-05,
+      "loss": 2.2481,
+      "step": 94696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.251791514076877e-05,
+      "loss": 2.2508,
+      "step": 94704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.251162078117255e-05,
+      "loss": 2.253,
+      "step": 94712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.250532635913716e-05,
+      "loss": 2.2483,
+      "step": 94720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.249903187481905e-05,
+      "loss": 2.2543,
+      "step": 94728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.249273732837474e-05,
+      "loss": 2.2578,
+      "step": 94736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.248644271996072e-05,
+      "loss": 2.2522,
+      "step": 94744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.248014804973343e-05,
+      "loss": 2.2456,
+      "step": 94752
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.247385331784939e-05,
+      "loss": 2.2521,
+      "step": 94760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.246755852446509e-05,
+      "loss": 2.2579,
+      "step": 94768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.246126366973699e-05,
+      "loss": 2.2459,
+      "step": 94776
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.245496875382162e-05,
+      "loss": 2.2635,
+      "step": 94784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.244867377687543e-05,
+      "loss": 2.2589,
+      "step": 94792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.244237873905494e-05,
+      "loss": 2.2753,
+      "step": 94800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.243608364051666e-05,
+      "loss": 2.2642,
+      "step": 94808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.242978848141705e-05,
+      "loss": 2.2431,
+      "step": 94816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.242349326191262e-05,
+      "loss": 2.2491,
+      "step": 94824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.241719798215988e-05,
+      "loss": 2.2455,
+      "step": 94832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.241090264231534e-05,
+      "loss": 2.248,
+      "step": 94840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.240460724253549e-05,
+      "loss": 2.2448,
+      "step": 94848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.239831178297682e-05,
+      "loss": 2.2557,
+      "step": 94856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.239201626379585e-05,
+      "loss": 2.2554,
+      "step": 94864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.23857206851491e-05,
+      "loss": 2.2676,
+      "step": 94872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.237942504719305e-05,
+      "loss": 2.2433,
+      "step": 94880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.237312935008424e-05,
+      "loss": 2.2463,
+      "step": 94888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.236683359397915e-05,
+      "loss": 2.2657,
+      "step": 94896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.236053777903435e-05,
+      "loss": 2.2495,
+      "step": 94904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.235424190540628e-05,
+      "loss": 2.2567,
+      "step": 94912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.234794597325149e-05,
+      "loss": 2.2559,
+      "step": 94920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.234164998272651e-05,
+      "loss": 2.2524,
+      "step": 94928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.233535393398786e-05,
+      "loss": 2.2613,
+      "step": 94936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.232905782719203e-05,
+      "loss": 2.2503,
+      "step": 94944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.232276166249557e-05,
+      "loss": 2.2569,
+      "step": 94952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.2316465440055e-05,
+      "loss": 2.247,
+      "step": 94960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.231016916002684e-05,
+      "loss": 2.2503,
+      "step": 94968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.230387282256762e-05,
+      "loss": 2.2599,
+      "step": 94976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.229757642783384e-05,
+      "loss": 2.2443,
+      "step": 94984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.229127997598206e-05,
+      "loss": 2.2801,
+      "step": 94992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.228498346716883e-05,
+      "loss": 2.2639,
+      "step": 95000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.227868690155062e-05,
+      "loss": 2.2505,
+      "step": 95008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.2272390279284e-05,
+      "loss": 2.2382,
+      "step": 95016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.226609360052553e-05,
+      "loss": 2.2581,
+      "step": 95024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.225979686543169e-05,
+      "loss": 2.2678,
+      "step": 95032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.225350007415906e-05,
+      "loss": 2.2572,
+      "step": 95040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.224720322686415e-05,
+      "loss": 2.2562,
+      "step": 95048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.224090632370355e-05,
+      "loss": 2.254,
+      "step": 95056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.223460936483374e-05,
+      "loss": 2.2454,
+      "step": 95064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.222831235041128e-05,
+      "loss": 2.2434,
+      "step": 95072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.222201528059276e-05,
+      "loss": 2.255,
+      "step": 95080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.221571815553466e-05,
+      "loss": 2.255,
+      "step": 95088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.220942097539358e-05,
+      "loss": 2.2551,
+      "step": 95096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.220312374032604e-05,
+      "loss": 2.2627,
+      "step": 95104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.219682645048858e-05,
+      "loss": 2.2486,
+      "step": 95112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.21905291060378e-05,
+      "loss": 2.2446,
+      "step": 95120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.21842317071302e-05,
+      "loss": 2.251,
+      "step": 95128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.217793425392235e-05,
+      "loss": 2.2531,
+      "step": 95136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.217163674657083e-05,
+      "loss": 2.2471,
+      "step": 95144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.216533918523216e-05,
+      "loss": 2.2601,
+      "step": 95152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.215904157006294e-05,
+      "loss": 2.2512,
+      "step": 95160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.215274390121966e-05,
+      "loss": 2.2633,
+      "step": 95168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.214644617885897e-05,
+      "loss": 2.2552,
+      "step": 95176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.214014840313734e-05,
+      "loss": 2.2678,
+      "step": 95184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.213385057421142e-05,
+      "loss": 2.2405,
+      "step": 95192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.212755269223772e-05,
+      "loss": 2.2528,
+      "step": 95200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.212125475737281e-05,
+      "loss": 2.2475,
+      "step": 95208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.211495676977331e-05,
+      "loss": 2.2537,
+      "step": 95216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.21086587295957e-05,
+      "loss": 2.2585,
+      "step": 95224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.210236063699662e-05,
+      "loss": 2.2527,
+      "step": 95232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.209606249213263e-05,
+      "loss": 2.2428,
+      "step": 95240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.208976429516026e-05,
+      "loss": 2.2541,
+      "step": 95248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.208346604623613e-05,
+      "loss": 2.2636,
+      "step": 95256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.20771677455168e-05,
+      "loss": 2.2519,
+      "step": 95264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.207086939315884e-05,
+      "loss": 2.2569,
+      "step": 95272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.206457098931886e-05,
+      "loss": 2.2457,
+      "step": 95280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.205827253415339e-05,
+      "loss": 2.258,
+      "step": 95288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.205197402781903e-05,
+      "loss": 2.2484,
+      "step": 95296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.20456754704724e-05,
+      "loss": 2.2695,
+      "step": 95304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.203937686227002e-05,
+      "loss": 2.2469,
+      "step": 95312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.20330782033685e-05,
+      "loss": 2.2358,
+      "step": 95320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.202677949392444e-05,
+      "loss": 2.2774,
+      "step": 95328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.202048073409441e-05,
+      "loss": 2.2535,
+      "step": 95336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.2014181924035e-05,
+      "loss": 2.2611,
+      "step": 95344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.20078830639028e-05,
+      "loss": 2.2598,
+      "step": 95352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.20015841538544e-05,
+      "loss": 2.2441,
+      "step": 95360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.19952851940464e-05,
+      "loss": 2.2411,
+      "step": 95368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.19889861846354e-05,
+      "loss": 2.2484,
+      "step": 95376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.198268712577795e-05,
+      "loss": 2.2623,
+      "step": 95384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.19763880176307e-05,
+      "loss": 2.2453,
+      "step": 95392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.197008886035021e-05,
+      "loss": 2.2518,
+      "step": 95400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.19637896540931e-05,
+      "loss": 2.2305,
+      "step": 95408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.195749039901595e-05,
+      "loss": 2.2571,
+      "step": 95416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.195119109527536e-05,
+      "loss": 2.2524,
+      "step": 95424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.194489174302796e-05,
+      "loss": 2.2673,
+      "step": 95432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.193859234243032e-05,
+      "loss": 2.2638,
+      "step": 95440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.193229289363906e-05,
+      "loss": 2.2476,
+      "step": 95448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.192599339681078e-05,
+      "loss": 2.2643,
+      "step": 95456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.19196938521021e-05,
+      "loss": 2.2565,
+      "step": 95464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.19133942596696e-05,
+      "loss": 2.2513,
+      "step": 95472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.19070946196699e-05,
+      "loss": 2.2508,
+      "step": 95480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.190079493225963e-05,
+      "loss": 2.2492,
+      "step": 95488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.189449519759539e-05,
+      "loss": 2.2641,
+      "step": 95496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.188819541583377e-05,
+      "loss": 2.2528,
+      "step": 95504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.18818955871314e-05,
+      "loss": 2.2436,
+      "step": 95512
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.18755957116449e-05,
+      "loss": 2.2525,
+      "step": 95520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.186929578953088e-05,
+      "loss": 2.2428,
+      "step": 95528
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.186299582094595e-05,
+      "loss": 2.2463,
+      "step": 95536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.185669580604674e-05,
+      "loss": 2.256,
+      "step": 95544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.185039574498988e-05,
+      "loss": 2.2579,
+      "step": 95552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.184409563793195e-05,
+      "loss": 2.2382,
+      "step": 95560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.18377954850296e-05,
+      "loss": 2.2477,
+      "step": 95568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.183149528643945e-05,
+      "loss": 2.2569,
+      "step": 95576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.182519504231814e-05,
+      "loss": 2.2488,
+      "step": 95584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.181889475282226e-05,
+      "loss": 2.2597,
+      "step": 95592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.181259441810845e-05,
+      "loss": 2.2472,
+      "step": 95600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.180629403833335e-05,
+      "loss": 2.2448,
+      "step": 95608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.179999361365356e-05,
+      "loss": 2.2628,
+      "step": 95616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.179369314422576e-05,
+      "loss": 2.2461,
+      "step": 95624
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.178739263020651e-05,
+      "loss": 2.2391,
+      "step": 95632
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.17810920717525e-05,
+      "loss": 2.2244,
+      "step": 95640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.177479146902034e-05,
+      "loss": 2.2486,
+      "step": 95648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.176849082216665e-05,
+      "loss": 2.2437,
+      "step": 95656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.17621901313481e-05,
+      "loss": 2.2513,
+      "step": 95664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.17558893967213e-05,
+      "loss": 2.2352,
+      "step": 95672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.174958861844287e-05,
+      "loss": 2.2451,
+      "step": 95680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.17432877966695e-05,
+      "loss": 2.2373,
+      "step": 95688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.173698693155778e-05,
+      "loss": 2.2614,
+      "step": 95696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.173068602326437e-05,
+      "loss": 2.2566,
+      "step": 95704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.172438507194592e-05,
+      "loss": 2.2616,
+      "step": 95712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.171808407775907e-05,
+      "loss": 2.2566,
+      "step": 95720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.171178304086044e-05,
+      "loss": 2.2448,
+      "step": 95728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.17054819614067e-05,
+      "loss": 2.2409,
+      "step": 95736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.16991808395545e-05,
+      "loss": 2.256,
+      "step": 95744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.169287967546046e-05,
+      "loss": 2.2431,
+      "step": 95752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.168657846928126e-05,
+      "loss": 2.2715,
+      "step": 95760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.168027722117351e-05,
+      "loss": 2.247,
+      "step": 95768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.167397593129389e-05,
+      "loss": 2.2377,
+      "step": 95776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.166767459979901e-05,
+      "loss": 2.2518,
+      "step": 95784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.166137322684559e-05,
+      "loss": 2.2409,
+      "step": 95792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.165507181259022e-05,
+      "loss": 2.2496,
+      "step": 95800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.164877035718961e-05,
+      "loss": 2.2735,
+      "step": 95808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.164246886080037e-05,
+      "loss": 2.2422,
+      "step": 95816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.163616732357915e-05,
+      "loss": 2.2422,
+      "step": 95824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.162986574568264e-05,
+      "loss": 2.2621,
+      "step": 95832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.162356412726749e-05,
+      "loss": 2.235,
+      "step": 95840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.161726246849036e-05,
+      "loss": 2.2541,
+      "step": 95848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.161096076950787e-05,
+      "loss": 2.2515,
+      "step": 95856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.160465903047676e-05,
+      "loss": 2.2385,
+      "step": 95864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.15983572515536e-05,
+      "loss": 2.2585,
+      "step": 95872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.159205543289513e-05,
+      "loss": 2.2606,
+      "step": 95880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.158575357465798e-05,
+      "loss": 2.2626,
+      "step": 95888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.157945167699881e-05,
+      "loss": 2.2493,
+      "step": 95896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.15731497400743e-05,
+      "loss": 2.2558,
+      "step": 95904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.156684776404111e-05,
+      "loss": 2.2447,
+      "step": 95912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.156054574905592e-05,
+      "loss": 2.242,
+      "step": 95920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.155424369527537e-05,
+      "loss": 2.2566,
+      "step": 95928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.154794160285616e-05,
+      "loss": 2.2422,
+      "step": 95936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.154163947195493e-05,
+      "loss": 2.2616,
+      "step": 95944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.153533730272838e-05,
+      "loss": 2.2636,
+      "step": 95952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.152903509533319e-05,
+      "loss": 2.2647,
+      "step": 95960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.1522732849926e-05,
+      "loss": 2.2611,
+      "step": 95968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.15164305666635e-05,
+      "loss": 2.2413,
+      "step": 95976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.151012824570237e-05,
+      "loss": 2.248,
+      "step": 95984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.150382588719927e-05,
+      "loss": 2.2396,
+      "step": 95992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.14975234913109e-05,
+      "loss": 2.2431,
+      "step": 96000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.149122105819391e-05,
+      "loss": 2.2583,
+      "step": 96008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.148491858800503e-05,
+      "loss": 2.2516,
+      "step": 96016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.147861608090089e-05,
+      "loss": 2.2475,
+      "step": 96024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.14723135370382e-05,
+      "loss": 2.2427,
+      "step": 96032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.14660109565736e-05,
+      "loss": 2.2436,
+      "step": 96040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.145970833966383e-05,
+      "loss": 2.2492,
+      "step": 96048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.145340568646556e-05,
+      "loss": 2.2533,
+      "step": 96056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.144710299713543e-05,
+      "loss": 2.2496,
+      "step": 96064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.144080027183016e-05,
+      "loss": 2.2482,
+      "step": 96072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.143449751070644e-05,
+      "loss": 2.2451,
+      "step": 96080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.142819471392097e-05,
+      "loss": 2.2368,
+      "step": 96088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.142189188163039e-05,
+      "loss": 2.2629,
+      "step": 96096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.141558901399142e-05,
+      "loss": 2.247,
+      "step": 96104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.140928611116077e-05,
+      "loss": 2.2532,
+      "step": 96112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.14029831732951e-05,
+      "loss": 2.2431,
+      "step": 96120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.139668020055111e-05,
+      "loss": 2.2368,
+      "step": 96128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.139037719308549e-05,
+      "loss": 2.2521,
+      "step": 96136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.138407415105493e-05,
+      "loss": 2.2518,
+      "step": 96144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.137777107461615e-05,
+      "loss": 2.2572,
+      "step": 96152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.137146796392582e-05,
+      "loss": 2.2451,
+      "step": 96160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.136516481914064e-05,
+      "loss": 2.2578,
+      "step": 96168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.13588616404173e-05,
+      "loss": 2.2482,
+      "step": 96176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.135255842791252e-05,
+      "loss": 2.2459,
+      "step": 96184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.134625518178299e-05,
+      "loss": 2.2572,
+      "step": 96192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.133995190218539e-05,
+      "loss": 2.2644,
+      "step": 96200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.133364858927643e-05,
+      "loss": 2.2312,
+      "step": 96208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.132734524321285e-05,
+      "loss": 2.2513,
+      "step": 96216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.132104186415128e-05,
+      "loss": 2.2646,
+      "step": 96224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.131473845224849e-05,
+      "loss": 2.2604,
+      "step": 96232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.130843500766115e-05,
+      "loss": 2.2597,
+      "step": 96240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.130213153054595e-05,
+      "loss": 2.2454,
+      "step": 96248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.129582802105964e-05,
+      "loss": 2.244,
+      "step": 96256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.128952447935887e-05,
+      "loss": 2.2694,
+      "step": 96264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.128322090560041e-05,
+      "loss": 2.2534,
+      "step": 96272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.127691729994092e-05,
+      "loss": 2.2306,
+      "step": 96280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.12706136625371e-05,
+      "loss": 2.2502,
+      "step": 96288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.126430999354571e-05,
+      "loss": 2.2607,
+      "step": 96296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.125800629312341e-05,
+      "loss": 2.2445,
+      "step": 96304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.125170256142696e-05,
+      "loss": 2.2526,
+      "step": 96312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.124539879861302e-05,
+      "loss": 2.2629,
+      "step": 96320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.123909500483835e-05,
+      "loss": 2.2503,
+      "step": 96328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.123279118025961e-05,
+      "loss": 2.2558,
+      "step": 96336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.122648732503357e-05,
+      "loss": 2.2432,
+      "step": 96344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.12201834393169e-05,
+      "loss": 2.2548,
+      "step": 96352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.121387952326634e-05,
+      "loss": 2.2621,
+      "step": 96360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.12075755770386e-05,
+      "loss": 2.2371,
+      "step": 96368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.120127160079039e-05,
+      "loss": 2.2497,
+      "step": 96376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.119496759467842e-05,
+      "loss": 2.2528,
+      "step": 96384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.118866355885945e-05,
+      "loss": 2.2445,
+      "step": 96392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.118235949349016e-05,
+      "loss": 2.2377,
+      "step": 96400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.117605539872728e-05,
+      "loss": 2.2568,
+      "step": 96408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.116975127472752e-05,
+      "loss": 2.2417,
+      "step": 96416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.116344712164763e-05,
+      "loss": 2.2368,
+      "step": 96424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.11571429396443e-05,
+      "loss": 2.2529,
+      "step": 96432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.115083872887427e-05,
+      "loss": 2.2537,
+      "step": 96440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.114453448949426e-05,
+      "loss": 2.2341,
+      "step": 96448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.1138230221661e-05,
+      "loss": 2.243,
+      "step": 96456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.11319259255312e-05,
+      "loss": 2.2313,
+      "step": 96464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.11256216012616e-05,
+      "loss": 2.2401,
+      "step": 96472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.111931724900892e-05,
+      "loss": 2.257,
+      "step": 96480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.11130128689299e-05,
+      "loss": 2.2651,
+      "step": 96488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.110670846118123e-05,
+      "loss": 2.2537,
+      "step": 96496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.110040402591968e-05,
+      "loss": 2.2345,
+      "step": 96504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.109409956330196e-05,
+      "loss": 2.2403,
+      "step": 96512
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.108779507348481e-05,
+      "loss": 2.2383,
+      "step": 96520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.108149055662495e-05,
+      "loss": 2.2539,
+      "step": 96528
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.10751860128791e-05,
+      "loss": 2.2512,
+      "step": 96536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.1068881442404e-05,
+      "loss": 2.2566,
+      "step": 96544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.106257684535641e-05,
+      "loss": 2.2491,
+      "step": 96552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.105627222189302e-05,
+      "loss": 2.2574,
+      "step": 96560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.104996757217061e-05,
+      "loss": 2.2485,
+      "step": 96568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.104366289634587e-05,
+      "loss": 2.2505,
+      "step": 96576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.103735819457554e-05,
+      "loss": 2.2461,
+      "step": 96584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.103105346701638e-05,
+      "loss": 2.2731,
+      "step": 96592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.102474871382513e-05,
+      "loss": 2.2264,
+      "step": 96600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.1018443935158495e-05,
+      "loss": 2.2582,
+      "step": 96608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.101213913117324e-05,
+      "loss": 2.2616,
+      "step": 96616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.100583430202609e-05,
+      "loss": 2.2707,
+      "step": 96624
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.099952944787377e-05,
+      "loss": 2.2569,
+      "step": 96632
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.099322456887306e-05,
+      "loss": 2.248,
+      "step": 96640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.098691966518069e-05,
+      "loss": 2.2309,
+      "step": 96648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.098061473695335e-05,
+      "loss": 2.2625,
+      "step": 96656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.097430978434783e-05,
+      "loss": 2.2444,
+      "step": 96664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.096800480752085e-05,
+      "loss": 2.2509,
+      "step": 96672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.096169980662918e-05,
+      "loss": 2.2553,
+      "step": 96680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.095539478182954e-05,
+      "loss": 2.2488,
+      "step": 96688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0949089733278664e-05,
+      "loss": 2.2501,
+      "step": 96696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.094278466113334e-05,
+      "loss": 2.2503,
+      "step": 96704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.093647956555026e-05,
+      "loss": 2.2515,
+      "step": 96712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0930174446686196e-05,
+      "loss": 2.2492,
+      "step": 96720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0923869304697887e-05,
+      "loss": 2.2513,
+      "step": 96728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.09175641397421e-05,
+      "loss": 2.2525,
+      "step": 96736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.091125895197556e-05,
+      "loss": 2.2525,
+      "step": 96744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0904953741555e-05,
+      "loss": 2.252,
+      "step": 96752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0898648508637204e-05,
+      "loss": 2.2485,
+      "step": 96760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.089234325337891e-05,
+      "loss": 2.2701,
+      "step": 96768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.088603797593686e-05,
+      "loss": 2.2628,
+      "step": 96776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0879732676467807e-05,
+      "loss": 2.256,
+      "step": 96784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.087342735512849e-05,
+      "loss": 2.252,
+      "step": 96792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0867122012075694e-05,
+      "loss": 2.2435,
+      "step": 96800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0860816647466126e-05,
+      "loss": 2.2422,
+      "step": 96808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.085451126145656e-05,
+      "loss": 2.252,
+      "step": 96816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0848205854203764e-05,
+      "loss": 2.256,
+      "step": 96824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.084190042586446e-05,
+      "loss": 2.2557,
+      "step": 96832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.083559497659543e-05,
+      "loss": 2.2707,
+      "step": 96840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.08292895065534e-05,
+      "loss": 2.2515,
+      "step": 96848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.082298401589515e-05,
+      "loss": 2.2432,
+      "step": 96856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.081667850477742e-05,
+      "loss": 2.2524,
+      "step": 96864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.081037297335699e-05,
+      "loss": 2.2554,
+      "step": 96872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0804067421790565e-05,
+      "loss": 2.2738,
+      "step": 96880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0797761850234944e-05,
+      "loss": 2.2664,
+      "step": 96888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.079145625884689e-05,
+      "loss": 2.2729,
+      "step": 96896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.078515064778314e-05,
+      "loss": 2.2573,
+      "step": 96904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.077884501720044e-05,
+      "loss": 2.2445,
+      "step": 96912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.077253936725558e-05,
+      "loss": 2.2527,
+      "step": 96920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.07662336981053e-05,
+      "loss": 2.2691,
+      "step": 96928
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.075992800990635e-05,
+      "loss": 2.2554,
+      "step": 96936
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0753622302815526e-05,
+      "loss": 2.2637,
+      "step": 96944
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.074731657698956e-05,
+      "loss": 2.275,
+      "step": 96952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.074101083258523e-05,
+      "loss": 2.2538,
+      "step": 96960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.073470506975927e-05,
+      "loss": 2.2624,
+      "step": 96968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.072839928866847e-05,
+      "loss": 2.247,
+      "step": 96976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0722093489469576e-05,
+      "loss": 2.2548,
+      "step": 96984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.071578767231937e-05,
+      "loss": 2.2682,
+      "step": 96992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0709481837374597e-05,
+      "loss": 2.2433,
+      "step": 97000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0703175984792004e-05,
+      "loss": 2.2538,
+      "step": 97008
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.06968701147284e-05,
+      "loss": 2.2483,
+      "step": 97016
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.069056422734052e-05,
+      "loss": 2.2505,
+      "step": 97024
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0684258322785136e-05,
+      "loss": 2.2375,
+      "step": 97032
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0677952401219015e-05,
+      "loss": 2.2625,
+      "step": 97040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0671646462798924e-05,
+      "loss": 2.2466,
+      "step": 97048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.066534050768163e-05,
+      "loss": 2.2521,
+      "step": 97056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0659034536023885e-05,
+      "loss": 2.249,
+      "step": 97064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0652728547982476e-05,
+      "loss": 2.2487,
+      "step": 97072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.064642254371415e-05,
+      "loss": 2.2513,
+      "step": 97080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.06401165233757e-05,
+      "loss": 2.2543,
+      "step": 97088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.063381048712387e-05,
+      "loss": 2.2501,
+      "step": 97096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0627504435115444e-05,
+      "loss": 2.2691,
+      "step": 97104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0621198367507195e-05,
+      "loss": 2.2609,
+      "step": 97112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.061489228445588e-05,
+      "loss": 2.2618,
+      "step": 97120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0608586186118274e-05,
+      "loss": 2.2437,
+      "step": 97128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.060228007265114e-05,
+      "loss": 2.2631,
+      "step": 97136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.059597394421128e-05,
+      "loss": 2.271,
+      "step": 97144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.058966780095541e-05,
+      "loss": 2.2575,
+      "step": 97152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.058336164304035e-05,
+      "loss": 2.2656,
+      "step": 97160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0577055470622846e-05,
+      "loss": 2.2555,
+      "step": 97168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0570749283859694e-05,
+      "loss": 2.269,
+      "step": 97176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.056444308290763e-05,
+      "loss": 2.2683,
+      "step": 97184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.055813686792347e-05,
+      "loss": 2.2672,
+      "step": 97192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.055183063906394e-05,
+      "loss": 2.2786,
+      "step": 97200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.054552439648587e-05,
+      "loss": 2.2625,
+      "step": 97208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.053921814034597e-05,
+      "loss": 2.2545,
+      "step": 97216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0532911870801045e-05,
+      "loss": 2.242,
+      "step": 97224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.052660558800789e-05,
+      "loss": 2.2526,
+      "step": 97232
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.052029929212326e-05,
+      "loss": 2.2372,
+      "step": 97240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0513992983303915e-05,
+      "loss": 2.2587,
+      "step": 97248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.050768666170666e-05,
+      "loss": 2.251,
+      "step": 97256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.050138032748827e-05,
+      "loss": 2.2634,
+      "step": 97264
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.049507398080549e-05,
+      "loss": 2.2438,
+      "step": 97272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.048876762181511e-05,
+      "loss": 2.2368,
+      "step": 97280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.048246125067394e-05,
+      "loss": 2.2539,
+      "step": 97288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.04761548675387e-05,
+      "loss": 2.2398,
+      "step": 97296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.046984847256623e-05,
+      "loss": 2.2523,
+      "step": 97304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.046354206591323e-05,
+      "loss": 2.2634,
+      "step": 97312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0457235647736554e-05,
+      "loss": 2.2503,
+      "step": 97320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.045092921819293e-05,
+      "loss": 2.2478,
+      "step": 97328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0444622777439176e-05,
+      "loss": 2.273,
+      "step": 97336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0438316325632036e-05,
+      "loss": 2.2392,
+      "step": 97344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0432009862928305e-05,
+      "loss": 2.2581,
+      "step": 97352
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.042570338948476e-05,
+      "loss": 2.2563,
+      "step": 97360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.041939690545819e-05,
+      "loss": 2.257,
+      "step": 97368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0413090411005335e-05,
+      "loss": 2.2466,
+      "step": 97376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.040678390628305e-05,
+      "loss": 2.2537,
+      "step": 97384
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.040047739144805e-05,
+      "loss": 2.2577,
+      "step": 97392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.039417086665713e-05,
+      "loss": 2.253,
+      "step": 97400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.038786433206709e-05,
+      "loss": 2.2555,
+      "step": 97408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0381557787834676e-05,
+      "loss": 2.2632,
+      "step": 97416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.037525123411671e-05,
+      "loss": 2.2636,
+      "step": 97424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.036894467106997e-05,
+      "loss": 2.254,
+      "step": 97432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.036263809885121e-05,
+      "loss": 2.2454,
+      "step": 97440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.035633151761723e-05,
+      "loss": 2.2644,
+      "step": 97448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.03500249275248e-05,
+      "loss": 2.2528,
+      "step": 97456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0343718328730714e-05,
+      "loss": 2.244,
+      "step": 97464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.033741172139176e-05,
+      "loss": 2.2509,
+      "step": 97472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.033110510566471e-05,
+      "loss": 2.2471,
+      "step": 97480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.032479848170636e-05,
+      "loss": 2.2532,
+      "step": 97488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.031849184967345e-05,
+      "loss": 2.2531,
+      "step": 97496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0312185209722824e-05,
+      "loss": 2.2541,
+      "step": 97504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.030587856201122e-05,
+      "loss": 2.2568,
+      "step": 97512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.029957190669546e-05,
+      "loss": 2.26,
+      "step": 97520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.029326524393231e-05,
+      "loss": 2.2458,
+      "step": 97528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0286958573878536e-05,
+      "loss": 2.2399,
+      "step": 97536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0280651896690965e-05,
+      "loss": 2.2606,
+      "step": 97544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.027434521252633e-05,
+      "loss": 2.2397,
+      "step": 97552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.026803852154144e-05,
+      "loss": 2.2415,
+      "step": 97560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0261731823893096e-05,
+      "loss": 2.2807,
+      "step": 97568
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0255425119738076e-05,
+      "loss": 2.2418,
+      "step": 97576
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0249118409233155e-05,
+      "loss": 2.2415,
+      "step": 97584
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.024281169253511e-05,
+      "loss": 2.2515,
+      "step": 97592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.023650496980075e-05,
+      "loss": 2.2491,
+      "step": 97600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.023019824118685e-05,
+      "loss": 2.2349,
+      "step": 97608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.02238915068502e-05,
+      "loss": 2.255,
+      "step": 97616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.021758476694757e-05,
+      "loss": 2.2473,
+      "step": 97624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.021127802163575e-05,
+      "loss": 2.2366,
+      "step": 97632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.020497127107156e-05,
+      "loss": 2.2561,
+      "step": 97640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.019866451541174e-05,
+      "loss": 2.2389,
+      "step": 97648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.019235775481311e-05,
+      "loss": 2.2471,
+      "step": 97656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.018605098943244e-05,
+      "loss": 2.2531,
+      "step": 97664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0179744219426524e-05,
+      "loss": 2.251,
+      "step": 97672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0173437444952143e-05,
+      "loss": 2.2509,
+      "step": 97680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.016713066616608e-05,
+      "loss": 2.2461,
+      "step": 97688
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.016082388322516e-05,
+      "loss": 2.2524,
+      "step": 97696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0154517096286126e-05,
+      "loss": 2.2568,
+      "step": 97704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.014821030550577e-05,
+      "loss": 2.255,
+      "step": 97712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0141903511040885e-05,
+      "loss": 2.2681,
+      "step": 97720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.013559671304827e-05,
+      "loss": 2.2244,
+      "step": 97728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.012928991168472e-05,
+      "loss": 2.2365,
+      "step": 97736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.012298310710698e-05,
+      "loss": 2.2518,
+      "step": 97744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.011667629947189e-05,
+      "loss": 2.2413,
+      "step": 97752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0110369488936204e-05,
+      "loss": 2.2535,
+      "step": 97760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.010406267565672e-05,
+      "loss": 2.2385,
+      "step": 97768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.009775585979023e-05,
+      "loss": 2.2451,
+      "step": 97776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.009144904149351e-05,
+      "loss": 2.243,
+      "step": 97784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0085142220923364e-05,
+      "loss": 2.2484,
+      "step": 97792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.007883539823658e-05,
+      "loss": 2.2513,
+      "step": 97800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.007252857358992e-05,
+      "loss": 2.2398,
+      "step": 97808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0066221747140205e-05,
+      "loss": 2.2373,
+      "step": 97816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0059914919044194e-05,
+      "loss": 2.2421,
+      "step": 97824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.005360808945871e-05,
+      "loss": 2.2428,
+      "step": 97832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.004730125854052e-05,
+      "loss": 2.2543,
+      "step": 97840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.004099442644641e-05,
+      "loss": 2.2461,
+      "step": 97848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0034687593333184e-05,
+      "loss": 2.2432,
+      "step": 97856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.002838075935762e-05,
+      "loss": 2.2627,
+      "step": 97864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.002207392467649e-05,
+      "loss": 2.2289,
+      "step": 97872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.00157670894466e-05,
+      "loss": 2.2556,
+      "step": 97880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0009460253824765e-05,
+      "loss": 2.2326,
+      "step": 97888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.000315341796774e-05,
+      "loss": 2.2563,
+      "step": 97896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9996846582032287e-05,
+      "loss": 2.2428,
+      "step": 97904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.999053974617526e-05,
+      "loss": 2.2519,
+      "step": 97912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.998423291055342e-05,
+      "loss": 2.2465,
+      "step": 97920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9977926075323535e-05,
+      "loss": 2.2525,
+      "step": 97928
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9971619240642405e-05,
+      "loss": 2.2644,
+      "step": 97936
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9965312406666826e-05,
+      "loss": 2.2371,
+      "step": 97944
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.995900557355361e-05,
+      "loss": 2.2484,
+      "step": 97952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.99526987414595e-05,
+      "loss": 2.2569,
+      "step": 97960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.99463919105413e-05,
+      "loss": 2.2319,
+      "step": 97968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.994008508095581e-05,
+      "loss": 2.2245,
+      "step": 97976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.993377825285982e-05,
+      "loss": 2.234,
+      "step": 97984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9927471426410086e-05,
+      "loss": 2.2441,
+      "step": 97992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9921164601763444e-05,
+      "loss": 2.2691,
+      "step": 98000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.991485777907666e-05,
+      "loss": 2.2536,
+      "step": 98008
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9908550958506516e-05,
+      "loss": 2.251,
+      "step": 98016
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9902244140209795e-05,
+      "loss": 2.2451,
+      "step": 98024
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.98959373243433e-05,
+      "loss": 2.253,
+      "step": 98032
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.988963051106381e-05,
+      "loss": 2.2552,
+      "step": 98040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9883323700528136e-05,
+      "loss": 2.2417,
+      "step": 98048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9877016892893025e-05,
+      "loss": 2.2297,
+      "step": 98056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9870710088315304e-05,
+      "loss": 2.2443,
+      "step": 98064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9864403286951754e-05,
+      "loss": 2.2484,
+      "step": 98072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9858096488959125e-05,
+      "loss": 2.251,
+      "step": 98080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.985178969449425e-05,
+      "loss": 2.2488,
+      "step": 98088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9845482903713904e-05,
+      "loss": 2.247,
+      "step": 98096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.983917611677485e-05,
+      "loss": 2.2457,
+      "step": 98104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9832869333833916e-05,
+      "loss": 2.2396,
+      "step": 98112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.982656255504786e-05,
+      "loss": 2.2755,
+      "step": 98120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9820255780573486e-05,
+      "loss": 2.2367,
+      "step": 98128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9813949010567563e-05,
+      "loss": 2.2524,
+      "step": 98136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9807642245186915e-05,
+      "loss": 2.232,
+      "step": 98144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.980133548458827e-05,
+      "loss": 2.2339,
+      "step": 98152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.979502872892846e-05,
+      "loss": 2.2537,
+      "step": 98160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.978872197836426e-05,
+      "loss": 2.2577,
+      "step": 98168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9782415233052455e-05,
+      "loss": 2.2585,
+      "step": 98176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9776108493149826e-05,
+      "loss": 2.2512,
+      "step": 98184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.976980175881317e-05,
+      "loss": 2.2498,
+      "step": 98192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.976349503019927e-05,
+      "loss": 2.2536,
+      "step": 98200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.975718830746491e-05,
+      "loss": 2.2331,
+      "step": 98208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.975088159076686e-05,
+      "loss": 2.255,
+      "step": 98216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.974457488026194e-05,
+      "loss": 2.2495,
+      "step": 98224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.973826817610693e-05,
+      "loss": 2.2541,
+      "step": 98232
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9731961478458564e-05,
+      "loss": 2.2545,
+      "step": 98240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9725654787473694e-05,
+      "loss": 2.251,
+      "step": 98248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.971934810330906e-05,
+      "loss": 2.2597,
+      "step": 98256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.971304142612149e-05,
+      "loss": 2.2403,
+      "step": 98264
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9706734756067716e-05,
+      "loss": 2.2471,
+      "step": 98272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.970042809330455e-05,
+      "loss": 2.2477,
+      "step": 98280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.969412143798878e-05,
+      "loss": 2.249,
+      "step": 98288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9687814790277186e-05,
+      "loss": 2.2451,
+      "step": 98296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.968150815032656e-05,
+      "loss": 2.2575,
+      "step": 98304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9675201518293664e-05,
+      "loss": 2.2506,
+      "step": 98312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.966889489433532e-05,
+      "loss": 2.2488,
+      "step": 98320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9662588278608267e-05,
+      "loss": 2.2583,
+      "step": 98328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.965628167126929e-05,
+      "loss": 2.258,
+      "step": 98336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.964997507247521e-05,
+      "loss": 2.2336,
+      "step": 98344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.964366848238279e-05,
+      "loss": 2.2424,
+      "step": 98352
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.963736190114881e-05,
+      "loss": 2.2501,
+      "step": 98360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9631055328930055e-05,
+      "loss": 2.2349,
+      "step": 98368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9624748765883284e-05,
+      "loss": 2.267,
+      "step": 98376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.961844221216533e-05,
+      "loss": 2.2554,
+      "step": 98384
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9612135667932934e-05,
+      "loss": 2.2431,
+      "step": 98392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9605829133342885e-05,
+      "loss": 2.2387,
+      "step": 98400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.959952260855197e-05,
+      "loss": 2.2351,
+      "step": 98408
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.959321609371697e-05,
+      "loss": 2.257,
+      "step": 98416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.958690958899467e-05,
+      "loss": 2.2561,
+      "step": 98424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.958060309454183e-05,
+      "loss": 2.2541,
+      "step": 98432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.957429661051526e-05,
+      "loss": 2.2311,
+      "step": 98440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9567990137071726e-05,
+      "loss": 2.2406,
+      "step": 98448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.956168367436799e-05,
+      "loss": 2.2467,
+      "step": 98456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.955537722256084e-05,
+      "loss": 2.2513,
+      "step": 98464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9549070781807074e-05,
+      "loss": 2.2608,
+      "step": 98472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.954276435226347e-05,
+      "loss": 2.253,
+      "step": 98480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9536457934086776e-05,
+      "loss": 2.2391,
+      "step": 98488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.95301515274338e-05,
+      "loss": 2.266,
+      "step": 98496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.952384513246132e-05,
+      "loss": 2.2559,
+      "step": 98504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.951753874932608e-05,
+      "loss": 2.2297,
+      "step": 98512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.95112323781849e-05,
+      "loss": 2.2516,
+      "step": 98520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.950492601919452e-05,
+      "loss": 2.226,
+      "step": 98528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9498619672511754e-05,
+      "loss": 2.2381,
+      "step": 98536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9492313338293346e-05,
+      "loss": 2.2411,
+      "step": 98544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.948600701669609e-05,
+      "loss": 2.2562,
+      "step": 98552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.947970070787676e-05,
+      "loss": 2.252,
+      "step": 98560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9473394411992125e-05,
+      "loss": 2.2449,
+      "step": 98568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9467088129198964e-05,
+      "loss": 2.2408,
+      "step": 98576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.946078185965405e-05,
+      "loss": 2.2619,
+      "step": 98584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9454475603514156e-05,
+      "loss": 2.2648,
+      "step": 98592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.944816936093608e-05,
+      "loss": 2.2396,
+      "step": 98600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9441863132076556e-05,
+      "loss": 2.2473,
+      "step": 98608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9435556917092374e-05,
+      "loss": 2.2539,
+      "step": 98616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9429250716140336e-05,
+      "loss": 2.2364,
+      "step": 98624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.942294452937716e-05,
+      "loss": 2.2577,
+      "step": 98632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9416638356959674e-05,
+      "loss": 2.246,
+      "step": 98640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9410332199044594e-05,
+      "loss": 2.2417,
+      "step": 98648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.940402605578874e-05,
+      "loss": 2.254,
+      "step": 98656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.939771992734886e-05,
+      "loss": 2.2605,
+      "step": 98664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9391413813881735e-05,
+      "loss": 2.2684,
+      "step": 98672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.938510771554413e-05,
+      "loss": 2.2455,
+      "step": 98680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.937880163249281e-05,
+      "loss": 2.2527,
+      "step": 98688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9372495564884566e-05,
+      "loss": 2.2458,
+      "step": 98696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9366189512876136e-05,
+      "loss": 2.2531,
+      "step": 98704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.935988347662431e-05,
+      "loss": 2.2436,
+      "step": 98712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.935357745628587e-05,
+      "loss": 2.2617,
+      "step": 98720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.934727145201755e-05,
+      "loss": 2.2271,
+      "step": 98728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.934096546397614e-05,
+      "loss": 2.245,
+      "step": 98736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.933465949231838e-05,
+      "loss": 2.2478,
+      "step": 98744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.93283535372011e-05,
+      "loss": 2.2613,
+      "step": 98752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9322047598780995e-05,
+      "loss": 2.2501,
+      "step": 98760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.931574167721487e-05,
+      "loss": 2.248,
+      "step": 98768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9309435772659484e-05,
+      "loss": 2.2515,
+      "step": 98776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.93031298852716e-05,
+      "loss": 2.2334,
+      "step": 98784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9296824015208006e-05,
+      "loss": 2.2609,
+      "step": 98792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9290518162625434e-05,
+      "loss": 2.2509,
+      "step": 98800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.928421232768064e-05,
+      "loss": 2.246,
+      "step": 98808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.927790651053045e-05,
+      "loss": 2.2576,
+      "step": 98816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9271600711331544e-05,
+      "loss": 2.2652,
+      "step": 98824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9265294930240734e-05,
+      "loss": 2.2642,
+      "step": 98832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9258989167414794e-05,
+      "loss": 2.2553,
+      "step": 98840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9252683423010446e-05,
+      "loss": 2.2676,
+      "step": 98848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9246377697184484e-05,
+      "loss": 2.2649,
+      "step": 98856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.924007199009365e-05,
+      "loss": 2.2474,
+      "step": 98864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.923376630189472e-05,
+      "loss": 2.2618,
+      "step": 98872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.922746063274444e-05,
+      "loss": 2.2541,
+      "step": 98880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9221154982799585e-05,
+      "loss": 2.2588,
+      "step": 98888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.921484935221688e-05,
+      "loss": 2.2585,
+      "step": 98896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.920854374115312e-05,
+      "loss": 2.2504,
+      "step": 98904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9202238149765065e-05,
+      "loss": 2.2558,
+      "step": 98912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.919593257820946e-05,
+      "loss": 2.256,
+      "step": 98920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9189627026643036e-05,
+      "loss": 2.2485,
+      "step": 98928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.918332149522259e-05,
+      "loss": 2.2394,
+      "step": 98936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.917701598410487e-05,
+      "loss": 2.2623,
+      "step": 98944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9170710493446626e-05,
+      "loss": 2.2546,
+      "step": 98952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.916440502340459e-05,
+      "loss": 2.2324,
+      "step": 98960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.915809957413556e-05,
+      "loss": 2.2509,
+      "step": 98968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.915179414579626e-05,
+      "loss": 2.2563,
+      "step": 98976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9145488738543455e-05,
+      "loss": 2.2496,
+      "step": 98984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.91391833525339e-05,
+      "loss": 2.2585,
+      "step": 98992
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.913287798792433e-05,
+      "loss": 2.2571,
+      "step": 99000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.912657264487151e-05,
+      "loss": 2.2371,
+      "step": 99008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.912026732353222e-05,
+      "loss": 2.2358,
+      "step": 99016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9113962024063155e-05,
+      "loss": 2.2678,
+      "step": 99024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.910765674662111e-05,
+      "loss": 2.252,
+      "step": 99032
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.91013514913628e-05,
+      "loss": 2.2364,
+      "step": 99040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.909504625844501e-05,
+      "loss": 2.2341,
+      "step": 99048
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9088741048024465e-05,
+      "loss": 2.2394,
+      "step": 99056
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.908243586025792e-05,
+      "loss": 2.2501,
+      "step": 99064
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.907613069530214e-05,
+      "loss": 2.2475,
+      "step": 99072
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.906982555331382e-05,
+      "loss": 2.2588,
+      "step": 99080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.906352043444976e-05,
+      "loss": 2.2536,
+      "step": 99088
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.905721533886668e-05,
+      "loss": 2.2339,
+      "step": 99096
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.905091026672134e-05,
+      "loss": 2.2633,
+      "step": 99104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.904460521817048e-05,
+      "loss": 2.254,
+      "step": 99112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.903830019337083e-05,
+      "loss": 2.2441,
+      "step": 99120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9031995192479163e-05,
+      "loss": 2.2474,
+      "step": 99128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.902569021565218e-05,
+      "loss": 2.2538,
+      "step": 99136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.901938526304667e-05,
+      "loss": 2.2573,
+      "step": 99144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9013080334819335e-05,
+      "loss": 2.2574,
+      "step": 99152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.900677543112695e-05,
+      "loss": 2.2469,
+      "step": 99160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.900047055212623e-05,
+      "loss": 2.249,
+      "step": 99168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8994165697973915e-05,
+      "loss": 2.2568,
+      "step": 99176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8987860868826775e-05,
+      "loss": 2.2572,
+      "step": 99184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.89815560648415e-05,
+      "loss": 2.259,
+      "step": 99192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8975251286174896e-05,
+      "loss": 2.2689,
+      "step": 99200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.896894653298361e-05,
+      "loss": 2.265,
+      "step": 99208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8962641805424455e-05,
+      "loss": 2.2523,
+      "step": 99216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.895633710365416e-05,
+      "loss": 2.2539,
+      "step": 99224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8950032427829414e-05,
+      "loss": 2.2564,
+      "step": 99232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.894372777810699e-05,
+      "loss": 2.2481,
+      "step": 99240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.893742315464361e-05,
+      "loss": 2.2532,
+      "step": 99248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.893111855759601e-05,
+      "loss": 2.2622,
+      "step": 99256
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.892481398712093e-05,
+      "loss": 2.241,
+      "step": 99264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.891850944337507e-05,
+      "loss": 2.2607,
+      "step": 99272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.891220492651521e-05,
+      "loss": 2.2469,
+      "step": 99280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8905900436698056e-05,
+      "loss": 2.2393,
+      "step": 99288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8899595974080334e-05,
+      "loss": 2.2334,
+      "step": 99296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8893291538818775e-05,
+      "loss": 2.2543,
+      "step": 99304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.888698713107013e-05,
+      "loss": 2.2625,
+      "step": 99312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.888068275099109e-05,
+      "loss": 2.2317,
+      "step": 99320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8874378398738415e-05,
+      "loss": 2.2532,
+      "step": 99328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.88680740744688e-05,
+      "loss": 2.2497,
+      "step": 99336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.886176977833902e-05,
+      "loss": 2.244,
+      "step": 99344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8855465510505744e-05,
+      "loss": 2.2539,
+      "step": 99352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.884916127112574e-05,
+      "loss": 2.244,
+      "step": 99360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8842857060355706e-05,
+      "loss": 2.2516,
+      "step": 99368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.883655287835239e-05,
+      "loss": 2.2557,
+      "step": 99376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.88302487252725e-05,
+      "loss": 2.2653,
+      "step": 99384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.882394460127274e-05,
+      "loss": 2.2561,
+      "step": 99392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8817640506509867e-05,
+      "loss": 2.2668,
+      "step": 99400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.881133644114056e-05,
+      "loss": 2.2412,
+      "step": 99408
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.880503240532158e-05,
+      "loss": 2.2475,
+      "step": 99416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.879872839920964e-05,
+      "loss": 2.2403,
+      "step": 99424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8792424422961425e-05,
+      "loss": 2.2538,
+      "step": 99432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.878612047673369e-05,
+      "loss": 2.2458,
+      "step": 99440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.877981656068312e-05,
+      "loss": 2.2435,
+      "step": 99448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8773512674966446e-05,
+      "loss": 2.2531,
+      "step": 99456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8767208819740404e-05,
+      "loss": 2.2628,
+      "step": 99464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.876090499516167e-05,
+      "loss": 2.2682,
+      "step": 99472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8754601201386994e-05,
+      "loss": 2.2501,
+      "step": 99480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.874829743857305e-05,
+      "loss": 2.2615,
+      "step": 99488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.874199370687658e-05,
+      "loss": 2.2452,
+      "step": 99496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8735690006454317e-05,
+      "loss": 2.2535,
+      "step": 99504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.872938633746291e-05,
+      "loss": 2.2606,
+      "step": 99512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.872308270005911e-05,
+      "loss": 2.2382,
+      "step": 99520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.871677909439961e-05,
+      "loss": 2.2347,
+      "step": 99528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.871047552064114e-05,
+      "loss": 2.2435,
+      "step": 99536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.870417197894038e-05,
+      "loss": 2.2351,
+      "step": 99544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.869786846945406e-05,
+      "loss": 2.2496,
+      "step": 99552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8691564992338875e-05,
+      "loss": 2.255,
+      "step": 99560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8685261547751526e-05,
+      "loss": 2.259,
+      "step": 99568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8678958135848727e-05,
+      "loss": 2.2429,
+      "step": 99576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.867265475678716e-05,
+      "loss": 2.258,
+      "step": 99584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.866635141072356e-05,
+      "loss": 2.2532,
+      "step": 99592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.866004809781462e-05,
+      "loss": 2.2702,
+      "step": 99600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.865374481821704e-05,
+      "loss": 2.2585,
+      "step": 99608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.864744157208748e-05,
+      "loss": 2.2415,
+      "step": 99616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8641138359582705e-05,
+      "loss": 2.2604,
+      "step": 99624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.863483518085937e-05,
+      "loss": 2.2486,
+      "step": 99632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.862853203607419e-05,
+      "loss": 2.2491,
+      "step": 99640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.862222892538386e-05,
+      "loss": 2.2688,
+      "step": 99648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.861592584894508e-05,
+      "loss": 2.242,
+      "step": 99656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.860962280691452e-05,
+      "loss": 2.2437,
+      "step": 99664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.860331979944891e-05,
+      "loss": 2.2482,
+      "step": 99672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.859701682670491e-05,
+      "loss": 2.257,
+      "step": 99680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.859071388883923e-05,
+      "loss": 2.2306,
+      "step": 99688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8584410986008586e-05,
+      "loss": 2.2534,
+      "step": 99696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.857810811836961e-05,
+      "loss": 2.2437,
+      "step": 99704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.857180528607905e-05,
+      "loss": 2.2479,
+      "step": 99712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8565502489293555e-05,
+      "loss": 2.2601,
+      "step": 99720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8559199728169855e-05,
+      "loss": 2.2498,
+      "step": 99728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8552897002864596e-05,
+      "loss": 2.2426,
+      "step": 99736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8546594313534466e-05,
+      "loss": 2.2543,
+      "step": 99744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.854029166033619e-05,
+      "loss": 2.2513,
+      "step": 99752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.85339890434264e-05,
+      "loss": 2.24,
+      "step": 99760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8527686462961816e-05,
+      "loss": 2.2455,
+      "step": 99768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.852138391909911e-05,
+      "loss": 2.2595,
+      "step": 99776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8515081411994975e-05,
+      "loss": 2.2402,
+      "step": 99784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8508778941806086e-05,
+      "loss": 2.2452,
+      "step": 99792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.850247650868911e-05,
+      "loss": 2.2499,
+      "step": 99800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.849617411280074e-05,
+      "loss": 2.2446,
+      "step": 99808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.848987175429766e-05,
+      "loss": 2.2566,
+      "step": 99816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8483569433336525e-05,
+      "loss": 2.2347,
+      "step": 99824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8477267150074024e-05,
+      "loss": 2.2644,
+      "step": 99832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.847096490466682e-05,
+      "loss": 2.2247,
+      "step": 99840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.846466269727162e-05,
+      "loss": 2.2522,
+      "step": 99848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8458360528045076e-05,
+      "loss": 2.2334,
+      "step": 99856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.845205839714385e-05,
+      "loss": 2.2492,
+      "step": 99864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8445756304724645e-05,
+      "loss": 2.2604,
+      "step": 99872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8439454250944105e-05,
+      "loss": 2.2527,
+      "step": 99880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8433152235958887e-05,
+      "loss": 2.2403,
+      "step": 99888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.842685025992569e-05,
+      "loss": 2.2535,
+      "step": 99896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.842054832300119e-05,
+      "loss": 2.2345,
+      "step": 99904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.841424642534203e-05,
+      "loss": 2.2377,
+      "step": 99912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8407944567104876e-05,
+      "loss": 2.2578,
+      "step": 99920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.84016427484464e-05,
+      "loss": 2.2531,
+      "step": 99928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.839534096952326e-05,
+      "loss": 2.2459,
+      "step": 99936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.838903923049214e-05,
+      "loss": 2.2359,
+      "step": 99944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8382737531509665e-05,
+      "loss": 2.2538,
+      "step": 99952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.837643587273252e-05,
+      "loss": 2.2514,
+      "step": 99960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.837013425431738e-05,
+      "loss": 2.2313,
+      "step": 99968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8363832676420873e-05,
+      "loss": 2.2552,
+      "step": 99976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.835753113919966e-05,
+      "loss": 2.2534,
+      "step": 99984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8351229642810414e-05,
+      "loss": 2.2492,
+      "step": 99992
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.8344928187409774e-05,
+      "loss": 2.2576,
+      "step": 100000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8338626773154424e-05,
+      "loss": 2.2427,
+      "step": 100008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8332325400200996e-05,
+      "loss": 2.2394,
+      "step": 100016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.832602406870613e-05,
+      "loss": 2.2547,
+      "step": 100024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.831972277882651e-05,
+      "loss": 2.2597,
+      "step": 100032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.831342153071877e-05,
+      "loss": 2.2641,
+      "step": 100040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.830712032453955e-05,
+      "loss": 2.2506,
+      "step": 100048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.830081916044551e-05,
+      "loss": 2.2459,
+      "step": 100056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8294518038593306e-05,
+      "loss": 2.2496,
+      "step": 100064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8288216959139566e-05,
+      "loss": 2.2529,
+      "step": 100072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.828191592224094e-05,
+      "loss": 2.2325,
+      "step": 100080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.827561492805409e-05,
+      "loss": 2.2611,
+      "step": 100088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.826931397673563e-05,
+      "loss": 2.2425,
+      "step": 100096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.826301306844224e-05,
+      "loss": 2.2426,
+      "step": 100104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.825671220333051e-05,
+      "loss": 2.2445,
+      "step": 100112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8250411381557135e-05,
+      "loss": 2.255,
+      "step": 100120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.824411060327873e-05,
+      "loss": 2.2495,
+      "step": 100128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.823780986865193e-05,
+      "loss": 2.2522,
+      "step": 100136
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8231509177833366e-05,
+      "loss": 2.2483,
+      "step": 100144
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.822520853097968e-05,
+      "loss": 2.2459,
+      "step": 100152
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8218907928247515e-05,
+      "loss": 2.2407,
+      "step": 100160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.82126073697935e-05,
+      "loss": 2.2656,
+      "step": 100168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.820630685577426e-05,
+      "loss": 2.2502,
+      "step": 100176
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.820000638634644e-05,
+      "loss": 2.2352,
+      "step": 100184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8193705961666665e-05,
+      "loss": 2.237,
+      "step": 100192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8187405581891555e-05,
+      "loss": 2.2374,
+      "step": 100200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8181105247177744e-05,
+      "loss": 2.2593,
+      "step": 100208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8174804957681886e-05,
+      "loss": 2.2459,
+      "step": 100216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.816850471356056e-05,
+      "loss": 2.2542,
+      "step": 100224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.81622045149704e-05,
+      "loss": 2.2482,
+      "step": 100232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8155904362068056e-05,
+      "loss": 2.2293,
+      "step": 100240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.814960425501014e-05,
+      "loss": 2.2539,
+      "step": 100248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8143304193953266e-05,
+      "loss": 2.2592,
+      "step": 100256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.813700417905406e-05,
+      "loss": 2.2534,
+      "step": 100264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8130704210469134e-05,
+      "loss": 2.2482,
+      "step": 100272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8124404288355114e-05,
+      "loss": 2.2332,
+      "step": 100280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8118104412868625e-05,
+      "loss": 2.2379,
+      "step": 100288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.811180458416625e-05,
+      "loss": 2.2426,
+      "step": 100296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8105504802404644e-05,
+      "loss": 2.2441,
+      "step": 100304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8099205067740384e-05,
+      "loss": 2.247,
+      "step": 100312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.809290538033012e-05,
+      "loss": 2.2196,
+      "step": 100320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.808660574033041e-05,
+      "loss": 2.2427,
+      "step": 100328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8080306147897926e-05,
+      "loss": 2.2516,
+      "step": 100336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.807400660318925e-05,
+      "loss": 2.2438,
+      "step": 100344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.806770710636096e-05,
+      "loss": 2.2406,
+      "step": 100352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.806140765756969e-05,
+      "loss": 2.2291,
+      "step": 100360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.805510825697206e-05,
+      "loss": 2.247,
+      "step": 100368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.804880890472466e-05,
+      "loss": 2.246,
+      "step": 100376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8042509600984076e-05,
+      "loss": 2.2411,
+      "step": 100384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8036210345906916e-05,
+      "loss": 2.2521,
+      "step": 100392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.802991113964981e-05,
+      "loss": 2.2322,
+      "step": 100400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.802361198236932e-05,
+      "loss": 2.2633,
+      "step": 100408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8017312874222065e-05,
+      "loss": 2.243,
+      "step": 100416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.801101381536462e-05,
+      "loss": 2.2389,
+      "step": 100424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8004714805953607e-05,
+      "loss": 2.2332,
+      "step": 100432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.799841584614561e-05,
+      "loss": 2.2352,
+      "step": 100440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.799211693609721e-05,
+      "loss": 2.2373,
+      "step": 100448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.798581807596502e-05,
+      "loss": 2.2545,
+      "step": 100456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.79795192659056e-05,
+      "loss": 2.2448,
+      "step": 100464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.797322050607558e-05,
+      "loss": 2.2376,
+      "step": 100472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.79669217966315e-05,
+      "loss": 2.261,
+      "step": 100480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.796062313772999e-05,
+      "loss": 2.2508,
+      "step": 100488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.795432452952763e-05,
+      "loss": 2.2464,
+      "step": 100496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7948025972180964e-05,
+      "loss": 2.256,
+      "step": 100504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.794172746584661e-05,
+      "loss": 2.2574,
+      "step": 100512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.793542901068116e-05,
+      "loss": 2.2332,
+      "step": 100520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.792913060684115e-05,
+      "loss": 2.247,
+      "step": 100528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7922832254483203e-05,
+      "loss": 2.2315,
+      "step": 100536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.791653395376387e-05,
+      "loss": 2.2415,
+      "step": 100544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7910235704839753e-05,
+      "loss": 2.2552,
+      "step": 100552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7903937507867404e-05,
+      "loss": 2.2477,
+      "step": 100560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7897639363003396e-05,
+      "loss": 2.2503,
+      "step": 100568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.789134127040432e-05,
+      "loss": 2.2433,
+      "step": 100576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.788504323022672e-05,
+      "loss": 2.2495,
+      "step": 100584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7878745242627196e-05,
+      "loss": 2.2519,
+      "step": 100592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.787244730776229e-05,
+      "loss": 2.2669,
+      "step": 100600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7866149425788594e-05,
+      "loss": 2.2552,
+      "step": 100608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.785985159686267e-05,
+      "loss": 2.2544,
+      "step": 100616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.785355382114106e-05,
+      "loss": 2.2594,
+      "step": 100624
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7847256098780356e-05,
+      "loss": 2.256,
+      "step": 100632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.784095842993709e-05,
+      "loss": 2.2442,
+      "step": 100640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.783466081476785e-05,
+      "loss": 2.2586,
+      "step": 100648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.78283632534292e-05,
+      "loss": 2.2471,
+      "step": 100656
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.782206574607766e-05,
+      "loss": 2.2561,
+      "step": 100664
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.781576829286981e-05,
+      "loss": 2.2574,
+      "step": 100672
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.780947089396221e-05,
+      "loss": 2.2558,
+      "step": 100680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.780317354951143e-05,
+      "loss": 2.2497,
+      "step": 100688
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.779687625967398e-05,
+      "loss": 2.2634,
+      "step": 100696
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.779057902460643e-05,
+      "loss": 2.2411,
+      "step": 100704
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7784281844465336e-05,
+      "loss": 2.2474,
+      "step": 100712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.777798471940725e-05,
+      "loss": 2.2497,
+      "step": 100720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.777168764958871e-05,
+      "loss": 2.2472,
+      "step": 100728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.776539063516626e-05,
+      "loss": 2.2421,
+      "step": 100736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.775909367629647e-05,
+      "loss": 2.2649,
+      "step": 100744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.775279677313585e-05,
+      "loss": 2.2484,
+      "step": 100752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.774649992584096e-05,
+      "loss": 2.2546,
+      "step": 100760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7740203134568316e-05,
+      "loss": 2.2587,
+      "step": 100768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.773390639947449e-05,
+      "loss": 2.2629,
+      "step": 100776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.772760972071601e-05,
+      "loss": 2.2596,
+      "step": 100784
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7721313098449394e-05,
+      "loss": 2.236,
+      "step": 100792
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.77150165328312e-05,
+      "loss": 2.2584,
+      "step": 100800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.770872002401795e-05,
+      "loss": 2.2409,
+      "step": 100808
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7702423572166176e-05,
+      "loss": 2.2473,
+      "step": 100816
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.76961271774324e-05,
+      "loss": 2.2461,
+      "step": 100824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.768983083997318e-05,
+      "loss": 2.2302,
+      "step": 100832
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.768353455994502e-05,
+      "loss": 2.2448,
+      "step": 100840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7677238337504447e-05,
+      "loss": 2.2475,
+      "step": 100848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7670942172807974e-05,
+      "loss": 2.2527,
+      "step": 100856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.766464606601216e-05,
+      "loss": 2.2589,
+      "step": 100864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.765835001727351e-05,
+      "loss": 2.257,
+      "step": 100872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7652054026748525e-05,
+      "loss": 2.2368,
+      "step": 100880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.764575809459374e-05,
+      "loss": 2.2415,
+      "step": 100888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.763946222096568e-05,
+      "loss": 2.2606,
+      "step": 100896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7633166406020855e-05,
+      "loss": 2.2371,
+      "step": 100904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.762687064991579e-05,
+      "loss": 2.2348,
+      "step": 100912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.762057495280696e-05,
+      "loss": 2.2429,
+      "step": 100920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7614279314850926e-05,
+      "loss": 2.2507,
+      "step": 100928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.760798373620416e-05,
+      "loss": 2.252,
+      "step": 100936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7601688217023206e-05,
+      "loss": 2.2387,
+      "step": 100944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.759539275746453e-05,
+      "loss": 2.2674,
+      "step": 100952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.758909735768467e-05,
+      "loss": 2.2573,
+      "step": 100960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.758280201784013e-05,
+      "loss": 2.2467,
+      "step": 100968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.757650673808739e-05,
+      "loss": 2.2477,
+      "step": 100976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7570211518582976e-05,
+      "loss": 2.2292,
+      "step": 100984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.756391635948335e-05,
+      "loss": 2.2571,
+      "step": 100992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.755762126094506e-05,
+      "loss": 2.2519,
+      "step": 101000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.755132622312458e-05,
+      "loss": 2.261,
+      "step": 101008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.754503124617841e-05,
+      "loss": 2.2455,
+      "step": 101016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.753873633026302e-05,
+      "loss": 2.2485,
+      "step": 101024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7532441475534926e-05,
+      "loss": 2.2382,
+      "step": 101032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7526146682150615e-05,
+      "loss": 2.2505,
+      "step": 101040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.751985195026659e-05,
+      "loss": 2.2373,
+      "step": 101048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.75135572800393e-05,
+      "loss": 2.2437,
+      "step": 101056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.750726267162527e-05,
+      "loss": 2.2409,
+      "step": 101064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7500968125180956e-05,
+      "loss": 2.243,
+      "step": 101072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7494673640862875e-05,
+      "loss": 2.2443,
+      "step": 101080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.748837921882747e-05,
+      "loss": 2.2404,
+      "step": 101088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.748208485923126e-05,
+      "loss": 2.2578,
+      "step": 101096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.747579056223068e-05,
+      "loss": 2.2509,
+      "step": 101104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.746949632798224e-05,
+      "loss": 2.232,
+      "step": 101112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.746320215664241e-05,
+      "loss": 2.2555,
+      "step": 101120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.745690804836765e-05,
+      "loss": 2.2453,
+      "step": 101128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.745061400331444e-05,
+      "loss": 2.2499,
+      "step": 101136
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.744432002163926e-05,
+      "loss": 2.2607,
+      "step": 101144
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7438026103498575e-05,
+      "loss": 2.2429,
+      "step": 101152
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7431732249048844e-05,
+      "loss": 2.2424,
+      "step": 101160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7425438458446525e-05,
+      "loss": 2.2397,
+      "step": 101168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.741914473184811e-05,
+      "loss": 2.2615,
+      "step": 101176
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.741285106941003e-05,
+      "loss": 2.2458,
+      "step": 101184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.740655747128878e-05,
+      "loss": 2.2366,
+      "step": 101192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.74002639376408e-05,
+      "loss": 2.2488,
+      "step": 101200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.739397046862254e-05,
+      "loss": 2.2535,
+      "step": 101208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.738767706439048e-05,
+      "loss": 2.2496,
+      "step": 101216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.738138372510104e-05,
+      "loss": 2.236,
+      "step": 101224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.737509045091072e-05,
+      "loss": 2.2642,
+      "step": 101232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.736879724197594e-05,
+      "loss": 2.2553,
+      "step": 101240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.736250409845314e-05,
+      "loss": 2.2423,
+      "step": 101248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7356211020498795e-05,
+      "loss": 2.2625,
+      "step": 101256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7349918008269346e-05,
+      "loss": 2.2475,
+      "step": 101264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7343625061921225e-05,
+      "loss": 2.2457,
+      "step": 101272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.733733218161088e-05,
+      "loss": 2.2557,
+      "step": 101280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.733103936749477e-05,
+      "loss": 2.2517,
+      "step": 101288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.732474661972931e-05,
+      "loss": 2.2674,
+      "step": 101296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.731845393847095e-05,
+      "loss": 2.2554,
+      "step": 101304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.731216132387612e-05,
+      "loss": 2.2614,
+      "step": 101312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7305868776101265e-05,
+      "loss": 2.2521,
+      "step": 101320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.729957629530283e-05,
+      "loss": 2.2619,
+      "step": 101328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7293283881637214e-05,
+      "loss": 2.2626,
+      "step": 101336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.728699153526087e-05,
+      "loss": 2.2684,
+      "step": 101344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7280699256330214e-05,
+      "loss": 2.2379,
+      "step": 101352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.727440704500169e-05,
+      "loss": 2.2419,
+      "step": 101360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.72681149014317e-05,
+      "loss": 2.2521,
+      "step": 101368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.72618228257767e-05,
+      "loss": 2.2369,
+      "step": 101376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.725553081819306e-05,
+      "loss": 2.241,
+      "step": 101384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7249238878837264e-05,
+      "loss": 2.242,
+      "step": 101392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7242947007865686e-05,
+      "loss": 2.2515,
+      "step": 101400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7236655205434754e-05,
+      "loss": 2.245,
+      "step": 101408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.723036347170088e-05,
+      "loss": 2.2454,
+      "step": 101416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.722407180682048e-05,
+      "loss": 2.2488,
+      "step": 101424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7217780210949976e-05,
+      "loss": 2.2462,
+      "step": 101432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7211488684245764e-05,
+      "loss": 2.2413,
+      "step": 101440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.720519722686426e-05,
+      "loss": 2.2605,
+      "step": 101448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.719890583896186e-05,
+      "loss": 2.256,
+      "step": 101456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7192614520694975e-05,
+      "loss": 2.2581,
+      "step": 101464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.718632327222002e-05,
+      "loss": 2.2449,
+      "step": 101472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7180032093693356e-05,
+      "loss": 2.2629,
+      "step": 101480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.717374098527144e-05,
+      "loss": 2.2635,
+      "step": 101488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.716744994711063e-05,
+      "loss": 2.2575,
+      "step": 101496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7161158979367326e-05,
+      "loss": 2.2783,
+      "step": 101504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.715486808219793e-05,
+      "loss": 2.2436,
+      "step": 101512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.714857725575884e-05,
+      "loss": 2.2379,
+      "step": 101520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7142286500206435e-05,
+      "loss": 2.245,
+      "step": 101528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.713599581569711e-05,
+      "loss": 2.2431,
+      "step": 101536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7129705202387244e-05,
+      "loss": 2.2387,
+      "step": 101544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.712341466043323e-05,
+      "loss": 2.2298,
+      "step": 101552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7117124189991457e-05,
+      "loss": 2.253,
+      "step": 101560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7110833791218286e-05,
+      "loss": 2.2295,
+      "step": 101568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7104543464270135e-05,
+      "loss": 2.2564,
+      "step": 101576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7098253209303346e-05,
+      "loss": 2.2471,
+      "step": 101584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.7091963026474306e-05,
+      "loss": 2.2551,
+      "step": 101592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.708567291593938e-05,
+      "loss": 2.2501,
+      "step": 101600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.707938287785497e-05,
+      "loss": 2.2491,
+      "step": 101608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7073092912377425e-05,
+      "loss": 2.2571,
+      "step": 101616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.706680301966312e-05,
+      "loss": 2.2574,
+      "step": 101624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7060513199868424e-05,
+      "loss": 2.2515,
+      "step": 101632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.705422345314968e-05,
+      "loss": 2.2347,
+      "step": 101640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7047933779663284e-05,
+      "loss": 2.2563,
+      "step": 101648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.704164417956561e-05,
+      "loss": 2.2451,
+      "step": 101656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.703535465301295e-05,
+      "loss": 2.2397,
+      "step": 101664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7029065200161736e-05,
+      "loss": 2.242,
+      "step": 101672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.702277582116829e-05,
+      "loss": 2.2369,
+      "step": 101680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.701648651618896e-05,
+      "loss": 2.2385,
+      "step": 101688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.701019728538011e-05,
+      "loss": 2.2412,
+      "step": 101696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.70039081288981e-05,
+      "loss": 2.2739,
+      "step": 101704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.699761904689929e-05,
+      "loss": 2.2501,
+      "step": 101712
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.699133003953998e-05,
+      "loss": 2.2512,
+      "step": 101720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.698504110697653e-05,
+      "loss": 2.2332,
+      "step": 101728
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6978752249365334e-05,
+      "loss": 2.2365,
+      "step": 101736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.697246346686267e-05,
+      "loss": 2.2568,
+      "step": 101744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.69661747596249e-05,
+      "loss": 2.2425,
+      "step": 101752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.695988612780838e-05,
+      "loss": 2.2425,
+      "step": 101760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6953597571569425e-05,
+      "loss": 2.2495,
+      "step": 101768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6947309091064384e-05,
+      "loss": 2.2597,
+      "step": 101776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.694102068644958e-05,
+      "loss": 2.255,
+      "step": 101784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.693473235788134e-05,
+      "loss": 2.2644,
+      "step": 101792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.692844410551599e-05,
+      "loss": 2.2572,
+      "step": 101800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6922155929509866e-05,
+      "loss": 2.2441,
+      "step": 101808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6915867830019306e-05,
+      "loss": 2.2329,
+      "step": 101816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.690957980720062e-05,
+      "loss": 2.2449,
+      "step": 101824
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.690329186121011e-05,
+      "loss": 2.2474,
+      "step": 101832
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6897003992204124e-05,
+      "loss": 2.2609,
+      "step": 101840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.689071620033894e-05,
+      "loss": 2.2608,
+      "step": 101848
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.688442848577093e-05,
+      "loss": 2.2569,
+      "step": 101856
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.687814084865637e-05,
+      "loss": 2.2465,
+      "step": 101864
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.687185328915159e-05,
+      "loss": 2.2318,
+      "step": 101872
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.686556580741287e-05,
+      "loss": 2.2536,
+      "step": 101880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.685927840359653e-05,
+      "loss": 2.2433,
+      "step": 101888
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.685299107785891e-05,
+      "loss": 2.2436,
+      "step": 101896
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6846703830356265e-05,
+      "loss": 2.2615,
+      "step": 101904
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.684041666124491e-05,
+      "loss": 2.2344,
+      "step": 101912
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.683412957068115e-05,
+      "loss": 2.2397,
+      "step": 101920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6827842558821306e-05,
+      "loss": 2.2403,
+      "step": 101928
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.682155562582164e-05,
+      "loss": 2.248,
+      "step": 101936
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.681526877183846e-05,
+      "loss": 2.2503,
+      "step": 101944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.680898199702807e-05,
+      "loss": 2.2421,
+      "step": 101952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6802695301546715e-05,
+      "loss": 2.2394,
+      "step": 101960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6796408685550716e-05,
+      "loss": 2.2474,
+      "step": 101968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.679012214919639e-05,
+      "loss": 2.2436,
+      "step": 101976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.678383569263996e-05,
+      "loss": 2.2335,
+      "step": 101984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6777549316037734e-05,
+      "loss": 2.2316,
+      "step": 101992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6771263019546005e-05,
+      "loss": 2.2516,
+      "step": 102000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.676497680332104e-05,
+      "loss": 2.239,
+      "step": 102008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.67586906675191e-05,
+      "loss": 2.25,
+      "step": 102016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.675240461229648e-05,
+      "loss": 2.2327,
+      "step": 102024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.674611863780945e-05,
+      "loss": 2.2278,
+      "step": 102032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6739832744214284e-05,
+      "loss": 2.2361,
+      "step": 102040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6733546931667226e-05,
+      "loss": 2.2459,
+      "step": 102048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.672726120032457e-05,
+      "loss": 2.2475,
+      "step": 102056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.672097555034256e-05,
+      "loss": 2.2343,
+      "step": 102064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6714689981877466e-05,
+      "loss": 2.2418,
+      "step": 102072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6708404495085545e-05,
+      "loss": 2.2607,
+      "step": 102080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6702119090123085e-05,
+      "loss": 2.2514,
+      "step": 102088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6695833767146265e-05,
+      "loss": 2.2219,
+      "step": 102096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.668954852631142e-05,
+      "loss": 2.2521,
+      "step": 102104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6683263367774786e-05,
+      "loss": 2.261,
+      "step": 102112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.667697829169256e-05,
+      "loss": 2.2464,
+      "step": 102120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6670693298221056e-05,
+      "loss": 2.248,
+      "step": 102128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6664408387516456e-05,
+      "loss": 2.2366,
+      "step": 102136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6658123559735074e-05,
+      "loss": 2.2437,
+      "step": 102144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6651838815033095e-05,
+      "loss": 2.2462,
+      "step": 102152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.664555415356678e-05,
+      "loss": 2.2376,
+      "step": 102160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.663926957549237e-05,
+      "loss": 2.2267,
+      "step": 102168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.663298508096611e-05,
+      "loss": 2.2402,
+      "step": 102176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6626700670144195e-05,
+      "loss": 2.2338,
+      "step": 102184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6620416343182894e-05,
+      "loss": 2.225,
+      "step": 102192
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6614132100238406e-05,
+      "loss": 2.2464,
+      "step": 102200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.660784794146698e-05,
+      "loss": 2.2538,
+      "step": 102208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.660156386702483e-05,
+      "loss": 2.2542,
+      "step": 102216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6595279877068206e-05,
+      "loss": 2.227,
+      "step": 102224
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6588995971753286e-05,
+      "loss": 2.2481,
+      "step": 102232
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6582712151236305e-05,
+      "loss": 2.2516,
+      "step": 102240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6576428415673497e-05,
+      "loss": 2.2416,
+      "step": 102248
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.657014476522107e-05,
+      "loss": 2.2475,
+      "step": 102256
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.656386120003521e-05,
+      "loss": 2.2478,
+      "step": 102264
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.655757772027215e-05,
+      "loss": 2.2409,
+      "step": 102272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.65512943260881e-05,
+      "loss": 2.2466,
+      "step": 102280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.654501101763925e-05,
+      "loss": 2.2376,
+      "step": 102288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.653872779508182e-05,
+      "loss": 2.2645,
+      "step": 102296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.653244465857203e-05,
+      "loss": 2.2539,
+      "step": 102304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.652616160826603e-05,
+      "loss": 2.2407,
+      "step": 102312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.651987864432003e-05,
+      "loss": 2.2563,
+      "step": 102320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.651359576689026e-05,
+      "loss": 2.2427,
+      "step": 102328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.650731297613288e-05,
+      "loss": 2.238,
+      "step": 102336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.650103027220409e-05,
+      "loss": 2.2236,
+      "step": 102344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6494747655260075e-05,
+      "loss": 2.2415,
+      "step": 102352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.648846512545705e-05,
+      "loss": 2.2318,
+      "step": 102360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6482182682951156e-05,
+      "loss": 2.2638,
+      "step": 102368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.64759003278986e-05,
+      "loss": 2.2381,
+      "step": 102376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.646961806045556e-05,
+      "loss": 2.2521,
+      "step": 102384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.64633358807782e-05,
+      "loss": 2.2584,
+      "step": 102392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.645705378902271e-05,
+      "loss": 2.2619,
+      "step": 102400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.645077178534526e-05,
+      "loss": 2.2366,
+      "step": 102408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.644448986990204e-05,
+      "loss": 2.2632,
+      "step": 102416
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.643820804284918e-05,
+      "loss": 2.2573,
+      "step": 102424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.643192630434288e-05,
+      "loss": 2.2435,
+      "step": 102432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6425644654539287e-05,
+      "loss": 2.2439,
+      "step": 102440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.641936309359457e-05,
+      "loss": 2.2321,
+      "step": 102448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6413081621664875e-05,
+      "loss": 2.2529,
+      "step": 102456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.64068002389064e-05,
+      "loss": 2.2373,
+      "step": 102464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6400518945475265e-05,
+      "loss": 2.228,
+      "step": 102472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.639423774152761e-05,
+      "loss": 2.2372,
+      "step": 102480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6387956627219626e-05,
+      "loss": 2.2241,
+      "step": 102488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.638167560270746e-05,
+      "loss": 2.244,
+      "step": 102496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.637539466814723e-05,
+      "loss": 2.2527,
+      "step": 102504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.636911382369509e-05,
+      "loss": 2.2589,
+      "step": 102512
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6362833069507204e-05,
+      "loss": 2.2501,
+      "step": 102520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6356552405739695e-05,
+      "loss": 2.2374,
+      "step": 102528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6350271832548686e-05,
+      "loss": 2.2563,
+      "step": 102536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.634399135009034e-05,
+      "loss": 2.2563,
+      "step": 102544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6337710958520774e-05,
+      "loss": 2.2509,
+      "step": 102552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.633143065799613e-05,
+      "loss": 2.2661,
+      "step": 102560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.63251504486725e-05,
+      "loss": 2.2434,
+      "step": 102568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.631887033070609e-05,
+      "loss": 2.2546,
+      "step": 102576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6312590304252953e-05,
+      "loss": 2.2584,
+      "step": 102584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.630631036946924e-05,
+      "loss": 2.251,
+      "step": 102592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6300030526511074e-05,
+      "loss": 2.2598,
+      "step": 102600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.629375077553454e-05,
+      "loss": 2.2377,
+      "step": 102608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.628747111669579e-05,
+      "loss": 2.2512,
+      "step": 102616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.628119155015092e-05,
+      "loss": 2.2465,
+      "step": 102624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.627491207605606e-05,
+      "loss": 2.2575,
+      "step": 102632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.62686326945673e-05,
+      "loss": 2.2662,
+      "step": 102640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.626235340584074e-05,
+      "loss": 2.237,
+      "step": 102648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.625607421003249e-05,
+      "loss": 2.2463,
+      "step": 102656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.624979510729866e-05,
+      "loss": 2.2485,
+      "step": 102664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6243516097795356e-05,
+      "loss": 2.2638,
+      "step": 102672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.623723718167867e-05,
+      "loss": 2.25,
+      "step": 102680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6230958359104655e-05,
+      "loss": 2.2635,
+      "step": 102688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.622467963022946e-05,
+      "loss": 2.2444,
+      "step": 102696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6218400995209154e-05,
+      "loss": 2.2614,
+      "step": 102704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.621212245419983e-05,
+      "loss": 2.2397,
+      "step": 102712
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6205844007357555e-05,
+      "loss": 2.2548,
+      "step": 102720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6199565654838414e-05,
+      "loss": 2.2508,
+      "step": 102728
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.619328739679852e-05,
+      "loss": 2.2626,
+      "step": 102736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6187009233393925e-05,
+      "loss": 2.2328,
+      "step": 102744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6180731164780685e-05,
+      "loss": 2.2369,
+      "step": 102752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.617445319111492e-05,
+      "loss": 2.2406,
+      "step": 102760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.616817531255267e-05,
+      "loss": 2.2422,
+      "step": 102768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.616189752925002e-05,
+      "loss": 2.2549,
+      "step": 102776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6155619841363004e-05,
+      "loss": 2.2475,
+      "step": 102784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.614934224904774e-05,
+      "loss": 2.2389,
+      "step": 102792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.614306475246025e-05,
+      "loss": 2.2456,
+      "step": 102800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.613678735175659e-05,
+      "loss": 2.2538,
+      "step": 102808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6130510047092845e-05,
+      "loss": 2.2574,
+      "step": 102816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.612423283862505e-05,
+      "loss": 2.2511,
+      "step": 102824
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.611795572650925e-05,
+      "loss": 2.2542,
+      "step": 102832
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.611167871090152e-05,
+      "loss": 2.2424,
+      "step": 102840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.610540179195789e-05,
+      "loss": 2.2637,
+      "step": 102848
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.609912496983441e-05,
+      "loss": 2.238,
+      "step": 102856
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6092848244687114e-05,
+      "loss": 2.2508,
+      "step": 102864
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.608657161667206e-05,
+      "loss": 2.2576,
+      "step": 102872
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.608029508594527e-05,
+      "loss": 2.2466,
+      "step": 102880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6074018652662774e-05,
+      "loss": 2.2255,
+      "step": 102888
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.606774231698064e-05,
+      "loss": 2.2618,
+      "step": 102896
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6061466079054844e-05,
+      "loss": 2.2587,
+      "step": 102904
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6055189939041455e-05,
+      "loss": 2.2378,
+      "step": 102912
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.604891389709648e-05,
+      "loss": 2.2319,
+      "step": 102920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6042637953375974e-05,
+      "loss": 2.2452,
+      "step": 102928
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.603636210803591e-05,
+      "loss": 2.2473,
+      "step": 102936
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6030086361232324e-05,
+      "loss": 2.2536,
+      "step": 102944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.602381071312126e-05,
+      "loss": 2.2586,
+      "step": 102952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.601753516385869e-05,
+      "loss": 2.2531,
+      "step": 102960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.601125971360064e-05,
+      "loss": 2.241,
+      "step": 102968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.600498436250312e-05,
+      "loss": 2.2581,
+      "step": 102976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5998709110722156e-05,
+      "loss": 2.2515,
+      "step": 102984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5992433958413714e-05,
+      "loss": 2.2649,
+      "step": 102992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.598615890573382e-05,
+      "loss": 2.2493,
+      "step": 103000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5979883952838485e-05,
+      "loss": 2.2705,
+      "step": 103008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.597360909988365e-05,
+      "loss": 2.2576,
+      "step": 103016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.596733434702537e-05,
+      "loss": 2.2382,
+      "step": 103024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.59610596944196e-05,
+      "loss": 2.2417,
+      "step": 103032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.595478514222234e-05,
+      "loss": 2.2722,
+      "step": 103040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5948510690589576e-05,
+      "loss": 2.2553,
+      "step": 103048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.594223633967729e-05,
+      "loss": 2.2603,
+      "step": 103056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.593596208964147e-05,
+      "loss": 2.2565,
+      "step": 103064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.592968794063809e-05,
+      "loss": 2.2469,
+      "step": 103072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5923413892823104e-05,
+      "loss": 2.2375,
+      "step": 103080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.591713994635253e-05,
+      "loss": 2.2818,
+      "step": 103088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5910866101382296e-05,
+      "loss": 2.2491,
+      "step": 103096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.59045923580684e-05,
+      "loss": 2.2446,
+      "step": 103104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.589831871656679e-05,
+      "loss": 2.256,
+      "step": 103112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.589204517703346e-05,
+      "loss": 2.2602,
+      "step": 103120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5885771739624325e-05,
+      "loss": 2.2528,
+      "step": 103128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.587949840449535e-05,
+      "loss": 2.2266,
+      "step": 103136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5873225171802555e-05,
+      "loss": 2.2518,
+      "step": 103144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5866952041701825e-05,
+      "loss": 2.2538,
+      "step": 103152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5860679014349105e-05,
+      "loss": 2.2532,
+      "step": 103160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.58544060899004e-05,
+      "loss": 2.2464,
+      "step": 103168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5848133268511614e-05,
+      "loss": 2.2417,
+      "step": 103176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.584186055033871e-05,
+      "loss": 2.243,
+      "step": 103184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.583558793553759e-05,
+      "loss": 2.2314,
+      "step": 103192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.582931542426426e-05,
+      "loss": 2.2523,
+      "step": 103200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.582304301667459e-05,
+      "loss": 2.2481,
+      "step": 103208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.581677071292455e-05,
+      "loss": 2.2445,
+      "step": 103216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5810498513170064e-05,
+      "loss": 2.2518,
+      "step": 103224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.580422641756704e-05,
+      "loss": 2.2391,
+      "step": 103232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.579795442627143e-05,
+      "loss": 2.2555,
+      "step": 103240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.579168253943914e-05,
+      "loss": 2.2682,
+      "step": 103248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.57854107572261e-05,
+      "loss": 2.2466,
+      "step": 103256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.57791390797882e-05,
+      "loss": 2.2415,
+      "step": 103264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.577286750728139e-05,
+      "loss": 2.2532,
+      "step": 103272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5766596039861585e-05,
+      "loss": 2.258,
+      "step": 103280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.576032467768465e-05,
+      "loss": 2.2551,
+      "step": 103288
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5754053420906535e-05,
+      "loss": 2.2373,
+      "step": 103296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5747782269683126e-05,
+      "loss": 2.2476,
+      "step": 103304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.574151122417033e-05,
+      "loss": 2.2469,
+      "step": 103312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.573524028452403e-05,
+      "loss": 2.2461,
+      "step": 103320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.572896945090014e-05,
+      "loss": 2.2349,
+      "step": 103328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.572269872345457e-05,
+      "loss": 2.2603,
+      "step": 103336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.571642810234315e-05,
+      "loss": 2.2616,
+      "step": 103344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.571015758772184e-05,
+      "loss": 2.2404,
+      "step": 103352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.570388717974648e-05,
+      "loss": 2.2586,
+      "step": 103360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5697616878572975e-05,
+      "loss": 2.2463,
+      "step": 103368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.569134668435718e-05,
+      "loss": 2.2399,
+      "step": 103376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5685076597255014e-05,
+      "loss": 2.2407,
+      "step": 103384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.567880661742232e-05,
+      "loss": 2.2541,
+      "step": 103392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.567253674501497e-05,
+      "loss": 2.25,
+      "step": 103400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5666266980188844e-05,
+      "loss": 2.251,
+      "step": 103408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5659997323099814e-05,
+      "loss": 2.2477,
+      "step": 103416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.565372777390373e-05,
+      "loss": 2.2251,
+      "step": 103424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5647458332756464e-05,
+      "loss": 2.2566,
+      "step": 103432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.564118899981388e-05,
+      "loss": 2.2469,
+      "step": 103440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.563491977523181e-05,
+      "loss": 2.2367,
+      "step": 103448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.562865065916614e-05,
+      "loss": 2.2572,
+      "step": 103456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5622381651772696e-05,
+      "loss": 2.2509,
+      "step": 103464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.561611275320733e-05,
+      "loss": 2.2415,
+      "step": 103472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5609843963625896e-05,
+      "loss": 2.2641,
+      "step": 103480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.560357528318424e-05,
+      "loss": 2.2488,
+      "step": 103488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5597306712038196e-05,
+      "loss": 2.2365,
+      "step": 103496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.55910382503436e-05,
+      "loss": 2.2313,
+      "step": 103504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.558476989825628e-05,
+      "loss": 2.2469,
+      "step": 103512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.55785016559321e-05,
+      "loss": 2.2418,
+      "step": 103520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5572233523526826e-05,
+      "loss": 2.251,
+      "step": 103528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.556596550119635e-05,
+      "loss": 2.2519,
+      "step": 103536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.555969758909646e-05,
+      "loss": 2.2465,
+      "step": 103544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5553429787383005e-05,
+      "loss": 2.2472,
+      "step": 103552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.554716209621176e-05,
+      "loss": 2.2654,
+      "step": 103560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.554089451573856e-05,
+      "loss": 2.2372,
+      "step": 103568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.553462704611927e-05,
+      "loss": 2.2538,
+      "step": 103576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.55283596875096e-05,
+      "loss": 2.254,
+      "step": 103584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.552209244006543e-05,
+      "loss": 2.2566,
+      "step": 103592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.551582530394255e-05,
+      "loss": 2.2547,
+      "step": 103600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.550955827929675e-05,
+      "loss": 2.2407,
+      "step": 103608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.550329136628383e-05,
+      "loss": 2.248,
+      "step": 103616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.549702456505959e-05,
+      "loss": 2.2591,
+      "step": 103624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.549075787577984e-05,
+      "loss": 2.2528,
+      "step": 103632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.548449129860034e-05,
+      "loss": 2.2419,
+      "step": 103640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.54782248336769e-05,
+      "loss": 2.2545,
+      "step": 103648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.54719584811653e-05,
+      "loss": 2.2364,
+      "step": 103656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5465692241221314e-05,
+      "loss": 2.2477,
+      "step": 103664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5459426114000725e-05,
+      "loss": 2.2575,
+      "step": 103672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5453160099659305e-05,
+      "loss": 2.246,
+      "step": 103680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.544689419835287e-05,
+      "loss": 2.2514,
+      "step": 103688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5440628410237114e-05,
+      "loss": 2.2352,
+      "step": 103696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.543436273546787e-05,
+      "loss": 2.2669,
+      "step": 103704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.542809717420089e-05,
+      "loss": 2.256,
+      "step": 103712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.542183172659191e-05,
+      "loss": 2.258,
+      "step": 103720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.54155663927967e-05,
+      "loss": 2.249,
+      "step": 103728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5409301172971054e-05,
+      "loss": 2.2605,
+      "step": 103736
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.540303606727066e-05,
+      "loss": 2.2399,
+      "step": 103744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.539677107585133e-05,
+      "loss": 2.25,
+      "step": 103752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5390506198868775e-05,
+      "loss": 2.2376,
+      "step": 103760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.538424143647878e-05,
+      "loss": 2.2566,
+      "step": 103768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5377976788837025e-05,
+      "loss": 2.2488,
+      "step": 103776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.53717122560993e-05,
+      "loss": 2.2668,
+      "step": 103784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5365447838421325e-05,
+      "loss": 2.2556,
+      "step": 103792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5359183535958834e-05,
+      "loss": 2.2515,
+      "step": 103800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.535291934886757e-05,
+      "loss": 2.2629,
+      "step": 103808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.534665527730324e-05,
+      "loss": 2.2464,
+      "step": 103816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5340391321421605e-05,
+      "loss": 2.2407,
+      "step": 103824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.533412748137834e-05,
+      "loss": 2.2585,
+      "step": 103832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.53278637573292e-05,
+      "loss": 2.2475,
+      "step": 103840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5321600149429896e-05,
+      "loss": 2.2348,
+      "step": 103848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5315336657836125e-05,
+      "loss": 2.2359,
+      "step": 103856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.530907328270362e-05,
+      "loss": 2.2516,
+      "step": 103864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.530281002418808e-05,
+      "loss": 2.2438,
+      "step": 103872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5296546882445214e-05,
+      "loss": 2.2433,
+      "step": 103880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.52902838576307e-05,
+      "loss": 2.248,
+      "step": 103888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.528402094990026e-05,
+      "loss": 2.2609,
+      "step": 103896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.527775815940961e-05,
+      "loss": 2.2357,
+      "step": 103904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5271495486314406e-05,
+      "loss": 2.2576,
+      "step": 103912
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.526523293077035e-05,
+      "loss": 2.2473,
+      "step": 103920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.525897049293316e-05,
+      "loss": 2.2498,
+      "step": 103928
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5252708172958465e-05,
+      "loss": 2.246,
+      "step": 103936
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.524644597100199e-05,
+      "loss": 2.2348,
+      "step": 103944
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.524018388721942e-05,
+      "loss": 2.2601,
+      "step": 103952
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5233921921766384e-05,
+      "loss": 2.2678,
+      "step": 103960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.522766007479859e-05,
+      "loss": 2.2414,
+      "step": 103968
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5221398346471706e-05,
+      "loss": 2.2467,
+      "step": 103976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.521513673694141e-05,
+      "loss": 2.2393,
+      "step": 103984
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.520887524636333e-05,
+      "loss": 2.2432,
+      "step": 103992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.520261387489316e-05,
+      "loss": 2.2404,
+      "step": 104000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5196352622686555e-05,
+      "loss": 2.247,
+      "step": 104008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.519009148989915e-05,
+      "loss": 2.2417,
+      "step": 104016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.51838304766866e-05,
+      "loss": 2.2384,
+      "step": 104024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5177569583204566e-05,
+      "loss": 2.2349,
+      "step": 104032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.517130880960872e-05,
+      "loss": 2.2426,
+      "step": 104040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5165048156054656e-05,
+      "loss": 2.2473,
+      "step": 104048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.515878762269805e-05,
+      "loss": 2.2466,
+      "step": 104056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5152527209694524e-05,
+      "loss": 2.234,
+      "step": 104064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5146266917199704e-05,
+      "loss": 2.2166,
+      "step": 104072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5140006745369255e-05,
+      "loss": 2.236,
+      "step": 104080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5133746694358765e-05,
+      "loss": 2.2457,
+      "step": 104088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5127486764323885e-05,
+      "loss": 2.2367,
+      "step": 104096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5121226955420226e-05,
+      "loss": 2.246,
+      "step": 104104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5114967267803405e-05,
+      "loss": 2.227,
+      "step": 104112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.510870770162906e-05,
+      "loss": 2.2365,
+      "step": 104120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5102448257052796e-05,
+      "loss": 2.234,
+      "step": 104128
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.509618893423021e-05,
+      "loss": 2.2288,
+      "step": 104136
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5089929733316906e-05,
+      "loss": 2.2461,
+      "step": 104144
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.508367065446851e-05,
+      "loss": 2.2438,
+      "step": 104152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.507741169784062e-05,
+      "loss": 2.2208,
+      "step": 104160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5071152863588814e-05,
+      "loss": 2.2322,
+      "step": 104168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.50648941518687e-05,
+      "loss": 2.2321,
+      "step": 104176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.505863556283588e-05,
+      "loss": 2.2469,
+      "step": 104184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.505237709664592e-05,
+      "loss": 2.2512,
+      "step": 104192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.504611875345446e-05,
+      "loss": 2.246,
+      "step": 104200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.503986053341701e-05,
+      "loss": 2.2477,
+      "step": 104208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.503360243668918e-05,
+      "loss": 2.2397,
+      "step": 104216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.502734446342657e-05,
+      "loss": 2.2507,
+      "step": 104224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5021086613784735e-05,
+      "loss": 2.2496,
+      "step": 104232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5014828887919225e-05,
+      "loss": 2.2523,
+      "step": 104240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.500857128598562e-05,
+      "loss": 2.2437,
+      "step": 104248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.500231380813954e-05,
+      "loss": 2.2521,
+      "step": 104256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.499605645453646e-05,
+      "loss": 2.2364,
+      "step": 104264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4989799225331984e-05,
+      "loss": 2.2305,
+      "step": 104272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.498354212068168e-05,
+      "loss": 2.2442,
+      "step": 104280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4977285140741055e-05,
+      "loss": 2.2387,
+      "step": 104288
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.497102828566569e-05,
+      "loss": 2.239,
+      "step": 104296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4964771555611147e-05,
+      "loss": 2.2566,
+      "step": 104304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4958514950732945e-05,
+      "loss": 2.2321,
+      "step": 104312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.495225847118661e-05,
+      "loss": 2.2455,
+      "step": 104320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.494600211712769e-05,
+      "loss": 2.2514,
+      "step": 104328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.493974588871175e-05,
+      "loss": 2.2365,
+      "step": 104336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4933489786094275e-05,
+      "loss": 2.2602,
+      "step": 104344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.492723380943081e-05,
+      "loss": 2.2469,
+      "step": 104352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.492097795887692e-05,
+      "loss": 2.2431,
+      "step": 104360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.491472223458804e-05,
+      "loss": 2.2432,
+      "step": 104368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.490846663671976e-05,
+      "loss": 2.2449,
+      "step": 104376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4902211165427555e-05,
+      "loss": 2.2294,
+      "step": 104384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4895955820866976e-05,
+      "loss": 2.2472,
+      "step": 104392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4889700603193484e-05,
+      "loss": 2.25,
+      "step": 104400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4883445512562615e-05,
+      "loss": 2.2443,
+      "step": 104408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4877190549129884e-05,
+      "loss": 2.257,
+      "step": 104416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.487093571305075e-05,
+      "loss": 2.2515,
+      "step": 104424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.486468100448072e-05,
+      "loss": 2.2374,
+      "step": 104432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.485842642357534e-05,
+      "loss": 2.2369,
+      "step": 104440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4852171970490016e-05,
+      "loss": 2.2392,
+      "step": 104448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.48459176453803e-05,
+      "loss": 2.2617,
+      "step": 104456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.483966344840161e-05,
+      "loss": 2.2392,
+      "step": 104464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.483340937970951e-05,
+      "loss": 2.2438,
+      "step": 104472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.482715543945941e-05,
+      "loss": 2.2423,
+      "step": 104480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.48209016278068e-05,
+      "loss": 2.2398,
+      "step": 104488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.481464794490718e-05,
+      "loss": 2.2349,
+      "step": 104496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.480839439091598e-05,
+      "loss": 2.2447,
+      "step": 104504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4802140965988666e-05,
+      "loss": 2.2292,
+      "step": 104512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.479588767028072e-05,
+      "loss": 2.2488,
+      "step": 104520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4789634503947606e-05,
+      "loss": 2.2427,
+      "step": 104528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4783381467144745e-05,
+      "loss": 2.2516,
+      "step": 104536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4777128560027615e-05,
+      "loss": 2.2349,
+      "step": 104544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.477087578275165e-05,
+      "loss": 2.2558,
+      "step": 104552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4764623135472305e-05,
+      "loss": 2.2298,
+      "step": 104560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4758370618345e-05,
+      "loss": 2.2464,
+      "step": 104568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.475211823152522e-05,
+      "loss": 2.2473,
+      "step": 104576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.474586597516833e-05,
+      "loss": 2.2344,
+      "step": 104584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.473961384942983e-05,
+      "loss": 2.2344,
+      "step": 104592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.473336185446511e-05,
+      "loss": 2.2445,
+      "step": 104600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4727109990429626e-05,
+      "loss": 2.2404,
+      "step": 104608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.472085825747875e-05,
+      "loss": 2.2593,
+      "step": 104616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.471460665576793e-05,
+      "loss": 2.2631,
+      "step": 104624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.470835518545261e-05,
+      "loss": 2.2304,
+      "step": 104632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4702103846688145e-05,
+      "loss": 2.2522,
+      "step": 104640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.469585263962997e-05,
+      "loss": 2.2466,
+      "step": 104648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4689601564433525e-05,
+      "loss": 2.2521,
+      "step": 104656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.468335062125417e-05,
+      "loss": 2.2497,
+      "step": 104664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.467709981024731e-05,
+      "loss": 2.2399,
+      "step": 104672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.467084913156833e-05,
+      "loss": 2.2399,
+      "step": 104680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.466459858537267e-05,
+      "loss": 2.2573,
+      "step": 104688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.465834817181566e-05,
+      "loss": 2.2471,
+      "step": 104696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4652097891052724e-05,
+      "loss": 2.2665,
+      "step": 104704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4645847743239244e-05,
+      "loss": 2.2514,
+      "step": 104712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.463959772853058e-05,
+      "loss": 2.2545,
+      "step": 104720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4633347847082115e-05,
+      "loss": 2.2506,
+      "step": 104728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4627098099049226e-05,
+      "loss": 2.2514,
+      "step": 104736
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.462084848458729e-05,
+      "loss": 2.2448,
+      "step": 104744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.461459900385165e-05,
+      "loss": 2.2327,
+      "step": 104752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.460834965699768e-05,
+      "loss": 2.2496,
+      "step": 104760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.460210044418077e-05,
+      "loss": 2.2401,
+      "step": 104768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.459585136555623e-05,
+      "loss": 2.2444,
+      "step": 104776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.458960242127943e-05,
+      "loss": 2.2498,
+      "step": 104784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.458335361150574e-05,
+      "loss": 2.2451,
+      "step": 104792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457710493639046e-05,
+      "loss": 2.2471,
+      "step": 104800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457085639608897e-05,
+      "loss": 2.2483,
+      "step": 104808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.45646079907566e-05,
+      "loss": 2.2462,
+      "step": 104816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.455835972054869e-05,
+      "loss": 2.2301,
+      "step": 104824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.455211158562055e-05,
+      "loss": 2.2361,
+      "step": 104832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454586358612754e-05,
+      "loss": 2.2337,
+      "step": 104840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453961572222499e-05,
+      "loss": 2.2604,
+      "step": 104848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453336799406818e-05,
+      "loss": 2.239,
+      "step": 104856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452712040181247e-05,
+      "loss": 2.25,
+      "step": 104864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452087294561315e-05,
+      "loss": 2.2485,
+      "step": 104872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.451462562562555e-05,
+      "loss": 2.2449,
+      "step": 104880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450837844200497e-05,
+      "loss": 2.2545,
+      "step": 104888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450213139490672e-05,
+      "loss": 2.2442,
+      "step": 104896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4495884484486106e-05,
+      "loss": 2.2469,
+      "step": 104904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448963771089841e-05,
+      "loss": 2.2428,
+      "step": 104912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448339107429894e-05,
+      "loss": 2.2457,
+      "step": 104920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447714457484298e-05,
+      "loss": 2.2471,
+      "step": 104928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447089821268585e-05,
+      "loss": 2.2417,
+      "step": 104936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.446465198798279e-05,
+      "loss": 2.2436,
+      "step": 104944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445840590088911e-05,
+      "loss": 2.2513,
+      "step": 104952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445215995156008e-05,
+      "loss": 2.2393,
+      "step": 104960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4445914140150964e-05,
+      "loss": 2.2443,
+      "step": 104968
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4439668466817075e-05,
+      "loss": 2.2382,
+      "step": 104976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.443342293171363e-05,
+      "loss": 2.2456,
+      "step": 104984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.442717753499592e-05,
+      "loss": 2.247,
+      "step": 104992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.442093227681919e-05,
+      "loss": 2.2372,
+      "step": 105000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4414687157338725e-05,
+      "loss": 2.2485,
+      "step": 105008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4408442176709776e-05,
+      "loss": 2.2512,
+      "step": 105016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440219733508756e-05,
+      "loss": 2.2399,
+      "step": 105024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439595263262735e-05,
+      "loss": 2.2545,
+      "step": 105032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4389708069484404e-05,
+      "loss": 2.2643,
+      "step": 105040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4383463645813925e-05,
+      "loss": 2.2457,
+      "step": 105048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.437721936177116e-05,
+      "loss": 2.2547,
+      "step": 105056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4370975217511376e-05,
+      "loss": 2.2493,
+      "step": 105064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4364731213189766e-05,
+      "loss": 2.263,
+      "step": 105072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4358487348961585e-05,
+      "loss": 2.2511,
+      "step": 105080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4352243624982015e-05,
+      "loss": 2.2559,
+      "step": 105088
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434600004140633e-05,
+      "loss": 2.2404,
+      "step": 105096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43397565983897e-05,
+      "loss": 2.2427,
+      "step": 105104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433351329608736e-05,
+      "loss": 2.2496,
+      "step": 105112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432727013465453e-05,
+      "loss": 2.2461,
+      "step": 105120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432102711424638e-05,
+      "loss": 2.2379,
+      "step": 105128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431478423501815e-05,
+      "loss": 2.2447,
+      "step": 105136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430854149712501e-05,
+      "loss": 2.2376,
+      "step": 105144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430229890072217e-05,
+      "loss": 2.2501,
+      "step": 105152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.429605644596483e-05,
+      "loss": 2.2321,
+      "step": 105160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.428981413300814e-05,
+      "loss": 2.2438,
+      "step": 105168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4283571962007326e-05,
+      "loss": 2.2501,
+      "step": 105176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.427732993311756e-05,
+      "loss": 2.2443,
+      "step": 105184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4271088046494015e-05,
+      "loss": 2.2543,
+      "step": 105192
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426484630229187e-05,
+      "loss": 2.2552,
+      "step": 105200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.425860470066626e-05,
+      "loss": 2.2466,
+      "step": 105208
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.425236324177241e-05,
+      "loss": 2.2543,
+      "step": 105216
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424612192576544e-05,
+      "loss": 2.2431,
+      "step": 105224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423988075280053e-05,
+      "loss": 2.2619,
+      "step": 105232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423363972303283e-05,
+      "loss": 2.2311,
+      "step": 105240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4227398836617494e-05,
+      "loss": 2.2697,
+      "step": 105248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.422115809370967e-05,
+      "loss": 2.2258,
+      "step": 105256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4214917494464504e-05,
+      "loss": 2.2576,
+      "step": 105264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.420867703903714e-05,
+      "loss": 2.2407,
+      "step": 105272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4202436727582714e-05,
+      "loss": 2.2596,
+      "step": 105280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.419619656025636e-05,
+      "loss": 2.2547,
+      "step": 105288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4189956537213206e-05,
+      "loss": 2.2566,
+      "step": 105296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.418371665860837e-05,
+      "loss": 2.2602,
+      "step": 105304
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.417747692459702e-05,
+      "loss": 2.2356,
+      "step": 105312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.417123733533425e-05,
+      "loss": 2.2486,
+      "step": 105320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416499789097515e-05,
+      "loss": 2.2426,
+      "step": 105328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4158758591674874e-05,
+      "loss": 2.2417,
+      "step": 105336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.41525194375885e-05,
+      "loss": 2.2367,
+      "step": 105344
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414628042887116e-05,
+      "loss": 2.2536,
+      "step": 105352
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4140041565677935e-05,
+      "loss": 2.2419,
+      "step": 105360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413380284816395e-05,
+      "loss": 2.2283,
+      "step": 105368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412756427648427e-05,
+      "loss": 2.2505,
+      "step": 105376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412132585079401e-05,
+      "loss": 2.2644,
+      "step": 105384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.411508757124826e-05,
+      "loss": 2.2461,
+      "step": 105392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410884943800207e-05,
+      "loss": 2.246,
+      "step": 105400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410261145121054e-05,
+      "loss": 2.2343,
+      "step": 105408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4096373611028784e-05,
+      "loss": 2.2463,
+      "step": 105416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4090135917611806e-05,
+      "loss": 2.2541,
+      "step": 105424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.408389837111474e-05,
+      "loss": 2.2601,
+      "step": 105432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.40776609716926e-05,
+      "loss": 2.2384,
+      "step": 105440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4071423719500504e-05,
+      "loss": 2.2498,
+      "step": 105448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406518661469344e-05,
+      "loss": 2.2519,
+      "step": 105456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4058949657426525e-05,
+      "loss": 2.2585,
+      "step": 105464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4052712847854804e-05,
+      "loss": 2.2452,
+      "step": 105472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404647618613329e-05,
+      "loss": 2.234,
+      "step": 105480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404023967241703e-05,
+      "loss": 2.2379,
+      "step": 105488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.403400330686112e-05,
+      "loss": 2.2528,
+      "step": 105496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402776708962052e-05,
+      "loss": 2.2472,
+      "step": 105504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402153102085032e-05,
+      "loss": 2.2451,
+      "step": 105512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401529510070551e-05,
+      "loss": 2.2503,
+      "step": 105520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4009059329341174e-05,
+      "loss": 2.2425,
+      "step": 105528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4002823706912265e-05,
+      "loss": 2.25,
+      "step": 105536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399658823357383e-05,
+      "loss": 2.2604,
+      "step": 105544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3990352909480905e-05,
+      "loss": 2.2649,
+      "step": 105552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398411773478846e-05,
+      "loss": 2.2457,
+      "step": 105560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3977882709651536e-05,
+      "loss": 2.2493,
+      "step": 105568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3971647834225117e-05,
+      "loss": 2.2547,
+      "step": 105576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396541310866421e-05,
+      "loss": 2.2405,
+      "step": 105584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3959178533123814e-05,
+      "loss": 2.235,
+      "step": 105592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3952944107758905e-05,
+      "loss": 2.2443,
+      "step": 105600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3946709832724495e-05,
+      "loss": 2.2507,
+      "step": 105608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3940475708175565e-05,
+      "loss": 2.2622,
+      "step": 105616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393424173426708e-05,
+      "loss": 2.251,
+      "step": 105624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.392800791115404e-05,
+      "loss": 2.261,
+      "step": 105632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.392177423899139e-05,
+      "loss": 2.2623,
+      "step": 105640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39155407179341e-05,
+      "loss": 2.2609,
+      "step": 105648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.390930734813716e-05,
+      "loss": 2.2545,
+      "step": 105656
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3903074129755545e-05,
+      "loss": 2.2417,
+      "step": 105664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389684106294417e-05,
+      "loss": 2.2433,
+      "step": 105672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3890608147858e-05,
+      "loss": 2.2421,
+      "step": 105680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.388437538465203e-05,
+      "loss": 2.2513,
+      "step": 105688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3878142773481156e-05,
+      "loss": 2.2786,
+      "step": 105696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.387191031450034e-05,
+      "loss": 2.2598,
+      "step": 105704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3865678007864506e-05,
+      "loss": 2.2572,
+      "step": 105712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.385944585372864e-05,
+      "loss": 2.2438,
+      "step": 105720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.385321385224762e-05,
+      "loss": 2.2429,
+      "step": 105728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384698200357638e-05,
+      "loss": 2.2575,
+      "step": 105736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384075030786988e-05,
+      "loss": 2.2481,
+      "step": 105744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3834518765283e-05,
+      "loss": 2.2525,
+      "step": 105752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.38282873759707e-05,
+      "loss": 2.2427,
+      "step": 105760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382205614008785e-05,
+      "loss": 2.2415,
+      "step": 105768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3815825057789374e-05,
+      "loss": 2.262,
+      "step": 105776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.380959412923019e-05,
+      "loss": 2.235,
+      "step": 105784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3803363354565186e-05,
+      "loss": 2.2395,
+      "step": 105792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3797132733949266e-05,
+      "loss": 2.2272,
+      "step": 105800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379090226753733e-05,
+      "loss": 2.2473,
+      "step": 105808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378467195548425e-05,
+      "loss": 2.2453,
+      "step": 105816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.377844179794492e-05,
+      "loss": 2.2481,
+      "step": 105824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3772211795074235e-05,
+      "loss": 2.2467,
+      "step": 105832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3765981947027045e-05,
+      "loss": 2.2528,
+      "step": 105840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3759752253958274e-05,
+      "loss": 2.2342,
+      "step": 105848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.375352271602274e-05,
+      "loss": 2.2593,
+      "step": 105856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.374729333337532e-05,
+      "loss": 2.2613,
+      "step": 105864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.374106410617091e-05,
+      "loss": 2.2522,
+      "step": 105872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373483503456436e-05,
+      "loss": 2.2443,
+      "step": 105880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.372860611871049e-05,
+      "loss": 2.2659,
+      "step": 105888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3722377358764176e-05,
+      "loss": 2.2505,
+      "step": 105896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.371614875488029e-05,
+      "loss": 2.2427,
+      "step": 105904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3709920307213626e-05,
+      "loss": 2.2464,
+      "step": 105912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.370369201591906e-05,
+      "loss": 2.2453,
+      "step": 105920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36974638811514e-05,
+      "loss": 2.2518,
+      "step": 105928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3691235903065526e-05,
+      "loss": 2.2454,
+      "step": 105936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.368500808181622e-05,
+      "loss": 2.2568,
+      "step": 105944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367878041755832e-05,
+      "loss": 2.2488,
+      "step": 105952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367255291044664e-05,
+      "loss": 2.2361,
+      "step": 105960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3666325560636024e-05,
+      "loss": 2.2468,
+      "step": 105968
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3660098368281244e-05,
+      "loss": 2.2579,
+      "step": 105976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3653871333537145e-05,
+      "loss": 2.2424,
+      "step": 105984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36476444565585e-05,
+      "loss": 2.2493,
+      "step": 105992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3641417737500146e-05,
+      "loss": 2.2524,
+      "step": 106000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3635191176516855e-05,
+      "loss": 2.2367,
+      "step": 106008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3628964773763424e-05,
+      "loss": 2.2207,
+      "step": 106016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362273852939463e-05,
+      "loss": 2.2342,
+      "step": 106024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.361651244356529e-05,
+      "loss": 2.2267,
+      "step": 106032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3610286516430166e-05,
+      "loss": 2.2497,
+      "step": 106040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360406074814403e-05,
+      "loss": 2.2328,
+      "step": 106048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.359783513886165e-05,
+      "loss": 2.2402,
+      "step": 106056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.359160968873782e-05,
+      "loss": 2.2471,
+      "step": 106064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.358538439792731e-05,
+      "loss": 2.2417,
+      "step": 106072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3579159266584834e-05,
+      "loss": 2.252,
+      "step": 106080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357293429486518e-05,
+      "loss": 2.2416,
+      "step": 106088
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356670948292313e-05,
+      "loss": 2.2405,
+      "step": 106096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356048483091338e-05,
+      "loss": 2.2501,
+      "step": 106104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.355426033899069e-05,
+      "loss": 2.2331,
+      "step": 106112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.354803600730984e-05,
+      "loss": 2.2477,
+      "step": 106120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.354181183602552e-05,
+      "loss": 2.2532,
+      "step": 106128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3535587825292485e-05,
+      "loss": 2.2284,
+      "step": 106136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352936397526544e-05,
+      "loss": 2.2576,
+      "step": 106144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352314028609918e-05,
+      "loss": 2.2397,
+      "step": 106152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3516916757948364e-05,
+      "loss": 2.2287,
+      "step": 106160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3510693390967705e-05,
+      "loss": 2.258,
+      "step": 106168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.350447018531195e-05,
+      "loss": 2.2459,
+      "step": 106176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349824714113579e-05,
+      "loss": 2.2329,
+      "step": 106184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349202425859394e-05,
+      "loss": 2.2466,
+      "step": 106192
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3485801537841105e-05,
+      "loss": 2.2569,
+      "step": 106200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3479578979031955e-05,
+      "loss": 2.2581,
+      "step": 106208
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347335658232121e-05,
+      "loss": 2.2484,
+      "step": 106216
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3467134347863544e-05,
+      "loss": 2.2539,
+      "step": 106224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.346091227581368e-05,
+      "loss": 2.2275,
+      "step": 106232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3454690366326247e-05,
+      "loss": 2.2476,
+      "step": 106240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.344846861955595e-05,
+      "loss": 2.2551,
+      "step": 106248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3442247035657475e-05,
+      "loss": 2.2348,
+      "step": 106256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.343602561478545e-05,
+      "loss": 2.2449,
+      "step": 106264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.342980435709457e-05,
+      "loss": 2.2469,
+      "step": 106272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3423583262739484e-05,
+      "loss": 2.2382,
+      "step": 106280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3417362331874885e-05,
+      "loss": 2.2528,
+      "step": 106288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3411141564655364e-05,
+      "loss": 2.2409,
+      "step": 106296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340492096123562e-05,
+      "loss": 2.2628,
+      "step": 106304
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.33987005217703e-05,
+      "loss": 2.2483,
+      "step": 106312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3392480246413986e-05,
+      "loss": 2.2388,
+      "step": 106320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338626013532138e-05,
+      "loss": 2.2499,
+      "step": 106328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338004018864708e-05,
+      "loss": 2.2473,
+      "step": 106336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337382040654573e-05,
+      "loss": 2.2528,
+      "step": 106344
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.336760078917194e-05,
+      "loss": 2.2326,
+      "step": 106352
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.336138133668034e-05,
+      "loss": 2.2514,
+      "step": 106360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3355162049225546e-05,
+      "loss": 2.2477,
+      "step": 106368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3348942926962176e-05,
+      "loss": 2.2452,
+      "step": 106376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.334272397004483e-05,
+      "loss": 2.2355,
+      "step": 106384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.33365051786281e-05,
+      "loss": 2.2374,
+      "step": 106392
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3330286552866606e-05,
+      "loss": 2.2272,
+      "step": 106400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.332406809291494e-05,
+      "loss": 2.2386,
+      "step": 106408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.331784979892769e-05,
+      "loss": 2.2478,
+      "step": 106416
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.331163167105944e-05,
+      "loss": 2.2541,
+      "step": 106424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.330541370946478e-05,
+      "loss": 2.2379,
+      "step": 106432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3299195914298286e-05,
+      "loss": 2.2394,
+      "step": 106440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.329297828571453e-05,
+      "loss": 2.2435,
+      "step": 106448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.328676082386809e-05,
+      "loss": 2.2293,
+      "step": 106456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.328054352891352e-05,
+      "loss": 2.2555,
+      "step": 106464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.327432640100541e-05,
+      "loss": 2.241,
+      "step": 106472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3268109440298274e-05,
+      "loss": 2.2575,
+      "step": 106480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.326189264694672e-05,
+      "loss": 2.2305,
+      "step": 106488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.325567602110525e-05,
+      "loss": 2.2451,
+      "step": 106496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324945956292846e-05,
+      "loss": 2.236,
+      "step": 106504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324324327257082e-05,
+      "loss": 2.2498,
+      "step": 106512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.323702715018694e-05,
+      "loss": 2.2475,
+      "step": 106520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.323081119593133e-05,
+      "loss": 2.2482,
+      "step": 106528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3224595409958497e-05,
+      "loss": 2.25,
+      "step": 106536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3218379792422966e-05,
+      "loss": 2.2473,
+      "step": 106544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.321216434347931e-05,
+      "loss": 2.2473,
+      "step": 106552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3205949063282e-05,
+      "loss": 2.2631,
+      "step": 106560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.319973395198554e-05,
+      "loss": 2.2586,
+      "step": 106568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3193519009744474e-05,
+      "loss": 2.2473,
+      "step": 106576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318730423671329e-05,
+      "loss": 2.2635,
+      "step": 106584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318108963304648e-05,
+      "loss": 2.2503,
+      "step": 106592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3174875198898545e-05,
+      "loss": 2.2422,
+      "step": 106600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316866093442399e-05,
+      "loss": 2.2463,
+      "step": 106608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316244683977727e-05,
+      "loss": 2.2246,
+      "step": 106616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3156232915112894e-05,
+      "loss": 2.2385,
+      "step": 106624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3150019160585354e-05,
+      "loss": 2.2451,
+      "step": 106632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.314380557634909e-05,
+      "loss": 2.2407,
+      "step": 106640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3137592162558595e-05,
+      "loss": 2.2343,
+      "step": 106648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3131378919368325e-05,
+      "loss": 2.2464,
+      "step": 106656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.312516584693275e-05,
+      "loss": 2.2498,
+      "step": 106664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3118952945406316e-05,
+      "loss": 2.2477,
+      "step": 106672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311274021494348e-05,
+      "loss": 2.2503,
+      "step": 106680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3106527655698726e-05,
+      "loss": 2.2436,
+      "step": 106688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310031526782644e-05,
+      "loss": 2.2494,
+      "step": 106696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309410305148108e-05,
+      "loss": 2.2491,
+      "step": 106704
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.308789100681712e-05,
+      "loss": 2.2363,
+      "step": 106712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.308167913398896e-05,
+      "loss": 2.2387,
+      "step": 106720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3075467433151026e-05,
+      "loss": 2.2503,
+      "step": 106728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306925590445775e-05,
+      "loss": 2.2328,
+      "step": 106736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306304454806356e-05,
+      "loss": 2.2539,
+      "step": 106744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3056833364122835e-05,
+      "loss": 2.2502,
+      "step": 106752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305062235279004e-05,
+      "loss": 2.249,
+      "step": 106760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.304441151421952e-05,
+      "loss": 2.2498,
+      "step": 106768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303820084856574e-05,
+      "loss": 2.2412,
+      "step": 106776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3031990355983054e-05,
+      "loss": 2.2648,
+      "step": 106784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302578003662588e-05,
+      "loss": 2.2543,
+      "step": 106792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30195698906486e-05,
+      "loss": 2.2375,
+      "step": 106800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.301335991820557e-05,
+      "loss": 2.2364,
+      "step": 106808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300715011945121e-05,
+      "loss": 2.2442,
+      "step": 106816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3000940494539894e-05,
+      "loss": 2.2537,
+      "step": 106824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299473104362598e-05,
+      "loss": 2.2396,
+      "step": 106832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.298852176686382e-05,
+      "loss": 2.2425,
+      "step": 106840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.298231266440781e-05,
+      "loss": 2.2447,
+      "step": 106848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29761037364123e-05,
+      "loss": 2.239,
+      "step": 106856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.296989498303162e-05,
+      "loss": 2.2459,
+      "step": 106864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2963686404420154e-05,
+      "loss": 2.228,
+      "step": 106872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295747800073223e-05,
+      "loss": 2.2536,
+      "step": 106880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295126977212219e-05,
+      "loss": 2.2423,
+      "step": 106888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294506171874438e-05,
+      "loss": 2.2481,
+      "step": 106896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.293885384075315e-05,
+      "loss": 2.2603,
+      "step": 106904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.293264613830277e-05,
+      "loss": 2.2475,
+      "step": 106912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29264386115476e-05,
+      "loss": 2.2318,
+      "step": 106920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292023126064196e-05,
+      "loss": 2.2469,
+      "step": 106928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29140240857402e-05,
+      "loss": 2.2387,
+      "step": 106936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.290781708699657e-05,
+      "loss": 2.2545,
+      "step": 106944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.290161026456539e-05,
+      "loss": 2.2453,
+      "step": 106952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.289540361860099e-05,
+      "loss": 2.2577,
+      "step": 106960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288919714925766e-05,
+      "loss": 2.2478,
+      "step": 106968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2882990856689686e-05,
+      "loss": 2.2463,
+      "step": 106976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2876784741051336e-05,
+      "loss": 2.2451,
+      "step": 106984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.287057880249694e-05,
+      "loss": 2.2404,
+      "step": 106992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.286437304118075e-05,
+      "loss": 2.2401,
+      "step": 107000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.285816745725703e-05,
+      "loss": 2.2451,
+      "step": 107008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28519620508801e-05,
+      "loss": 2.2519,
+      "step": 107016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284575682220416e-05,
+      "loss": 2.2466,
+      "step": 107024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2839551771383534e-05,
+      "loss": 2.2522,
+      "step": 107032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283334689857244e-05,
+      "loss": 2.2386,
+      "step": 107040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.282714220392515e-05,
+      "loss": 2.2453,
+      "step": 107048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.282093768759592e-05,
+      "loss": 2.2382,
+      "step": 107056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281473334973897e-05,
+      "loss": 2.2464,
+      "step": 107064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280852919050856e-05,
+      "loss": 2.2494,
+      "step": 107072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280232521005892e-05,
+      "loss": 2.2485,
+      "step": 107080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2796121408544284e-05,
+      "loss": 2.2541,
+      "step": 107088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2789917786118906e-05,
+      "loss": 2.2347,
+      "step": 107096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.278371434293694e-05,
+      "loss": 2.2377,
+      "step": 107104
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277751107915265e-05,
+      "loss": 2.2443,
+      "step": 107112
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277130799492025e-05,
+      "loss": 2.2503,
+      "step": 107120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276510509039396e-05,
+      "loss": 2.2288,
+      "step": 107128
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2758902365727955e-05,
+      "loss": 2.2391,
+      "step": 107136
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.275269982107645e-05,
+      "loss": 2.2287,
+      "step": 107144
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274649745659366e-05,
+      "loss": 2.2493,
+      "step": 107152
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2740295272433735e-05,
+      "loss": 2.2337,
+      "step": 107160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.273409326875087e-05,
+      "loss": 2.2474,
+      "step": 107168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.272789144569929e-05,
+      "loss": 2.2354,
+      "step": 107176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.272168980343315e-05,
+      "loss": 2.2308,
+      "step": 107184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.27154883421066e-05,
+      "loss": 2.2467,
+      "step": 107192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.270928706187381e-05,
+      "loss": 2.2419,
+      "step": 107200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2703085962889e-05,
+      "loss": 2.2448,
+      "step": 107208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2696885045306274e-05,
+      "loss": 2.2428,
+      "step": 107216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269068430927979e-05,
+      "loss": 2.2287,
+      "step": 107224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2684483754963746e-05,
+      "loss": 2.2458,
+      "step": 107232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.267828338251224e-05,
+      "loss": 2.2387,
+      "step": 107240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2672083192079426e-05,
+      "loss": 2.2402,
+      "step": 107248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.266588318381945e-05,
+      "loss": 2.2392,
+      "step": 107256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265968335788645e-05,
+      "loss": 2.2389,
+      "step": 107264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265348371443453e-05,
+      "loss": 2.2454,
+      "step": 107272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.264728425361782e-05,
+      "loss": 2.247,
+      "step": 107280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.264108497559048e-05,
+      "loss": 2.2452,
+      "step": 107288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263488588050658e-05,
+      "loss": 2.2464,
+      "step": 107296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2628686968520235e-05,
+      "loss": 2.2383,
+      "step": 107304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2622488239785575e-05,
+      "loss": 2.2512,
+      "step": 107312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.261628969445667e-05,
+      "loss": 2.2438,
+      "step": 107320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2610091332687615e-05,
+      "loss": 2.2564,
+      "step": 107328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2603893154632546e-05,
+      "loss": 2.2349,
+      "step": 107336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259769516044553e-05,
+      "loss": 2.2547,
+      "step": 107344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2591497350280614e-05,
+      "loss": 2.2296,
+      "step": 107352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2585299724291914e-05,
+      "loss": 2.2388,
+      "step": 107360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.257910228263352e-05,
+      "loss": 2.2603,
+      "step": 107368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2572905025459466e-05,
+      "loss": 2.2363,
+      "step": 107376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25667079529238e-05,
+      "loss": 2.2421,
+      "step": 107384
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256051106518064e-05,
+      "loss": 2.2502,
+      "step": 107392
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2554314362383996e-05,
+      "loss": 2.239,
+      "step": 107400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2548117844687925e-05,
+      "loss": 2.2325,
+      "step": 107408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2541921512246485e-05,
+      "loss": 2.2479,
+      "step": 107416
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.253572536521372e-05,
+      "loss": 2.23,
+      "step": 107424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252952940374365e-05,
+      "loss": 2.242,
+      "step": 107432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2523333627990325e-05,
+      "loss": 2.235,
+      "step": 107440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.251713803810777e-05,
+      "loss": 2.2521,
+      "step": 107448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2510942634249984e-05,
+      "loss": 2.2306,
+      "step": 107456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.250474741657102e-05,
+      "loss": 2.249,
+      "step": 107464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249855238522487e-05,
+      "loss": 2.2573,
+      "step": 107472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249235754036556e-05,
+      "loss": 2.2298,
+      "step": 107480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.248616288214707e-05,
+      "loss": 2.2541,
+      "step": 107488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.247996841072342e-05,
+      "loss": 2.2371,
+      "step": 107496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2473774126248595e-05,
+      "loss": 2.2438,
+      "step": 107504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.24675800288766e-05,
+      "loss": 2.2407,
+      "step": 107512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.246138611876139e-05,
+      "loss": 2.247,
+      "step": 107520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2455192396057004e-05,
+      "loss": 2.253,
+      "step": 107528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244899886091734e-05,
+      "loss": 2.2651,
+      "step": 107536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244280551349643e-05,
+      "loss": 2.2462,
+      "step": 107544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.243661235394821e-05,
+      "loss": 2.2551,
+      "step": 107552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2430419382426686e-05,
+      "loss": 2.2501,
+      "step": 107560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242422659908576e-05,
+      "loss": 2.2573,
+      "step": 107568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.241803400407941e-05,
+      "loss": 2.2378,
+      "step": 107576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.241184159756161e-05,
+      "loss": 2.2518,
+      "step": 107584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.240564937968626e-05,
+      "loss": 2.2512,
+      "step": 107592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2399457350607314e-05,
+      "loss": 2.2475,
+      "step": 107600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239326551047873e-05,
+      "loss": 2.246,
+      "step": 107608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2387073859454406e-05,
+      "loss": 2.2398,
+      "step": 107616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238088239768828e-05,
+      "loss": 2.2579,
+      "step": 107624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2374691125334266e-05,
+      "loss": 2.2408,
+      "step": 107632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23685000425463e-05,
+      "loss": 2.2442,
+      "step": 107640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.236230914947826e-05,
+      "loss": 2.2504,
+      "step": 107648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235611844628408e-05,
+      "loss": 2.239,
+      "step": 107656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234992793311767e-05,
+      "loss": 2.2504,
+      "step": 107664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2343737610132895e-05,
+      "loss": 2.2481,
+      "step": 107672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.233754747748366e-05,
+      "loss": 2.2541,
+      "step": 107680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2331357535323854e-05,
+      "loss": 2.252,
+      "step": 107688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2325167783807374e-05,
+      "loss": 2.2487,
+      "step": 107696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2318978223088066e-05,
+      "loss": 2.2578,
+      "step": 107704
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231278885331983e-05,
+      "loss": 2.2567,
+      "step": 107712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2306599674656534e-05,
+      "loss": 2.2408,
+      "step": 107720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230041068725202e-05,
+      "loss": 2.2462,
+      "step": 107728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2294221891260165e-05,
+      "loss": 2.2605,
+      "step": 107736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228803328683483e-05,
+      "loss": 2.2536,
+      "step": 107744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2281844874129835e-05,
+      "loss": 2.2602,
+      "step": 107752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2275656653299034e-05,
+      "loss": 2.24,
+      "step": 107760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226946862449629e-05,
+      "loss": 2.2392,
+      "step": 107768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226328078787544e-05,
+      "loss": 2.2423,
+      "step": 107776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225709314359027e-05,
+      "loss": 2.2349,
+      "step": 107784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2250905691794636e-05,
+      "loss": 2.2346,
+      "step": 107792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224471843264238e-05,
+      "loss": 2.2555,
+      "step": 107800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.223853136628727e-05,
+      "loss": 2.2434,
+      "step": 107808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2232344492883144e-05,
+      "loss": 2.2386,
+      "step": 107816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22261578125838e-05,
+      "loss": 2.2369,
+      "step": 107824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2219971325543085e-05,
+      "loss": 2.2414,
+      "step": 107832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221378503191472e-05,
+      "loss": 2.2342,
+      "step": 107840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.220759893185252e-05,
+      "loss": 2.2489,
+      "step": 107848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2201413025510315e-05,
+      "loss": 2.2451,
+      "step": 107856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.219522731304185e-05,
+      "loss": 2.242,
+      "step": 107864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.218904179460089e-05,
+      "loss": 2.2308,
+      "step": 107872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2182856470341245e-05,
+      "loss": 2.2543,
+      "step": 107880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217667134041666e-05,
+      "loss": 2.2383,
+      "step": 107888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217048640498091e-05,
+      "loss": 2.2345,
+      "step": 107896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.216430166418774e-05,
+      "loss": 2.2477,
+      "step": 107904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2158117118190916e-05,
+      "loss": 2.2362,
+      "step": 107912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.215193276714417e-05,
+      "loss": 2.2435,
+      "step": 107920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.214574861120127e-05,
+      "loss": 2.2524,
+      "step": 107928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.213956465051595e-05,
+      "loss": 2.2434,
+      "step": 107936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21333808852419e-05,
+      "loss": 2.2569,
+      "step": 107944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212719731553291e-05,
+      "loss": 2.2267,
+      "step": 107952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2121013941542675e-05,
+      "loss": 2.2345,
+      "step": 107960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.211483076342491e-05,
+      "loss": 2.2596,
+      "step": 107968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.210864778133333e-05,
+      "loss": 2.2473,
+      "step": 107976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2102464995421654e-05,
+      "loss": 2.2535,
+      "step": 107984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209628240584361e-05,
+      "loss": 2.2541,
+      "step": 107992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.209010001275285e-05,
+      "loss": 2.2342,
+      "step": 108000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.208391781630307e-05,
+      "loss": 2.243,
+      "step": 108008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.207773581664803e-05,
+      "loss": 2.2403,
+      "step": 108016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2071554013941334e-05,
+      "loss": 2.2519,
+      "step": 108024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2065372408336696e-05,
+      "loss": 2.2556,
+      "step": 108032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2059190999987796e-05,
+      "loss": 2.2471,
+      "step": 108040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.20530097890483e-05,
+      "loss": 2.2296,
+      "step": 108048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.204682877567187e-05,
+      "loss": 2.2438,
+      "step": 108056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.204064796001218e-05,
+      "loss": 2.2287,
+      "step": 108064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.203446734222286e-05,
+      "loss": 2.2427,
+      "step": 108072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.202828692245758e-05,
+      "loss": 2.2481,
+      "step": 108080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.202210670086998e-05,
+      "loss": 2.2385,
+      "step": 108088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.2015926677613716e-05,
+      "loss": 2.2479,
+      "step": 108096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.200974685284239e-05,
+      "loss": 2.2634,
+      "step": 108104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.200356722670967e-05,
+      "loss": 2.243,
+      "step": 108112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.199738779936916e-05,
+      "loss": 2.2223,
+      "step": 108120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.199120857097448e-05,
+      "loss": 2.2586,
+      "step": 108128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1985029541679254e-05,
+      "loss": 2.2535,
+      "step": 108136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1978850711637104e-05,
+      "loss": 2.2574,
+      "step": 108144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1972672081001627e-05,
+      "loss": 2.2357,
+      "step": 108152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.196649364992641e-05,
+      "loss": 2.2511,
+      "step": 108160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1960315418565063e-05,
+      "loss": 2.2481,
+      "step": 108168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.195413738707118e-05,
+      "loss": 2.2447,
+      "step": 108176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1947959555598366e-05,
+      "loss": 2.2513,
+      "step": 108184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1941781924300146e-05,
+      "loss": 2.2449,
+      "step": 108192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.193560449333015e-05,
+      "loss": 2.2392,
+      "step": 108200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.192942726284193e-05,
+      "loss": 2.2463,
+      "step": 108208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.192325023298905e-05,
+      "loss": 2.2338,
+      "step": 108216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1917073403925067e-05,
+      "loss": 2.2368,
+      "step": 108224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.191089677580356e-05,
+      "loss": 2.2345,
+      "step": 108232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.190472034877807e-05,
+      "loss": 2.242,
+      "step": 108240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1898544123002125e-05,
+      "loss": 2.2567,
+      "step": 108248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.189236809862929e-05,
+      "loss": 2.2407,
+      "step": 108256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.188619227581311e-05,
+      "loss": 2.2398,
+      "step": 108264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.188001665470708e-05,
+      "loss": 2.2386,
+      "step": 108272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.187384123546475e-05,
+      "loss": 2.2398,
+      "step": 108280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.186766601823965e-05,
+      "loss": 2.2385,
+      "step": 108288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.186149100318526e-05,
+      "loss": 2.2321,
+      "step": 108296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.185531619045513e-05,
+      "loss": 2.2337,
+      "step": 108304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.184914158020275e-05,
+      "loss": 2.2455,
+      "step": 108312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1842967172581636e-05,
+      "loss": 2.252,
+      "step": 108320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.183679296774526e-05,
+      "loss": 2.2442,
+      "step": 108328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1830618965847134e-05,
+      "loss": 2.235,
+      "step": 108336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1824445167040746e-05,
+      "loss": 2.2358,
+      "step": 108344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.181827157147956e-05,
+      "loss": 2.2614,
+      "step": 108352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.181209817931705e-05,
+      "loss": 2.2345,
+      "step": 108360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.180592499070672e-05,
+      "loss": 2.232,
+      "step": 108368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.179975200580199e-05,
+      "loss": 2.234,
+      "step": 108376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.179357922475636e-05,
+      "loss": 2.2424,
+      "step": 108384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.178740664772326e-05,
+      "loss": 2.2437,
+      "step": 108392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.178123427485617e-05,
+      "loss": 2.2427,
+      "step": 108400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.177506210630851e-05,
+      "loss": 2.2407,
+      "step": 108408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1768890142233716e-05,
+      "loss": 2.2352,
+      "step": 108416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.176271838278527e-05,
+      "loss": 2.2305,
+      "step": 108424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.175654682811655e-05,
+      "loss": 2.238,
+      "step": 108432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.175037547838098e-05,
+      "loss": 2.2306,
+      "step": 108440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.174420433373204e-05,
+      "loss": 2.2458,
+      "step": 108448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1738033394323085e-05,
+      "loss": 2.2426,
+      "step": 108456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.173186266030754e-05,
+      "loss": 2.2567,
+      "step": 108464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.172569213183882e-05,
+      "loss": 2.2534,
+      "step": 108472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1719521809070336e-05,
+      "loss": 2.2528,
+      "step": 108480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.171335169215545e-05,
+      "loss": 2.247,
+      "step": 108488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1707181781247584e-05,
+      "loss": 2.2624,
+      "step": 108496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.170101207650011e-05,
+      "loss": 2.2456,
+      "step": 108504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1694842578066397e-05,
+      "loss": 2.2419,
+      "step": 108512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.168867328609983e-05,
+      "loss": 2.2347,
+      "step": 108520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.168250420075378e-05,
+      "loss": 2.251,
+      "step": 108528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1676335322181613e-05,
+      "loss": 2.2219,
+      "step": 108536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.167016665053668e-05,
+      "loss": 2.2462,
+      "step": 108544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.166399818597234e-05,
+      "loss": 2.2306,
+      "step": 108552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.165782992864194e-05,
+      "loss": 2.2367,
+      "step": 108560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.165166187869883e-05,
+      "loss": 2.2296,
+      "step": 108568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1645494036296326e-05,
+      "loss": 2.2453,
+      "step": 108576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1639326401587804e-05,
+      "loss": 2.2515,
+      "step": 108584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1633158974726536e-05,
+      "loss": 2.2376,
+      "step": 108592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.16269917558659e-05,
+      "loss": 2.242,
+      "step": 108600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.162082474515917e-05,
+      "loss": 2.2501,
+      "step": 108608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1614657942759706e-05,
+      "loss": 2.2362,
+      "step": 108616
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.160849134882075e-05,
+      "loss": 2.2622,
+      "step": 108624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.160232496349564e-05,
+      "loss": 2.239,
+      "step": 108632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.159615878693771e-05,
+      "loss": 2.2518,
+      "step": 108640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.158999281930019e-05,
+      "loss": 2.271,
+      "step": 108648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.158382706073639e-05,
+      "loss": 2.2458,
+      "step": 108656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.15776615113996e-05,
+      "loss": 2.2495,
+      "step": 108664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.15714961714431e-05,
+      "loss": 2.2411,
+      "step": 108672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.156533104102013e-05,
+      "loss": 2.2437,
+      "step": 108680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1559166120284e-05,
+      "loss": 2.2347,
+      "step": 108688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1553001409387925e-05,
+      "loss": 2.2444,
+      "step": 108696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.15468369084852e-05,
+      "loss": 2.253,
+      "step": 108704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1540672617729036e-05,
+      "loss": 2.2491,
+      "step": 108712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.153450853727273e-05,
+      "loss": 2.2399,
+      "step": 108720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.152834466726947e-05,
+      "loss": 2.2418,
+      "step": 108728
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1522181007872515e-05,
+      "loss": 2.2337,
+      "step": 108736
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.15160175592351e-05,
+      "loss": 2.2456,
+      "step": 108744
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.150985432151043e-05,
+      "loss": 2.2418,
+      "step": 108752
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.150369129485173e-05,
+      "loss": 2.2336,
+      "step": 108760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.149752847941223e-05,
+      "loss": 2.2402,
+      "step": 108768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.149136587534513e-05,
+      "loss": 2.2486,
+      "step": 108776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.148520348280361e-05,
+      "loss": 2.2404,
+      "step": 108784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.14790413019409e-05,
+      "loss": 2.2446,
+      "step": 108792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.147287933291018e-05,
+      "loss": 2.2257,
+      "step": 108800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1466717575864646e-05,
+      "loss": 2.2374,
+      "step": 108808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.146055603095744e-05,
+      "loss": 2.247,
+      "step": 108816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.145439469834179e-05,
+      "loss": 2.2362,
+      "step": 108824
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1448233578170856e-05,
+      "loss": 2.2275,
+      "step": 108832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.144207267059778e-05,
+      "loss": 2.2342,
+      "step": 108840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.143591197577573e-05,
+      "loss": 2.2506,
+      "step": 108848
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.142975149385789e-05,
+      "loss": 2.249,
+      "step": 108856
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.142359122499738e-05,
+      "loss": 2.247,
+      "step": 108864
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.141743116934734e-05,
+      "loss": 2.2444,
+      "step": 108872
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1411271327060936e-05,
+      "loss": 2.2513,
+      "step": 108880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1405111698291296e-05,
+      "loss": 2.2437,
+      "step": 108888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1398952283191536e-05,
+      "loss": 2.2307,
+      "step": 108896
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.13927930819148e-05,
+      "loss": 2.2518,
+      "step": 108904
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.13866340946142e-05,
+      "loss": 2.2552,
+      "step": 108912
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.138047532144281e-05,
+      "loss": 2.2477,
+      "step": 108920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.137431676255379e-05,
+      "loss": 2.2486,
+      "step": 108928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.136815841810022e-05,
+      "loss": 2.2612,
+      "step": 108936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.13620002882352e-05,
+      "loss": 2.2367,
+      "step": 108944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1355842373111815e-05,
+      "loss": 2.242,
+      "step": 108952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.134968467288317e-05,
+      "loss": 2.256,
+      "step": 108960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.134352718770233e-05,
+      "loss": 2.2487,
+      "step": 108968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1337369917722376e-05,
+      "loss": 2.2453,
+      "step": 108976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.133121286309638e-05,
+      "loss": 2.2613,
+      "step": 108984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1325056023977416e-05,
+      "loss": 2.2421,
+      "step": 108992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1318899400518504e-05,
+      "loss": 2.2389,
+      "step": 109000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1312742992872745e-05,
+      "loss": 2.2503,
+      "step": 109008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.130658680119316e-05,
+      "loss": 2.2445,
+      "step": 109016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1300430825632826e-05,
+      "loss": 2.228,
+      "step": 109024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.129427506634473e-05,
+      "loss": 2.2521,
+      "step": 109032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.128811952348194e-05,
+      "loss": 2.2573,
+      "step": 109040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.12819641971975e-05,
+      "loss": 2.2569,
+      "step": 109048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1275809087644375e-05,
+      "loss": 2.2354,
+      "step": 109056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.126965419497562e-05,
+      "loss": 2.2492,
+      "step": 109064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.126349951934426e-05,
+      "loss": 2.2326,
+      "step": 109072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.125734506090327e-05,
+      "loss": 2.2545,
+      "step": 109080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.125119081980567e-05,
+      "loss": 2.2438,
+      "step": 109088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1245036796204424e-05,
+      "loss": 2.2468,
+      "step": 109096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.123888299025259e-05,
+      "loss": 2.2616,
+      "step": 109104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1232729402103075e-05,
+      "loss": 2.2424,
+      "step": 109112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.122657603190889e-05,
+      "loss": 2.2483,
+      "step": 109120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1220422879823024e-05,
+      "loss": 2.2366,
+      "step": 109128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.121426994599843e-05,
+      "loss": 2.237,
+      "step": 109136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1208117230588045e-05,
+      "loss": 2.2192,
+      "step": 109144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1201964733744876e-05,
+      "loss": 2.2275,
+      "step": 109152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.119581245562185e-05,
+      "loss": 2.2356,
+      "step": 109160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.11896603963719e-05,
+      "loss": 2.2611,
+      "step": 109168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.118350855614799e-05,
+      "loss": 2.2623,
+      "step": 109176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.117735693510306e-05,
+      "loss": 2.2605,
+      "step": 109184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.117120553339001e-05,
+      "loss": 2.2348,
+      "step": 109192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.116505435116178e-05,
+      "loss": 2.2452,
+      "step": 109200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1158903388571313e-05,
+      "loss": 2.2444,
+      "step": 109208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.115275264577146e-05,
+      "loss": 2.2327,
+      "step": 109216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.11466021229152e-05,
+      "loss": 2.2259,
+      "step": 109224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.114045182015539e-05,
+      "loss": 2.2309,
+      "step": 109232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.113430173764495e-05,
+      "loss": 2.2492,
+      "step": 109240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.112815187553675e-05,
+      "loss": 2.2139,
+      "step": 109248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.11220022339837e-05,
+      "loss": 2.236,
+      "step": 109256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1115852813138685e-05,
+      "loss": 2.2414,
+      "step": 109264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1109703613154544e-05,
+      "loss": 2.2513,
+      "step": 109272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.110355463418418e-05,
+      "loss": 2.2361,
+      "step": 109280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1097405876380454e-05,
+      "loss": 2.2272,
+      "step": 109288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1091257339896194e-05,
+      "loss": 2.2373,
+      "step": 109296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1085109024884284e-05,
+      "loss": 2.2335,
+      "step": 109304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.107896093149757e-05,
+      "loss": 2.2647,
+      "step": 109312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1072813059888894e-05,
+      "loss": 2.2577,
+      "step": 109320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1066665410211064e-05,
+      "loss": 2.2502,
+      "step": 109328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.106051798261695e-05,
+      "loss": 2.2464,
+      "step": 109336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.105437077725936e-05,
+      "loss": 2.2311,
+      "step": 109344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1048223794291126e-05,
+      "loss": 2.242,
+      "step": 109352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1042077033865035e-05,
+      "loss": 2.2425,
+      "step": 109360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.103593049613391e-05,
+      "loss": 2.238,
+      "step": 109368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1029784181250584e-05,
+      "loss": 2.2447,
+      "step": 109376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.102363808936781e-05,
+      "loss": 2.2479,
+      "step": 109384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.101749222063841e-05,
+      "loss": 2.2396,
+      "step": 109392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.101134657521516e-05,
+      "loss": 2.2243,
+      "step": 109400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.100520115325084e-05,
+      "loss": 2.2265,
+      "step": 109408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0999055954898224e-05,
+      "loss": 2.2445,
+      "step": 109416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0992910980310104e-05,
+      "loss": 2.2482,
+      "step": 109424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.09867662296392e-05,
+      "loss": 2.2301,
+      "step": 109432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0980621703038324e-05,
+      "loss": 2.2461,
+      "step": 109440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.097447740066019e-05,
+      "loss": 2.2419,
+      "step": 109448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.096833332265758e-05,
+      "loss": 2.2359,
+      "step": 109456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.09621894691832e-05,
+      "loss": 2.2271,
+      "step": 109464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.095604584038979e-05,
+      "loss": 2.2401,
+      "step": 109472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.094990243643012e-05,
+      "loss": 2.2316,
+      "step": 109480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.094375925745689e-05,
+      "loss": 2.2398,
+      "step": 109488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0937616303622804e-05,
+      "loss": 2.2407,
+      "step": 109496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.093147357508061e-05,
+      "loss": 2.2305,
+      "step": 109504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0925331071982984e-05,
+      "loss": 2.2319,
+      "step": 109512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.091918879448265e-05,
+      "loss": 2.2279,
+      "step": 109520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.091304674273231e-05,
+      "loss": 2.2373,
+      "step": 109528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.090690491688465e-05,
+      "loss": 2.2472,
+      "step": 109536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0900763317092345e-05,
+      "loss": 2.2416,
+      "step": 109544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.089462194350808e-05,
+      "loss": 2.2317,
+      "step": 109552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0888480796284545e-05,
+      "loss": 2.2386,
+      "step": 109560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0882339875574406e-05,
+      "loss": 2.2303,
+      "step": 109568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0876199181530304e-05,
+      "loss": 2.2415,
+      "step": 109576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0870058714304915e-05,
+      "loss": 2.2357,
+      "step": 109584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0863918474050895e-05,
+      "loss": 2.2265,
+      "step": 109592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.08577784609209e-05,
+      "loss": 2.2499,
+      "step": 109600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0851638675067536e-05,
+      "loss": 2.2472,
+      "step": 109608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.084549911664348e-05,
+      "loss": 2.2303,
+      "step": 109616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.083935978580133e-05,
+      "loss": 2.2423,
+      "step": 109624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.083322068269373e-05,
+      "loss": 2.2279,
+      "step": 109632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.082708180747331e-05,
+      "loss": 2.2372,
+      "step": 109640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.082094316029264e-05,
+      "loss": 2.2377,
+      "step": 109648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.081480474130435e-05,
+      "loss": 2.2499,
+      "step": 109656
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.080866655066105e-05,
+      "loss": 2.2334,
+      "step": 109664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.080252858851535e-05,
+      "loss": 2.2371,
+      "step": 109672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.079639085501979e-05,
+      "loss": 2.2398,
+      "step": 109680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0790253350327e-05,
+      "loss": 2.2412,
+      "step": 109688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.078411607458957e-05,
+      "loss": 2.2387,
+      "step": 109696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.077797902796002e-05,
+      "loss": 2.2398,
+      "step": 109704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0771842210590946e-05,
+      "loss": 2.2519,
+      "step": 109712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.076570562263489e-05,
+      "loss": 2.2479,
+      "step": 109720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.075956926424448e-05,
+      "loss": 2.2433,
+      "step": 109728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0753433135572195e-05,
+      "loss": 2.2297,
+      "step": 109736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0747297236770584e-05,
+      "loss": 2.2227,
+      "step": 109744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0741161567992215e-05,
+      "loss": 2.2436,
+      "step": 109752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.073502612938961e-05,
+      "loss": 2.2512,
+      "step": 109760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0728890921115276e-05,
+      "loss": 2.2485,
+      "step": 109768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0722755943321774e-05,
+      "loss": 2.2483,
+      "step": 109776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.071662119616159e-05,
+      "loss": 2.2467,
+      "step": 109784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.071048667978724e-05,
+      "loss": 2.239,
+      "step": 109792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.070435239435125e-05,
+      "loss": 2.2544,
+      "step": 109800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.06982183400061e-05,
+      "loss": 2.2384,
+      "step": 109808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.069208451690428e-05,
+      "loss": 2.2483,
+      "step": 109816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.068595092519828e-05,
+      "loss": 2.243,
+      "step": 109824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.067981756504059e-05,
+      "loss": 2.2459,
+      "step": 109832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.067368443658367e-05,
+      "loss": 2.2479,
+      "step": 109840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0667551539980016e-05,
+      "loss": 2.2442,
+      "step": 109848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0661418875382074e-05,
+      "loss": 2.2307,
+      "step": 109856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0655286442942305e-05,
+      "loss": 2.2343,
+      "step": 109864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.064915424281316e-05,
+      "loss": 2.2421,
+      "step": 109872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0643022275147075e-05,
+      "loss": 2.2466,
+      "step": 109880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.063689054009655e-05,
+      "loss": 2.2175,
+      "step": 109888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.063075903781395e-05,
+      "loss": 2.2506,
+      "step": 109896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.062462776845172e-05,
+      "loss": 2.2419,
+      "step": 109904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.061849673216232e-05,
+      "loss": 2.2356,
+      "step": 109912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.061236592909813e-05,
+      "loss": 2.2455,
+      "step": 109920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.060623535941158e-05,
+      "loss": 2.222,
+      "step": 109928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.060010502325505e-05,
+      "loss": 2.2406,
+      "step": 109936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0593974920780995e-05,
+      "loss": 2.2374,
+      "step": 109944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.058784505214177e-05,
+      "loss": 2.2351,
+      "step": 109952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.058171541748977e-05,
+      "loss": 2.2619,
+      "step": 109960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0575586016977385e-05,
+      "loss": 2.2312,
+      "step": 109968
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.056945685075698e-05,
+      "loss": 2.25,
+      "step": 109976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0563327918980954e-05,
+      "loss": 2.2467,
+      "step": 109984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0557199221801644e-05,
+      "loss": 2.2486,
+      "step": 109992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.055107075937142e-05,
+      "loss": 2.2411,
+      "step": 110000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.054494253184263e-05,
+      "loss": 2.2351,
+      "step": 110008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.053881453936763e-05,
+      "loss": 2.2445,
+      "step": 110016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.053268678209877e-05,
+      "loss": 2.2243,
+      "step": 110024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.052655926018839e-05,
+      "loss": 2.2392,
+      "step": 110032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.052043197378878e-05,
+      "loss": 2.2394,
+      "step": 110040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0514304923052334e-05,
+      "loss": 2.24,
+      "step": 110048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.050817810813129e-05,
+      "loss": 2.2396,
+      "step": 110056
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.050205152917802e-05,
+      "loss": 2.241,
+      "step": 110064
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.049592518634483e-05,
+      "loss": 2.2414,
+      "step": 110072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0489799079783995e-05,
+      "loss": 2.2512,
+      "step": 110080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.048367320964782e-05,
+      "loss": 2.2517,
+      "step": 110088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.047754757608858e-05,
+      "loss": 2.2315,
+      "step": 110096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.047142217925862e-05,
+      "loss": 2.2499,
+      "step": 110104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.046529701931013e-05,
+      "loss": 2.2531,
+      "step": 110112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0459172096395436e-05,
+      "loss": 2.249,
+      "step": 110120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.045304741066682e-05,
+      "loss": 2.223,
+      "step": 110128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.044692296227649e-05,
+      "loss": 2.2577,
+      "step": 110136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0440798751376725e-05,
+      "loss": 2.2527,
+      "step": 110144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0434674778119786e-05,
+      "loss": 2.2386,
+      "step": 110152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0428551042657914e-05,
+      "loss": 2.2534,
+      "step": 110160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0422427545143324e-05,
+      "loss": 2.2497,
+      "step": 110168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.041630428572825e-05,
+      "loss": 2.2382,
+      "step": 110176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.041018126456496e-05,
+      "loss": 2.2361,
+      "step": 110184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0404058481805614e-05,
+      "loss": 2.2366,
+      "step": 110192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.039793593760245e-05,
+      "loss": 2.2461,
+      "step": 110200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.03918136321077e-05,
+      "loss": 2.2436,
+      "step": 110208
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.03856915654735e-05,
+      "loss": 2.2542,
+      "step": 110216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.037956973785212e-05,
+      "loss": 2.251,
+      "step": 110224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0373448149395706e-05,
+      "loss": 2.2448,
+      "step": 110232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.036732680025644e-05,
+      "loss": 2.2425,
+      "step": 110240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.036120569058653e-05,
+      "loss": 2.2443,
+      "step": 110248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.035508482053812e-05,
+      "loss": 2.2454,
+      "step": 110256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.034896419026339e-05,
+      "loss": 2.2516,
+      "step": 110264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0342843799914476e-05,
+      "loss": 2.2495,
+      "step": 110272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0336723649643545e-05,
+      "loss": 2.2466,
+      "step": 110280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0330603739602766e-05,
+      "loss": 2.2373,
+      "step": 110288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.032448406994428e-05,
+      "loss": 2.2466,
+      "step": 110296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.031836464082016e-05,
+      "loss": 2.2342,
+      "step": 110304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.03122454523826e-05,
+      "loss": 2.2348,
+      "step": 110312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0306126504783736e-05,
+      "loss": 2.2528,
+      "step": 110320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.030000779817563e-05,
+      "loss": 2.2636,
+      "step": 110328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.02938893327104e-05,
+      "loss": 2.2466,
+      "step": 110336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0287771108540195e-05,
+      "loss": 2.2463,
+      "step": 110344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.028165312581708e-05,
+      "loss": 2.2441,
+      "step": 110352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0275535384693165e-05,
+      "loss": 2.2562,
+      "step": 110360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.026941788532053e-05,
+      "loss": 2.253,
+      "step": 110368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.026330062785125e-05,
+      "loss": 2.2555,
+      "step": 110376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0257183612437405e-05,
+      "loss": 2.2322,
+      "step": 110384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.025106683923109e-05,
+      "loss": 2.2359,
+      "step": 110392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.024495030838434e-05,
+      "loss": 2.2468,
+      "step": 110400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0238834020049206e-05,
+      "loss": 2.2298,
+      "step": 110408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.023271797437776e-05,
+      "loss": 2.2458,
+      "step": 110416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.022660217152204e-05,
+      "loss": 2.2294,
+      "step": 110424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.02204866116341e-05,
+      "loss": 2.2345,
+      "step": 110432
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0214371294865944e-05,
+      "loss": 2.2585,
+      "step": 110440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.020825622136962e-05,
+      "loss": 2.222,
+      "step": 110448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0202141391297153e-05,
+      "loss": 2.2288,
+      "step": 110456
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.019602680480053e-05,
+      "loss": 2.2368,
+      "step": 110464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.01899124620318e-05,
+      "loss": 2.2483,
+      "step": 110472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0183798363142954e-05,
+      "loss": 2.2516,
+      "step": 110480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0177684508285974e-05,
+      "loss": 2.2443,
+      "step": 110488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.017157089761284e-05,
+      "loss": 2.2449,
+      "step": 110496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.016545753127557e-05,
+      "loss": 2.2382,
+      "step": 110504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0159344409426134e-05,
+      "loss": 2.2574,
+      "step": 110512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0153231532216494e-05,
+      "loss": 2.2289,
+      "step": 110520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0147118899798616e-05,
+      "loss": 2.2462,
+      "step": 110528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.014100651232449e-05,
+      "loss": 2.247,
+      "step": 110536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.013489436994603e-05,
+      "loss": 2.2488,
+      "step": 110544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0128782472815205e-05,
+      "loss": 2.2436,
+      "step": 110552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.012267082108397e-05,
+      "loss": 2.2419,
+      "step": 110560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.011655941490423e-05,
+      "loss": 2.258,
+      "step": 110568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.011044825442793e-05,
+      "loss": 2.253,
+      "step": 110576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.010433733980701e-05,
+      "loss": 2.2456,
+      "step": 110584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.009822667119337e-05,
+      "loss": 2.2591,
+      "step": 110592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.009211624873891e-05,
+      "loss": 2.2537,
+      "step": 110600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.008600607259557e-05,
+      "loss": 2.2444,
+      "step": 110608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.007989614291523e-05,
+      "loss": 2.2505,
+      "step": 110616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.007378645984977e-05,
+      "loss": 2.2457,
+      "step": 110624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0067677023551106e-05,
+      "loss": 2.237,
+      "step": 110632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0061567834171096e-05,
+      "loss": 2.2478,
+      "step": 110640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.005545889186164e-05,
+      "loss": 2.239,
+      "step": 110648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.004935019677457e-05,
+      "loss": 2.2475,
+      "step": 110656
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0043241749061784e-05,
+      "loss": 2.2508,
+      "step": 110664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.003713354887514e-05,
+      "loss": 2.2438,
+      "step": 110672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.003102559636645e-05,
+      "loss": 2.2417,
+      "step": 110680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0024917891687585e-05,
+      "loss": 2.2504,
+      "step": 110688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.001881043499039e-05,
+      "loss": 2.2594,
+      "step": 110696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.001270322642669e-05,
+      "loss": 2.2296,
+      "step": 110704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.0006596266148295e-05,
+      "loss": 2.2494,
+      "step": 110712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.000048955430703e-05,
+      "loss": 2.2293,
+      "step": 110720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.999438309105475e-05,
+      "loss": 2.2431,
+      "step": 110728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9988276876543214e-05,
+      "loss": 2.249,
+      "step": 110736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.998217091092422e-05,
+      "loss": 2.239,
+      "step": 110744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.997606519434961e-05,
+      "loss": 2.2405,
+      "step": 110752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.996995972697113e-05,
+      "loss": 2.2425,
+      "step": 110760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9963854508940565e-05,
+      "loss": 2.2514,
+      "step": 110768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.99577495404097e-05,
+      "loss": 2.2447,
+      "step": 110776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.995164482153034e-05,
+      "loss": 2.2379,
+      "step": 110784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.99455403524542e-05,
+      "loss": 2.2483,
+      "step": 110792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.993943613333305e-05,
+      "loss": 2.2136,
+      "step": 110800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.993333216431866e-05,
+      "loss": 2.2412,
+      "step": 110808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9927228445562744e-05,
+      "loss": 2.2345,
+      "step": 110816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9921124977217064e-05,
+      "loss": 2.2569,
+      "step": 110824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.991502175943336e-05,
+      "loss": 2.2343,
+      "step": 110832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9908918792363335e-05,
+      "loss": 2.2472,
+      "step": 110840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.990281607615872e-05,
+      "loss": 2.2344,
+      "step": 110848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.989671361097123e-05,
+      "loss": 2.2304,
+      "step": 110856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.989061139695258e-05,
+      "loss": 2.2386,
+      "step": 110864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9884509434254464e-05,
+      "loss": 2.2461,
+      "step": 110872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.987840772302857e-05,
+      "loss": 2.2413,
+      "step": 110880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.987230626342662e-05,
+      "loss": 2.2426,
+      "step": 110888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.986620505560024e-05,
+      "loss": 2.2354,
+      "step": 110896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.986010409970115e-05,
+      "loss": 2.2422,
+      "step": 110904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.985400339588103e-05,
+      "loss": 2.2433,
+      "step": 110912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.98479029442915e-05,
+      "loss": 2.2198,
+      "step": 110920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9841802745084245e-05,
+      "loss": 2.25,
+      "step": 110928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.983570279841092e-05,
+      "loss": 2.2316,
+      "step": 110936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9829603104423173e-05,
+      "loss": 2.2477,
+      "step": 110944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.982350366327262e-05,
+      "loss": 2.2489,
+      "step": 110952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9817404475110896e-05,
+      "loss": 2.2348,
+      "step": 110960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.981130554008966e-05,
+      "loss": 2.2505,
+      "step": 110968
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.980520685836051e-05,
+      "loss": 2.2339,
+      "step": 110976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9799108430075054e-05,
+      "loss": 2.2421,
+      "step": 110984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.97930102553849e-05,
+      "loss": 2.2417,
+      "step": 110992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9786912334441685e-05,
+      "loss": 2.2277,
+      "step": 111000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9780814667396945e-05,
+      "loss": 2.2452,
+      "step": 111008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.97747172544023e-05,
+      "loss": 2.2373,
+      "step": 111016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.976862009560935e-05,
+      "loss": 2.2485,
+      "step": 111024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9762523191169644e-05,
+      "loss": 2.2458,
+      "step": 111032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9756426541234756e-05,
+      "loss": 2.2357,
+      "step": 111040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9750330145956265e-05,
+      "loss": 2.2588,
+      "step": 111048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.974423400548571e-05,
+      "loss": 2.2373,
+      "step": 111056
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9738138119974636e-05,
+      "loss": 2.2319,
+      "step": 111064
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9732042489574614e-05,
+      "loss": 2.2321,
+      "step": 111072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9725947114437166e-05,
+      "loss": 2.2416,
+      "step": 111080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.971985199471382e-05,
+      "loss": 2.2471,
+      "step": 111088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.97137571305561e-05,
+      "loss": 2.2404,
+      "step": 111096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.970766252211556e-05,
+      "loss": 2.2333,
+      "step": 111104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.970156816954366e-05,
+      "loss": 2.2383,
+      "step": 111112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9695474072991934e-05,
+      "loss": 2.243,
+      "step": 111120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.968938023261188e-05,
+      "loss": 2.232,
+      "step": 111128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.968328664855501e-05,
+      "loss": 2.2548,
+      "step": 111136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.967719332097276e-05,
+      "loss": 2.2474,
+      "step": 111144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9671100250016655e-05,
+      "loss": 2.2509,
+      "step": 111152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.966500743583817e-05,
+      "loss": 2.2359,
+      "step": 111160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9658914878588744e-05,
+      "loss": 2.2493,
+      "step": 111168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9652822578419853e-05,
+      "loss": 2.2416,
+      "step": 111176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.964673053548297e-05,
+      "loss": 2.2335,
+      "step": 111184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9640638749929515e-05,
+      "loss": 2.2377,
+      "step": 111192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9634547221910946e-05,
+      "loss": 2.2161,
+      "step": 111200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9628455951578685e-05,
+      "loss": 2.2277,
+      "step": 111208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.962236493908419e-05,
+      "loss": 2.2536,
+      "step": 111216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.961627418457886e-05,
+      "loss": 2.2372,
+      "step": 111224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.961018368821411e-05,
+      "loss": 2.2548,
+      "step": 111232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.960409345014137e-05,
+      "loss": 2.244,
+      "step": 111240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9598003470512027e-05,
+      "loss": 2.2403,
+      "step": 111248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9591913749477484e-05,
+      "loss": 2.2403,
+      "step": 111256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9585824287189144e-05,
+      "loss": 2.2535,
+      "step": 111264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.957973508379837e-05,
+      "loss": 2.2478,
+      "step": 111272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9573646139456556e-05,
+      "loss": 2.227,
+      "step": 111280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9567557454315065e-05,
+      "loss": 2.2244,
+      "step": 111288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.956146902852527e-05,
+      "loss": 2.2265,
+      "step": 111296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.955538086223852e-05,
+      "loss": 2.2369,
+      "step": 111304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.954929295560618e-05,
+      "loss": 2.2461,
+      "step": 111312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.95432053087796e-05,
+      "loss": 2.2331,
+      "step": 111320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.953711792191009e-05,
+      "loss": 2.2333,
+      "step": 111328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9531030795149e-05,
+      "loss": 2.2499,
+      "step": 111336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9524943928647663e-05,
+      "loss": 2.247,
+      "step": 111344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9518857322557424e-05,
+      "loss": 2.2236,
+      "step": 111352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9512770977029536e-05,
+      "loss": 2.2319,
+      "step": 111360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9506684892215346e-05,
+      "loss": 2.2182,
+      "step": 111368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9500599068266155e-05,
+      "loss": 2.2415,
+      "step": 111376
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.949451350533324e-05,
+      "loss": 2.2534,
+      "step": 111384
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.948842820356789e-05,
+      "loss": 2.2381,
+      "step": 111392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.948234316312141e-05,
+      "loss": 2.2186,
+      "step": 111400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.947625838414506e-05,
+      "loss": 2.2254,
+      "step": 111408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.947017386679009e-05,
+      "loss": 2.2459,
+      "step": 111416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.946408961120779e-05,
+      "loss": 2.2477,
+      "step": 111424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.945800561754941e-05,
+      "loss": 2.2292,
+      "step": 111432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9451921885966184e-05,
+      "loss": 2.2427,
+      "step": 111440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.944583841660937e-05,
+      "loss": 2.2378,
+      "step": 111448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9439755209630196e-05,
+      "loss": 2.2292,
+      "step": 111456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9433672265179884e-05,
+      "loss": 2.2436,
+      "step": 111464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9427589583409674e-05,
+      "loss": 2.2393,
+      "step": 111472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.942150716447076e-05,
+      "loss": 2.2381,
+      "step": 111480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9415425008514385e-05,
+      "loss": 2.2245,
+      "step": 111488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.940934311569172e-05,
+      "loss": 2.2528,
+      "step": 111496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9403261486153966e-05,
+      "loss": 2.2387,
+      "step": 111504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9397180120052345e-05,
+      "loss": 2.245,
+      "step": 111512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.939109901753799e-05,
+      "loss": 2.2375,
+      "step": 111520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.938501817876211e-05,
+      "loss": 2.2598,
+      "step": 111528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.937893760387589e-05,
+      "loss": 2.2504,
+      "step": 111536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.937285729303045e-05,
+      "loss": 2.2354,
+      "step": 111544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.936677724637697e-05,
+      "loss": 2.2452,
+      "step": 111552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.93606974640666e-05,
+      "loss": 2.2374,
+      "step": 111560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.935461794625051e-05,
+      "loss": 2.2466,
+      "step": 111568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9348538693079784e-05,
+      "loss": 2.2473,
+      "step": 111576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9342459704705586e-05,
+      "loss": 2.2582,
+      "step": 111584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.933638098127905e-05,
+      "loss": 2.2432,
+      "step": 111592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9330302522951274e-05,
+      "loss": 2.2404,
+      "step": 111600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.932422432987336e-05,
+      "loss": 2.2429,
+      "step": 111608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.931814640219644e-05,
+      "loss": 2.2491,
+      "step": 111616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9312068740071596e-05,
+      "loss": 2.2417,
+      "step": 111624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.930599134364992e-05,
+      "loss": 2.2216,
+      "step": 111632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.92999142130825e-05,
+      "loss": 2.2511,
+      "step": 111640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9293837348520414e-05,
+      "loss": 2.235,
+      "step": 111648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.928776075011473e-05,
+      "loss": 2.2393,
+      "step": 111656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.928168441801652e-05,
+      "loss": 2.2621,
+      "step": 111664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.927560835237685e-05,
+      "loss": 2.2513,
+      "step": 111672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.926953255334674e-05,
+      "loss": 2.2542,
+      "step": 111680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.926345702107725e-05,
+      "loss": 2.244,
+      "step": 111688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9257381755719436e-05,
+      "loss": 2.2561,
+      "step": 111696
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.925130675742431e-05,
+      "loss": 2.2299,
+      "step": 111704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9245232026342916e-05,
+      "loss": 2.23,
+      "step": 111712
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.923915756262625e-05,
+      "loss": 2.2497,
+      "step": 111720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.923308336642535e-05,
+      "loss": 2.2485,
+      "step": 111728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.922700943789118e-05,
+      "loss": 2.2458,
+      "step": 111736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9220935777174776e-05,
+      "loss": 2.2267,
+      "step": 111744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9214862384427134e-05,
+      "loss": 2.2252,
+      "step": 111752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.920878925979921e-05,
+      "loss": 2.2479,
+      "step": 111760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.920271640344198e-05,
+      "loss": 2.2453,
+      "step": 111768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9196643815506435e-05,
+      "loss": 2.2399,
+      "step": 111776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.919057149614357e-05,
+      "loss": 2.2408,
+      "step": 111784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9184499445504284e-05,
+      "loss": 2.2467,
+      "step": 111792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.917842766373955e-05,
+      "loss": 2.2616,
+      "step": 111800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.917235615100034e-05,
+      "loss": 2.2413,
+      "step": 111808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.916628490743756e-05,
+      "loss": 2.226,
+      "step": 111816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9160213933202144e-05,
+      "loss": 2.2451,
+      "step": 111824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9154143228445025e-05,
+      "loss": 2.2461,
+      "step": 111832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.914807279331715e-05,
+      "loss": 2.2455,
+      "step": 111840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9142002627969373e-05,
+      "loss": 2.2265,
+      "step": 111848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9135932732552635e-05,
+      "loss": 2.2452,
+      "step": 111856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.912986310721784e-05,
+      "loss": 2.2543,
+      "step": 111864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.912379375211585e-05,
+      "loss": 2.2351,
+      "step": 111872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.911772466739758e-05,
+      "loss": 2.2316,
+      "step": 111880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.91116558532139e-05,
+      "loss": 2.2515,
+      "step": 111888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.910558730971567e-05,
+      "loss": 2.2552,
+      "step": 111896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.909951903705376e-05,
+      "loss": 2.2575,
+      "step": 111904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.909345103537903e-05,
+      "loss": 2.2638,
+      "step": 111912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.908738330484234e-05,
+      "loss": 2.245,
+      "step": 111920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.908131584559451e-05,
+      "loss": 2.2562,
+      "step": 111928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.90752486577864e-05,
+      "loss": 2.2279,
+      "step": 111936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.906918174156885e-05,
+      "loss": 2.2458,
+      "step": 111944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9063115097092645e-05,
+      "loss": 2.2338,
+      "step": 111952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.905704872450863e-05,
+      "loss": 2.2579,
+      "step": 111960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.905098262396763e-05,
+      "loss": 2.2262,
+      "step": 111968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.904491679562042e-05,
+      "loss": 2.2398,
+      "step": 111976
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.903885123961779e-05,
+      "loss": 2.2363,
+      "step": 111984
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.903278595611056e-05,
+      "loss": 2.2466,
+      "step": 111992
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.902672094524952e-05,
+      "loss": 2.2316,
+      "step": 112000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.902065620718541e-05,
+      "loss": 2.2531,
+      "step": 112008
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9014591742069016e-05,
+      "loss": 2.2436,
+      "step": 112016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.900852755005111e-05,
+      "loss": 2.2535,
+      "step": 112024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9002463631282444e-05,
+      "loss": 2.2502,
+      "step": 112032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.899639998591376e-05,
+      "loss": 2.2305,
+      "step": 112040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.89903366140958e-05,
+      "loss": 2.2427,
+      "step": 112048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8984273515979326e-05,
+      "loss": 2.2603,
+      "step": 112056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.897821069171504e-05,
+      "loss": 2.2485,
+      "step": 112064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8972148141453665e-05,
+      "loss": 2.2286,
+      "step": 112072
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8966085865345946e-05,
+      "loss": 2.2404,
+      "step": 112080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.896002386354255e-05,
+      "loss": 2.2517,
+      "step": 112088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8953962136194206e-05,
+      "loss": 2.2523,
+      "step": 112096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8947900683451606e-05,
+      "loss": 2.2352,
+      "step": 112104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.894183950546543e-05,
+      "loss": 2.2541,
+      "step": 112112
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.893577860238636e-05,
+      "loss": 2.2438,
+      "step": 112120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.892971797436507e-05,
+      "loss": 2.2396,
+      "step": 112128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.892365762155226e-05,
+      "loss": 2.2645,
+      "step": 112136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8917597544098545e-05,
+      "loss": 2.2509,
+      "step": 112144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.89115377421546e-05,
+      "loss": 2.2574,
+      "step": 112152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8905478215871084e-05,
+      "loss": 2.231,
+      "step": 112160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.88994189653986e-05,
+      "loss": 2.2372,
+      "step": 112168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.889335999088782e-05,
+      "loss": 2.2219,
+      "step": 112176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.888730129248935e-05,
+      "loss": 2.2281,
+      "step": 112184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.888124287035384e-05,
+      "loss": 2.2266,
+      "step": 112192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.887518472463186e-05,
+      "loss": 2.2497,
+      "step": 112200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8869126855474024e-05,
+      "loss": 2.2431,
+      "step": 112208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.886306926303097e-05,
+      "loss": 2.2449,
+      "step": 112216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.885701194745325e-05,
+      "loss": 2.2442,
+      "step": 112224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.885095490889146e-05,
+      "loss": 2.2368,
+      "step": 112232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.88448981474962e-05,
+      "loss": 2.2518,
+      "step": 112240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8838841663418006e-05,
+      "loss": 2.2379,
+      "step": 112248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883278545680746e-05,
+      "loss": 2.2559,
+      "step": 112256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.882672952781512e-05,
+      "loss": 2.247,
+      "step": 112264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.882067387659155e-05,
+      "loss": 2.2455,
+      "step": 112272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8814618503287266e-05,
+      "loss": 2.2627,
+      "step": 112280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8808563408052834e-05,
+      "loss": 2.2468,
+      "step": 112288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.880250859103877e-05,
+      "loss": 2.2512,
+      "step": 112296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.879645405239559e-05,
+      "loss": 2.2584,
+      "step": 112304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.879039979227382e-05,
+      "loss": 2.2619,
+      "step": 112312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8784345810823986e-05,
+      "loss": 2.242,
+      "step": 112320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.877829210819656e-05,
+      "loss": 2.2358,
+      "step": 112328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.877223868454205e-05,
+      "loss": 2.2388,
+      "step": 112336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.876618554001094e-05,
+      "loss": 2.2512,
+      "step": 112344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.876013267475373e-05,
+      "loss": 2.2589,
+      "step": 112352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8754080088920875e-05,
+      "loss": 2.2425,
+      "step": 112360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.874802778266285e-05,
+      "loss": 2.2406,
+      "step": 112368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8741975756130144e-05,
+      "loss": 2.2452,
+      "step": 112376
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8735924009473157e-05,
+      "loss": 2.2527,
+      "step": 112384
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8729872542842354e-05,
+      "loss": 2.2557,
+      "step": 112392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.87238213563882e-05,
+      "loss": 2.2513,
+      "step": 112400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.871777045026112e-05,
+      "loss": 2.244,
+      "step": 112408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.871171982461152e-05,
+      "loss": 2.2614,
+      "step": 112416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.870566947958982e-05,
+      "loss": 2.2414,
+      "step": 112424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.869961941534647e-05,
+      "loss": 2.2573,
+      "step": 112432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.869356963203184e-05,
+      "loss": 2.2417,
+      "step": 112440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8687520129796326e-05,
+      "loss": 2.2339,
+      "step": 112448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.868147090879035e-05,
+      "loss": 2.225,
+      "step": 112456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.867542196916426e-05,
+      "loss": 2.2359,
+      "step": 112464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.866937331106846e-05,
+      "loss": 2.2574,
+      "step": 112472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.86633249346533e-05,
+      "loss": 2.2507,
+      "step": 112480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8657276840069164e-05,
+      "loss": 2.2243,
+      "step": 112488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.86512290274664e-05,
+      "loss": 2.2345,
+      "step": 112496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.864518149699534e-05,
+      "loss": 2.2474,
+      "step": 112504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.863913424880636e-05,
+      "loss": 2.2312,
+      "step": 112512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.863308728304977e-05,
+      "loss": 2.2416,
+      "step": 112520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8627040599875906e-05,
+      "loss": 2.2512,
+      "step": 112528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.862099419943509e-05,
+      "loss": 2.2484,
+      "step": 112536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.861494808187764e-05,
+      "loss": 2.2377,
+      "step": 112544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.860890224735384e-05,
+      "loss": 2.2418,
+      "step": 112552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.860285669601402e-05,
+      "loss": 2.2425,
+      "step": 112560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.859681142800848e-05,
+      "loss": 2.2428,
+      "step": 112568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.859076644348745e-05,
+      "loss": 2.2458,
+      "step": 112576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8584721742601266e-05,
+      "loss": 2.2401,
+      "step": 112584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.857867732550019e-05,
+      "loss": 2.2299,
+      "step": 112592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.857263319233447e-05,
+      "loss": 2.2259,
+      "step": 112600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8566589343254356e-05,
+      "loss": 2.2388,
+      "step": 112608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.856054577841012e-05,
+      "loss": 2.2287,
+      "step": 112616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.855450249795202e-05,
+      "loss": 2.2298,
+      "step": 112624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8548459502030254e-05,
+      "loss": 2.2305,
+      "step": 112632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.854241679079506e-05,
+      "loss": 2.2384,
+      "step": 112640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.85363743643967e-05,
+      "loss": 2.2395,
+      "step": 112648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.853033222298534e-05,
+      "loss": 2.2234,
+      "step": 112656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.85242903667112e-05,
+      "loss": 2.2448,
+      "step": 112664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8518248795724486e-05,
+      "loss": 2.25,
+      "step": 112672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8512207510175424e-05,
+      "loss": 2.2387,
+      "step": 112680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.850616651021416e-05,
+      "loss": 2.2671,
+      "step": 112688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.850012579599087e-05,
+      "loss": 2.2231,
+      "step": 112696
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.849408536765577e-05,
+      "loss": 2.2384,
+      "step": 112704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8488045225358974e-05,
+      "loss": 2.2421,
+      "step": 112712
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.848200536925067e-05,
+      "loss": 2.2411,
+      "step": 112720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.847596579948102e-05,
+      "loss": 2.2222,
+      "step": 112728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.846992651620014e-05,
+      "loss": 2.2319,
+      "step": 112736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8463887519558185e-05,
+      "loss": 2.2396,
+      "step": 112744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8457848809705276e-05,
+      "loss": 2.2443,
+      "step": 112752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845181038679155e-05,
+      "loss": 2.2278,
+      "step": 112760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.84457722509671e-05,
+      "loss": 2.2418,
+      "step": 112768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8439734402382056e-05,
+      "loss": 2.2341,
+      "step": 112776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843369684118652e-05,
+      "loss": 2.2359,
+      "step": 112784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842765956753056e-05,
+      "loss": 2.2362,
+      "step": 112792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842162258156429e-05,
+      "loss": 2.2475,
+      "step": 112800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8415585883437797e-05,
+      "loss": 2.2196,
+      "step": 112808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.840954947330112e-05,
+      "loss": 2.2389,
+      "step": 112816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8403513351304334e-05,
+      "loss": 2.2313,
+      "step": 112824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839747751759752e-05,
+      "loss": 2.2369,
+      "step": 112832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8391441972330734e-05,
+      "loss": 2.2392,
+      "step": 112840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8385406715653986e-05,
+      "loss": 2.2433,
+      "step": 112848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.837937174771732e-05,
+      "loss": 2.2369,
+      "step": 112856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.83733370686708e-05,
+      "loss": 2.2353,
+      "step": 112864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8367302678664406e-05,
+      "loss": 2.2451,
+      "step": 112872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836126857784818e-05,
+      "loss": 2.2288,
+      "step": 112880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8355234766372114e-05,
+      "loss": 2.2415,
+      "step": 112888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.834920124438624e-05,
+      "loss": 2.2288,
+      "step": 112896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8343168012040506e-05,
+      "loss": 2.2438,
+      "step": 112904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8337135069484936e-05,
+      "loss": 2.2432,
+      "step": 112912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8331102416869495e-05,
+      "loss": 2.2452,
+      "step": 112920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.832507005434416e-05,
+      "loss": 2.2388,
+      "step": 112928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.831903798205888e-05,
+      "loss": 2.242,
+      "step": 112936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.831300620016364e-05,
+      "loss": 2.2266,
+      "step": 112944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.830697470880837e-05,
+      "loss": 2.2507,
+      "step": 112952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.830094350814303e-05,
+      "loss": 2.2451,
+      "step": 112960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8294912598317535e-05,
+      "loss": 2.2442,
+      "step": 112968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8288881979481835e-05,
+      "loss": 2.2541,
+      "step": 112976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.828285165178583e-05,
+      "loss": 2.2463,
+      "step": 112984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.827682161537946e-05,
+      "loss": 2.2374,
+      "step": 112992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8270791870412636e-05,
+      "loss": 2.2455,
+      "step": 113000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.826476241703521e-05,
+      "loss": 2.2429,
+      "step": 113008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.825873325539711e-05,
+      "loss": 2.2604,
+      "step": 113016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.825270438564823e-05,
+      "loss": 2.2603,
+      "step": 113024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.824667580793845e-05,
+      "loss": 2.243,
+      "step": 113032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8240647522417615e-05,
+      "loss": 2.2571,
+      "step": 113040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.823461952923558e-05,
+      "loss": 2.2544,
+      "step": 113048
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.822859182854227e-05,
+      "loss": 2.2413,
+      "step": 113056
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.822256442048745e-05,
+      "loss": 2.24,
+      "step": 113064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8216537305221014e-05,
+      "loss": 2.2469,
+      "step": 113072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.821051048289278e-05,
+      "loss": 2.2407,
+      "step": 113080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.820448395365258e-05,
+      "loss": 2.2384,
+      "step": 113088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.819845771765023e-05,
+      "loss": 2.2325,
+      "step": 113096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.819243177503555e-05,
+      "loss": 2.2537,
+      "step": 113104
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.818640612595834e-05,
+      "loss": 2.2567,
+      "step": 113112
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.818038077056839e-05,
+      "loss": 2.2483,
+      "step": 113120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.817435570901551e-05,
+      "loss": 2.2417,
+      "step": 113128
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.816833094144947e-05,
+      "loss": 2.2557,
+      "step": 113136
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.816230646802005e-05,
+      "loss": 2.2427,
+      "step": 113144
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.815628228887702e-05,
+      "loss": 2.2447,
+      "step": 113152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8150258404170143e-05,
+      "loss": 2.24,
+      "step": 113160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.814423481404917e-05,
+      "loss": 2.2396,
+      "step": 113168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.813821151866384e-05,
+      "loss": 2.2423,
+      "step": 113176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8132188518163915e-05,
+      "loss": 2.261,
+      "step": 113184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.812616581269912e-05,
+      "loss": 2.2422,
+      "step": 113192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8120143402419164e-05,
+      "loss": 2.2409,
+      "step": 113200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.811412128747377e-05,
+      "loss": 2.2504,
+      "step": 113208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.810809946801269e-05,
+      "loss": 2.2439,
+      "step": 113216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.810207794418555e-05,
+      "loss": 2.2439,
+      "step": 113224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.80960567161421e-05,
+      "loss": 2.2276,
+      "step": 113232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.809003578403202e-05,
+      "loss": 2.2581,
+      "step": 113240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8084015148005e-05,
+      "loss": 2.2532,
+      "step": 113248
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8077994808210676e-05,
+      "loss": 2.2526,
+      "step": 113256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.807197476479873e-05,
+      "loss": 2.2446,
+      "step": 113264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.806595501791886e-05,
+      "loss": 2.2432,
+      "step": 113272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.805993556772068e-05,
+      "loss": 2.2386,
+      "step": 113280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8053916414353835e-05,
+      "loss": 2.2519,
+      "step": 113288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8047897557967975e-05,
+      "loss": 2.2389,
+      "step": 113296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.804187899871271e-05,
+      "loss": 2.243,
+      "step": 113304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.803586073673768e-05,
+      "loss": 2.2405,
+      "step": 113312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.802984277219249e-05,
+      "loss": 2.2396,
+      "step": 113320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.802382510522677e-05,
+      "loss": 2.24,
+      "step": 113328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8017807735990084e-05,
+      "loss": 2.2409,
+      "step": 113336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8011790664632045e-05,
+      "loss": 2.2456,
+      "step": 113344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.800577389130224e-05,
+      "loss": 2.2286,
+      "step": 113352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.799975741615024e-05,
+      "loss": 2.2408,
+      "step": 113360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7993741239325615e-05,
+      "loss": 2.2578,
+      "step": 113368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7987725360977934e-05,
+      "loss": 2.2403,
+      "step": 113376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7981709781256746e-05,
+      "loss": 2.2306,
+      "step": 113384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.79756945003116e-05,
+      "loss": 2.2439,
+      "step": 113392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.796967951829203e-05,
+      "loss": 2.2408,
+      "step": 113400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.79636648353476e-05,
+      "loss": 2.2326,
+      "step": 113408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.795765045162779e-05,
+      "loss": 2.2384,
+      "step": 113416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.795163636728215e-05,
+      "loss": 2.2394,
+      "step": 113424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.794562258246019e-05,
+      "loss": 2.2455,
+      "step": 113432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7939609097311396e-05,
+      "loss": 2.238,
+      "step": 113440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.793359591198526e-05,
+      "loss": 2.2399,
+      "step": 113448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.79275830266313e-05,
+      "loss": 2.2298,
+      "step": 113456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.792157044139899e-05,
+      "loss": 2.2356,
+      "step": 113464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.791555815643778e-05,
+      "loss": 2.2498,
+      "step": 113472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7909546171897136e-05,
+      "loss": 2.2515,
+      "step": 113480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.790353448792655e-05,
+      "loss": 2.2434,
+      "step": 113488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7897523104675454e-05,
+      "loss": 2.2228,
+      "step": 113496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7891512022293284e-05,
+      "loss": 2.2347,
+      "step": 113504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.78855012409295e-05,
+      "loss": 2.236,
+      "step": 113512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.787949076073349e-05,
+      "loss": 2.2362,
+      "step": 113520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.787348058185471e-05,
+      "loss": 2.2405,
+      "step": 113528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.786747070444256e-05,
+      "loss": 2.2406,
+      "step": 113536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7861461128646465e-05,
+      "loss": 2.2278,
+      "step": 113544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.785545185461579e-05,
+      "loss": 2.2489,
+      "step": 113552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.784944288249995e-05,
+      "loss": 2.2461,
+      "step": 113560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.784343421244833e-05,
+      "loss": 2.2459,
+      "step": 113568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.783742584461028e-05,
+      "loss": 2.2366,
+      "step": 113576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7831417779135204e-05,
+      "loss": 2.2375,
+      "step": 113584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7825410016172446e-05,
+      "loss": 2.2315,
+      "step": 113592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.781940255587136e-05,
+      "loss": 2.2379,
+      "step": 113600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.78133953983813e-05,
+      "loss": 2.2495,
+      "step": 113608
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7807388543851584e-05,
+      "loss": 2.2467,
+      "step": 113616
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.780138199243158e-05,
+      "loss": 2.2502,
+      "step": 113624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.779537574427056e-05,
+      "loss": 2.2521,
+      "step": 113632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.778936979951788e-05,
+      "loss": 2.2577,
+      "step": 113640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7783364158322854e-05,
+      "loss": 2.2394,
+      "step": 113648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.777735882083474e-05,
+      "loss": 2.2323,
+      "step": 113656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7771353787202856e-05,
+      "loss": 2.2442,
+      "step": 113664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.77653490575765e-05,
+      "loss": 2.2448,
+      "step": 113672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7759344632104946e-05,
+      "loss": 2.2328,
+      "step": 113680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.775334051093745e-05,
+      "loss": 2.2369,
+      "step": 113688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.774733669422327e-05,
+      "loss": 2.2421,
+      "step": 113696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7741333182111704e-05,
+      "loss": 2.2438,
+      "step": 113704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7735329974751946e-05,
+      "loss": 2.2367,
+      "step": 113712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.772932707229327e-05,
+      "loss": 2.236,
+      "step": 113720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.77233244748849e-05,
+      "loss": 2.2515,
+      "step": 113728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7717322182676063e-05,
+      "loss": 2.2391,
+      "step": 113736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7711320195815965e-05,
+      "loss": 2.237,
+      "step": 113744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.770531851445383e-05,
+      "loss": 2.2351,
+      "step": 113752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7699317138738866e-05,
+      "loss": 2.2481,
+      "step": 113760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7693316068820245e-05,
+      "loss": 2.2524,
+      "step": 113768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7687315304847166e-05,
+      "loss": 2.2369,
+      "step": 113776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.768131484696883e-05,
+      "loss": 2.2417,
+      "step": 113784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.767531469533437e-05,
+      "loss": 2.2345,
+      "step": 113792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7669314850092965e-05,
+      "loss": 2.2403,
+      "step": 113800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.766331531139378e-05,
+      "loss": 2.2415,
+      "step": 113808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.765731607938598e-05,
+      "loss": 2.2399,
+      "step": 113816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7651317154218674e-05,
+      "loss": 2.2476,
+      "step": 113824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7645318536041005e-05,
+      "loss": 2.2403,
+      "step": 113832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.763932022500212e-05,
+      "loss": 2.2342,
+      "step": 113840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7633322221251095e-05,
+      "loss": 2.2309,
+      "step": 113848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.762732452493708e-05,
+      "loss": 2.252,
+      "step": 113856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.762132713620918e-05,
+      "loss": 2.2267,
+      "step": 113864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.761533005521646e-05,
+      "loss": 2.2428,
+      "step": 113872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.760933328210802e-05,
+      "loss": 2.2415,
+      "step": 113880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.760333681703294e-05,
+      "loss": 2.2086,
+      "step": 113888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.759734066014032e-05,
+      "loss": 2.2345,
+      "step": 113896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.759134481157919e-05,
+      "loss": 2.2359,
+      "step": 113904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7585349271498624e-05,
+      "loss": 2.2503,
+      "step": 113912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.757935404004767e-05,
+      "loss": 2.2436,
+      "step": 113920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7573359117375363e-05,
+      "loss": 2.2386,
+      "step": 113928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7567364503630736e-05,
+      "loss": 2.2503,
+      "step": 113936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.756137019896283e-05,
+      "loss": 2.2478,
+      "step": 113944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7555376203520655e-05,
+      "loss": 2.2259,
+      "step": 113952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.754938251745323e-05,
+      "loss": 2.2471,
+      "step": 113960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.754338914090955e-05,
+      "loss": 2.2241,
+      "step": 113968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.753739607403861e-05,
+      "loss": 2.242,
+      "step": 113976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.75314033169894e-05,
+      "loss": 2.2452,
+      "step": 113984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7525410869910905e-05,
+      "loss": 2.2317,
+      "step": 113992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.751941873295209e-05,
+      "loss": 2.2265,
+      "step": 114000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.751342690626193e-05,
+      "loss": 2.2218,
+      "step": 114008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7507435389989386e-05,
+      "loss": 2.2357,
+      "step": 114016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.750144418428339e-05,
+      "loss": 2.2292,
+      "step": 114024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.74954532892929e-05,
+      "loss": 2.2385,
+      "step": 114032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.748946270516684e-05,
+      "loss": 2.2337,
+      "step": 114040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7483472432054136e-05,
+      "loss": 2.2568,
+      "step": 114048
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.747748247010374e-05,
+      "loss": 2.2402,
+      "step": 114056
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7471492819464504e-05,
+      "loss": 2.2497,
+      "step": 114064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7465503480285363e-05,
+      "loss": 2.2425,
+      "step": 114072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.745951445271522e-05,
+      "loss": 2.2479,
+      "step": 114080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.745352573690296e-05,
+      "loss": 2.2365,
+      "step": 114088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.744753733299745e-05,
+      "loss": 2.2326,
+      "step": 114096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7441549241147567e-05,
+      "loss": 2.2425,
+      "step": 114104
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.743556146150219e-05,
+      "loss": 2.2255,
+      "step": 114112
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7429573994210163e-05,
+      "loss": 2.2398,
+      "step": 114120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.742358683942032e-05,
+      "loss": 2.2389,
+      "step": 114128
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.741759999728155e-05,
+      "loss": 2.2393,
+      "step": 114136
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.741161346794263e-05,
+      "loss": 2.2417,
+      "step": 114144
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.740562725155242e-05,
+      "loss": 2.2477,
+      "step": 114152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.739964134825973e-05,
+      "loss": 2.2419,
+      "step": 114160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7393655758213374e-05,
+      "loss": 2.2349,
+      "step": 114168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.738767048156214e-05,
+      "loss": 2.2466,
+      "step": 114176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.738168551845483e-05,
+      "loss": 2.2424,
+      "step": 114184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7375700869040253e-05,
+      "loss": 2.2373,
+      "step": 114192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7369716533467156e-05,
+      "loss": 2.2433,
+      "step": 114200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.736373251188432e-05,
+      "loss": 2.2269,
+      "step": 114208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7357748804440526e-05,
+      "loss": 2.2635,
+      "step": 114216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.73517654112845e-05,
+      "loss": 2.2373,
+      "step": 114224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.734578233256501e-05,
+      "loss": 2.2195,
+      "step": 114232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.733979956843079e-05,
+      "loss": 2.2453,
+      "step": 114240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.733381711903058e-05,
+      "loss": 2.2549,
+      "step": 114248
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.732783498451309e-05,
+      "loss": 2.2452,
+      "step": 114256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.732185316502704e-05,
+      "loss": 2.2322,
+      "step": 114264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.731587166072116e-05,
+      "loss": 2.2494,
+      "step": 114272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.730989047174412e-05,
+      "loss": 2.2413,
+      "step": 114280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.730390959824462e-05,
+      "loss": 2.2258,
+      "step": 114288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.729792904037136e-05,
+      "loss": 2.2523,
+      "step": 114296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7291948798273026e-05,
+      "loss": 2.2405,
+      "step": 114304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7285968872098245e-05,
+      "loss": 2.2356,
+      "step": 114312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.727998926199571e-05,
+      "loss": 2.2263,
+      "step": 114320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7274009968114084e-05,
+      "loss": 2.2254,
+      "step": 114328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7268030990601993e-05,
+      "loss": 2.233,
+      "step": 114336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.726205232960807e-05,
+      "loss": 2.2291,
+      "step": 114344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7256073985280965e-05,
+      "loss": 2.22,
+      "step": 114352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.725009595776928e-05,
+      "loss": 2.245,
+      "step": 114360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.724411824722164e-05,
+      "loss": 2.2364,
+      "step": 114368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.723814085378665e-05,
+      "loss": 2.2261,
+      "step": 114376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.723216377761292e-05,
+      "loss": 2.2259,
+      "step": 114384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7226187018849034e-05,
+      "loss": 2.2428,
+      "step": 114392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7220210577643564e-05,
+      "loss": 2.212,
+      "step": 114400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.72142344541451e-05,
+      "loss": 2.2399,
+      "step": 114408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.72082586485022e-05,
+      "loss": 2.2401,
+      "step": 114416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7202283160863416e-05,
+      "loss": 2.251,
+      "step": 114424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.719630799137731e-05,
+      "loss": 2.2148,
+      "step": 114432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7190333140192435e-05,
+      "loss": 2.2426,
+      "step": 114440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.718435860745731e-05,
+      "loss": 2.2385,
+      "step": 114448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7178384393320466e-05,
+      "loss": 2.2372,
+      "step": 114456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7172410497930436e-05,
+      "loss": 2.2225,
+      "step": 114464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.71664369214357e-05,
+      "loss": 2.2351,
+      "step": 114472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7160463663984786e-05,
+      "loss": 2.2445,
+      "step": 114480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.715449072572621e-05,
+      "loss": 2.2361,
+      "step": 114488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.714851810680841e-05,
+      "loss": 2.258,
+      "step": 114496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.714254580737989e-05,
+      "loss": 2.2394,
+      "step": 114504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.713657382758913e-05,
+      "loss": 2.2387,
+      "step": 114512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7130602167584606e-05,
+      "loss": 2.225,
+      "step": 114520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7124630827514735e-05,
+      "loss": 2.2393,
+      "step": 114528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.711865980752798e-05,
+      "loss": 2.2542,
+      "step": 114536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.711268910777282e-05,
+      "loss": 2.2391,
+      "step": 114544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7106718728397625e-05,
+      "loss": 2.2211,
+      "step": 114552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.710074866955085e-05,
+      "loss": 2.2296,
+      "step": 114560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.709477893138094e-05,
+      "loss": 2.2308,
+      "step": 114568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.708880951403624e-05,
+      "loss": 2.2336,
+      "step": 114576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7082840417665196e-05,
+      "loss": 2.2311,
+      "step": 114584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7076871642416185e-05,
+      "loss": 2.2513,
+      "step": 114592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.707090318843761e-05,
+      "loss": 2.2343,
+      "step": 114600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7064935055877815e-05,
+      "loss": 2.236,
+      "step": 114608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.705896724488519e-05,
+      "loss": 2.2324,
+      "step": 114616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7052999755608096e-05,
+      "loss": 2.2256,
+      "step": 114624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.704703258819487e-05,
+      "loss": 2.2282,
+      "step": 114632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.704106574279388e-05,
+      "loss": 2.2382,
+      "step": 114640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7035099219553444e-05,
+      "loss": 2.2283,
+      "step": 114648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7029133018621894e-05,
+      "loss": 2.2426,
+      "step": 114656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.702316714014756e-05,
+      "loss": 2.2426,
+      "step": 114664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7017201584278737e-05,
+      "loss": 2.2184,
+      "step": 114672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.701123635116376e-05,
+      "loss": 2.226,
+      "step": 114680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7005271440950874e-05,
+      "loss": 2.228,
+      "step": 114688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.699930685378841e-05,
+      "loss": 2.242,
+      "step": 114696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6993342589824655e-05,
+      "loss": 2.2385,
+      "step": 114704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.698737864920784e-05,
+      "loss": 2.2388,
+      "step": 114712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.698141503208625e-05,
+      "loss": 2.2434,
+      "step": 114720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.697545173860815e-05,
+      "loss": 2.2279,
+      "step": 114728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.69694887689218e-05,
+      "loss": 2.249,
+      "step": 114736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.696352612317539e-05,
+      "loss": 2.2379,
+      "step": 114744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.69575638015172e-05,
+      "loss": 2.2523,
+      "step": 114752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.695160180409544e-05,
+      "loss": 2.2523,
+      "step": 114760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.694564013105831e-05,
+      "loss": 2.2487,
+      "step": 114768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.693967878255404e-05,
+      "loss": 2.2426,
+      "step": 114776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6933717758730814e-05,
+      "loss": 2.2566,
+      "step": 114784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.692775705973684e-05,
+      "loss": 2.2527,
+      "step": 114792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.692179668572029e-05,
+      "loss": 2.2386,
+      "step": 114800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6915836636829335e-05,
+      "loss": 2.2596,
+      "step": 114808
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.690987691321216e-05,
+      "loss": 2.2296,
+      "step": 114816
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6903917515016906e-05,
+      "loss": 2.2402,
+      "step": 114824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.689795844239173e-05,
+      "loss": 2.2498,
+      "step": 114832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.689199969548479e-05,
+      "loss": 2.2464,
+      "step": 114840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6886041274444194e-05,
+      "loss": 2.2441,
+      "step": 114848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6880083179418084e-05,
+      "loss": 2.2388,
+      "step": 114856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.687412541055458e-05,
+      "loss": 2.2309,
+      "step": 114864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.68681679680018e-05,
+      "loss": 2.2473,
+      "step": 114872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6862210851907826e-05,
+      "loss": 2.2492,
+      "step": 114880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6856254062420766e-05,
+      "loss": 2.2413,
+      "step": 114888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.685029759968873e-05,
+      "loss": 2.2338,
+      "step": 114896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.684434146385973e-05,
+      "loss": 2.255,
+      "step": 114904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.683838565508189e-05,
+      "loss": 2.2429,
+      "step": 114912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6832430173503286e-05,
+      "loss": 2.2564,
+      "step": 114920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.682647501927192e-05,
+      "loss": 2.2434,
+      "step": 114928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6820520192535866e-05,
+      "loss": 2.264,
+      "step": 114936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6814565693443155e-05,
+      "loss": 2.2385,
+      "step": 114944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6808611522141833e-05,
+      "loss": 2.2498,
+      "step": 114952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6802657678779895e-05,
+      "loss": 2.2462,
+      "step": 114960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6796704163505375e-05,
+      "loss": 2.245,
+      "step": 114968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.679075097646627e-05,
+      "loss": 2.2473,
+      "step": 114976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.678479811781058e-05,
+      "loss": 2.2413,
+      "step": 114984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.677884558768629e-05,
+      "loss": 2.2377,
+      "step": 114992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.677289338624138e-05,
+      "loss": 2.2397,
+      "step": 115000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.676694151362383e-05,
+      "loss": 2.2394,
+      "step": 115008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.676098996998159e-05,
+      "loss": 2.2238,
+      "step": 115016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.675503875546263e-05,
+      "loss": 2.2623,
+      "step": 115024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.67490878702149e-05,
+      "loss": 2.2383,
+      "step": 115032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.674313731438631e-05,
+      "loss": 2.2449,
+      "step": 115040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.673718708812483e-05,
+      "loss": 2.2303,
+      "step": 115048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.673123719157837e-05,
+      "loss": 2.2485,
+      "step": 115056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.672528762489483e-05,
+      "loss": 2.231,
+      "step": 115064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.671933838822214e-05,
+      "loss": 2.248,
+      "step": 115072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.671338948170818e-05,
+      "loss": 2.2343,
+      "step": 115080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6707440905500856e-05,
+      "loss": 2.2366,
+      "step": 115088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.670149265974804e-05,
+      "loss": 2.244,
+      "step": 115096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.66955447445976e-05,
+      "loss": 2.2477,
+      "step": 115104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.668959716019743e-05,
+      "loss": 2.2447,
+      "step": 115112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.668364990669536e-05,
+      "loss": 2.2318,
+      "step": 115120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.667770298423923e-05,
+      "loss": 2.2521,
+      "step": 115128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6671756392976906e-05,
+      "loss": 2.2321,
+      "step": 115136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.666581013305623e-05,
+      "loss": 2.2471,
+      "step": 115144
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6659864204625005e-05,
+      "loss": 2.2533,
+      "step": 115152
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6653918607831026e-05,
+      "loss": 2.2535,
+      "step": 115160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.664797334282216e-05,
+      "loss": 2.2333,
+      "step": 115168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6642028409746166e-05,
+      "loss": 2.2221,
+      "step": 115176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.663608380875084e-05,
+      "loss": 2.224,
+      "step": 115184
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.663013953998398e-05,
+      "loss": 2.2576,
+      "step": 115192
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.662419560359333e-05,
+      "loss": 2.2256,
+      "step": 115200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.661825199972669e-05,
+      "loss": 2.2493,
+      "step": 115208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6612308728531814e-05,
+      "loss": 2.2352,
+      "step": 115216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.660636579015644e-05,
+      "loss": 2.2155,
+      "step": 115224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.660042318474831e-05,
+      "loss": 2.246,
+      "step": 115232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6594480912455164e-05,
+      "loss": 2.2371,
+      "step": 115240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.658853897342473e-05,
+      "loss": 2.2372,
+      "step": 115248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.658259736780473e-05,
+      "loss": 2.2402,
+      "step": 115256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.657665609574284e-05,
+      "loss": 2.234,
+      "step": 115264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6570715157386825e-05,
+      "loss": 2.2398,
+      "step": 115272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6564774552884304e-05,
+      "loss": 2.2293,
+      "step": 115280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6558834282382995e-05,
+      "loss": 2.2436,
+      "step": 115288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.655289434603059e-05,
+      "loss": 2.2407,
+      "step": 115296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.654695474397475e-05,
+      "loss": 2.2437,
+      "step": 115304
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.654101547636309e-05,
+      "loss": 2.2473,
+      "step": 115312
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.653507654334331e-05,
+      "loss": 2.2383,
+      "step": 115320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.652913794506306e-05,
+      "loss": 2.2481,
+      "step": 115328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.652319968166993e-05,
+      "loss": 2.2422,
+      "step": 115336
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6517261753311566e-05,
+      "loss": 2.2281,
+      "step": 115344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6511324160135586e-05,
+      "loss": 2.2434,
+      "step": 115352
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.650538690228963e-05,
+      "loss": 2.2348,
+      "step": 115360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.649944997992124e-05,
+      "loss": 2.2511,
+      "step": 115368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.649351339317804e-05,
+      "loss": 2.2249,
+      "step": 115376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6487577142207625e-05,
+      "loss": 2.2387,
+      "step": 115384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.648164122715755e-05,
+      "loss": 2.2337,
+      "step": 115392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.647570564817539e-05,
+      "loss": 2.2208,
+      "step": 115400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6469770405408717e-05,
+      "loss": 2.2271,
+      "step": 115408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.646383549900506e-05,
+      "loss": 2.2417,
+      "step": 115416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6457900929111966e-05,
+      "loss": 2.2377,
+      "step": 115424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.645196669587698e-05,
+      "loss": 2.2313,
+      "step": 115432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.644603279944763e-05,
+      "loss": 2.2368,
+      "step": 115440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6440099239971414e-05,
+      "loss": 2.2311,
+      "step": 115448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6434166017595846e-05,
+      "loss": 2.2242,
+      "step": 115456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.642823313246845e-05,
+      "loss": 2.2385,
+      "step": 115464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.64223005847367e-05,
+      "loss": 2.2396,
+      "step": 115472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6416368374548067e-05,
+      "loss": 2.2351,
+      "step": 115480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6410436502050045e-05,
+      "loss": 2.2385,
+      "step": 115488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6404504967390105e-05,
+      "loss": 2.2424,
+      "step": 115496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6398573770715685e-05,
+      "loss": 2.2293,
+      "step": 115504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6392642912174255e-05,
+      "loss": 2.2474,
+      "step": 115512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.638671239191326e-05,
+      "loss": 2.224,
+      "step": 115520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.63807822100801e-05,
+      "loss": 2.2476,
+      "step": 115528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6374852366822225e-05,
+      "loss": 2.2377,
+      "step": 115536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.636892286228707e-05,
+      "loss": 2.23,
+      "step": 115544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6362993696622004e-05,
+      "loss": 2.2375,
+      "step": 115552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.635706486997444e-05,
+      "loss": 2.2361,
+      "step": 115560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.635113638249178e-05,
+      "loss": 2.2554,
+      "step": 115568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.634520823432142e-05,
+      "loss": 2.2343,
+      "step": 115576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6339280425610695e-05,
+      "loss": 2.2479,
+      "step": 115584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.633335295650699e-05,
+      "loss": 2.2272,
+      "step": 115592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.632742582715768e-05,
+      "loss": 2.2412,
+      "step": 115600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.632149903771008e-05,
+      "loss": 2.2362,
+      "step": 115608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.631557258831155e-05,
+      "loss": 2.2404,
+      "step": 115616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6309646479109435e-05,
+      "loss": 2.2353,
+      "step": 115624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.630372071025103e-05,
+      "loss": 2.2554,
+      "step": 115632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.629779528188367e-05,
+      "loss": 2.253,
+      "step": 115640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.629187019415465e-05,
+      "loss": 2.2513,
+      "step": 115648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.628594544721129e-05,
+      "loss": 2.2603,
+      "step": 115656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.628002104120086e-05,
+      "loss": 2.2522,
+      "step": 115664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.627409697627064e-05,
+      "loss": 2.2561,
+      "step": 115672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6268173252567916e-05,
+      "loss": 2.2431,
+      "step": 115680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.626224987023994e-05,
+      "loss": 2.2368,
+      "step": 115688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.625632682943397e-05,
+      "loss": 2.2512,
+      "step": 115696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.625040413029727e-05,
+      "loss": 2.2313,
+      "step": 115704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.624448177297706e-05,
+      "loss": 2.2323,
+      "step": 115712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.623855975762057e-05,
+      "loss": 2.2252,
+      "step": 115720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.623263808437503e-05,
+      "loss": 2.2328,
+      "step": 115728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6226716753387674e-05,
+      "loss": 2.2331,
+      "step": 115736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6220795764805655e-05,
+      "loss": 2.2412,
+      "step": 115744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.621487511877621e-05,
+      "loss": 2.2285,
+      "step": 115752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6208954815446525e-05,
+      "loss": 2.2295,
+      "step": 115760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.620303485496376e-05,
+      "loss": 2.2342,
+      "step": 115768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.61971152374751e-05,
+      "loss": 2.2343,
+      "step": 115776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.619119596312769e-05,
+      "loss": 2.2371,
+      "step": 115784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.618527703206872e-05,
+      "loss": 2.2343,
+      "step": 115792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.61793584444453e-05,
+      "loss": 2.2329,
+      "step": 115800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.617344020040459e-05,
+      "loss": 2.2432,
+      "step": 115808
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.61675223000937e-05,
+      "loss": 2.242,
+      "step": 115816
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.616160474365976e-05,
+      "loss": 2.2469,
+      "step": 115824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6155687531249876e-05,
+      "loss": 2.239,
+      "step": 115832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.614977066301115e-05,
+      "loss": 2.249,
+      "step": 115840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.614385413909069e-05,
+      "loss": 2.2216,
+      "step": 115848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.613793795963556e-05,
+      "loss": 2.2504,
+      "step": 115856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6132022124792854e-05,
+      "loss": 2.2372,
+      "step": 115864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.612610663470964e-05,
+      "loss": 2.2255,
+      "step": 115872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6120191489532956e-05,
+      "loss": 2.2458,
+      "step": 115880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6114276689409875e-05,
+      "loss": 2.2428,
+      "step": 115888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.610836223448745e-05,
+      "loss": 2.2348,
+      "step": 115896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.610244812491267e-05,
+      "loss": 2.2419,
+      "step": 115904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.609653436083261e-05,
+      "loss": 2.2465,
+      "step": 115912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6090620942394244e-05,
+      "loss": 2.2398,
+      "step": 115920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6084707869744627e-05,
+      "loss": 2.2396,
+      "step": 115928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6078795143030725e-05,
+      "loss": 2.2322,
+      "step": 115936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.607288276239953e-05,
+      "loss": 2.2419,
+      "step": 115944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.606697072799805e-05,
+      "loss": 2.2451,
+      "step": 115952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.606105903997322e-05,
+      "loss": 2.2184,
+      "step": 115960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.605514769847203e-05,
+      "loss": 2.2335,
+      "step": 115968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6049236703641455e-05,
+      "loss": 2.2405,
+      "step": 115976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.60433260556284e-05,
+      "loss": 2.2346,
+      "step": 115984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6037415754579825e-05,
+      "loss": 2.2329,
+      "step": 115992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.603150580064265e-05,
+      "loss": 2.2337,
+      "step": 116000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.602559619396384e-05,
+      "loss": 2.2297,
+      "step": 116008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.601968693469026e-05,
+      "loss": 2.2421,
+      "step": 116016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6013778022968826e-05,
+      "loss": 2.2388,
+      "step": 116024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6007869458946444e-05,
+      "loss": 2.2347,
+      "step": 116032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6001961242769994e-05,
+      "loss": 2.2343,
+      "step": 116040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5996053374586365e-05,
+      "loss": 2.2303,
+      "step": 116048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.599014585454241e-05,
+      "loss": 2.2532,
+      "step": 116056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5984238682785015e-05,
+      "loss": 2.2308,
+      "step": 116064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5978331859461006e-05,
+      "loss": 2.2383,
+      "step": 116072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5972425384717244e-05,
+      "loss": 2.2317,
+      "step": 116080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.596651925870057e-05,
+      "loss": 2.2218,
+      "step": 116088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.596061348155779e-05,
+      "loss": 2.2267,
+      "step": 116096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5954708053435735e-05,
+      "loss": 2.2246,
+      "step": 116104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5948802974481226e-05,
+      "loss": 2.2299,
+      "step": 116112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.594289824484104e-05,
+      "loss": 2.2342,
+      "step": 116120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.593699386466198e-05,
+      "loss": 2.2386,
+      "step": 116128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.593108983409082e-05,
+      "loss": 2.2351,
+      "step": 116136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.592518615327438e-05,
+      "loss": 2.2324,
+      "step": 116144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5919282822359355e-05,
+      "loss": 2.2291,
+      "step": 116152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5913379841492555e-05,
+      "loss": 2.2338,
+      "step": 116160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5907477210820725e-05,
+      "loss": 2.2151,
+      "step": 116168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.590157493049058e-05,
+      "loss": 2.2396,
+      "step": 116176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5895673000648856e-05,
+      "loss": 2.2272,
+      "step": 116184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.588977142144228e-05,
+      "loss": 2.2443,
+      "step": 116192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.58838701930176e-05,
+      "loss": 2.2364,
+      "step": 116200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5877969315521466e-05,
+      "loss": 2.2381,
+      "step": 116208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.587206878910059e-05,
+      "loss": 2.2393,
+      "step": 116216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.58661686139017e-05,
+      "loss": 2.2331,
+      "step": 116224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.586026879007142e-05,
+      "loss": 2.2267,
+      "step": 116232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5854369317756456e-05,
+      "loss": 2.238,
+      "step": 116240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.584847019710345e-05,
+      "loss": 2.2483,
+      "step": 116248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.584257142825907e-05,
+      "loss": 2.2518,
+      "step": 116256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5836673011369954e-05,
+      "loss": 2.2556,
+      "step": 116264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.583077494658272e-05,
+      "loss": 2.2425,
+      "step": 116272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5824877234044025e-05,
+      "loss": 2.2345,
+      "step": 116280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.581897987390047e-05,
+      "loss": 2.2476,
+      "step": 116288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5813082866298665e-05,
+      "loss": 2.2513,
+      "step": 116296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.580718621138522e-05,
+      "loss": 2.2349,
+      "step": 116304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.58012899093067e-05,
+      "loss": 2.2432,
+      "step": 116312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.579539396020973e-05,
+      "loss": 2.2435,
+      "step": 116320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.578949836424085e-05,
+      "loss": 2.2473,
+      "step": 116328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.578360312154663e-05,
+      "loss": 2.247,
+      "step": 116336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.577770823227363e-05,
+      "loss": 2.2435,
+      "step": 116344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5771813696568396e-05,
+      "loss": 2.2145,
+      "step": 116352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.576591951457749e-05,
+      "loss": 2.2237,
+      "step": 116360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.57600256864474e-05,
+      "loss": 2.2432,
+      "step": 116368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.575413221232467e-05,
+      "loss": 2.2253,
+      "step": 116376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.574823909235583e-05,
+      "loss": 2.2582,
+      "step": 116384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.574234632668735e-05,
+      "loss": 2.2318,
+      "step": 116392
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.573645391546573e-05,
+      "loss": 2.2386,
+      "step": 116400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.573056185883748e-05,
+      "loss": 2.2549,
+      "step": 116408
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.572467015694908e-05,
+      "loss": 2.2314,
+      "step": 116416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.571877880994696e-05,
+      "loss": 2.2351,
+      "step": 116424
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5712887817977595e-05,
+      "loss": 2.2417,
+      "step": 116432
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.570699718118747e-05,
+      "loss": 2.2439,
+      "step": 116440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5701106899722976e-05,
+      "loss": 2.2313,
+      "step": 116448
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.569521697373056e-05,
+      "loss": 2.2459,
+      "step": 116456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.568932740335668e-05,
+      "loss": 2.2644,
+      "step": 116464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.568343818874772e-05,
+      "loss": 2.2448,
+      "step": 116472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.567754933005008e-05,
+      "loss": 2.2374,
+      "step": 116480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.567166082741017e-05,
+      "loss": 2.2318,
+      "step": 116488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5665772680974404e-05,
+      "loss": 2.2532,
+      "step": 116496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.565988489088912e-05,
+      "loss": 2.2368,
+      "step": 116504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.56539974573007e-05,
+      "loss": 2.2328,
+      "step": 116512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5648110380355525e-05,
+      "loss": 2.2426,
+      "step": 116520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.564222366019993e-05,
+      "loss": 2.2528,
+      "step": 116528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.563633729698027e-05,
+      "loss": 2.2354,
+      "step": 116536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5630451290842866e-05,
+      "loss": 2.24,
+      "step": 116544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.562456564193407e-05,
+      "loss": 2.2415,
+      "step": 116552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.561868035040018e-05,
+      "loss": 2.2497,
+      "step": 116560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5612795416387496e-05,
+      "loss": 2.2498,
+      "step": 116568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.560691084004236e-05,
+      "loss": 2.2364,
+      "step": 116576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.560102662151101e-05,
+      "loss": 2.2475,
+      "step": 116584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.559514276093977e-05,
+      "loss": 2.2394,
+      "step": 116592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5589259258474904e-05,
+      "loss": 2.238,
+      "step": 116600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5583376114262646e-05,
+      "loss": 2.249,
+      "step": 116608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5577493328449286e-05,
+      "loss": 2.2634,
+      "step": 116616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.557161090118105e-05,
+      "loss": 2.2547,
+      "step": 116624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.556572883260421e-05,
+      "loss": 2.2533,
+      "step": 116632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5559847122864954e-05,
+      "loss": 2.2438,
+      "step": 116640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.555396577210951e-05,
+      "loss": 2.2466,
+      "step": 116648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5548084780484125e-05,
+      "loss": 2.2372,
+      "step": 116656
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5542204148134946e-05,
+      "loss": 2.2383,
+      "step": 116664
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.553632387520821e-05,
+      "loss": 2.2331,
+      "step": 116672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.553044396185008e-05,
+      "loss": 2.2359,
+      "step": 116680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5524564408206735e-05,
+      "loss": 2.2511,
+      "step": 116688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.551868521442435e-05,
+      "loss": 2.2359,
+      "step": 116696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.551280638064906e-05,
+      "loss": 2.2325,
+      "step": 116704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5506927907027054e-05,
+      "loss": 2.2398,
+      "step": 116712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.550104979370443e-05,
+      "loss": 2.2333,
+      "step": 116720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5495172040827335e-05,
+      "loss": 2.2314,
+      "step": 116728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.54892946485419e-05,
+      "loss": 2.2615,
+      "step": 116736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5483417616994214e-05,
+      "loss": 2.23,
+      "step": 116744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.547754094633039e-05,
+      "loss": 2.229,
+      "step": 116752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.547166463669654e-05,
+      "loss": 2.2399,
+      "step": 116760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.546578868823874e-05,
+      "loss": 2.2347,
+      "step": 116768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5459913101103056e-05,
+      "loss": 2.2515,
+      "step": 116776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5454037875435565e-05,
+      "loss": 2.2511,
+      "step": 116784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.544816301138234e-05,
+      "loss": 2.2248,
+      "step": 116792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5442288509089404e-05,
+      "loss": 2.2568,
+      "step": 116800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.543641436870279e-05,
+      "loss": 2.2473,
+      "step": 116808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.543054059036858e-05,
+      "loss": 2.2513,
+      "step": 116816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.542466717423275e-05,
+      "loss": 2.2269,
+      "step": 116824
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.541879412044133e-05,
+      "loss": 2.2413,
+      "step": 116832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5412921429140316e-05,
+      "loss": 2.2369,
+      "step": 116840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5407049100475734e-05,
+      "loss": 2.241,
+      "step": 116848
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5401177134593534e-05,
+      "loss": 2.2367,
+      "step": 116856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.539530553163971e-05,
+      "loss": 2.2443,
+      "step": 116864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.538943429176023e-05,
+      "loss": 2.2411,
+      "step": 116872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.538356341510105e-05,
+      "loss": 2.2471,
+      "step": 116880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5377692901808124e-05,
+      "loss": 2.2403,
+      "step": 116888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5371822752027386e-05,
+      "loss": 2.2458,
+      "step": 116896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.536595296590479e-05,
+      "loss": 2.2467,
+      "step": 116904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.536008354358623e-05,
+      "loss": 2.2357,
+      "step": 116912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.535421448521765e-05,
+      "loss": 2.2235,
+      "step": 116920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5348345790944943e-05,
+      "loss": 2.2341,
+      "step": 116928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5342477460913993e-05,
+      "loss": 2.2363,
+      "step": 116936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.53366094952707e-05,
+      "loss": 2.2351,
+      "step": 116944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.533074189416095e-05,
+      "loss": 2.233,
+      "step": 116952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.532487465773059e-05,
+      "loss": 2.2363,
+      "step": 116960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.531900778612549e-05,
+      "loss": 2.2487,
+      "step": 116968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.531314127949151e-05,
+      "loss": 2.2471,
+      "step": 116976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5307275137974495e-05,
+      "loss": 2.2231,
+      "step": 116984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.530140936172025e-05,
+      "loss": 2.2541,
+      "step": 116992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5295543950874624e-05,
+      "loss": 2.2606,
+      "step": 117000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.528967890558344e-05,
+      "loss": 2.2468,
+      "step": 117008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5283814225992476e-05,
+      "loss": 2.2322,
+      "step": 117016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.527794991224753e-05,
+      "loss": 2.2367,
+      "step": 117024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.527208596449442e-05,
+      "loss": 2.2355,
+      "step": 117032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.52662223828789e-05,
+      "loss": 2.2452,
+      "step": 117040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.526035916754674e-05,
+      "loss": 2.2294,
+      "step": 117048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.52544963186437e-05,
+      "loss": 2.2341,
+      "step": 117056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.524863383631556e-05,
+      "loss": 2.2247,
+      "step": 117064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.524277172070801e-05,
+      "loss": 2.2409,
+      "step": 117072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.523690997196682e-05,
+      "loss": 2.2172,
+      "step": 117080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.523104859023772e-05,
+      "loss": 2.2171,
+      "step": 117088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5225187575666386e-05,
+      "loss": 2.2251,
+      "step": 117096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.521932692839856e-05,
+      "loss": 2.2337,
+      "step": 117104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5213466648579915e-05,
+      "loss": 2.2384,
+      "step": 117112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.520760673635616e-05,
+      "loss": 2.2311,
+      "step": 117120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.520174719187296e-05,
+      "loss": 2.2299,
+      "step": 117128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5195888015275974e-05,
+      "loss": 2.2392,
+      "step": 117136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.519002920671089e-05,
+      "loss": 2.2331,
+      "step": 117144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.518417076632332e-05,
+      "loss": 2.2414,
+      "step": 117152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.517831269425894e-05,
+      "loss": 2.2433,
+      "step": 117160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5172454990663385e-05,
+      "loss": 2.2398,
+      "step": 117168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.516659765568223e-05,
+      "loss": 2.2221,
+      "step": 117176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.516074068946114e-05,
+      "loss": 2.2423,
+      "step": 117184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5154884092145685e-05,
+      "loss": 2.2226,
+      "step": 117192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.514902786388151e-05,
+      "loss": 2.2513,
+      "step": 117200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.514317200481414e-05,
+      "loss": 2.2447,
+      "step": 117208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.513731651508918e-05,
+      "loss": 2.2311,
+      "step": 117216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.513146139485223e-05,
+      "loss": 2.2216,
+      "step": 117224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.512560664424879e-05,
+      "loss": 2.2415,
+      "step": 117232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.511975226342444e-05,
+      "loss": 2.2339,
+      "step": 117240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5113898252524716e-05,
+      "loss": 2.2387,
+      "step": 117248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.510804461169518e-05,
+      "loss": 2.2342,
+      "step": 117256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5102191341081295e-05,
+      "loss": 2.2349,
+      "step": 117264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.50963384408286e-05,
+      "loss": 2.2507,
+      "step": 117272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.509048591108263e-05,
+      "loss": 2.2339,
+      "step": 117280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.508463375198884e-05,
+      "loss": 2.228,
+      "step": 117288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.507878196369273e-05,
+      "loss": 2.2358,
+      "step": 117296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.507293054633978e-05,
+      "loss": 2.2401,
+      "step": 117304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5067079500075435e-05,
+      "loss": 2.2246,
+      "step": 117312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5061228825045176e-05,
+      "loss": 2.2411,
+      "step": 117320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5055378521394446e-05,
+      "loss": 2.2552,
+      "step": 117328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.504952858926869e-05,
+      "loss": 2.2335,
+      "step": 117336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.504367902881332e-05,
+      "loss": 2.2403,
+      "step": 117344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.503782984017377e-05,
+      "loss": 2.2216,
+      "step": 117352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.503198102349546e-05,
+      "loss": 2.2261,
+      "step": 117360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.502613257892377e-05,
+      "loss": 2.2364,
+      "step": 117368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.50202845066041e-05,
+      "loss": 2.2359,
+      "step": 117376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.501443680668185e-05,
+      "loss": 2.2423,
+      "step": 117384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.500858947930238e-05,
+      "loss": 2.2465,
+      "step": 117392
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.500274252461105e-05,
+      "loss": 2.2302,
+      "step": 117400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.499689594275322e-05,
+      "loss": 2.2308,
+      "step": 117408
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.499104973387426e-05,
+      "loss": 2.2183,
+      "step": 117416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.498520389811947e-05,
+      "loss": 2.2202,
+      "step": 117424
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.497935843563419e-05,
+      "loss": 2.2311,
+      "step": 117432
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.4973513346563776e-05,
+      "loss": 2.2485,
+      "step": 117440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.496766863105348e-05,
+      "loss": 2.2447,
+      "step": 117448
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.4961824289248625e-05,
+      "loss": 2.2244,
+      "step": 117456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.495598032129451e-05,
+      "loss": 2.2429,
+      "step": 117464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.495013672733643e-05,
+      "loss": 2.2393,
+      "step": 117472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.494429350751963e-05,
+      "loss": 2.2417,
+      "step": 117480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.493845066198937e-05,
+      "loss": 2.2525,
+      "step": 117488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.4932608190890936e-05,
+      "loss": 2.2457,
+      "step": 117496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.4926766094369545e-05,
+      "loss": 2.2392,
+      "step": 117504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.492092437257043e-05,
+      "loss": 2.2451,
+      "step": 117512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.491508302563884e-05,
+      "loss": 2.2312,
+      "step": 117520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.490924205371997e-05,
+      "loss": 2.2394,
+      "step": 117528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.490340145695904e-05,
+      "loss": 2.2487,
+      "step": 117536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.4897561235501236e-05,
+      "loss": 2.2376,
+      "step": 117544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.4891721389491756e-05,
+      "loss": 2.257,
+      "step": 117552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.488588191907578e-05,
+      "loss": 2.2365,
+      "step": 117560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.4880042824398484e-05,
+      "loss": 2.2533,
+      "step": 117568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.4874204105605014e-05,
+      "loss": 2.2405,
+      "step": 117576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.4868365762840525e-05,
+      "loss": 2.2293,
+      "step": 117584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.486252779625015e-05,
+      "loss": 2.2355,
+      "step": 117592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.4856690205979046e-05,
+      "loss": 2.2406,
+      "step": 117600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4850852992172324e-05,
+      "loss": 2.2366,
+      "step": 117608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.484501615497509e-05,
+      "loss": 2.2365,
+      "step": 117616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.483917969453246e-05,
+      "loss": 2.235,
+      "step": 117624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.483334361098953e-05,
+      "loss": 2.2584,
+      "step": 117632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.482750790449136e-05,
+      "loss": 2.2641,
+      "step": 117640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.482167257518306e-05,
+      "loss": 2.2448,
+      "step": 117648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.481583762320969e-05,
+      "loss": 2.2481,
+      "step": 117656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.481000304871629e-05,
+      "loss": 2.2441,
+      "step": 117664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.480416885184791e-05,
+      "loss": 2.2368,
+      "step": 117672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.47983350327496e-05,
+      "loss": 2.2277,
+      "step": 117680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.47925015915664e-05,
+      "loss": 2.2278,
+      "step": 117688
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.478666852844331e-05,
+      "loss": 2.2266,
+      "step": 117696
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.478083584352534e-05,
+      "loss": 2.2658,
+      "step": 117704
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4775003536957515e-05,
+      "loss": 2.2495,
+      "step": 117712
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.476917160888479e-05,
+      "loss": 2.2432,
+      "step": 117720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4763340059452174e-05,
+      "loss": 2.2433,
+      "step": 117728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4757508888804636e-05,
+      "loss": 2.2519,
+      "step": 117736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.475167809708715e-05,
+      "loss": 2.2469,
+      "step": 117744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4745847684444644e-05,
+      "loss": 2.2377,
+      "step": 117752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.474001765102207e-05,
+      "loss": 2.2402,
+      "step": 117760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4734187996964386e-05,
+      "loss": 2.2456,
+      "step": 117768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4728358722416486e-05,
+      "loss": 2.249,
+      "step": 117776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.472252982752331e-05,
+      "loss": 2.2385,
+      "step": 117784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.471670131242976e-05,
+      "loss": 2.252,
+      "step": 117792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4710873177280715e-05,
+      "loss": 2.2401,
+      "step": 117800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4705045422221085e-05,
+      "loss": 2.233,
+      "step": 117808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.469921804739574e-05,
+      "loss": 2.2531,
+      "step": 117816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.469339105294957e-05,
+      "loss": 2.2426,
+      "step": 117824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.46875644390274e-05,
+      "loss": 2.251,
+      "step": 117832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4681738205774085e-05,
+      "loss": 2.2638,
+      "step": 117840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4675912353334505e-05,
+      "loss": 2.2392,
+      "step": 117848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.467008688185345e-05,
+      "loss": 2.2333,
+      "step": 117856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4664261791475736e-05,
+      "loss": 2.2625,
+      "step": 117864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4658437082346226e-05,
+      "loss": 2.2235,
+      "step": 117872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.465261275460967e-05,
+      "loss": 2.2447,
+      "step": 117880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.464678880841089e-05,
+      "loss": 2.2401,
+      "step": 117888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.464096524389465e-05,
+      "loss": 2.2416,
+      "step": 117896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4635142061205764e-05,
+      "loss": 2.2376,
+      "step": 117904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.462931926048895e-05,
+      "loss": 2.2377,
+      "step": 117912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4623496841889e-05,
+      "loss": 2.2507,
+      "step": 117920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.461767480555063e-05,
+      "loss": 2.2312,
+      "step": 117928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4611853151618596e-05,
+      "loss": 2.2395,
+      "step": 117936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4606031880237605e-05,
+      "loss": 2.2458,
+      "step": 117944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.46002109915524e-05,
+      "loss": 2.2518,
+      "step": 117952
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4594390485707686e-05,
+      "loss": 2.2493,
+      "step": 117960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.458857036284814e-05,
+      "loss": 2.2306,
+      "step": 117968
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4582750623118466e-05,
+      "loss": 2.247,
+      "step": 117976
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4576931266663354e-05,
+      "loss": 2.2255,
+      "step": 117984
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.457111229362745e-05,
+      "loss": 2.2352,
+      "step": 117992
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.456529370415542e-05,
+      "loss": 2.2227,
+      "step": 118000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.455947549839195e-05,
+      "loss": 2.2492,
+      "step": 118008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4553657676481605e-05,
+      "loss": 2.2379,
+      "step": 118016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.454784023856909e-05,
+      "loss": 2.2192,
+      "step": 118024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.454202318479899e-05,
+      "loss": 2.2482,
+      "step": 118032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.453620651531595e-05,
+      "loss": 2.2611,
+      "step": 118040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.453039023026453e-05,
+      "loss": 2.2324,
+      "step": 118048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.452457432978934e-05,
+      "loss": 2.2404,
+      "step": 118056
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.451875881403499e-05,
+      "loss": 2.247,
+      "step": 118064
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.451294368314603e-05,
+      "loss": 2.2509,
+      "step": 118072
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4507128937267004e-05,
+      "loss": 2.248,
+      "step": 118080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.45013145765425e-05,
+      "loss": 2.2484,
+      "step": 118088
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.449550060111707e-05,
+      "loss": 2.2445,
+      "step": 118096
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.448968701113523e-05,
+      "loss": 2.256,
+      "step": 118104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4483873806741496e-05,
+      "loss": 2.2427,
+      "step": 118112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.447806098808043e-05,
+      "loss": 2.2343,
+      "step": 118120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.447224855529648e-05,
+      "loss": 2.2435,
+      "step": 118128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.446643650853419e-05,
+      "loss": 2.2505,
+      "step": 118136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4460624847938024e-05,
+      "loss": 2.25,
+      "step": 118144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.445481357365246e-05,
+      "loss": 2.2341,
+      "step": 118152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4449002685821984e-05,
+      "loss": 2.2433,
+      "step": 118160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4443192184591044e-05,
+      "loss": 2.236,
+      "step": 118168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.44373820701041e-05,
+      "loss": 2.2642,
+      "step": 118176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4431572342505565e-05,
+      "loss": 2.2467,
+      "step": 118184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4425763001939905e-05,
+      "loss": 2.2243,
+      "step": 118192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.441995404855152e-05,
+      "loss": 2.2247,
+      "step": 118200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.441414548248483e-05,
+      "loss": 2.2446,
+      "step": 118208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.440833730388422e-05,
+      "loss": 2.2256,
+      "step": 118216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4402529512894116e-05,
+      "loss": 2.2389,
+      "step": 118224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.439672210965885e-05,
+      "loss": 2.2474,
+      "step": 118232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.439091509432283e-05,
+      "loss": 2.2366,
+      "step": 118240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.438510846703041e-05,
+      "loss": 2.2333,
+      "step": 118248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.437930222792597e-05,
+      "loss": 2.2411,
+      "step": 118256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.43734963771538e-05,
+      "loss": 2.233,
+      "step": 118264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4367690914858275e-05,
+      "loss": 2.2323,
+      "step": 118272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.436188584118372e-05,
+      "loss": 2.2353,
+      "step": 118280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.435608115627442e-05,
+      "loss": 2.2581,
+      "step": 118288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.43502768602747e-05,
+      "loss": 2.249,
+      "step": 118296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.434447295332885e-05,
+      "loss": 2.2316,
+      "step": 118304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.433866943558118e-05,
+      "loss": 2.2388,
+      "step": 118312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.433286630717593e-05,
+      "loss": 2.2267,
+      "step": 118320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.432706356825738e-05,
+      "loss": 2.2405,
+      "step": 118328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.432126121896979e-05,
+      "loss": 2.2485,
+      "step": 118336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.431545925945741e-05,
+      "loss": 2.2314,
+      "step": 118344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.430965768986446e-05,
+      "loss": 2.2203,
+      "step": 118352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.43038565103352e-05,
+      "loss": 2.2372,
+      "step": 118360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.429805572101381e-05,
+      "loss": 2.2396,
+      "step": 118368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.429225532204452e-05,
+      "loss": 2.2396,
+      "step": 118376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.428645531357153e-05,
+      "loss": 2.2355,
+      "step": 118384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.428065569573903e-05,
+      "loss": 2.2398,
+      "step": 118392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.427485646869118e-05,
+      "loss": 2.2485,
+      "step": 118400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.426905763257216e-05,
+      "loss": 2.2359,
+      "step": 118408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4263259187526143e-05,
+      "loss": 2.2271,
+      "step": 118416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.425746113369727e-05,
+      "loss": 2.2311,
+      "step": 118424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.425166347122967e-05,
+      "loss": 2.2345,
+      "step": 118432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4245866200267486e-05,
+      "loss": 2.2389,
+      "step": 118440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.424006932095484e-05,
+      "loss": 2.2264,
+      "step": 118448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.423427283343583e-05,
+      "loss": 2.2236,
+      "step": 118456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4228476737854574e-05,
+      "loss": 2.2392,
+      "step": 118464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.422268103435517e-05,
+      "loss": 2.2486,
+      "step": 118472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4216885723081655e-05,
+      "loss": 2.2449,
+      "step": 118480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.421109080417815e-05,
+      "loss": 2.2341,
+      "step": 118488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.420529627778871e-05,
+      "loss": 2.2335,
+      "step": 118496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4199502144057355e-05,
+      "loss": 2.2234,
+      "step": 118504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.419370840312815e-05,
+      "loss": 2.2574,
+      "step": 118512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.418791505514513e-05,
+      "loss": 2.221,
+      "step": 118520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.418212210025234e-05,
+      "loss": 2.2505,
+      "step": 118528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.417632953859374e-05,
+      "loss": 2.2443,
+      "step": 118536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4170537370313366e-05,
+      "loss": 2.237,
+      "step": 118544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4164745595555225e-05,
+      "loss": 2.2271,
+      "step": 118552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4158954214463277e-05,
+      "loss": 2.2384,
+      "step": 118560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.41531632271815e-05,
+      "loss": 2.2409,
+      "step": 118568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.414737263385387e-05,
+      "loss": 2.2408,
+      "step": 118576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.414158243462433e-05,
+      "loss": 2.2314,
+      "step": 118584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.413579262963683e-05,
+      "loss": 2.2336,
+      "step": 118592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.41300032190353e-05,
+      "loss": 2.2383,
+      "step": 118600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.412421420296369e-05,
+      "loss": 2.2481,
+      "step": 118608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.411842558156588e-05,
+      "loss": 2.2281,
+      "step": 118616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4112637354985785e-05,
+      "loss": 2.2464,
+      "step": 118624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.410684952336733e-05,
+      "loss": 2.2392,
+      "step": 118632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.410106208685436e-05,
+      "loss": 2.2334,
+      "step": 118640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.409527504559078e-05,
+      "loss": 2.2248,
+      "step": 118648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4089488399720444e-05,
+      "loss": 2.2376,
+      "step": 118656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4083702149387215e-05,
+      "loss": 2.2288,
+      "step": 118664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.407791629473493e-05,
+      "loss": 2.2391,
+      "step": 118672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.407213083590743e-05,
+      "loss": 2.2342,
+      "step": 118680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.406634577304856e-05,
+      "loss": 2.2271,
+      "step": 118688
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4060561106302116e-05,
+      "loss": 2.2522,
+      "step": 118696
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.405477683581189e-05,
+      "loss": 2.2226,
+      "step": 118704
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.404899296172173e-05,
+      "loss": 2.2393,
+      "step": 118712
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4043209484175386e-05,
+      "loss": 2.2253,
+      "step": 118720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.403742640331664e-05,
+      "loss": 2.2342,
+      "step": 118728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.403164371928925e-05,
+      "loss": 2.2261,
+      "step": 118736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.402586143223702e-05,
+      "loss": 2.2224,
+      "step": 118744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.402007954230365e-05,
+      "loss": 2.2447,
+      "step": 118752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4014298049632896e-05,
+      "loss": 2.2467,
+      "step": 118760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4008516954368496e-05,
+      "loss": 2.2425,
+      "step": 118768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4002736256654144e-05,
+      "loss": 2.2268,
+      "step": 118776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3996955956633565e-05,
+      "loss": 2.2385,
+      "step": 118784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3991176054450464e-05,
+      "loss": 2.2352,
+      "step": 118792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.398539655024852e-05,
+      "loss": 2.2399,
+      "step": 118800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3979617444171414e-05,
+      "loss": 2.2274,
+      "step": 118808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.397383873636281e-05,
+      "loss": 2.2419,
+      "step": 118816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3968060426966385e-05,
+      "loss": 2.2312,
+      "step": 118824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3962282516125765e-05,
+      "loss": 2.2269,
+      "step": 118832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3956505003984604e-05,
+      "loss": 2.2443,
+      "step": 118840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3950727890686545e-05,
+      "loss": 2.2357,
+      "step": 118848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.394495117637516e-05,
+      "loss": 2.232,
+      "step": 118856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.393917486119411e-05,
+      "loss": 2.2361,
+      "step": 118864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.393339894528697e-05,
+      "loss": 2.2439,
+      "step": 118872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.392762342879735e-05,
+      "loss": 2.2398,
+      "step": 118880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.392184831186879e-05,
+      "loss": 2.2352,
+      "step": 118888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3916073594644895e-05,
+      "loss": 2.2429,
+      "step": 118896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.391029927726923e-05,
+      "loss": 2.2409,
+      "step": 118904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.390452535988532e-05,
+      "loss": 2.2325,
+      "step": 118912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.38987518426367e-05,
+      "loss": 2.2411,
+      "step": 118920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.389297872566694e-05,
+      "loss": 2.2531,
+      "step": 118928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.388720600911953e-05,
+      "loss": 2.2405,
+      "step": 118936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.388143369313798e-05,
+      "loss": 2.2421,
+      "step": 118944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.387566177786577e-05,
+      "loss": 2.2462,
+      "step": 118952
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.386989026344647e-05,
+      "loss": 2.2607,
+      "step": 118960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.386411915002348e-05,
+      "loss": 2.2363,
+      "step": 118968
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3858348437740296e-05,
+      "loss": 2.2367,
+      "step": 118976
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.385257812674039e-05,
+      "loss": 2.2622,
+      "step": 118984
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3846808217167196e-05,
+      "loss": 2.2296,
+      "step": 118992
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.384103870916416e-05,
+      "loss": 2.2415,
+      "step": 119000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3835269602874716e-05,
+      "loss": 2.2218,
+      "step": 119008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.38295008984423e-05,
+      "loss": 2.2212,
+      "step": 119016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3823732596010294e-05,
+      "loss": 2.2352,
+      "step": 119024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.38179646957221e-05,
+      "loss": 2.2451,
+      "step": 119032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.381219719772114e-05,
+      "loss": 2.2341,
+      "step": 119040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.380643010215076e-05,
+      "loss": 2.2446,
+      "step": 119048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.380066340915434e-05,
+      "loss": 2.2503,
+      "step": 119056
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3794897118875275e-05,
+      "loss": 2.2366,
+      "step": 119064
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.378913123145685e-05,
+      "loss": 2.2427,
+      "step": 119072
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.378336574704245e-05,
+      "loss": 2.2333,
+      "step": 119080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.37776006657754e-05,
+      "loss": 2.2441,
+      "step": 119088
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.377183598779903e-05,
+      "loss": 2.2401,
+      "step": 119096
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.376607171325662e-05,
+      "loss": 2.2468,
+      "step": 119104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3760307842291495e-05,
+      "loss": 2.2295,
+      "step": 119112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3754544375046954e-05,
+      "loss": 2.2089,
+      "step": 119120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3748781311666246e-05,
+      "loss": 2.2498,
+      "step": 119128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.374301865229265e-05,
+      "loss": 2.2333,
+      "step": 119136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.373725639706944e-05,
+      "loss": 2.2448,
+      "step": 119144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3731494546139885e-05,
+      "loss": 2.2472,
+      "step": 119152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3725733099647176e-05,
+      "loss": 2.2299,
+      "step": 119160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.371997205773457e-05,
+      "loss": 2.2277,
+      "step": 119168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.37142114205453e-05,
+      "loss": 2.2376,
+      "step": 119176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.370845118822254e-05,
+      "loss": 2.2361,
+      "step": 119184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3702691360909524e-05,
+      "loss": 2.2381,
+      "step": 119192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.369693193874943e-05,
+      "loss": 2.2589,
+      "step": 119200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3691172921885435e-05,
+      "loss": 2.2403,
+      "step": 119208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.368541431046071e-05,
+      "loss": 2.2474,
+      "step": 119216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.367965610461841e-05,
+      "loss": 2.23,
+      "step": 119224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.36738983045017e-05,
+      "loss": 2.2291,
+      "step": 119232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.366814091025371e-05,
+      "loss": 2.2353,
+      "step": 119240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.366238392201756e-05,
+      "loss": 2.245,
+      "step": 119248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3656627339936384e-05,
+      "loss": 2.2144,
+      "step": 119256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3650871164153284e-05,
+      "loss": 2.2309,
+      "step": 119264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.364511539481136e-05,
+      "loss": 2.242,
+      "step": 119272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3639360032053724e-05,
+      "loss": 2.2557,
+      "step": 119280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3633605076023405e-05,
+      "loss": 2.2385,
+      "step": 119288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3627850526863505e-05,
+      "loss": 2.225,
+      "step": 119296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.362209638471708e-05,
+      "loss": 2.2499,
+      "step": 119304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.361634264972719e-05,
+      "loss": 2.2424,
+      "step": 119312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3610589322036846e-05,
+      "loss": 2.2473,
+      "step": 119320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.360483640178909e-05,
+      "loss": 2.2518,
+      "step": 119328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.359908388912695e-05,
+      "loss": 2.2506,
+      "step": 119336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.359333178419342e-05,
+      "loss": 2.2318,
+      "step": 119344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.35875800871315e-05,
+      "loss": 2.2351,
+      "step": 119352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.358182879808417e-05,
+      "loss": 2.244,
+      "step": 119360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.357607791719445e-05,
+      "loss": 2.2307,
+      "step": 119368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3570327444605255e-05,
+      "loss": 2.2227,
+      "step": 119376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3564577380459556e-05,
+      "loss": 2.2409,
+      "step": 119384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.355882772490032e-05,
+      "loss": 2.2342,
+      "step": 119392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3553078478070465e-05,
+      "loss": 2.2367,
+      "step": 119400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3547329640112924e-05,
+      "loss": 2.2443,
+      "step": 119408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.354158121117062e-05,
+      "loss": 2.2355,
+      "step": 119416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3535833191386445e-05,
+      "loss": 2.2384,
+      "step": 119424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3530085580903304e-05,
+      "loss": 2.2284,
+      "step": 119432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.352433837986409e-05,
+      "loss": 2.2476,
+      "step": 119440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.351859158841167e-05,
+      "loss": 2.2391,
+      "step": 119448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.351284520668891e-05,
+      "loss": 2.2429,
+      "step": 119456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.350709923483867e-05,
+      "loss": 2.2418,
+      "step": 119464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.350135367300381e-05,
+      "loss": 2.24,
+      "step": 119472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.349560852132712e-05,
+      "loss": 2.2211,
+      "step": 119480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3489863779951465e-05,
+      "loss": 2.2465,
+      "step": 119488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.348411944901965e-05,
+      "loss": 2.2397,
+      "step": 119496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.347837552867449e-05,
+      "loss": 2.2429,
+      "step": 119504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.347263201905876e-05,
+      "loss": 2.2405,
+      "step": 119512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.346688892031524e-05,
+      "loss": 2.2344,
+      "step": 119520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.346114623258674e-05,
+      "loss": 2.2427,
+      "step": 119528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3455403956015986e-05,
+      "loss": 2.2406,
+      "step": 119536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.344966209074574e-05,
+      "loss": 2.2486,
+      "step": 119544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.344392063691876e-05,
+      "loss": 2.2427,
+      "step": 119552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.343817959467777e-05,
+      "loss": 2.2552,
+      "step": 119560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3432438964165486e-05,
+      "loss": 2.2353,
+      "step": 119568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.342669874552463e-05,
+      "loss": 2.251,
+      "step": 119576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3420958938897914e-05,
+      "loss": 2.2508,
+      "step": 119584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.341521954442801e-05,
+      "loss": 2.2474,
+      "step": 119592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.340948056225761e-05,
+      "loss": 2.2453,
+      "step": 119600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.340374199252939e-05,
+      "loss": 2.2543,
+      "step": 119608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3398003835385995e-05,
+      "loss": 2.2389,
+      "step": 119616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.339226609097009e-05,
+      "loss": 2.2222,
+      "step": 119624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.338652875942433e-05,
+      "loss": 2.2276,
+      "step": 119632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.338079184089131e-05,
+      "loss": 2.2299,
+      "step": 119640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.337505533551367e-05,
+      "loss": 2.2461,
+      "step": 119648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.336931924343403e-05,
+      "loss": 2.2367,
+      "step": 119656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3363583564794975e-05,
+      "loss": 2.2367,
+      "step": 119664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.33578482997391e-05,
+      "loss": 2.2421,
+      "step": 119672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.335211344840897e-05,
+      "loss": 2.2253,
+      "step": 119680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.33463790109472e-05,
+      "loss": 2.2359,
+      "step": 119688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.334064498749628e-05,
+      "loss": 2.2182,
+      "step": 119696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.333491137819882e-05,
+      "loss": 2.2401,
+      "step": 119704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3329178183197314e-05,
+      "loss": 2.2467,
+      "step": 119712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.332344540263433e-05,
+      "loss": 2.2355,
+      "step": 119720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.331771303665235e-05,
+      "loss": 2.2497,
+      "step": 119728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3311981085393894e-05,
+      "loss": 2.2222,
+      "step": 119736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.330624954900147e-05,
+      "loss": 2.2194,
+      "step": 119744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3300518427617556e-05,
+      "loss": 2.2418,
+      "step": 119752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3294787721384604e-05,
+      "loss": 2.2475,
+      "step": 119760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3289057430445136e-05,
+      "loss": 2.242,
+      "step": 119768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3283327554941564e-05,
+      "loss": 2.2459,
+      "step": 119776
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.327759809501632e-05,
+      "loss": 2.2358,
+      "step": 119784
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.32718690508119e-05,
+      "loss": 2.2397,
+      "step": 119792
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.326614042247068e-05,
+      "loss": 2.2152,
+      "step": 119800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3260412210135074e-05,
+      "loss": 2.2361,
+      "step": 119808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.325468441394751e-05,
+      "loss": 2.2397,
+      "step": 119816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.324895703405037e-05,
+      "loss": 2.2237,
+      "step": 119824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3243230070586035e-05,
+      "loss": 2.2385,
+      "step": 119832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.323750352369688e-05,
+      "loss": 2.2285,
+      "step": 119840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3231777393525265e-05,
+      "loss": 2.2374,
+      "step": 119848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3226051680213564e-05,
+      "loss": 2.2396,
+      "step": 119856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.322032638390408e-05,
+      "loss": 2.2275,
+      "step": 119864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.321460150473916e-05,
+      "loss": 2.2333,
+      "step": 119872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3208877042861164e-05,
+      "loss": 2.2449,
+      "step": 119880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.320315299841233e-05,
+      "loss": 2.2402,
+      "step": 119888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.319742937153502e-05,
+      "loss": 2.2254,
+      "step": 119896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3191706162371505e-05,
+      "loss": 2.2383,
+      "step": 119904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.318598337106403e-05,
+      "loss": 2.2535,
+      "step": 119912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3180260997754926e-05,
+      "loss": 2.2504,
+      "step": 119920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3174539042586396e-05,
+      "loss": 2.2439,
+      "step": 119928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.316881750570073e-05,
+      "loss": 2.2384,
+      "step": 119936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.316309638724015e-05,
+      "loss": 2.2489,
+      "step": 119944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.315737568734687e-05,
+      "loss": 2.2351,
+      "step": 119952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.315165540616313e-05,
+      "loss": 2.2338,
+      "step": 119960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.314593554383113e-05,
+      "loss": 2.2322,
+      "step": 119968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3140216100493054e-05,
+      "loss": 2.2465,
+      "step": 119976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.313449707629111e-05,
+      "loss": 2.2401,
+      "step": 119984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.312877847136745e-05,
+      "loss": 2.2424,
+      "step": 119992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3123060285864246e-05,
+      "loss": 2.2424,
+      "step": 120000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.311734251992367e-05,
+      "loss": 2.2302,
+      "step": 120008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.311162517368785e-05,
+      "loss": 2.2293,
+      "step": 120016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.310590824729891e-05,
+      "loss": 2.237,
+      "step": 120024
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.310019174089901e-05,
+      "loss": 2.2457,
+      "step": 120032
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3094475654630223e-05,
+      "loss": 2.2348,
+      "step": 120040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.308875998863467e-05,
+      "loss": 2.2471,
+      "step": 120048
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.308304474305444e-05,
+      "loss": 2.2331,
+      "step": 120056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.307732991803161e-05,
+      "loss": 2.2436,
+      "step": 120064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.307161551370829e-05,
+      "loss": 2.2329,
+      "step": 120072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3065901530226463e-05,
+      "loss": 2.2487,
+      "step": 120080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.306018796772823e-05,
+      "loss": 2.2293,
+      "step": 120088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.305447482635566e-05,
+      "loss": 2.2336,
+      "step": 120096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.304876210625072e-05,
+      "loss": 2.2543,
+      "step": 120104
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.304304980755546e-05,
+      "loss": 2.2569,
+      "step": 120112
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.303733793041189e-05,
+      "loss": 2.2294,
+      "step": 120120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.303162647496199e-05,
+      "loss": 2.2579,
+      "step": 120128
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3025915441347776e-05,
+      "loss": 2.2319,
+      "step": 120136
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.302020482971119e-05,
+      "loss": 2.2499,
+      "step": 120144
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.301449464019423e-05,
+      "loss": 2.256,
+      "step": 120152
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.300878487293884e-05,
+      "loss": 2.2389,
+      "step": 120160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.300307552808695e-05,
+      "loss": 2.2488,
+      "step": 120168
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2997366605780525e-05,
+      "loss": 2.2589,
+      "step": 120176
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2991658106161474e-05,
+      "loss": 2.2407,
+      "step": 120184
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.298595002937169e-05,
+      "loss": 2.2353,
+      "step": 120192
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.298024237555312e-05,
+      "loss": 2.2327,
+      "step": 120200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.297453514484763e-05,
+      "loss": 2.2398,
+      "step": 120208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.29688283373971e-05,
+      "loss": 2.2373,
+      "step": 120216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2963121953343415e-05,
+      "loss": 2.2394,
+      "step": 120224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.295741599282843e-05,
+      "loss": 2.2436,
+      "step": 120232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.295171045599398e-05,
+      "loss": 2.2442,
+      "step": 120240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2946005342981935e-05,
+      "loss": 2.2276,
+      "step": 120248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2940300653934114e-05,
+      "loss": 2.2337,
+      "step": 120256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2934596388992335e-05,
+      "loss": 2.2668,
+      "step": 120264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2928892548298386e-05,
+      "loss": 2.239,
+      "step": 120272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.292318913199408e-05,
+      "loss": 2.2422,
+      "step": 120280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.291748614022124e-05,
+      "loss": 2.2296,
+      "step": 120288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2911783573121586e-05,
+      "loss": 2.2286,
+      "step": 120296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2906081430836905e-05,
+      "loss": 2.2441,
+      "step": 120304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2900379713509e-05,
+      "loss": 2.2256,
+      "step": 120312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.289467842127953e-05,
+      "loss": 2.2392,
+      "step": 120320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.288897755429029e-05,
+      "loss": 2.2344,
+      "step": 120328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.288327711268298e-05,
+      "loss": 2.2452,
+      "step": 120336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2877577096599325e-05,
+      "loss": 2.2176,
+      "step": 120344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.287187750618103e-05,
+      "loss": 2.2541,
+      "step": 120352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.286617834156977e-05,
+      "loss": 2.2463,
+      "step": 120360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.286047960290725e-05,
+      "loss": 2.2361,
+      "step": 120368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2854781290335135e-05,
+      "loss": 2.238,
+      "step": 120376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.284908340399506e-05,
+      "loss": 2.2314,
+      "step": 120384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.284338594402871e-05,
+      "loss": 2.2317,
+      "step": 120392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.283768891057771e-05,
+      "loss": 2.2364,
+      "step": 120400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.283199230378368e-05,
+      "loss": 2.2312,
+      "step": 120408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2826296123788264e-05,
+      "loss": 2.2352,
+      "step": 120416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2820600370733054e-05,
+      "loss": 2.2366,
+      "step": 120424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.281490504475962e-05,
+      "loss": 2.2354,
+      "step": 120432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2809210146009606e-05,
+      "loss": 2.2376,
+      "step": 120440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.280351567462454e-05,
+      "loss": 2.2498,
+      "step": 120448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2797821630745994e-05,
+      "loss": 2.2458,
+      "step": 120456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.279212801451555e-05,
+      "loss": 2.2287,
+      "step": 120464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.278643482607474e-05,
+      "loss": 2.2295,
+      "step": 120472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2780742065565083e-05,
+      "loss": 2.2275,
+      "step": 120480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2775049733128094e-05,
+      "loss": 2.2439,
+      "step": 120488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2769357828905304e-05,
+      "loss": 2.2272,
+      "step": 120496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.276366635303824e-05,
+      "loss": 2.2526,
+      "step": 120504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2757975305668344e-05,
+      "loss": 2.2385,
+      "step": 120512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.27522846869371e-05,
+      "loss": 2.2371,
+      "step": 120520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2746594496986036e-05,
+      "loss": 2.245,
+      "step": 120528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.274090473595653e-05,
+      "loss": 2.2336,
+      "step": 120536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.273521540399009e-05,
+      "loss": 2.2528,
+      "step": 120544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2729526501228124e-05,
+      "loss": 2.2311,
+      "step": 120552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2723838027812066e-05,
+      "loss": 2.2538,
+      "step": 120560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.271814998388335e-05,
+      "loss": 2.2269,
+      "step": 120568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2712462369583335e-05,
+      "loss": 2.2515,
+      "step": 120576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2706775185053464e-05,
+      "loss": 2.2517,
+      "step": 120584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2701088430435104e-05,
+      "loss": 2.2275,
+      "step": 120592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2695402105869615e-05,
+      "loss": 2.2432,
+      "step": 120600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.268971621149838e-05,
+      "loss": 2.2357,
+      "step": 120608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.268403074746274e-05,
+      "loss": 2.2279,
+      "step": 120616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.267834571390403e-05,
+      "loss": 2.2361,
+      "step": 120624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.26726611109636e-05,
+      "loss": 2.2376,
+      "step": 120632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.266697693878277e-05,
+      "loss": 2.2478,
+      "step": 120640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.266129319750281e-05,
+      "loss": 2.2326,
+      "step": 120648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2655609887265064e-05,
+      "loss": 2.2275,
+      "step": 120656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.26499270082108e-05,
+      "loss": 2.2508,
+      "step": 120664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.264424456048129e-05,
+      "loss": 2.2244,
+      "step": 120672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.263856254421781e-05,
+      "loss": 2.2497,
+      "step": 120680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2632880959561615e-05,
+      "loss": 2.2488,
+      "step": 120688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.262719980665397e-05,
+      "loss": 2.2215,
+      "step": 120696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.262151908563605e-05,
+      "loss": 2.2582,
+      "step": 120704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.261583879664913e-05,
+      "loss": 2.243,
+      "step": 120712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2610158939834425e-05,
+      "loss": 2.248,
+      "step": 120720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2604479515333094e-05,
+      "loss": 2.2442,
+      "step": 120728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2598800523286366e-05,
+      "loss": 2.2544,
+      "step": 120736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.259312196383541e-05,
+      "loss": 2.2344,
+      "step": 120744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.258744383712139e-05,
+      "loss": 2.2337,
+      "step": 120752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2581766143285476e-05,
+      "loss": 2.2331,
+      "step": 120760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2576088882468795e-05,
+      "loss": 2.244,
+      "step": 120768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.257041205481252e-05,
+      "loss": 2.237,
+      "step": 120776
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.256473566045775e-05,
+      "loss": 2.2368,
+      "step": 120784
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.25590596995456e-05,
+      "loss": 2.2397,
+      "step": 120792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2553384172217195e-05,
+      "loss": 2.2382,
+      "step": 120800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2547709078613606e-05,
+      "loss": 2.2513,
+      "step": 120808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.254203441887592e-05,
+      "loss": 2.2378,
+      "step": 120816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.253636019314524e-05,
+      "loss": 2.2435,
+      "step": 120824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.25306864015626e-05,
+      "loss": 2.2299,
+      "step": 120832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.252501304426904e-05,
+      "loss": 2.2358,
+      "step": 120840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2519340121405626e-05,
+      "loss": 2.237,
+      "step": 120848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.251366763311338e-05,
+      "loss": 2.2352,
+      "step": 120856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.250799557953331e-05,
+      "loss": 2.2487,
+      "step": 120864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2502323960806444e-05,
+      "loss": 2.2536,
+      "step": 120872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.249665277707377e-05,
+      "loss": 2.2294,
+      "step": 120880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2490982028476266e-05,
+      "loss": 2.2296,
+      "step": 120888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.24853117151549e-05,
+      "loss": 2.2442,
+      "step": 120896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2479641837250656e-05,
+      "loss": 2.2395,
+      "step": 120904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2473972394904504e-05,
+      "loss": 2.238,
+      "step": 120912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2468303388257334e-05,
+      "loss": 2.2343,
+      "step": 120920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2462634817450114e-05,
+      "loss": 2.2327,
+      "step": 120928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.245696668262379e-05,
+      "loss": 2.2314,
+      "step": 120936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.245129898391921e-05,
+      "loss": 2.2238,
+      "step": 120944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2445631721477325e-05,
+      "loss": 2.2421,
+      "step": 120952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2439964895439e-05,
+      "loss": 2.2497,
+      "step": 120960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.243429850594511e-05,
+      "loss": 2.224,
+      "step": 120968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.242863255313654e-05,
+      "loss": 2.2223,
+      "step": 120976
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.242296703715412e-05,
+      "loss": 2.2326,
+      "step": 120984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.241730195813873e-05,
+      "loss": 2.2326,
+      "step": 120992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.241163731623118e-05,
+      "loss": 2.2274,
+      "step": 121000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2405973111572284e-05,
+      "loss": 2.2239,
+      "step": 121008
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2400309344302893e-05,
+      "loss": 2.2325,
+      "step": 121016
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2394646014563784e-05,
+      "loss": 2.2412,
+      "step": 121024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.238898312249573e-05,
+      "loss": 2.2328,
+      "step": 121032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.238332066823956e-05,
+      "loss": 2.2379,
+      "step": 121040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2377658651936e-05,
+      "loss": 2.2459,
+      "step": 121048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2371997073725814e-05,
+      "loss": 2.2265,
+      "step": 121056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.236633593374977e-05,
+      "loss": 2.2424,
+      "step": 121064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2360675232148604e-05,
+      "loss": 2.2198,
+      "step": 121072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.235501496906301e-05,
+      "loss": 2.2285,
+      "step": 121080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.234935514463374e-05,
+      "loss": 2.238,
+      "step": 121088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2343695759001476e-05,
+      "loss": 2.2291,
+      "step": 121096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.233803681230692e-05,
+      "loss": 2.2323,
+      "step": 121104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.233237830469073e-05,
+      "loss": 2.245,
+      "step": 121112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.232672023629361e-05,
+      "loss": 2.2391,
+      "step": 121120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.232106260725622e-05,
+      "loss": 2.2356,
+      "step": 121128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.231540541771918e-05,
+      "loss": 2.2399,
+      "step": 121136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.230974866782314e-05,
+      "loss": 2.2585,
+      "step": 121144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.230409235770877e-05,
+      "loss": 2.2262,
+      "step": 121152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2298436487516616e-05,
+      "loss": 2.238,
+      "step": 121160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2292781057387327e-05,
+      "loss": 2.2274,
+      "step": 121168
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2287126067461493e-05,
+      "loss": 2.2311,
+      "step": 121176
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.228147151787967e-05,
+      "loss": 2.2387,
+      "step": 121184
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2275817408782484e-05,
+      "loss": 2.2426,
+      "step": 121192
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2270163740310434e-05,
+      "loss": 2.2349,
+      "step": 121200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.226451051260413e-05,
+      "loss": 2.2347,
+      "step": 121208
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.225885772580407e-05,
+      "loss": 2.233,
+      "step": 121216
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.22532053800508e-05,
+      "loss": 2.2367,
+      "step": 121224
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2247553475484835e-05,
+      "loss": 2.2302,
+      "step": 121232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.224190201224669e-05,
+      "loss": 2.241,
+      "step": 121240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2236250990476836e-05,
+      "loss": 2.247,
+      "step": 121248
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.22306004103158e-05,
+      "loss": 2.2117,
+      "step": 121256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.222495027190402e-05,
+      "loss": 2.2494,
+      "step": 121264
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.221930057538196e-05,
+      "loss": 2.2305,
+      "step": 121272
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2213651320890095e-05,
+      "loss": 2.246,
+      "step": 121280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.220800250856885e-05,
+      "loss": 2.2372,
+      "step": 121288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2202354138558654e-05,
+      "loss": 2.2278,
+      "step": 121296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2196706210999944e-05,
+      "loss": 2.2454,
+      "step": 121304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2191058726033113e-05,
+      "loss": 2.2324,
+      "step": 121312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.218541168379856e-05,
+      "loss": 2.2285,
+      "step": 121320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.217976508443666e-05,
+      "loss": 2.2475,
+      "step": 121328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.217411892808779e-05,
+      "loss": 2.2335,
+      "step": 121336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.216847321489236e-05,
+      "loss": 2.2391,
+      "step": 121344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.216282794499065e-05,
+      "loss": 2.2504,
+      "step": 121352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.215718311852305e-05,
+      "loss": 2.2515,
+      "step": 121360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.21515387356299e-05,
+      "loss": 2.2317,
+      "step": 121368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2145894796451476e-05,
+      "loss": 2.2445,
+      "step": 121376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.214025130112812e-05,
+      "loss": 2.2447,
+      "step": 121384
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.213460824980011e-05,
+      "loss": 2.2518,
+      "step": 121392
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.212896564260775e-05,
+      "loss": 2.2458,
+      "step": 121400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2123323479691326e-05,
+      "loss": 2.244,
+      "step": 121408
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.211768176119106e-05,
+      "loss": 2.2442,
+      "step": 121416
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.211204048724724e-05,
+      "loss": 2.2585,
+      "step": 121424
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2106399658000114e-05,
+      "loss": 2.2514,
+      "step": 121432
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2100759273589876e-05,
+      "loss": 2.2506,
+      "step": 121440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.209511933415678e-05,
+      "loss": 2.2344,
+      "step": 121448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.208947983984105e-05,
+      "loss": 2.2302,
+      "step": 121456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2083840790782825e-05,
+      "loss": 2.2492,
+      "step": 121464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.207820218712235e-05,
+      "loss": 2.2405,
+      "step": 121472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2072564028999775e-05,
+      "loss": 2.2437,
+      "step": 121480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.206692631655526e-05,
+      "loss": 2.2183,
+      "step": 121488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.206128904992898e-05,
+      "loss": 2.2277,
+      "step": 121496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.205565222926107e-05,
+      "loss": 2.2195,
+      "step": 121504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.205001585469164e-05,
+      "loss": 2.2614,
+      "step": 121512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.204437992636085e-05,
+      "loss": 2.2192,
+      "step": 121520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.203874444440879e-05,
+      "loss": 2.2332,
+      "step": 121528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2033109408975554e-05,
+      "loss": 2.2423,
+      "step": 121536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.202747482020122e-05,
+      "loss": 2.2504,
+      "step": 121544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2021840678225875e-05,
+      "loss": 2.2396,
+      "step": 121552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.201620698318962e-05,
+      "loss": 2.2316,
+      "step": 121560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.201057373523245e-05,
+      "loss": 2.2436,
+      "step": 121568
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.200494093449445e-05,
+      "loss": 2.236,
+      "step": 121576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.199930858111564e-05,
+      "loss": 2.2343,
+      "step": 121584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1993676675236016e-05,
+      "loss": 2.2296,
+      "step": 121592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.198804521699563e-05,
+      "loss": 2.2278,
+      "step": 121600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.198241420653444e-05,
+      "loss": 2.2328,
+      "step": 121608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1976783643992476e-05,
+      "loss": 2.258,
+      "step": 121616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.197115352950969e-05,
+      "loss": 2.2484,
+      "step": 121624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1965523863226035e-05,
+      "loss": 2.243,
+      "step": 121632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.195989464528149e-05,
+      "loss": 2.227,
+      "step": 121640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1954265875816004e-05,
+      "loss": 2.227,
+      "step": 121648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.194863755496947e-05,
+      "loss": 2.2371,
+      "step": 121656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1943009682881844e-05,
+      "loss": 2.2414,
+      "step": 121664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.193738225969302e-05,
+      "loss": 2.2354,
+      "step": 121672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.193175528554289e-05,
+      "loss": 2.2476,
+      "step": 121680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1926128760571356e-05,
+      "loss": 2.238,
+      "step": 121688
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.192050268491831e-05,
+      "loss": 2.2525,
+      "step": 121696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.191487705872356e-05,
+      "loss": 2.2447,
+      "step": 121704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.190925188212702e-05,
+      "loss": 2.2366,
+      "step": 121712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.190362715526851e-05,
+      "loss": 2.2515,
+      "step": 121720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.189800287828784e-05,
+      "loss": 2.2434,
+      "step": 121728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1892379051324865e-05,
+      "loss": 2.2573,
+      "step": 121736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1886755674519363e-05,
+      "loss": 2.2427,
+      "step": 121744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.188113274801117e-05,
+      "loss": 2.24,
+      "step": 121752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.187551027194003e-05,
+      "loss": 2.2427,
+      "step": 121760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.186988824644574e-05,
+      "loss": 2.2439,
+      "step": 121768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1864266671668096e-05,
+      "loss": 2.2449,
+      "step": 121776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.185864554774677e-05,
+      "loss": 2.2404,
+      "step": 121784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.185302487482158e-05,
+      "loss": 2.2501,
+      "step": 121792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.184740465303223e-05,
+      "loss": 2.2181,
+      "step": 121800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.18417848825184e-05,
+      "loss": 2.2366,
+      "step": 121808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.183616556341987e-05,
+      "loss": 2.2384,
+      "step": 121816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.183054669587629e-05,
+      "loss": 2.2462,
+      "step": 121824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.182492828002736e-05,
+      "loss": 2.2434,
+      "step": 121832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.181931031601276e-05,
+      "loss": 2.2345,
+      "step": 121840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1813692803972124e-05,
+      "loss": 2.234,
+      "step": 121848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.180807574404515e-05,
+      "loss": 2.226,
+      "step": 121856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1802459136371443e-05,
+      "loss": 2.2346,
+      "step": 121864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.179684298109064e-05,
+      "loss": 2.2369,
+      "step": 121872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1791227278342366e-05,
+      "loss": 2.2328,
+      "step": 121880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.178561202826623e-05,
+      "loss": 2.245,
+      "step": 121888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.177999723100181e-05,
+      "loss": 2.2469,
+      "step": 121896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1774382886688714e-05,
+      "loss": 2.2212,
+      "step": 121904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1768768995466504e-05,
+      "loss": 2.235,
+      "step": 121912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.176315555747473e-05,
+      "loss": 2.2282,
+      "step": 121920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.175754257285297e-05,
+      "loss": 2.2592,
+      "step": 121928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.175193004174075e-05,
+      "loss": 2.2463,
+      "step": 121936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.174631796427759e-05,
+      "loss": 2.2395,
+      "step": 121944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.174070634060301e-05,
+      "loss": 2.2552,
+      "step": 121952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1735095170856506e-05,
+      "loss": 2.2057,
+      "step": 121960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.172948445517763e-05,
+      "loss": 2.2506,
+      "step": 121968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1723874193705774e-05,
+      "loss": 2.2627,
+      "step": 121976
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.171826438658046e-05,
+      "loss": 2.2533,
+      "step": 121984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.171265503394118e-05,
+      "loss": 2.2254,
+      "step": 121992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.170704613592732e-05,
+      "loss": 2.2503,
+      "step": 122000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.170143769267836e-05,
+      "loss": 2.2346,
+      "step": 122008
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.169582970433371e-05,
+      "loss": 2.2175,
+      "step": 122016
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.169022217103276e-05,
+      "loss": 2.2408,
+      "step": 122024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.168461509291497e-05,
+      "loss": 2.2507,
+      "step": 122032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.167900847011969e-05,
+      "loss": 2.2369,
+      "step": 122040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1673402302786325e-05,
+      "loss": 2.2449,
+      "step": 122048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1667796591054234e-05,
+      "loss": 2.2532,
+      "step": 122056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1662191335062765e-05,
+      "loss": 2.2404,
+      "step": 122064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1656586534951287e-05,
+      "loss": 2.2463,
+      "step": 122072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.165098219085913e-05,
+      "loss": 2.2463,
+      "step": 122080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.16453783029256e-05,
+      "loss": 2.2362,
+      "step": 122088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1639774871290036e-05,
+      "loss": 2.2386,
+      "step": 122096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1634171896091736e-05,
+      "loss": 2.2457,
+      "step": 122104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.162856937746996e-05,
+      "loss": 2.2276,
+      "step": 122112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.162296731556403e-05,
+      "loss": 2.2511,
+      "step": 122120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.161736571051319e-05,
+      "loss": 2.2544,
+      "step": 122128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1611764562456686e-05,
+      "loss": 2.229,
+      "step": 122136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1606163871533796e-05,
+      "loss": 2.2368,
+      "step": 122144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1600563637883736e-05,
+      "loss": 2.2302,
+      "step": 122152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.159496386164573e-05,
+      "loss": 2.2376,
+      "step": 122160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.158936454295897e-05,
+      "loss": 2.2391,
+      "step": 122168
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.158376568196267e-05,
+      "loss": 2.2528,
+      "step": 122176
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.157816727879606e-05,
+      "loss": 2.2337,
+      "step": 122184
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.157256933359824e-05,
+      "loss": 2.239,
+      "step": 122192
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1566971846508415e-05,
+      "loss": 2.2482,
+      "step": 122200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.156137481766578e-05,
+      "loss": 2.246,
+      "step": 122208
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.155577824720939e-05,
+      "loss": 2.2281,
+      "step": 122216
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1550182135278444e-05,
+      "loss": 2.2354,
+      "step": 122224
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1544586482012044e-05,
+      "loss": 2.2471,
+      "step": 122232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.153899128754927e-05,
+      "loss": 2.2429,
+      "step": 122240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.153339655202926e-05,
+      "loss": 2.2327,
+      "step": 122248
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.152780227559108e-05,
+      "loss": 2.2298,
+      "step": 122256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.152220845837381e-05,
+      "loss": 2.2432,
+      "step": 122264
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.151661510051652e-05,
+      "loss": 2.2305,
+      "step": 122272
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.151102220215823e-05,
+      "loss": 2.2426,
+      "step": 122280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.150542976343802e-05,
+      "loss": 2.2307,
+      "step": 122288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.149983778449491e-05,
+      "loss": 2.2449,
+      "step": 122296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.149424626546788e-05,
+      "loss": 2.2442,
+      "step": 122304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1488655206495985e-05,
+      "loss": 2.2303,
+      "step": 122312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1483064607718205e-05,
+      "loss": 2.2339,
+      "step": 122320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.147747446927349e-05,
+      "loss": 2.2283,
+      "step": 122328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1471884791300864e-05,
+      "loss": 2.236,
+      "step": 122336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.146629557393925e-05,
+      "loss": 2.2575,
+      "step": 122344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1460706817327594e-05,
+      "loss": 2.2457,
+      "step": 122352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1455118521604866e-05,
+      "loss": 2.225,
+      "step": 122360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.144953068690997e-05,
+      "loss": 2.2405,
+      "step": 122368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1443943313381823e-05,
+      "loss": 2.2378,
+      "step": 122376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1438356401159316e-05,
+      "loss": 2.2292,
+      "step": 122384
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1432769950381344e-05,
+      "loss": 2.2218,
+      "step": 122392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1427183961186834e-05,
+      "loss": 2.2282,
+      "step": 122400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.142159843371458e-05,
+      "loss": 2.2281,
+      "step": 122408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1416013368103483e-05,
+      "loss": 2.2527,
+      "step": 122416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.141042876449239e-05,
+      "loss": 2.2338,
+      "step": 122424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1404844623020096e-05,
+      "loss": 2.2343,
+      "step": 122432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.139926094382547e-05,
+      "loss": 2.2111,
+      "step": 122440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1393677727047295e-05,
+      "loss": 2.2446,
+      "step": 122448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.138809497282439e-05,
+      "loss": 2.2448,
+      "step": 122456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1382512681295534e-05,
+      "loss": 2.2264,
+      "step": 122464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1376930852599474e-05,
+      "loss": 2.2328,
+      "step": 122472
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1371349486875035e-05,
+      "loss": 2.2516,
+      "step": 122480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1365768584260934e-05,
+      "loss": 2.241,
+      "step": 122488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.136018814489589e-05,
+      "loss": 2.2281,
+      "step": 122496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1354608168918686e-05,
+      "loss": 2.2124,
+      "step": 122504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1349028656468013e-05,
+      "loss": 2.2379,
+      "step": 122512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.134344960768257e-05,
+      "loss": 2.2491,
+      "step": 122520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.133787102270107e-05,
+      "loss": 2.2368,
+      "step": 122528
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.133229290166219e-05,
+      "loss": 2.2341,
+      "step": 122536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.132671524470459e-05,
+      "loss": 2.2224,
+      "step": 122544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.132113805196696e-05,
+      "loss": 2.2247,
+      "step": 122552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.131556132358794e-05,
+      "loss": 2.2351,
+      "step": 122560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1309985059706134e-05,
+      "loss": 2.2219,
+      "step": 122568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.130440926046022e-05,
+      "loss": 2.2503,
+      "step": 122576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.129883392598879e-05,
+      "loss": 2.22,
+      "step": 122584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.129325905643045e-05,
+      "loss": 2.2228,
+      "step": 122592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1287684651923764e-05,
+      "loss": 2.2316,
+      "step": 122600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1282110712607344e-05,
+      "loss": 2.2315,
+      "step": 122608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.127653723861977e-05,
+      "loss": 2.229,
+      "step": 122616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.127096423009956e-05,
+      "loss": 2.2566,
+      "step": 122624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1265391687185296e-05,
+      "loss": 2.2266,
+      "step": 122632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1259819610015496e-05,
+      "loss": 2.2559,
+      "step": 122640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1254247998728665e-05,
+      "loss": 2.2341,
+      "step": 122648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1248676853463355e-05,
+      "loss": 2.2454,
+      "step": 122656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.124310617435802e-05,
+      "loss": 2.2303,
+      "step": 122664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1237535961551194e-05,
+      "loss": 2.2297,
+      "step": 122672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1231966215181315e-05,
+      "loss": 2.2327,
+      "step": 122680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.122639693538686e-05,
+      "loss": 2.2267,
+      "step": 122688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1220828122306286e-05,
+      "loss": 2.2388,
+      "step": 122696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.121525977607804e-05,
+      "loss": 2.2367,
+      "step": 122704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1209691896840526e-05,
+      "loss": 2.2268,
+      "step": 122712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1204124484732195e-05,
+      "loss": 2.2479,
+      "step": 122720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.119855753989144e-05,
+      "loss": 2.2347,
+      "step": 122728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1192991062456644e-05,
+      "loss": 2.233,
+      "step": 122736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.118742505256621e-05,
+      "loss": 2.2379,
+      "step": 122744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1181859510358493e-05,
+      "loss": 2.236,
+      "step": 122752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1176294435971854e-05,
+      "loss": 2.2169,
+      "step": 122760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1170729829544666e-05,
+      "loss": 2.2362,
+      "step": 122768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1165165691215234e-05,
+      "loss": 2.2289,
+      "step": 122776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.11596020211219e-05,
+      "loss": 2.2343,
+      "step": 122784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.115403881940296e-05,
+      "loss": 2.2336,
+      "step": 122792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.114847608619674e-05,
+      "loss": 2.2465,
+      "step": 122800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.114291382164153e-05,
+      "loss": 2.2352,
+      "step": 122808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.113735202587558e-05,
+      "loss": 2.2277,
+      "step": 122816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.113179069903717e-05,
+      "loss": 2.2431,
+      "step": 122824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.11262298412646e-05,
+      "loss": 2.2383,
+      "step": 122832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1120669452696036e-05,
+      "loss": 2.2315,
+      "step": 122840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.111510953346976e-05,
+      "loss": 2.228,
+      "step": 122848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.110955008372399e-05,
+      "loss": 2.2337,
+      "step": 122856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.11039911035969e-05,
+      "loss": 2.2431,
+      "step": 122864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.109843259322673e-05,
+      "loss": 2.2247,
+      "step": 122872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1092874552751634e-05,
+      "loss": 2.2262,
+      "step": 122880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.108731698230981e-05,
+      "loss": 2.2399,
+      "step": 122888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.108175988203941e-05,
+      "loss": 2.2306,
+      "step": 122896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1076203252078557e-05,
+      "loss": 2.2304,
+      "step": 122904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.107064709256544e-05,
+      "loss": 2.2369,
+      "step": 122912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.106509140363815e-05,
+      "loss": 2.257,
+      "step": 122920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.105953618543481e-05,
+      "loss": 2.2479,
+      "step": 122928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1053981438093536e-05,
+      "loss": 2.2414,
+      "step": 122936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1048427161752405e-05,
+      "loss": 2.2342,
+      "step": 122944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1042873356549486e-05,
+      "loss": 2.2446,
+      "step": 122952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1037320022622884e-05,
+      "loss": 2.2489,
+      "step": 122960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.103176716011063e-05,
+      "loss": 2.2391,
+      "step": 122968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.102621476915076e-05,
+      "loss": 2.2279,
+      "step": 122976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.102066284988133e-05,
+      "loss": 2.2431,
+      "step": 122984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1015111402440346e-05,
+      "loss": 2.2276,
+      "step": 122992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1009560426965834e-05,
+      "loss": 2.2324,
+      "step": 123000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.100400992359576e-05,
+      "loss": 2.236,
+      "step": 123008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.099845989246813e-05,
+      "loss": 2.2307,
+      "step": 123016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.099291033372096e-05,
+      "loss": 2.2542,
+      "step": 123024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.098736124749213e-05,
+      "loss": 2.2397,
+      "step": 123032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.098181263391963e-05,
+      "loss": 2.2219,
+      "step": 123040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0976264493141434e-05,
+      "loss": 2.251,
+      "step": 123048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.097071682529541e-05,
+      "loss": 2.2532,
+      "step": 123056
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0965169630519516e-05,
+      "loss": 2.2298,
+      "step": 123064
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0959622908951645e-05,
+      "loss": 2.2551,
+      "step": 123072
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.095407666072966e-05,
+      "loss": 2.231,
+      "step": 123080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.094853088599149e-05,
+      "loss": 2.2367,
+      "step": 123088
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.094298558487496e-05,
+      "loss": 2.2474,
+      "step": 123096
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0937440757517965e-05,
+      "loss": 2.2339,
+      "step": 123104
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.093189640405832e-05,
+      "loss": 2.2384,
+      "step": 123112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.092635252463387e-05,
+      "loss": 2.2425,
+      "step": 123120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0920809119382444e-05,
+      "loss": 2.2447,
+      "step": 123128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.091526618844185e-05,
+      "loss": 2.2433,
+      "step": 123136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0909723731949856e-05,
+      "loss": 2.241,
+      "step": 123144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.090418175004429e-05,
+      "loss": 2.2334,
+      "step": 123152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.089864024286291e-05,
+      "loss": 2.2409,
+      "step": 123160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.089309921054346e-05,
+      "loss": 2.2269,
+      "step": 123168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.088755865322372e-05,
+      "loss": 2.2285,
+      "step": 123176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0882018571041426e-05,
+      "loss": 2.2407,
+      "step": 123184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.087647896413427e-05,
+      "loss": 2.2331,
+      "step": 123192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.087093983264001e-05,
+      "loss": 2.2477,
+      "step": 123200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.086540117669633e-05,
+      "loss": 2.2523,
+      "step": 123208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0859862996440934e-05,
+      "loss": 2.2208,
+      "step": 123216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0854325292011466e-05,
+      "loss": 2.2256,
+      "step": 123224
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.084878806354562e-05,
+      "loss": 2.2279,
+      "step": 123232
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.084325131118109e-05,
+      "loss": 2.2478,
+      "step": 123240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.083771503505544e-05,
+      "loss": 2.2519,
+      "step": 123248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.083217923530635e-05,
+      "loss": 2.2269,
+      "step": 123256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.082664391207147e-05,
+      "loss": 2.2449,
+      "step": 123264
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.082110906548833e-05,
+      "loss": 2.2325,
+      "step": 123272
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.08155746956946e-05,
+      "loss": 2.2393,
+      "step": 123280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.081004080282783e-05,
+      "loss": 2.2345,
+      "step": 123288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.080450738702558e-05,
+      "loss": 2.2386,
+      "step": 123296
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0798974448425445e-05,
+      "loss": 2.2444,
+      "step": 123304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.079344198716495e-05,
+      "loss": 2.2401,
+      "step": 123312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.078791000338165e-05,
+      "loss": 2.2385,
+      "step": 123320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.078237849721307e-05,
+      "loss": 2.2349,
+      "step": 123328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.07768474687967e-05,
+      "loss": 2.245,
+      "step": 123336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0771316918270076e-05,
+      "loss": 2.2479,
+      "step": 123344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.076578684577066e-05,
+      "loss": 2.2291,
+      "step": 123352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.076025725143594e-05,
+      "loss": 2.2398,
+      "step": 123360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0754728135403395e-05,
+      "loss": 2.2468,
+      "step": 123368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.074919949781046e-05,
+      "loss": 2.2317,
+      "step": 123376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.074367133879457e-05,
+      "loss": 2.2227,
+      "step": 123384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.07381436584932e-05,
+      "loss": 2.2427,
+      "step": 123392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.073261645704373e-05,
+      "loss": 2.2499,
+      "step": 123400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0727089734583565e-05,
+      "loss": 2.2374,
+      "step": 123408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0721563491250126e-05,
+      "loss": 2.2336,
+      "step": 123416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0716037727180776e-05,
+      "loss": 2.2402,
+      "step": 123424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.071051244251289e-05,
+      "loss": 2.2616,
+      "step": 123432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0704987637383824e-05,
+      "loss": 2.238,
+      "step": 123440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.069946331193093e-05,
+      "loss": 2.2394,
+      "step": 123448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.069393946629156e-05,
+      "loss": 2.2257,
+      "step": 123456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0688416100603e-05,
+      "loss": 2.2353,
+      "step": 123464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.068289321500261e-05,
+      "loss": 2.2324,
+      "step": 123472
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.067737080962764e-05,
+      "loss": 2.2334,
+      "step": 123480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0671848884615396e-05,
+      "loss": 2.22,
+      "step": 123488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.066632744010317e-05,
+      "loss": 2.2394,
+      "step": 123496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.06608064762282e-05,
+      "loss": 2.2184,
+      "step": 123504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.065528599312777e-05,
+      "loss": 2.2401,
+      "step": 123512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.064976599093909e-05,
+      "loss": 2.2171,
+      "step": 123520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.064424646979939e-05,
+      "loss": 2.2376,
+      "step": 123528
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0638727429845905e-05,
+      "loss": 2.2176,
+      "step": 123536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.063320887121584e-05,
+      "loss": 2.229,
+      "step": 123544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0627690794046344e-05,
+      "loss": 2.2297,
+      "step": 123552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.062217319847466e-05,
+      "loss": 2.2245,
+      "step": 123560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.061665608463792e-05,
+      "loss": 2.2233,
+      "step": 123568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.061113945267325e-05,
+      "loss": 2.2394,
+      "step": 123576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0605623302717865e-05,
+      "loss": 2.2398,
+      "step": 123584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.060010763490885e-05,
+      "loss": 2.2284,
+      "step": 123592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0594592449383326e-05,
+      "loss": 2.2365,
+      "step": 123600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.058907774627842e-05,
+      "loss": 2.2306,
+      "step": 123608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0583563525731234e-05,
+      "loss": 2.2292,
+      "step": 123616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.057804978787883e-05,
+      "loss": 2.2133,
+      "step": 123624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0572536532858266e-05,
+      "loss": 2.2447,
+      "step": 123632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.056702376080664e-05,
+      "loss": 2.2326,
+      "step": 123640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.056151147186099e-05,
+      "loss": 2.2498,
+      "step": 123648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.055599966615833e-05,
+      "loss": 2.2168,
+      "step": 123656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.055048834383569e-05,
+      "loss": 2.2289,
+      "step": 123664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0544977505030136e-05,
+      "loss": 2.2322,
+      "step": 123672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.053946714987859e-05,
+      "loss": 2.2432,
+      "step": 123680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.053395727851808e-05,
+      "loss": 2.2314,
+      "step": 123688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.052844789108558e-05,
+      "loss": 2.235,
+      "step": 123696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.052293898771804e-05,
+      "loss": 2.2287,
+      "step": 123704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0517430568552436e-05,
+      "loss": 2.2323,
+      "step": 123712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.051192263372567e-05,
+      "loss": 2.2382,
+      "step": 123720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.050641518337472e-05,
+      "loss": 2.2235,
+      "step": 123728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.050090821763647e-05,
+      "loss": 2.2241,
+      "step": 123736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.049540173664781e-05,
+      "loss": 2.2303,
+      "step": 123744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.048989574054567e-05,
+      "loss": 2.2282,
+      "step": 123752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0484390229466925e-05,
+      "loss": 2.2314,
+      "step": 123760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.04788852035484e-05,
+      "loss": 2.2324,
+      "step": 123768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0473380662927e-05,
+      "loss": 2.2292,
+      "step": 123776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.046787660773954e-05,
+      "loss": 2.2287,
+      "step": 123784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0462373038122846e-05,
+      "loss": 2.2247,
+      "step": 123792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.045686995421377e-05,
+      "loss": 2.2315,
+      "step": 123800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.04513673561491e-05,
+      "loss": 2.2152,
+      "step": 123808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.04458652440656e-05,
+      "loss": 2.22,
+      "step": 123816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.044036361810011e-05,
+      "loss": 2.2171,
+      "step": 123824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.043486247838938e-05,
+      "loss": 2.2292,
+      "step": 123832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.042936182507016e-05,
+      "loss": 2.2336,
+      "step": 123840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.042386165827918e-05,
+      "loss": 2.2426,
+      "step": 123848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0418361978153206e-05,
+      "loss": 2.2272,
+      "step": 123856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.041286278482898e-05,
+      "loss": 2.2471,
+      "step": 123864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.040736407844314e-05,
+      "loss": 2.2336,
+      "step": 123872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0401865859132435e-05,
+      "loss": 2.2225,
+      "step": 123880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.039636812703358e-05,
+      "loss": 2.2559,
+      "step": 123888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0390870882283177e-05,
+      "loss": 2.2306,
+      "step": 123896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.038537412501794e-05,
+      "loss": 2.2272,
+      "step": 123904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.037987785537451e-05,
+      "loss": 2.2466,
+      "step": 123912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0374382073489495e-05,
+      "loss": 2.2509,
+      "step": 123920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0368886779499546e-05,
+      "loss": 2.2284,
+      "step": 123928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0363391973541285e-05,
+      "loss": 2.2243,
+      "step": 123936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.03578976557513e-05,
+      "loss": 2.233,
+      "step": 123944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.035240382626619e-05,
+      "loss": 2.2183,
+      "step": 123952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.03469104852225e-05,
+      "loss": 2.2453,
+      "step": 123960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0341417632756854e-05,
+      "loss": 2.2506,
+      "step": 123968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0335925269005754e-05,
+      "loss": 2.2343,
+      "step": 123976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.033043339410575e-05,
+      "loss": 2.2323,
+      "step": 123984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.032494200819341e-05,
+      "loss": 2.2395,
+      "step": 123992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.031945111140518e-05,
+      "loss": 2.2427,
+      "step": 124000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0313960703877614e-05,
+      "loss": 2.2469,
+      "step": 124008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0308470785747207e-05,
+      "loss": 2.2619,
+      "step": 124016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0302981357150424e-05,
+      "loss": 2.2324,
+      "step": 124024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.029749241822372e-05,
+      "loss": 2.2361,
+      "step": 124032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0292003969103576e-05,
+      "loss": 2.2181,
+      "step": 124040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.028651600992643e-05,
+      "loss": 2.239,
+      "step": 124048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0281028540828707e-05,
+      "loss": 2.2327,
+      "step": 124056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0275541561946814e-05,
+      "loss": 2.2314,
+      "step": 124064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0270055073417166e-05,
+      "loss": 2.2437,
+      "step": 124072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.026456907537619e-05,
+      "loss": 2.2573,
+      "step": 124080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.025908356796022e-05,
+      "loss": 2.2264,
+      "step": 124088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0253598551305636e-05,
+      "loss": 2.2418,
+      "step": 124096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.024811402554884e-05,
+      "loss": 2.2432,
+      "step": 124104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0242629990826134e-05,
+      "loss": 2.2432,
+      "step": 124112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.023714644727386e-05,
+      "loss": 2.2255,
+      "step": 124120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0231663395028345e-05,
+      "loss": 2.2374,
+      "step": 124128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.022618083422588e-05,
+      "loss": 2.2445,
+      "step": 124136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.02206987650028e-05,
+      "loss": 2.2254,
+      "step": 124144
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0215217187495345e-05,
+      "loss": 2.2248,
+      "step": 124152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.020973610183983e-05,
+      "loss": 2.2294,
+      "step": 124160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.02042555081725e-05,
+      "loss": 2.2495,
+      "step": 124168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0198775406629575e-05,
+      "loss": 2.2379,
+      "step": 124176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0193295797347336e-05,
+      "loss": 2.2443,
+      "step": 124184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.018781668046198e-05,
+      "loss": 2.2094,
+      "step": 124192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.018233805610971e-05,
+      "loss": 2.2578,
+      "step": 124200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.017685992442676e-05,
+      "loss": 2.225,
+      "step": 124208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.017138228554929e-05,
+      "loss": 2.2411,
+      "step": 124216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0165905139613474e-05,
+      "loss": 2.2219,
+      "step": 124224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.01604284867555e-05,
+      "loss": 2.2235,
+      "step": 124232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0154952327111486e-05,
+      "loss": 2.2157,
+      "step": 124240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.014947666081758e-05,
+      "loss": 2.2432,
+      "step": 124248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.014400148800992e-05,
+      "loss": 2.234,
+      "step": 124256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.013852680882461e-05,
+      "loss": 2.2436,
+      "step": 124264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.013305262339775e-05,
+      "loss": 2.2422,
+      "step": 124272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.012757893186542e-05,
+      "loss": 2.2462,
+      "step": 124280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.012210573436371e-05,
+      "loss": 2.2423,
+      "step": 124288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0116633031028706e-05,
+      "loss": 2.2267,
+      "step": 124296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0111160821996416e-05,
+      "loss": 2.2296,
+      "step": 124304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.01056891074029e-05,
+      "loss": 2.2221,
+      "step": 124312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.010021788738419e-05,
+      "loss": 2.2492,
+      "step": 124320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.009474716207629e-05,
+      "loss": 2.2333,
+      "step": 124328
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0089276931615215e-05,
+      "loss": 2.234,
+      "step": 124336
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0083807196136953e-05,
+      "loss": 2.2392,
+      "step": 124344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.007833795577746e-05,
+      "loss": 2.2155,
+      "step": 124352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0072869210672736e-05,
+      "loss": 2.2347,
+      "step": 124360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.006740096095871e-05,
+      "loss": 2.2311,
+      "step": 124368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.006193320677134e-05,
+      "loss": 2.2335,
+      "step": 124376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.005646594824655e-05,
+      "loss": 2.2512,
+      "step": 124384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.005099918552024e-05,
+      "loss": 2.2398,
+      "step": 124392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0045532918728346e-05,
+      "loss": 2.2206,
+      "step": 124400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0040067148006735e-05,
+      "loss": 2.2333,
+      "step": 124408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.003460187349128e-05,
+      "loss": 2.2588,
+      "step": 124416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0029137095317885e-05,
+      "loss": 2.2463,
+      "step": 124424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.002367281362238e-05,
+      "loss": 2.2298,
+      "step": 124432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.00182090285406e-05,
+      "loss": 2.2275,
+      "step": 124440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.001274574020839e-05,
+      "loss": 2.2241,
+      "step": 124448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0007282948761573e-05,
+      "loss": 2.2312,
+      "step": 124456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.000182065433592e-05,
+      "loss": 2.245,
+      "step": 124464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.999635885706727e-05,
+      "loss": 2.2327,
+      "step": 124472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.999089755709138e-05,
+      "loss": 2.2398,
+      "step": 124480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.998543675454403e-05,
+      "loss": 2.2454,
+      "step": 124488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.997997644956095e-05,
+      "loss": 2.2379,
+      "step": 124496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.997451664227789e-05,
+      "loss": 2.2529,
+      "step": 124504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.996905733283063e-05,
+      "loss": 2.2184,
+      "step": 124512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.996359852135483e-05,
+      "loss": 2.2231,
+      "step": 124520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9958140207986225e-05,
+      "loss": 2.2361,
+      "step": 124528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.995268239286051e-05,
+      "loss": 2.2423,
+      "step": 124536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.994722507611334e-05,
+      "loss": 2.236,
+      "step": 124544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9941768257880424e-05,
+      "loss": 2.2325,
+      "step": 124552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9936311938297375e-05,
+      "loss": 2.2346,
+      "step": 124560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.993085611749989e-05,
+      "loss": 2.2342,
+      "step": 124568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9925400795623566e-05,
+      "loss": 2.2272,
+      "step": 124576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.991994597280401e-05,
+      "loss": 2.2387,
+      "step": 124584
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.991449164917688e-05,
+      "loss": 2.2394,
+      "step": 124592
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9909037824877725e-05,
+      "loss": 2.2647,
+      "step": 124600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.990358450004214e-05,
+      "loss": 2.234,
+      "step": 124608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.989813167480571e-05,
+      "loss": 2.2404,
+      "step": 124616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.989267934930399e-05,
+      "loss": 2.2244,
+      "step": 124624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.988722752367249e-05,
+      "loss": 2.245,
+      "step": 124632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.988177619804679e-05,
+      "loss": 2.2337,
+      "step": 124640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.987632537256239e-05,
+      "loss": 2.2351,
+      "step": 124648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.987087504735478e-05,
+      "loss": 2.2364,
+      "step": 124656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.98654252225595e-05,
+      "loss": 2.2354,
+      "step": 124664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9859975898312e-05,
+      "loss": 2.2265,
+      "step": 124672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.985452707474777e-05,
+      "loss": 2.2458,
+      "step": 124680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.984907875200223e-05,
+      "loss": 2.2368,
+      "step": 124688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.984363093021086e-05,
+      "loss": 2.2321,
+      "step": 124696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.983818360950911e-05,
+      "loss": 2.2471,
+      "step": 124704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.983273679003235e-05,
+      "loss": 2.2337,
+      "step": 124712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.982729047191602e-05,
+      "loss": 2.2402,
+      "step": 124720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.982184465529554e-05,
+      "loss": 2.2198,
+      "step": 124728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.981639934030623e-05,
+      "loss": 2.2275,
+      "step": 124736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.981095452708352e-05,
+      "loss": 2.2354,
+      "step": 124744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.980551021576275e-05,
+      "loss": 2.214,
+      "step": 124752
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.980006640647924e-05,
+      "loss": 2.2215,
+      "step": 124760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.979462309936836e-05,
+      "loss": 2.2389,
+      "step": 124768
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.97891802945654e-05,
+      "loss": 2.2317,
+      "step": 124776
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.97837379922057e-05,
+      "loss": 2.245,
+      "step": 124784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.977829619242455e-05,
+      "loss": 2.2277,
+      "step": 124792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.97728548953572e-05,
+      "loss": 2.2316,
+      "step": 124800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9767414101138966e-05,
+      "loss": 2.2406,
+      "step": 124808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.976197380990509e-05,
+      "loss": 2.2166,
+      "step": 124816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.97565340217908e-05,
+      "loss": 2.24,
+      "step": 124824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975109473693137e-05,
+      "loss": 2.2296,
+      "step": 124832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.974565595546197e-05,
+      "loss": 2.2308,
+      "step": 124840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.974021767751784e-05,
+      "loss": 2.2496,
+      "step": 124848
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.973477990323418e-05,
+      "loss": 2.2357,
+      "step": 124856
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.972934263274617e-05,
+      "loss": 2.2213,
+      "step": 124864
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.972390586618896e-05,
+      "loss": 2.2415,
+      "step": 124872
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.971846960369774e-05,
+      "loss": 2.2514,
+      "step": 124880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.971303384540764e-05,
+      "loss": 2.2369,
+      "step": 124888
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.97075985914538e-05,
+      "loss": 2.2278,
+      "step": 124896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9702163841971315e-05,
+      "loss": 2.2198,
+      "step": 124904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.969672959709532e-05,
+      "loss": 2.226,
+      "step": 124912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.969129585696092e-05,
+      "loss": 2.2515,
+      "step": 124920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.968586262170317e-05,
+      "loss": 2.2411,
+      "step": 124928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.968042989145715e-05,
+      "loss": 2.2448,
+      "step": 124936
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9674997666357954e-05,
+      "loss": 2.239,
+      "step": 124944
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.966956594654056e-05,
+      "loss": 2.2257,
+      "step": 124952
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.966413473214006e-05,
+      "loss": 2.2414,
+      "step": 124960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.965870402329145e-05,
+      "loss": 2.2421,
+      "step": 124968
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9653273820129725e-05,
+      "loss": 2.23,
+      "step": 124976
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.964784412278991e-05,
+      "loss": 2.2511,
+      "step": 124984
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.964241493140696e-05,
+      "loss": 2.2355,
+      "step": 124992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.963698624611587e-05,
+      "loss": 2.237,
+      "step": 125000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.963155806705159e-05,
+      "loss": 2.221,
+      "step": 125008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.962613039434905e-05,
+      "loss": 2.2399,
+      "step": 125016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.96207032281432e-05,
+      "loss": 2.2356,
+      "step": 125024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.961527656856896e-05,
+      "loss": 2.2237,
+      "step": 125032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.960985041576122e-05,
+      "loss": 2.2388,
+      "step": 125040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.960442476985489e-05,
+      "loss": 2.2356,
+      "step": 125048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.959899963098486e-05,
+      "loss": 2.2352,
+      "step": 125056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.959357499928597e-05,
+      "loss": 2.231,
+      "step": 125064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.958815087489311e-05,
+      "loss": 2.2479,
+      "step": 125072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.958272725794111e-05,
+      "loss": 2.2404,
+      "step": 125080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9577304148564784e-05,
+      "loss": 2.2217,
+      "step": 125088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.957188154689898e-05,
+      "loss": 2.2302,
+      "step": 125096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.95664594530785e-05,
+      "loss": 2.2383,
+      "step": 125104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.956103786723813e-05,
+      "loss": 2.2454,
+      "step": 125112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.955561678951263e-05,
+      "loss": 2.2599,
+      "step": 125120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.955019622003679e-05,
+      "loss": 2.2369,
+      "step": 125128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.954477615894541e-05,
+      "loss": 2.2226,
+      "step": 125136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.953935660637315e-05,
+      "loss": 2.2286,
+      "step": 125144
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.953393756245479e-05,
+      "loss": 2.248,
+      "step": 125152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.952851902732505e-05,
+      "loss": 2.2334,
+      "step": 125160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.95231010011186e-05,
+      "loss": 2.2394,
+      "step": 125168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.951768348397018e-05,
+      "loss": 2.2295,
+      "step": 125176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.951226647601444e-05,
+      "loss": 2.2171,
+      "step": 125184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.950684997738604e-05,
+      "loss": 2.2353,
+      "step": 125192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.950143398821966e-05,
+      "loss": 2.2297,
+      "step": 125200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9496018508649924e-05,
+      "loss": 2.2246,
+      "step": 125208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.949060353881148e-05,
+      "loss": 2.2501,
+      "step": 125216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9485189078838936e-05,
+      "loss": 2.2284,
+      "step": 125224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.947977512886687e-05,
+      "loss": 2.234,
+      "step": 125232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.947436168902992e-05,
+      "loss": 2.2552,
+      "step": 125240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.946894875946264e-05,
+      "loss": 2.2373,
+      "step": 125248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.946353634029958e-05,
+      "loss": 2.2288,
+      "step": 125256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9458124431675325e-05,
+      "loss": 2.2355,
+      "step": 125264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9452713033724405e-05,
+      "loss": 2.2276,
+      "step": 125272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9447302146581325e-05,
+      "loss": 2.2446,
+      "step": 125280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.944189177038064e-05,
+      "loss": 2.2434,
+      "step": 125288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.943648190525683e-05,
+      "loss": 2.2406,
+      "step": 125296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.943107255134437e-05,
+      "loss": 2.2352,
+      "step": 125304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9425663708777766e-05,
+      "loss": 2.2321,
+      "step": 125312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.942025537769147e-05,
+      "loss": 2.24,
+      "step": 125320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.941484755821994e-05,
+      "loss": 2.2374,
+      "step": 125328
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.94094402504976e-05,
+      "loss": 2.2294,
+      "step": 125336
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.940403345465888e-05,
+      "loss": 2.2479,
+      "step": 125344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.939862717083822e-05,
+      "loss": 2.2276,
+      "step": 125352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.939322139916998e-05,
+      "loss": 2.2245,
+      "step": 125360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9387816139788585e-05,
+      "loss": 2.2283,
+      "step": 125368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.938241139282839e-05,
+      "loss": 2.2287,
+      "step": 125376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.937700715842374e-05,
+      "loss": 2.2373,
+      "step": 125384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.937160343670903e-05,
+      "loss": 2.2419,
+      "step": 125392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.936620022781855e-05,
+      "loss": 2.2474,
+      "step": 125400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.936079753188667e-05,
+      "loss": 2.2495,
+      "step": 125408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9355395349047674e-05,
+      "loss": 2.23,
+      "step": 125416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.934999367943586e-05,
+      "loss": 2.2438,
+      "step": 125424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.934459252318552e-05,
+      "loss": 2.2477,
+      "step": 125432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9339191880430934e-05,
+      "loss": 2.2418,
+      "step": 125440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.933379175130634e-05,
+      "loss": 2.2414,
+      "step": 125448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.932839213594602e-05,
+      "loss": 2.224,
+      "step": 125456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.932299303448419e-05,
+      "loss": 2.2277,
+      "step": 125464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.931759444705506e-05,
+      "loss": 2.2237,
+      "step": 125472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.931219637379286e-05,
+      "loss": 2.2345,
+      "step": 125480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.930679881483179e-05,
+      "loss": 2.2304,
+      "step": 125488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9301401770306005e-05,
+      "loss": 2.2391,
+      "step": 125496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.929600524034971e-05,
+      "loss": 2.2278,
+      "step": 125504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9290609225097045e-05,
+      "loss": 2.2238,
+      "step": 125512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.928521372468215e-05,
+      "loss": 2.236,
+      "step": 125520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.927981873923916e-05,
+      "loss": 2.2324,
+      "step": 125528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.927442426890222e-05,
+      "loss": 2.2472,
+      "step": 125536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9269030313805414e-05,
+      "loss": 2.2379,
+      "step": 125544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.926363687408282e-05,
+      "loss": 2.2239,
+      "step": 125552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.925824394986855e-05,
+      "loss": 2.2173,
+      "step": 125560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.925285154129668e-05,
+      "loss": 2.2298,
+      "step": 125568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9247459648501226e-05,
+      "loss": 2.2409,
+      "step": 125576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.924206827161627e-05,
+      "loss": 2.236,
+      "step": 125584
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.923667741077582e-05,
+      "loss": 2.2244,
+      "step": 125592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9231287066113886e-05,
+      "loss": 2.228,
+      "step": 125600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.922589723776451e-05,
+      "loss": 2.2264,
+      "step": 125608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.922050792586163e-05,
+      "loss": 2.2416,
+      "step": 125616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.921511913053928e-05,
+      "loss": 2.2185,
+      "step": 125624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9209730851931393e-05,
+      "loss": 2.2296,
+      "step": 125632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.920434309017192e-05,
+      "loss": 2.2238,
+      "step": 125640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9198955845394825e-05,
+      "loss": 2.2285,
+      "step": 125648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.919356911773402e-05,
+      "loss": 2.2399,
+      "step": 125656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.918818290732341e-05,
+      "loss": 2.2309,
+      "step": 125664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.918279721429693e-05,
+      "loss": 2.2265,
+      "step": 125672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.917741203878842e-05,
+      "loss": 2.2405,
+      "step": 125680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.917202738093179e-05,
+      "loss": 2.2237,
+      "step": 125688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.91666432408609e-05,
+      "loss": 2.2308,
+      "step": 125696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.916125961870961e-05,
+      "loss": 2.2317,
+      "step": 125704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.915587651461172e-05,
+      "loss": 2.2457,
+      "step": 125712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.915049392870109e-05,
+      "loss": 2.2458,
+      "step": 125720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9145111861111525e-05,
+      "loss": 2.2227,
+      "step": 125728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.913973031197682e-05,
+      "loss": 2.2119,
+      "step": 125736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9134349281430745e-05,
+      "loss": 2.2379,
+      "step": 125744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9128968769607085e-05,
+      "loss": 2.2311,
+      "step": 125752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.912358877663964e-05,
+      "loss": 2.2424,
+      "step": 125760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9118209302662076e-05,
+      "loss": 2.244,
+      "step": 125768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9112830347808186e-05,
+      "loss": 2.243,
+      "step": 125776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.91074519122117e-05,
+      "loss": 2.2556,
+      "step": 125784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9102073996006275e-05,
+      "loss": 2.2335,
+      "step": 125792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.909669659932565e-05,
+      "loss": 2.2355,
+      "step": 125800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.90913197223035e-05,
+      "loss": 2.2453,
+      "step": 125808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9085943365073466e-05,
+      "loss": 2.2391,
+      "step": 125816
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.908056752776924e-05,
+      "loss": 2.2458,
+      "step": 125824
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9075192210524435e-05,
+      "loss": 2.2372,
+      "step": 125832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.906981741347272e-05,
+      "loss": 2.2318,
+      "step": 125840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.906444313674769e-05,
+      "loss": 2.2335,
+      "step": 125848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.905906938048294e-05,
+      "loss": 2.242,
+      "step": 125856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.905369614481209e-05,
+      "loss": 2.2259,
+      "step": 125864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.90483234298687e-05,
+      "loss": 2.2258,
+      "step": 125872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.904295123578632e-05,
+      "loss": 2.223,
+      "step": 125880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9037579562698566e-05,
+      "loss": 2.2349,
+      "step": 125888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9032208410738904e-05,
+      "loss": 2.2389,
+      "step": 125896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9026837780040886e-05,
+      "loss": 2.2316,
+      "step": 125904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.902146767073806e-05,
+      "loss": 2.2432,
+      "step": 125912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.901609808296389e-05,
+      "loss": 2.225,
+      "step": 125920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9010729016851865e-05,
+      "loss": 2.2439,
+      "step": 125928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.900536047253549e-05,
+      "loss": 2.2339,
+      "step": 125936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.89999924501482e-05,
+      "loss": 2.2281,
+      "step": 125944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8994624949823465e-05,
+      "loss": 2.2472,
+      "step": 125952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8989257971694695e-05,
+      "loss": 2.2378,
+      "step": 125960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.898389151589532e-05,
+      "loss": 2.2272,
+      "step": 125968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8978525582558796e-05,
+      "loss": 2.241,
+      "step": 125976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.897316017181845e-05,
+      "loss": 2.2307,
+      "step": 125984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.896779528380772e-05,
+      "loss": 2.2277,
+      "step": 125992
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8962430918659964e-05,
+      "loss": 2.2322,
+      "step": 126000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.895706707650851e-05,
+      "loss": 2.231,
+      "step": 126008
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8951703757486755e-05,
+      "loss": 2.2405,
+      "step": 126016
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8946340961728e-05,
+      "loss": 2.2336,
+      "step": 126024
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8940978689365555e-05,
+      "loss": 2.2354,
+      "step": 126032
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.893561694053276e-05,
+      "loss": 2.2239,
+      "step": 126040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.893025571536288e-05,
+      "loss": 2.2202,
+      "step": 126048
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8924895013989214e-05,
+      "loss": 2.2274,
+      "step": 126056
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8919534836545025e-05,
+      "loss": 2.2526,
+      "step": 126064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8914175183163556e-05,
+      "loss": 2.2154,
+      "step": 126072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.890881605397806e-05,
+      "loss": 2.2292,
+      "step": 126080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.890345744912178e-05,
+      "loss": 2.2433,
+      "step": 126088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8898099368727886e-05,
+      "loss": 2.2274,
+      "step": 126096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.889274181292963e-05,
+      "loss": 2.2475,
+      "step": 126104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.888738478186018e-05,
+      "loss": 2.2308,
+      "step": 126112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.88820282756527e-05,
+      "loss": 2.2253,
+      "step": 126120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.887667229444038e-05,
+      "loss": 2.2359,
+      "step": 126128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8871316838356354e-05,
+      "loss": 2.2311,
+      "step": 126136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8865961907533756e-05,
+      "loss": 2.234,
+      "step": 126144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.886060750210572e-05,
+      "loss": 2.2234,
+      "step": 126152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8855253622205357e-05,
+      "loss": 2.2238,
+      "step": 126160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.884990026796576e-05,
+      "loss": 2.2378,
+      "step": 126168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.884454743952e-05,
+      "loss": 2.2345,
+      "step": 126176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8839195137001174e-05,
+      "loss": 2.2341,
+      "step": 126184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.883384336054235e-05,
+      "loss": 2.2322,
+      "step": 126192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.882849211027653e-05,
+      "loss": 2.2462,
+      "step": 126200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.882314138633678e-05,
+      "loss": 2.2326,
+      "step": 126208
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.881779118885612e-05,
+      "loss": 2.2379,
+      "step": 126216
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.881244151796753e-05,
+      "loss": 2.2232,
+      "step": 126224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.880709237380404e-05,
+      "loss": 2.2292,
+      "step": 126232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8801743756498606e-05,
+      "loss": 2.212,
+      "step": 126240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.879639566618418e-05,
+      "loss": 2.2425,
+      "step": 126248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.879104810299376e-05,
+      "loss": 2.2359,
+      "step": 126256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.878570106706025e-05,
+      "loss": 2.2323,
+      "step": 126264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8780354558516605e-05,
+      "loss": 2.2241,
+      "step": 126272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.877500857749572e-05,
+      "loss": 2.2338,
+      "step": 126280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.876966312413049e-05,
+      "loss": 2.2371,
+      "step": 126288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.876431819855384e-05,
+      "loss": 2.2261,
+      "step": 126296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.875897380089861e-05,
+      "loss": 2.2337,
+      "step": 126304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.875362993129767e-05,
+      "loss": 2.244,
+      "step": 126312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8748286589883887e-05,
+      "loss": 2.2358,
+      "step": 126320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.874294377679008e-05,
+      "loss": 2.2426,
+      "step": 126328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.873760149214906e-05,
+      "loss": 2.2377,
+      "step": 126336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.873225973609367e-05,
+      "loss": 2.2122,
+      "step": 126344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.872691850875669e-05,
+      "loss": 2.2354,
+      "step": 126352
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.872157781027091e-05,
+      "loss": 2.2112,
+      "step": 126360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.871623764076906e-05,
+      "loss": 2.2259,
+      "step": 126368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.871089800038396e-05,
+      "loss": 2.2302,
+      "step": 126376
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.870555888924832e-05,
+      "loss": 2.2297,
+      "step": 126384
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8700220307494855e-05,
+      "loss": 2.2403,
+      "step": 126392
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.869488225525631e-05,
+      "loss": 2.2329,
+      "step": 126400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.868954473266541e-05,
+      "loss": 2.2269,
+      "step": 126408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.868420773985479e-05,
+      "loss": 2.2222,
+      "step": 126416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8678871276957176e-05,
+      "loss": 2.2324,
+      "step": 126424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.867353534410521e-05,
+      "loss": 2.2318,
+      "step": 126432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8668199941431524e-05,
+      "loss": 2.2356,
+      "step": 126440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8662865069068804e-05,
+      "loss": 2.2277,
+      "step": 126448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.865753072714964e-05,
+      "loss": 2.2269,
+      "step": 126456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8652196915806674e-05,
+      "loss": 2.2391,
+      "step": 126464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.864686363517247e-05,
+      "loss": 2.2311,
+      "step": 126472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.864153088537964e-05,
+      "loss": 2.232,
+      "step": 126480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.863619866656075e-05,
+      "loss": 2.2384,
+      "step": 126488
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.863086697884837e-05,
+      "loss": 2.2309,
+      "step": 126496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8625535822375006e-05,
+      "loss": 2.221,
+      "step": 126504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.862020519727324e-05,
+      "loss": 2.2436,
+      "step": 126512
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8614875103675566e-05,
+      "loss": 2.2327,
+      "step": 126520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8609545541714494e-05,
+      "loss": 2.218,
+      "step": 126528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.860421651152253e-05,
+      "loss": 2.2503,
+      "step": 126536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.859888801323214e-05,
+      "loss": 2.2338,
+      "step": 126544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.859356004697578e-05,
+      "loss": 2.2533,
+      "step": 126552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8588232612885934e-05,
+      "loss": 2.2212,
+      "step": 126560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.858290571109503e-05,
+      "loss": 2.2197,
+      "step": 126568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8577579341735494e-05,
+      "loss": 2.241,
+      "step": 126576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.857225350493972e-05,
+      "loss": 2.2221,
+      "step": 126584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8566928200840146e-05,
+      "loss": 2.2294,
+      "step": 126592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8561603429569136e-05,
+      "loss": 2.2343,
+      "step": 126600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.855627919125906e-05,
+      "loss": 2.2296,
+      "step": 126608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.855095548604229e-05,
+      "loss": 2.2274,
+      "step": 126616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.854563231405121e-05,
+      "loss": 2.2421,
+      "step": 126624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.854030967541808e-05,
+      "loss": 2.2292,
+      "step": 126632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.853498757027527e-05,
+      "loss": 2.2357,
+      "step": 126640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8529665998755084e-05,
+      "loss": 2.2342,
+      "step": 126648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.852434496098979e-05,
+      "loss": 2.2129,
+      "step": 126656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8519024457111705e-05,
+      "loss": 2.2233,
+      "step": 126664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8513704487253075e-05,
+      "loss": 2.261,
+      "step": 126672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.850838505154617e-05,
+      "loss": 2.2364,
+      "step": 126680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.850306615012322e-05,
+      "loss": 2.2336,
+      "step": 126688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8497747783116455e-05,
+      "loss": 2.2512,
+      "step": 126696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.84924299506581e-05,
+      "loss": 2.2496,
+      "step": 126704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.848711265288035e-05,
+      "loss": 2.2296,
+      "step": 126712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.848179588991539e-05,
+      "loss": 2.2399,
+      "step": 126720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.847647966189541e-05,
+      "loss": 2.2392,
+      "step": 126728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.847116396895254e-05,
+      "loss": 2.2525,
+      "step": 126736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.846584881121895e-05,
+      "loss": 2.2693,
+      "step": 126744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.846053418882678e-05,
+      "loss": 2.2498,
+      "step": 126752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.845522010190816e-05,
+      "loss": 2.2354,
+      "step": 126760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8449906550595155e-05,
+      "loss": 2.245,
+      "step": 126768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.844459353501992e-05,
+      "loss": 2.2451,
+      "step": 126776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.84392810553145e-05,
+      "loss": 2.2376,
+      "step": 126784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.843396911161097e-05,
+      "loss": 2.2385,
+      "step": 126792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.842865770404138e-05,
+      "loss": 2.2472,
+      "step": 126800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.842334683273778e-05,
+      "loss": 2.2517,
+      "step": 126808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8418036497832226e-05,
+      "loss": 2.2148,
+      "step": 126816
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.841272669945668e-05,
+      "loss": 2.248,
+      "step": 126824
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8407417437743165e-05,
+      "loss": 2.2347,
+      "step": 126832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8402108712823705e-05,
+      "loss": 2.2309,
+      "step": 126840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.839680052483022e-05,
+      "loss": 2.2442,
+      "step": 126848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8391492873894705e-05,
+      "loss": 2.2365,
+      "step": 126856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8386185760149106e-05,
+      "loss": 2.2582,
+      "step": 126864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.838087918372534e-05,
+      "loss": 2.2394,
+      "step": 126872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.837557314475536e-05,
+      "loss": 2.2298,
+      "step": 126880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.837026764337104e-05,
+      "loss": 2.2474,
+      "step": 126888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.836496267970431e-05,
+      "loss": 2.2383,
+      "step": 126896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.835965825388703e-05,
+      "loss": 2.2328,
+      "step": 126904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.835435436605106e-05,
+      "loss": 2.243,
+      "step": 126912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.834905101632829e-05,
+      "loss": 2.2521,
+      "step": 126920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8343748204850536e-05,
+      "loss": 2.2398,
+      "step": 126928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.833844593174962e-05,
+      "loss": 2.2336,
+      "step": 126936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.83331441971574e-05,
+      "loss": 2.2286,
+      "step": 126944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.832784300120561e-05,
+      "loss": 2.2435,
+      "step": 126952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.832254234402608e-05,
+      "loss": 2.2341,
+      "step": 126960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.831724222575059e-05,
+      "loss": 2.2363,
+      "step": 126968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.83119426465109e-05,
+      "loss": 2.2202,
+      "step": 126976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.830664360643872e-05,
+      "loss": 2.2307,
+      "step": 126984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.830134510566584e-05,
+      "loss": 2.2456,
+      "step": 126992
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.829604714432395e-05,
+      "loss": 2.2415,
+      "step": 127000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.829074972254476e-05,
+      "loss": 2.2326,
+      "step": 127008
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8285452840459965e-05,
+      "loss": 2.2352,
+      "step": 127016
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.828015649820124e-05,
+      "loss": 2.2346,
+      "step": 127024
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8274860695900286e-05,
+      "loss": 2.2462,
+      "step": 127032
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.826956543368871e-05,
+      "loss": 2.2264,
+      "step": 127040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.826427071169819e-05,
+      "loss": 2.2461,
+      "step": 127048
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.825897653006034e-05,
+      "loss": 2.2516,
+      "step": 127056
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.825368288890675e-05,
+      "loss": 2.2294,
+      "step": 127064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8248389788369066e-05,
+      "loss": 2.2288,
+      "step": 127072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.824309722857885e-05,
+      "loss": 2.2407,
+      "step": 127080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.823780520966767e-05,
+      "loss": 2.2304,
+      "step": 127088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.82325137317671e-05,
+      "loss": 2.229,
+      "step": 127096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.822722279500866e-05,
+      "loss": 2.2223,
+      "step": 127104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.822193239952393e-05,
+      "loss": 2.2341,
+      "step": 127112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8216642545444404e-05,
+      "loss": 2.2205,
+      "step": 127120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.821135323290157e-05,
+      "loss": 2.2244,
+      "step": 127128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.820606446202695e-05,
+      "loss": 2.2333,
+      "step": 127136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.820077623295202e-05,
+      "loss": 2.232,
+      "step": 127144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.819548854580822e-05,
+      "loss": 2.2261,
+      "step": 127152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.819020140072704e-05,
+      "loss": 2.2457,
+      "step": 127160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.81849147978399e-05,
+      "loss": 2.2371,
+      "step": 127168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8179628737278205e-05,
+      "loss": 2.2238,
+      "step": 127176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.817434321917341e-05,
+      "loss": 2.2261,
+      "step": 127184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8169058243656886e-05,
+      "loss": 2.2342,
+      "step": 127192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.816377381086003e-05,
+      "loss": 2.2409,
+      "step": 127200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8158489920914184e-05,
+      "loss": 2.2218,
+      "step": 127208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.815320657395075e-05,
+      "loss": 2.2333,
+      "step": 127216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.814792377010105e-05,
+      "loss": 2.2307,
+      "step": 127224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.81426415094964e-05,
+      "loss": 2.2279,
+      "step": 127232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.813735979226813e-05,
+      "loss": 2.2232,
+      "step": 127240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8132078618547585e-05,
+      "loss": 2.226,
+      "step": 127248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.812679798846599e-05,
+      "loss": 2.2286,
+      "step": 127256
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.812151790215466e-05,
+      "loss": 2.2203,
+      "step": 127264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.811623835974486e-05,
+      "loss": 2.235,
+      "step": 127272
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8110959361367806e-05,
+      "loss": 2.2381,
+      "step": 127280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.810568090715477e-05,
+      "loss": 2.2415,
+      "step": 127288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.810040299723697e-05,
+      "loss": 2.237,
+      "step": 127296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8095125631745594e-05,
+      "loss": 2.2161,
+      "step": 127304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.808984881081186e-05,
+      "loss": 2.2326,
+      "step": 127312
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.808457253456694e-05,
+      "loss": 2.2341,
+      "step": 127320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.807929680314201e-05,
+      "loss": 2.2222,
+      "step": 127328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.807402161666823e-05,
+      "loss": 2.2321,
+      "step": 127336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.806874697527671e-05,
+      "loss": 2.2281,
+      "step": 127344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8063472879098615e-05,
+      "loss": 2.2208,
+      "step": 127352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8058199328265055e-05,
+      "loss": 2.2355,
+      "step": 127360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8052926322907093e-05,
+      "loss": 2.2244,
+      "step": 127368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.804765386315588e-05,
+      "loss": 2.2329,
+      "step": 127376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.804238194914244e-05,
+      "loss": 2.2363,
+      "step": 127384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.803711058099783e-05,
+      "loss": 2.2409,
+      "step": 127392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8031839758853135e-05,
+      "loss": 2.2299,
+      "step": 127400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8026569482839376e-05,
+      "loss": 2.2293,
+      "step": 127408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.802129975308755e-05,
+      "loss": 2.244,
+      "step": 127416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.801603056972868e-05,
+      "loss": 2.224,
+      "step": 127424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8010761932893755e-05,
+      "loss": 2.2212,
+      "step": 127432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.800549384271377e-05,
+      "loss": 2.2329,
+      "step": 127440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.800022629931965e-05,
+      "loss": 2.2274,
+      "step": 127448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.799495930284238e-05,
+      "loss": 2.2499,
+      "step": 127456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.798969285341292e-05,
+      "loss": 2.2302,
+      "step": 127464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7984426951162136e-05,
+      "loss": 2.2298,
+      "step": 127472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7979161596220966e-05,
+      "loss": 2.2318,
+      "step": 127480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7973896788720334e-05,
+      "loss": 2.2259,
+      "step": 127488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.796863252879107e-05,
+      "loss": 2.2277,
+      "step": 127496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.796336881656409e-05,
+      "loss": 2.2213,
+      "step": 127504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7958105652170225e-05,
+      "loss": 2.2175,
+      "step": 127512
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7952843035740336e-05,
+      "loss": 2.2357,
+      "step": 127520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.794758096740524e-05,
+      "loss": 2.2236,
+      "step": 127528
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.794231944729574e-05,
+      "loss": 2.2257,
+      "step": 127536
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.793705847554267e-05,
+      "loss": 2.2098,
+      "step": 127544
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.793179805227681e-05,
+      "loss": 2.2384,
+      "step": 127552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.792653817762891e-05,
+      "loss": 2.2355,
+      "step": 127560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.792127885172977e-05,
+      "loss": 2.2277,
+      "step": 127568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.791602007471009e-05,
+      "loss": 2.2162,
+      "step": 127576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.791076184670062e-05,
+      "loss": 2.2111,
+      "step": 127584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.790550416783212e-05,
+      "loss": 2.2283,
+      "step": 127592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7900247038235245e-05,
+      "loss": 2.2291,
+      "step": 127600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7894990458040705e-05,
+      "loss": 2.2403,
+      "step": 127608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.78897344273792e-05,
+      "loss": 2.2342,
+      "step": 127616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.788447894638137e-05,
+      "loss": 2.2338,
+      "step": 127624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.787922401517788e-05,
+      "loss": 2.2103,
+      "step": 127632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.787396963389935e-05,
+      "loss": 2.2402,
+      "step": 127640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.786871580267643e-05,
+      "loss": 2.2407,
+      "step": 127648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.786346252163972e-05,
+      "loss": 2.2278,
+      "step": 127656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7858209790919794e-05,
+      "loss": 2.2182,
+      "step": 127664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.785295761064726e-05,
+      "loss": 2.232,
+      "step": 127672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.784770598095271e-05,
+      "loss": 2.2206,
+      "step": 127680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7842454901966656e-05,
+      "loss": 2.2375,
+      "step": 127688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.783720437381967e-05,
+      "loss": 2.2331,
+      "step": 127696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.783195439664227e-05,
+      "loss": 2.2283,
+      "step": 127704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.782670497056495e-05,
+      "loss": 2.2352,
+      "step": 127712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.782145609571826e-05,
+      "loss": 2.2176,
+      "step": 127720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7816207772232624e-05,
+      "loss": 2.2298,
+      "step": 127728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7810960000238584e-05,
+      "loss": 2.2341,
+      "step": 127736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.780571277986656e-05,
+      "loss": 2.2243,
+      "step": 127744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7800466111246995e-05,
+      "loss": 2.2376,
+      "step": 127752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.779521999451035e-05,
+      "loss": 2.2484,
+      "step": 127760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.778997442978702e-05,
+      "loss": 2.242,
+      "step": 127768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.778472941720741e-05,
+      "loss": 2.2434,
+      "step": 127776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7779484956901946e-05,
+      "loss": 2.2285,
+      "step": 127784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.777424104900095e-05,
+      "loss": 2.2251,
+      "step": 127792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7768997693634815e-05,
+      "loss": 2.2297,
+      "step": 127800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7763754890933906e-05,
+      "loss": 2.2491,
+      "step": 127808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7758512641028545e-05,
+      "loss": 2.239,
+      "step": 127816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.775327094404904e-05,
+      "loss": 2.2311,
+      "step": 127824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.774802980012573e-05,
+      "loss": 2.2215,
+      "step": 127832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.774278920938889e-05,
+      "loss": 2.2379,
+      "step": 127840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.773754917196882e-05,
+      "loss": 2.2383,
+      "step": 127848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.773230968799576e-05,
+      "loss": 2.244,
+      "step": 127856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7727070757599974e-05,
+      "loss": 2.2296,
+      "step": 127864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7721832380911736e-05,
+      "loss": 2.2405,
+      "step": 127872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.771659455806122e-05,
+      "loss": 2.2293,
+      "step": 127880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.771135728917867e-05,
+      "loss": 2.2387,
+      "step": 127888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.770612057439429e-05,
+      "loss": 2.2326,
+      "step": 127896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.770088441383823e-05,
+      "loss": 2.2366,
+      "step": 127904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.76956488076407e-05,
+      "loss": 2.236,
+      "step": 127912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7690413755931846e-05,
+      "loss": 2.2256,
+      "step": 127920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7685179258841795e-05,
+      "loss": 2.2198,
+      "step": 127928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7679945316500706e-05,
+      "loss": 2.2345,
+      "step": 127936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.767471192903866e-05,
+      "loss": 2.2396,
+      "step": 127944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.76694790965858e-05,
+      "loss": 2.2253,
+      "step": 127952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7664246819272205e-05,
+      "loss": 2.2415,
+      "step": 127960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.765901509722791e-05,
+      "loss": 2.2301,
+      "step": 127968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.765378393058303e-05,
+      "loss": 2.2085,
+      "step": 127976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7648553319467585e-05,
+      "loss": 2.2443,
+      "step": 127984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.76433232640116e-05,
+      "loss": 2.2272,
+      "step": 127992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.763809376434513e-05,
+      "loss": 2.2225,
+      "step": 128000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.763286482059814e-05,
+      "loss": 2.2199,
+      "step": 128008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7627636432900634e-05,
+      "loss": 2.2239,
+      "step": 128016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.76224086013826e-05,
+      "loss": 2.218,
+      "step": 128024
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7617181326174015e-05,
+      "loss": 2.2219,
+      "step": 128032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.761195460740479e-05,
+      "loss": 2.2301,
+      "step": 128040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.760672844520491e-05,
+      "loss": 2.2317,
+      "step": 128048
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7601502839704264e-05,
+      "loss": 2.228,
+      "step": 128056
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.759627779103277e-05,
+      "loss": 2.2358,
+      "step": 128064
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.759105329932031e-05,
+      "loss": 2.2249,
+      "step": 128072
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7585829364696786e-05,
+      "loss": 2.2391,
+      "step": 128080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.758060598729208e-05,
+      "loss": 2.2203,
+      "step": 128088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7575383167236e-05,
+      "loss": 2.2337,
+      "step": 128096
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.757016090465843e-05,
+      "loss": 2.2302,
+      "step": 128104
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7564939199689166e-05,
+      "loss": 2.2377,
+      "step": 128112
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.755971805245801e-05,
+      "loss": 2.2442,
+      "step": 128120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.755449746309481e-05,
+      "loss": 2.2353,
+      "step": 128128
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7549277431729326e-05,
+      "loss": 2.2241,
+      "step": 128136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7544057958491296e-05,
+      "loss": 2.2393,
+      "step": 128144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.753883904351052e-05,
+      "loss": 2.2403,
+      "step": 128152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.753362068691672e-05,
+      "loss": 2.2473,
+      "step": 128160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.752840288883964e-05,
+      "loss": 2.2329,
+      "step": 128168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.752318564940898e-05,
+      "loss": 2.2305,
+      "step": 128176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.751796896875444e-05,
+      "loss": 2.2473,
+      "step": 128184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.751275284700573e-05,
+      "loss": 2.2215,
+      "step": 128192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.75075372842925e-05,
+      "loss": 2.2394,
+      "step": 128200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7502322280744403e-05,
+      "loss": 2.2349,
+      "step": 128208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.749710783649112e-05,
+      "loss": 2.2303,
+      "step": 128216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.749189395166225e-05,
+      "loss": 2.2303,
+      "step": 128224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.748668062638742e-05,
+      "loss": 2.2426,
+      "step": 128232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7481467860796236e-05,
+      "loss": 2.2345,
+      "step": 128240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.747625565501829e-05,
+      "loss": 2.2261,
+      "step": 128248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.747104400918317e-05,
+      "loss": 2.2423,
+      "step": 128256
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7465832923420395e-05,
+      "loss": 2.2264,
+      "step": 128264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.746062239785956e-05,
+      "loss": 2.2319,
+      "step": 128272
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.745541243263018e-05,
+      "loss": 2.2384,
+      "step": 128280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.745020302786177e-05,
+      "loss": 2.2454,
+      "step": 128288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.744499418368383e-05,
+      "loss": 2.2379,
+      "step": 128296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7439785900225904e-05,
+      "loss": 2.2458,
+      "step": 128304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.74345781776174e-05,
+      "loss": 2.2334,
+      "step": 128312
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.742937101598783e-05,
+      "loss": 2.2313,
+      "step": 128320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.742416441546662e-05,
+      "loss": 2.2448,
+      "step": 128328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.741895837618321e-05,
+      "loss": 2.2095,
+      "step": 128336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.741375289826705e-05,
+      "loss": 2.2231,
+      "step": 128344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.740854798184751e-05,
+      "loss": 2.2348,
+      "step": 128352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7403343627054e-05,
+      "loss": 2.2409,
+      "step": 128360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.739813983401591e-05,
+      "loss": 2.2243,
+      "step": 128368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7392936602862586e-05,
+      "loss": 2.2334,
+      "step": 128376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.738773393372341e-05,
+      "loss": 2.2329,
+      "step": 128384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7382531826727705e-05,
+      "loss": 2.2292,
+      "step": 128392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.737733028200479e-05,
+      "loss": 2.2237,
+      "step": 128400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7372129299684e-05,
+      "loss": 2.2255,
+      "step": 128408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.736692887989459e-05,
+      "loss": 2.234,
+      "step": 128416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7361729022765865e-05,
+      "loss": 2.2342,
+      "step": 128424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.735652972842711e-05,
+      "loss": 2.2345,
+      "step": 128432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.735133099700757e-05,
+      "loss": 2.2264,
+      "step": 128440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.734613282863647e-05,
+      "loss": 2.2303,
+      "step": 128448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7340935223443066e-05,
+      "loss": 2.2404,
+      "step": 128456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.733573818155655e-05,
+      "loss": 2.2554,
+      "step": 128464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.733054170310613e-05,
+      "loss": 2.2359,
+      "step": 128472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.732534578822096e-05,
+      "loss": 2.2409,
+      "step": 128480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.732015043703026e-05,
+      "loss": 2.2393,
+      "step": 128488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.731495564966317e-05,
+      "loss": 2.2444,
+      "step": 128496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7309761426248805e-05,
+      "loss": 2.2336,
+      "step": 128504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.730456776691632e-05,
+      "loss": 2.2443,
+      "step": 128512
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7299374671794854e-05,
+      "loss": 2.2291,
+      "step": 128520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.729418214101346e-05,
+      "loss": 2.2256,
+      "step": 128528
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.728899017470125e-05,
+      "loss": 2.2306,
+      "step": 128536
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.72837987729873e-05,
+      "loss": 2.2418,
+      "step": 128544
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7278607936000645e-05,
+      "loss": 2.2293,
+      "step": 128552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.727341766387036e-05,
+      "loss": 2.2245,
+      "step": 128560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7268227956725446e-05,
+      "loss": 2.2308,
+      "step": 128568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7263038814694947e-05,
+      "loss": 2.2417,
+      "step": 128576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7257850237907856e-05,
+      "loss": 2.2341,
+      "step": 128584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.725266222649315e-05,
+      "loss": 2.232,
+      "step": 128592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.724747478057982e-05,
+      "loss": 2.2191,
+      "step": 128600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7242287900296824e-05,
+      "loss": 2.2337,
+      "step": 128608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.723710158577309e-05,
+      "loss": 2.245,
+      "step": 128616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.723191583713759e-05,
+      "loss": 2.2381,
+      "step": 128624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7226730654519186e-05,
+      "loss": 2.2179,
+      "step": 128632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.722154603804682e-05,
+      "loss": 2.2462,
+      "step": 128640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.721636198784939e-05,
+      "loss": 2.2252,
+      "step": 128648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.721117850405575e-05,
+      "loss": 2.225,
+      "step": 128656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.720599558679476e-05,
+      "loss": 2.2347,
+      "step": 128664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.720081323619529e-05,
+      "loss": 2.24,
+      "step": 128672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.719563145238616e-05,
+      "loss": 2.2397,
+      "step": 128680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.719045023549619e-05,
+      "loss": 2.2241,
+      "step": 128688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.718526958565418e-05,
+      "loss": 2.2312,
+      "step": 128696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.718008950298894e-05,
+      "loss": 2.233,
+      "step": 128704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.717490998762922e-05,
+      "loss": 2.2234,
+      "step": 128712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7169731039703804e-05,
+      "loss": 2.2416,
+      "step": 128720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.716455265934144e-05,
+      "loss": 2.227,
+      "step": 128728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.715937484667086e-05,
+      "loss": 2.2468,
+      "step": 128736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7154197601820774e-05,
+      "loss": 2.2326,
+      "step": 128744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.714902092491991e-05,
+      "loss": 2.2367,
+      "step": 128752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.714384481609695e-05,
+      "loss": 2.2189,
+      "step": 128760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.713866927548055e-05,
+      "loss": 2.2433,
+      "step": 128768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7133494303199424e-05,
+      "loss": 2.2456,
+      "step": 128776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.712831989938217e-05,
+      "loss": 2.2331,
+      "step": 128784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.712314606415748e-05,
+      "loss": 2.2286,
+      "step": 128792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.711797279765393e-05,
+      "loss": 2.2357,
+      "step": 128800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7112800100000135e-05,
+      "loss": 2.2218,
+      "step": 128808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.710762797132471e-05,
+      "loss": 2.2169,
+      "step": 128816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7102456411756226e-05,
+      "loss": 2.2409,
+      "step": 128824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7097285421423225e-05,
+      "loss": 2.2314,
+      "step": 128832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7092115000454306e-05,
+      "loss": 2.2439,
+      "step": 128840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.708694514897796e-05,
+      "loss": 2.2382,
+      "step": 128848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.708177586712271e-05,
+      "loss": 2.2284,
+      "step": 128856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7076607155017115e-05,
+      "loss": 2.2189,
+      "step": 128864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7071439012789634e-05,
+      "loss": 2.2245,
+      "step": 128872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.706627144056873e-05,
+      "loss": 2.221,
+      "step": 128880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7061104438482917e-05,
+      "loss": 2.2379,
+      "step": 128888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.705593800666061e-05,
+      "loss": 2.2222,
+      "step": 128896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.705077214523027e-05,
+      "loss": 2.2302,
+      "step": 128904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.704560685432028e-05,
+      "loss": 2.2354,
+      "step": 128912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.704044213405909e-05,
+      "loss": 2.2305,
+      "step": 128920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.703527798457511e-05,
+      "loss": 2.2257,
+      "step": 128928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.703011440599667e-05,
+      "loss": 2.2413,
+      "step": 128936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.702495139845217e-05,
+      "loss": 2.2303,
+      "step": 128944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.701978896206997e-05,
+      "loss": 2.2362,
+      "step": 128952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7014627096978364e-05,
+      "loss": 2.2246,
+      "step": 128960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7009465803305735e-05,
+      "loss": 2.2222,
+      "step": 128968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.700430508118036e-05,
+      "loss": 2.2305,
+      "step": 128976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6999144930730526e-05,
+      "loss": 2.2263,
+      "step": 128984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.699398535208454e-05,
+      "loss": 2.2323,
+      "step": 128992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6988826345370656e-05,
+      "loss": 2.23,
+      "step": 129000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.698366791071713e-05,
+      "loss": 2.2245,
+      "step": 129008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6978510048252214e-05,
+      "loss": 2.2347,
+      "step": 129016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6973352758104106e-05,
+      "loss": 2.2351,
+      "step": 129024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6968196040401044e-05,
+      "loss": 2.2435,
+      "step": 129032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6963039895271214e-05,
+      "loss": 2.2137,
+      "step": 129040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6957884322842776e-05,
+      "loss": 2.2408,
+      "step": 129048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.695272932324394e-05,
+      "loss": 2.2469,
+      "step": 129056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.694757489660284e-05,
+      "loss": 2.2282,
+      "step": 129064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6942421043047595e-05,
+      "loss": 2.2316,
+      "step": 129072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6937267762706365e-05,
+      "loss": 2.2287,
+      "step": 129080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.693211505570725e-05,
+      "loss": 2.2328,
+      "step": 129088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6926962922178346e-05,
+      "loss": 2.232,
+      "step": 129096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.692181136224771e-05,
+      "loss": 2.248,
+      "step": 129104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6916660376043444e-05,
+      "loss": 2.2358,
+      "step": 129112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6911509963693594e-05,
+      "loss": 2.2366,
+      "step": 129120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.690636012532619e-05,
+      "loss": 2.2234,
+      "step": 129128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.690121086106926e-05,
+      "loss": 2.2347,
+      "step": 129136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6896062171050846e-05,
+      "loss": 2.2179,
+      "step": 129144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.68909140553989e-05,
+      "loss": 2.226,
+      "step": 129152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6885766514241426e-05,
+      "loss": 2.2317,
+      "step": 129160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6880619547706384e-05,
+      "loss": 2.24,
+      "step": 129168
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.687547315592174e-05,
+      "loss": 2.2254,
+      "step": 129176
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.687032733901543e-05,
+      "loss": 2.2209,
+      "step": 129184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6865182097115375e-05,
+      "loss": 2.2233,
+      "step": 129192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.686003743034947e-05,
+      "loss": 2.2334,
+      "step": 129200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6854893338845664e-05,
+      "loss": 2.2356,
+      "step": 129208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6849749822731777e-05,
+      "loss": 2.2395,
+      "step": 129216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.684460688213573e-05,
+      "loss": 2.2116,
+      "step": 129224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.683946451718535e-05,
+      "loss": 2.2256,
+      "step": 129232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.683432272800846e-05,
+      "loss": 2.2155,
+      "step": 129240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.682918151473293e-05,
+      "loss": 2.2294,
+      "step": 129248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.682404087748654e-05,
+      "loss": 2.2292,
+      "step": 129256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.681890081639708e-05,
+      "loss": 2.2517,
+      "step": 129264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.681376133159237e-05,
+      "loss": 2.2304,
+      "step": 129272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.680862242320014e-05,
+      "loss": 2.2363,
+      "step": 129280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.680348409134816e-05,
+      "loss": 2.2298,
+      "step": 129288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.679834633616417e-05,
+      "loss": 2.2187,
+      "step": 129296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.67932091577759e-05,
+      "loss": 2.2263,
+      "step": 129304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.678807255631105e-05,
+      "loss": 2.2491,
+      "step": 129312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.678293653189731e-05,
+      "loss": 2.2457,
+      "step": 129320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.677780108466239e-05,
+      "loss": 2.2111,
+      "step": 129328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.677266621473394e-05,
+      "loss": 2.2392,
+      "step": 129336
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.676753192223959e-05,
+      "loss": 2.2307,
+      "step": 129344
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.676239820730701e-05,
+      "loss": 2.2402,
+      "step": 129352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.675726507006385e-05,
+      "loss": 2.2373,
+      "step": 129360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.675213251063765e-05,
+      "loss": 2.2249,
+      "step": 129368
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6747000529156064e-05,
+      "loss": 2.2358,
+      "step": 129376
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.674186912574665e-05,
+      "loss": 2.2369,
+      "step": 129384
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6736738300536974e-05,
+      "loss": 2.2478,
+      "step": 129392
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.67316080536546e-05,
+      "loss": 2.245,
+      "step": 129400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.672647838522705e-05,
+      "loss": 2.2508,
+      "step": 129408
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.672134929538187e-05,
+      "loss": 2.2168,
+      "step": 129416
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6716220784246556e-05,
+      "loss": 2.2312,
+      "step": 129424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.67110928519486e-05,
+      "loss": 2.2184,
+      "step": 129432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6705965498615494e-05,
+      "loss": 2.2214,
+      "step": 129440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6700838724374706e-05,
+      "loss": 2.2312,
+      "step": 129448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.669571252935367e-05,
+      "loss": 2.2432,
+      "step": 129456
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.669058691367985e-05,
+      "loss": 2.2292,
+      "step": 129464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.668546187748064e-05,
+      "loss": 2.2361,
+      "step": 129472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.668033742088345e-05,
+      "loss": 2.2207,
+      "step": 129480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.667521354401571e-05,
+      "loss": 2.2289,
+      "step": 129488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.667009024700479e-05,
+      "loss": 2.2392,
+      "step": 129496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.666496752997801e-05,
+      "loss": 2.2268,
+      "step": 129504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6659845393062774e-05,
+      "loss": 2.2338,
+      "step": 129512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.66547238363864e-05,
+      "loss": 2.2501,
+      "step": 129520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.664960286007622e-05,
+      "loss": 2.2296,
+      "step": 129528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.664448246425952e-05,
+      "loss": 2.2319,
+      "step": 129536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.663936264906361e-05,
+      "loss": 2.2249,
+      "step": 129544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.663424341461576e-05,
+      "loss": 2.2273,
+      "step": 129552
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6629124761043244e-05,
+      "loss": 2.2311,
+      "step": 129560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.66240066884733e-05,
+      "loss": 2.237,
+      "step": 129568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.66188891970332e-05,
+      "loss": 2.2311,
+      "step": 129576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6613772286850115e-05,
+      "loss": 2.2269,
+      "step": 129584
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660865595805128e-05,
+      "loss": 2.2396,
+      "step": 129592
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660354021076389e-05,
+      "loss": 2.2317,
+      "step": 129600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6598425045115096e-05,
+      "loss": 2.2326,
+      "step": 129608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6593310461232094e-05,
+      "loss": 2.2466,
+      "step": 129616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6588196459242005e-05,
+      "loss": 2.2298,
+      "step": 129624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6583083039272004e-05,
+      "loss": 2.238,
+      "step": 129632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657797020144918e-05,
+      "loss": 2.23,
+      "step": 129640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657285794590063e-05,
+      "loss": 2.2441,
+      "step": 129648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656774627275347e-05,
+      "loss": 2.2344,
+      "step": 129656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6562635182134774e-05,
+      "loss": 2.2324,
+      "step": 129664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655752467417158e-05,
+      "loss": 2.2234,
+      "step": 129672
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655241474899098e-05,
+      "loss": 2.2245,
+      "step": 129680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654730540671995e-05,
+      "loss": 2.2379,
+      "step": 129688
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6542196647485544e-05,
+      "loss": 2.2452,
+      "step": 129696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653708847141477e-05,
+      "loss": 2.2371,
+      "step": 129704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653198087863461e-05,
+      "loss": 2.2214,
+      "step": 129712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6526873869272025e-05,
+      "loss": 2.2221,
+      "step": 129720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6521767443454006e-05,
+      "loss": 2.2249,
+      "step": 129728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6516661601307485e-05,
+      "loss": 2.2344,
+      "step": 129736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651155634295938e-05,
+      "loss": 2.2302,
+      "step": 129744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650645166853661e-05,
+      "loss": 2.2372,
+      "step": 129752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650134757816609e-05,
+      "loss": 2.2233,
+      "step": 129760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.649624407197474e-05,
+      "loss": 2.2328,
+      "step": 129768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6491141150089365e-05,
+      "loss": 2.2386,
+      "step": 129776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.648603881263688e-05,
+      "loss": 2.2444,
+      "step": 129784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.648093705974409e-05,
+      "loss": 2.2271,
+      "step": 129792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6475835891537844e-05,
+      "loss": 2.2453,
+      "step": 129800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6470735308144974e-05,
+      "loss": 2.2355,
+      "step": 129808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6465635309692255e-05,
+      "loss": 2.2126,
+      "step": 129816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646053589630647e-05,
+      "loss": 2.2128,
+      "step": 129824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6455437068114417e-05,
+      "loss": 2.2233,
+      "step": 129832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.645033882524283e-05,
+      "loss": 2.2381,
+      "step": 129840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644524116781847e-05,
+      "loss": 2.2276,
+      "step": 129848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6440144095968064e-05,
+      "loss": 2.2157,
+      "step": 129856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.643504760981831e-05,
+      "loss": 2.2336,
+      "step": 129864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.642995170949592e-05,
+      "loss": 2.2329,
+      "step": 129872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.642485639512758e-05,
+      "loss": 2.2393,
+      "step": 129880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6419761666839935e-05,
+      "loss": 2.2469,
+      "step": 129888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.641466752475969e-05,
+      "loss": 2.2479,
+      "step": 129896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6409573969013434e-05,
+      "loss": 2.2356,
+      "step": 129904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.640448099972782e-05,
+      "loss": 2.227,
+      "step": 129912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6399388617029454e-05,
+      "loss": 2.248,
+      "step": 129920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.639429682104495e-05,
+      "loss": 2.2381,
+      "step": 129928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6389205611900864e-05,
+      "loss": 2.2371,
+      "step": 129936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.638411498972377e-05,
+      "loss": 2.2206,
+      "step": 129944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.637902495464024e-05,
+      "loss": 2.2413,
+      "step": 129952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6373935506776793e-05,
+      "loss": 2.2234,
+      "step": 129960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.636884664625996e-05,
+      "loss": 2.2155,
+      "step": 129968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.636375837321624e-05,
+      "loss": 2.2282,
+      "step": 129976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6358670687772165e-05,
+      "loss": 2.2341,
+      "step": 129984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6353583590054166e-05,
+      "loss": 2.2135,
+      "step": 129992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.634849708018875e-05,
+      "loss": 2.2119,
+      "step": 130000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6343411158302345e-05,
+      "loss": 2.2489,
+      "step": 130008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.633832582452137e-05,
+      "loss": 2.2073,
+      "step": 130016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6333241078972295e-05,
+      "loss": 2.2232,
+      "step": 130024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6328156921781506e-05,
+      "loss": 2.2233,
+      "step": 130032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.632307335307537e-05,
+      "loss": 2.2364,
+      "step": 130040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6317990372980306e-05,
+      "loss": 2.2392,
+      "step": 130048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6312907981622635e-05,
+      "loss": 2.2416,
+      "step": 130056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6307826179128755e-05,
+      "loss": 2.2243,
+      "step": 130064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.630274496562497e-05,
+      "loss": 2.2466,
+      "step": 130072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.629766434123758e-05,
+      "loss": 2.2433,
+      "step": 130080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.629258430609294e-05,
+      "loss": 2.2305,
+      "step": 130088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.628750486031731e-05,
+      "loss": 2.2251,
+      "step": 130096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6282426004036953e-05,
+      "loss": 2.2168,
+      "step": 130104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.627734773737817e-05,
+      "loss": 2.217,
+      "step": 130112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.627227006046718e-05,
+      "loss": 2.238,
+      "step": 130120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6267192973430194e-05,
+      "loss": 2.2333,
+      "step": 130128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6262116476393485e-05,
+      "loss": 2.237,
+      "step": 130136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6257040569483206e-05,
+      "loss": 2.2367,
+      "step": 130144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6251965252825574e-05,
+      "loss": 2.2434,
+      "step": 130152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6246890526546735e-05,
+      "loss": 2.2231,
+      "step": 130160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6241816390772876e-05,
+      "loss": 2.2393,
+      "step": 130168
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6236742845630125e-05,
+      "loss": 2.2201,
+      "step": 130176
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.62316698912446e-05,
+      "loss": 2.2482,
+      "step": 130184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.622659752774242e-05,
+      "loss": 2.2463,
+      "step": 130192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.622152575524974e-05,
+      "loss": 2.2333,
+      "step": 130200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.621645457389255e-05,
+      "loss": 2.2148,
+      "step": 130208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6211383983797e-05,
+      "loss": 2.2347,
+      "step": 130216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6206313985089094e-05,
+      "loss": 2.232,
+      "step": 130224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.620124457789489e-05,
+      "loss": 2.2463,
+      "step": 130232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.619617576234043e-05,
+      "loss": 2.2352,
+      "step": 130240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6191107538551705e-05,
+      "loss": 2.2478,
+      "step": 130248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6186039906654704e-05,
+      "loss": 2.2255,
+      "step": 130256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6180972866775444e-05,
+      "loss": 2.2405,
+      "step": 130264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6175906419039854e-05,
+      "loss": 2.2266,
+      "step": 130272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.617084056357391e-05,
+      "loss": 2.2348,
+      "step": 130280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6165775300503556e-05,
+      "loss": 2.2354,
+      "step": 130288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.616071062995468e-05,
+      "loss": 2.2444,
+      "step": 130296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6155646552053245e-05,
+      "loss": 2.2294,
+      "step": 130304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.615058306692508e-05,
+      "loss": 2.2152,
+      "step": 130312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.61455201746961e-05,
+      "loss": 2.241,
+      "step": 130320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6140457875492177e-05,
+      "loss": 2.2169,
+      "step": 130328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.613539616943915e-05,
+      "loss": 2.2248,
+      "step": 130336
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6130335056662836e-05,
+      "loss": 2.2248,
+      "step": 130344
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.612527453728908e-05,
+      "loss": 2.2372,
+      "step": 130352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.612021461144368e-05,
+      "loss": 2.2175,
+      "step": 130360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.611515527925243e-05,
+      "loss": 2.2513,
+      "step": 130368
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6110096540841075e-05,
+      "loss": 2.2471,
+      "step": 130376
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.610503839633542e-05,
+      "loss": 2.2205,
+      "step": 130384
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.609998084586119e-05,
+      "loss": 2.2306,
+      "step": 130392
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6094923889544117e-05,
+      "loss": 2.2253,
+      "step": 130400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6089867527509905e-05,
+      "loss": 2.2381,
+      "step": 130408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6084811759884296e-05,
+      "loss": 2.217,
+      "step": 130416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.607975658679292e-05,
+      "loss": 2.2222,
+      "step": 130424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.60747020083615e-05,
+      "loss": 2.2217,
+      "step": 130432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.606964802471567e-05,
+      "loss": 2.2326,
+      "step": 130440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6064594635981065e-05,
+      "loss": 2.2207,
+      "step": 130448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.605954184228333e-05,
+      "loss": 2.2346,
+      "step": 130456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.605448964374806e-05,
+      "loss": 2.239,
+      "step": 130464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.604943804050088e-05,
+      "loss": 2.2162,
+      "step": 130472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.604438703266735e-05,
+      "loss": 2.236,
+      "step": 130480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.603933662037304e-05,
+      "loss": 2.2343,
+      "step": 130488
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.603428680374352e-05,
+      "loss": 2.243,
+      "step": 130496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.602923758290432e-05,
+      "loss": 2.2367,
+      "step": 130504
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6024188957980955e-05,
+      "loss": 2.2319,
+      "step": 130512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.601914092909896e-05,
+      "loss": 2.226,
+      "step": 130520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.601409349638379e-05,
+      "loss": 2.2248,
+      "step": 130528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.600904665996094e-05,
+      "loss": 2.2395,
+      "step": 130536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.600400041995591e-05,
+      "loss": 2.2344,
+      "step": 130544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.599895477649411e-05,
+      "loss": 2.2332,
+      "step": 130552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.599390972970098e-05,
+      "loss": 2.225,
+      "step": 130560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5988865279701956e-05,
+      "loss": 2.251,
+      "step": 130568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5983821426622445e-05,
+      "loss": 2.2216,
+      "step": 130576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.597877817058781e-05,
+      "loss": 2.2351,
+      "step": 130584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.597373551172345e-05,
+      "loss": 2.2332,
+      "step": 130592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5968693450154726e-05,
+      "loss": 2.2249,
+      "step": 130600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.596365198600697e-05,
+      "loss": 2.2401,
+      "step": 130608
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5958611119405525e-05,
+      "loss": 2.2397,
+      "step": 130616
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.595357085047571e-05,
+      "loss": 2.2339,
+      "step": 130624
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.594853117934282e-05,
+      "loss": 2.2282,
+      "step": 130632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5943492106132134e-05,
+      "loss": 2.2355,
+      "step": 130640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.593845363096895e-05,
+      "loss": 2.2334,
+      "step": 130648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.59334157539785e-05,
+      "loss": 2.2353,
+      "step": 130656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.592837847528603e-05,
+      "loss": 2.2185,
+      "step": 130664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.592334179501678e-05,
+      "loss": 2.2154,
+      "step": 130672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5918305713295944e-05,
+      "loss": 2.2372,
+      "step": 130680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5913270230248734e-05,
+      "loss": 2.2367,
+      "step": 130688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.590823534600034e-05,
+      "loss": 2.219,
+      "step": 130696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.59032010606759e-05,
+      "loss": 2.2299,
+      "step": 130704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5898167374400595e-05,
+      "loss": 2.2388,
+      "step": 130712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.589313428729956e-05,
+      "loss": 2.2336,
+      "step": 130720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5888101799497894e-05,
+      "loss": 2.2379,
+      "step": 130728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.588306991112075e-05,
+      "loss": 2.2415,
+      "step": 130736
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5878038622293166e-05,
+      "loss": 2.2277,
+      "step": 130744
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.587300793314025e-05,
+      "loss": 2.2369,
+      "step": 130752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.586797784378708e-05,
+      "loss": 2.2473,
+      "step": 130760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.586294835435868e-05,
+      "loss": 2.2331,
+      "step": 130768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.58579194649801e-05,
+      "loss": 2.2473,
+      "step": 130776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.585289117577634e-05,
+      "loss": 2.2163,
+      "step": 130784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5847863486872415e-05,
+      "loss": 2.2285,
+      "step": 130792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.584283639839333e-05,
+      "loss": 2.2285,
+      "step": 130800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5837809910464015e-05,
+      "loss": 2.2282,
+      "step": 130808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5832784023209465e-05,
+      "loss": 2.2433,
+      "step": 130816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.582775873675463e-05,
+      "loss": 2.2335,
+      "step": 130824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.582273405122442e-05,
+      "loss": 2.2352,
+      "step": 130832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5817709966743753e-05,
+      "loss": 2.2241,
+      "step": 130840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.581268648343753e-05,
+      "loss": 2.2258,
+      "step": 130848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.580766360143062e-05,
+      "loss": 2.2222,
+      "step": 130856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.580264132084792e-05,
+      "loss": 2.2308,
+      "step": 130864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5797619641814275e-05,
+      "loss": 2.2274,
+      "step": 130872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.579259856445452e-05,
+      "loss": 2.2307,
+      "step": 130880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5787578088893474e-05,
+      "loss": 2.2427,
+      "step": 130888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.578255821525595e-05,
+      "loss": 2.2427,
+      "step": 130896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5777538943666753e-05,
+      "loss": 2.2451,
+      "step": 130904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5772520274250666e-05,
+      "loss": 2.2399,
+      "step": 130912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.576750220713242e-05,
+      "loss": 2.2427,
+      "step": 130920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.576248474243681e-05,
+      "loss": 2.2364,
+      "step": 130928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.575746788028854e-05,
+      "loss": 2.2318,
+      "step": 130936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5752451620812334e-05,
+      "loss": 2.2235,
+      "step": 130944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.574743596413292e-05,
+      "loss": 2.235,
+      "step": 130952
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.574242091037494e-05,
+      "loss": 2.224,
+      "step": 130960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.57374064596631e-05,
+      "loss": 2.2426,
+      "step": 130968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5732392612122076e-05,
+      "loss": 2.2364,
+      "step": 130976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5727379367876494e-05,
+      "loss": 2.245,
+      "step": 130984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5722366727050983e-05,
+      "loss": 2.2434,
+      "step": 130992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.571735468977015e-05,
+      "loss": 2.2277,
+      "step": 131000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.571234325615861e-05,
+      "loss": 2.2427,
+      "step": 131008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5707332426340946e-05,
+      "loss": 2.2502,
+      "step": 131016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5702322200441706e-05,
+      "loss": 2.2197,
+      "step": 131024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.569731257858546e-05,
+      "loss": 2.2271,
+      "step": 131032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.569230356089679e-05,
+      "loss": 2.2342,
+      "step": 131040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.568729514750014e-05,
+      "loss": 2.227,
+      "step": 131048
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5682287338520085e-05,
+      "loss": 2.2214,
+      "step": 131056
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5677280134081084e-05,
+      "loss": 2.2416,
+      "step": 131064
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.567227353430762e-05,
+      "loss": 2.2156,
+      "step": 131072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.566726753932418e-05,
+      "loss": 2.2305,
+      "step": 131080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5662262149255196e-05,
+      "loss": 2.2269,
+      "step": 131088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.565725736422509e-05,
+      "loss": 2.2359,
+      "step": 131096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.565225318435832e-05,
+      "loss": 2.2318,
+      "step": 131104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.564724960977925e-05,
+      "loss": 2.2353,
+      "step": 131112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.56422466406123e-05,
+      "loss": 2.2289,
+      "step": 131120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.563724427698183e-05,
+      "loss": 2.2095,
+      "step": 131128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5632242519012183e-05,
+      "loss": 2.2371,
+      "step": 131136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.562724136682776e-05,
+      "loss": 2.2384,
+      "step": 131144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.562224082055281e-05,
+      "loss": 2.236,
+      "step": 131152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.561724088031169e-05,
+      "loss": 2.2331,
+      "step": 131160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.561224154622872e-05,
+      "loss": 2.2052,
+      "step": 131168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.560724281842815e-05,
+      "loss": 2.2224,
+      "step": 131176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.560224469703426e-05,
+      "loss": 2.2249,
+      "step": 131184
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5597247182171316e-05,
+      "loss": 2.2243,
+      "step": 131192
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.559225027396354e-05,
+      "loss": 2.2386,
+      "step": 131200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5587253972535165e-05,
+      "loss": 2.2328,
+      "step": 131208
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.558225827801038e-05,
+      "loss": 2.2449,
+      "step": 131216
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.55772631905134e-05,
+      "loss": 2.2291,
+      "step": 131224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5572268710168406e-05,
+      "loss": 2.2375,
+      "step": 131232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.556727483709954e-05,
+      "loss": 2.2252,
+      "step": 131240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5562281571430955e-05,
+      "loss": 2.2283,
+      "step": 131248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.555728891328682e-05,
+      "loss": 2.2374,
+      "step": 131256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.55522968627912e-05,
+      "loss": 2.2252,
+      "step": 131264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.554730542006824e-05,
+      "loss": 2.2236,
+      "step": 131272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.554231458524201e-05,
+      "loss": 2.2291,
+      "step": 131280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.553732435843657e-05,
+      "loss": 2.2373,
+      "step": 131288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5532334739775996e-05,
+      "loss": 2.2376,
+      "step": 131296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.552734572938434e-05,
+      "loss": 2.2182,
+      "step": 131304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5522357327385586e-05,
+      "loss": 2.2255,
+      "step": 131312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.55173695339038e-05,
+      "loss": 2.2209,
+      "step": 131320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5512382349062935e-05,
+      "loss": 2.2334,
+      "step": 131328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5507395772987005e-05,
+      "loss": 2.2425,
+      "step": 131336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.550240980579997e-05,
+      "loss": 2.2324,
+      "step": 131344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.549742444762576e-05,
+      "loss": 2.2303,
+      "step": 131352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5492439698588356e-05,
+      "loss": 2.226,
+      "step": 131360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.548745555881162e-05,
+      "loss": 2.2275,
+      "step": 131368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5482472028419495e-05,
+      "loss": 2.2508,
+      "step": 131376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.547748910753587e-05,
+      "loss": 2.2334,
+      "step": 131384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.547250679628462e-05,
+      "loss": 2.2111,
+      "step": 131392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5467525094789594e-05,
+      "loss": 2.2356,
+      "step": 131400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.546254400317466e-05,
+      "loss": 2.224,
+      "step": 131408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.545756352156364e-05,
+      "loss": 2.2334,
+      "step": 131416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.545258365008033e-05,
+      "loss": 2.2134,
+      "step": 131424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.544760438884854e-05,
+      "loss": 2.228,
+      "step": 131432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.544262573799207e-05,
+      "loss": 2.2304,
+      "step": 131440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.543764769763468e-05,
+      "loss": 2.2416,
+      "step": 131448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.543267026790012e-05,
+      "loss": 2.2403,
+      "step": 131456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5427693448912135e-05,
+      "loss": 2.2346,
+      "step": 131464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5422717240794454e-05,
+      "loss": 2.2401,
+      "step": 131472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.541774164367076e-05,
+      "loss": 2.2356,
+      "step": 131480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.541276665766479e-05,
+      "loss": 2.2306,
+      "step": 131488
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.540779228290019e-05,
+      "loss": 2.2304,
+      "step": 131496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.540281851950062e-05,
+      "loss": 2.2475,
+      "step": 131504
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5397845367589766e-05,
+      "loss": 2.2267,
+      "step": 131512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.539287282729121e-05,
+      "loss": 2.2171,
+      "step": 131520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.538790089872862e-05,
+      "loss": 2.2169,
+      "step": 131528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.538292958202557e-05,
+      "loss": 2.23,
+      "step": 131536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.537795887730564e-05,
+      "loss": 2.2357,
+      "step": 131544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.537298878469242e-05,
+      "loss": 2.2341,
+      "step": 131552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.536801930430948e-05,
+      "loss": 2.2327,
+      "step": 131560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.536305043628032e-05,
+      "loss": 2.2423,
+      "step": 131568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.535808218072851e-05,
+      "loss": 2.235,
+      "step": 131576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.535311453777753e-05,
+      "loss": 2.2136,
+      "step": 131584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.534814750755088e-05,
+      "loss": 2.2175,
+      "step": 131592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5343181090172057e-05,
+      "loss": 2.2435,
+      "step": 131600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.533821528576452e-05,
+      "loss": 2.2336,
+      "step": 131608
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5333250094451715e-05,
+      "loss": 2.2425,
+      "step": 131616
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.532828551635708e-05,
+      "loss": 2.2258,
+      "step": 131624
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.532332155160405e-05,
+      "loss": 2.234,
+      "step": 131632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5318358200316e-05,
+      "loss": 2.2327,
+      "step": 131640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5313395462616334e-05,
+      "loss": 2.2355,
+      "step": 131648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.530843333862843e-05,
+      "loss": 2.2138,
+      "step": 131656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5303471828475654e-05,
+      "loss": 2.2371,
+      "step": 131664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.529851093228131e-05,
+      "loss": 2.2268,
+      "step": 131672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.529355065016879e-05,
+      "loss": 2.2247,
+      "step": 131680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.528859098226136e-05,
+      "loss": 2.2429,
+      "step": 131688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5283631928682316e-05,
+      "loss": 2.2139,
+      "step": 131696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.527867348955497e-05,
+      "loss": 2.2254,
+      "step": 131704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5273715665002574e-05,
+      "loss": 2.2267,
+      "step": 131712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.526875845514837e-05,
+      "loss": 2.2435,
+      "step": 131720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.526380186011562e-05,
+      "loss": 2.2486,
+      "step": 131728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.525884588002752e-05,
+      "loss": 2.2356,
+      "step": 131736
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.525389051500729e-05,
+      "loss": 2.2236,
+      "step": 131744
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.524893576517813e-05,
+      "loss": 2.2263,
+      "step": 131752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.524398163066318e-05,
+      "loss": 2.2419,
+      "step": 131760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.523902811158565e-05,
+      "loss": 2.2334,
+      "step": 131768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.523407520806865e-05,
+      "loss": 2.2283,
+      "step": 131776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.522912292023531e-05,
+      "loss": 2.2268,
+      "step": 131784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.522417124820878e-05,
+      "loss": 2.2214,
+      "step": 131792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.52192201921121e-05,
+      "loss": 2.2292,
+      "step": 131800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.52142697520684e-05,
+      "loss": 2.2362,
+      "step": 131808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5209319928200734e-05,
+      "loss": 2.2237,
+      "step": 131816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.520437072063216e-05,
+      "loss": 2.2264,
+      "step": 131824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.519942212948572e-05,
+      "loss": 2.2395,
+      "step": 131832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.51944741548844e-05,
+      "loss": 2.2279,
+      "step": 131840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.518952679695126e-05,
+      "loss": 2.2305,
+      "step": 131848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.518458005580926e-05,
+      "loss": 2.2291,
+      "step": 131856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.517963393158138e-05,
+      "loss": 2.2376,
+      "step": 131864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.517468842439058e-05,
+      "loss": 2.2353,
+      "step": 131872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5169743534359836e-05,
+      "loss": 2.2273,
+      "step": 131880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.516479926161203e-05,
+      "loss": 2.2497,
+      "step": 131888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.515985560627011e-05,
+      "loss": 2.2218,
+      "step": 131896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5154912568456974e-05,
+      "loss": 2.2263,
+      "step": 131904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5149970148295484e-05,
+      "loss": 2.2381,
+      "step": 131912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5145028345908536e-05,
+      "loss": 2.2197,
+      "step": 131920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5140087161418964e-05,
+      "loss": 2.2338,
+      "step": 131928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.513514659494961e-05,
+      "loss": 2.2337,
+      "step": 131936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.513020664662331e-05,
+      "loss": 2.238,
+      "step": 131944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.512526731656286e-05,
+      "loss": 2.223,
+      "step": 131952
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.512032860489106e-05,
+      "loss": 2.2202,
+      "step": 131960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5115390511730685e-05,
+      "loss": 2.2381,
+      "step": 131968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5110453037204486e-05,
+      "loss": 2.2392,
+      "step": 131976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5105516181435225e-05,
+      "loss": 2.2332,
+      "step": 131984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.510057994454561e-05,
+      "loss": 2.2364,
+      "step": 131992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5095644326658375e-05,
+      "loss": 2.2258,
+      "step": 132000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.509070932789624e-05,
+      "loss": 2.221,
+      "step": 132008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.508577494838184e-05,
+      "loss": 2.2277,
+      "step": 132016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.508084118823787e-05,
+      "loss": 2.2265,
+      "step": 132024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5075908047587e-05,
+      "loss": 2.2399,
+      "step": 132032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.507097552655185e-05,
+      "loss": 2.2239,
+      "step": 132040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.506604362525503e-05,
+      "loss": 2.2244,
+      "step": 132048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.506111234381917e-05,
+      "loss": 2.2304,
+      "step": 132056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.505618168236686e-05,
+      "loss": 2.2251,
+      "step": 132064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.505125164102068e-05,
+      "loss": 2.242,
+      "step": 132072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.504632221990316e-05,
+      "loss": 2.2252,
+      "step": 132080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.504139341913687e-05,
+      "loss": 2.2191,
+      "step": 132088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.503646523884437e-05,
+      "loss": 2.2214,
+      "step": 132096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.503153767914812e-05,
+      "loss": 2.2328,
+      "step": 132104
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.502661074017066e-05,
+      "loss": 2.2312,
+      "step": 132112
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.502168442203446e-05,
+      "loss": 2.2222,
+      "step": 132120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.501675872486196e-05,
+      "loss": 2.2285,
+      "step": 132128
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.501183364877567e-05,
+      "loss": 2.22,
+      "step": 132136
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5006909193897995e-05,
+      "loss": 2.2161,
+      "step": 132144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.500198536035135e-05,
+      "loss": 2.2107,
+      "step": 132152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4997062148258156e-05,
+      "loss": 2.2113,
+      "step": 132160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.49921395577408e-05,
+      "loss": 2.2308,
+      "step": 132168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4987217588921675e-05,
+      "loss": 2.2417,
+      "step": 132176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.498229624192312e-05,
+      "loss": 2.2292,
+      "step": 132184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.497737551686747e-05,
+      "loss": 2.2337,
+      "step": 132192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.497245541387711e-05,
+      "loss": 2.2358,
+      "step": 132200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.496753593307428e-05,
+      "loss": 2.2436,
+      "step": 132208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.496261707458133e-05,
+      "loss": 2.2273,
+      "step": 132216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4957698838520524e-05,
+      "loss": 2.227,
+      "step": 132224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.495278122501414e-05,
+      "loss": 2.2352,
+      "step": 132232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4947864234184416e-05,
+      "loss": 2.2396,
+      "step": 132240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.494294786615362e-05,
+      "loss": 2.2384,
+      "step": 132248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.493803212104394e-05,
+      "loss": 2.2256,
+      "step": 132256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.493311699897759e-05,
+      "loss": 2.2285,
+      "step": 132264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4928202500076754e-05,
+      "loss": 2.2385,
+      "step": 132272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4923288624463635e-05,
+      "loss": 2.2355,
+      "step": 132280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.491837537226037e-05,
+      "loss": 2.2325,
+      "step": 132288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4913462743589096e-05,
+      "loss": 2.2357,
+      "step": 132296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.490855073857197e-05,
+      "loss": 2.2477,
+      "step": 132304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4903639357331077e-05,
+      "loss": 2.2149,
+      "step": 132312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.489872859998852e-05,
+      "loss": 2.2238,
+      "step": 132320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4893818466666396e-05,
+      "loss": 2.2247,
+      "step": 132328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.488890895748676e-05,
+      "loss": 2.2256,
+      "step": 132336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4884000072571655e-05,
+      "loss": 2.2285,
+      "step": 132344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.487909181204314e-05,
+      "loss": 2.231,
+      "step": 132352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.487418417602322e-05,
+      "loss": 2.2214,
+      "step": 132360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.486927716463388e-05,
+      "loss": 2.2226,
+      "step": 132368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4864370777997156e-05,
+      "loss": 2.2237,
+      "step": 132376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4859465016234984e-05,
+      "loss": 2.2252,
+      "step": 132384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4854559879469326e-05,
+      "loss": 2.2304,
+      "step": 132392
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.484965536782214e-05,
+      "loss": 2.2433,
+      "step": 132400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.484475148141534e-05,
+      "loss": 2.2254,
+      "step": 132408
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.483984822037086e-05,
+      "loss": 2.2272,
+      "step": 132416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.483494558481054e-05,
+      "loss": 2.2296,
+      "step": 132424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4830043574856306e-05,
+      "loss": 2.2235,
+      "step": 132432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.482514219063003e-05,
+      "loss": 2.2389,
+      "step": 132440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.482024143225354e-05,
+      "loss": 2.2252,
+      "step": 132448
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.481534129984865e-05,
+      "loss": 2.2311,
+      "step": 132456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4810441793537226e-05,
+      "loss": 2.23,
+      "step": 132464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.480554291344105e-05,
+      "loss": 2.2157,
+      "step": 132472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.480064465968189e-05,
+      "loss": 2.2309,
+      "step": 132480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.479574703238152e-05,
+      "loss": 2.2337,
+      "step": 132488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.479085003166173e-05,
+      "loss": 2.2222,
+      "step": 132496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.478595365764422e-05,
+      "loss": 2.2561,
+      "step": 132504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.478105791045073e-05,
+      "loss": 2.2259,
+      "step": 132512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.477616279020298e-05,
+      "loss": 2.2105,
+      "step": 132520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.477126829702265e-05,
+      "loss": 2.2314,
+      "step": 132528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.476637443103142e-05,
+      "loss": 2.2338,
+      "step": 132536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4761481192350954e-05,
+      "loss": 2.2172,
+      "step": 132544
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.47565885811029e-05,
+      "loss": 2.2046,
+      "step": 132552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.475169659740888e-05,
+      "loss": 2.226,
+      "step": 132560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.474680524139052e-05,
+      "loss": 2.229,
+      "step": 132568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.47419145131694e-05,
+      "loss": 2.2336,
+      "step": 132576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.473702441286714e-05,
+      "loss": 2.2161,
+      "step": 132584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.473213494060528e-05,
+      "loss": 2.2287,
+      "step": 132592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.472724609650537e-05,
+      "loss": 2.2196,
+      "step": 132600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.472235788068897e-05,
+      "loss": 2.2271,
+      "step": 132608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4717470293277594e-05,
+      "loss": 2.2131,
+      "step": 132616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.471258333439272e-05,
+      "loss": 2.2409,
+      "step": 132624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.47076970041559e-05,
+      "loss": 2.2147,
+      "step": 132632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4702811302688544e-05,
+      "loss": 2.2262,
+      "step": 132640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.469792623011213e-05,
+      "loss": 2.204,
+      "step": 132648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.469304178654812e-05,
+      "loss": 2.2135,
+      "step": 132656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.468815797211794e-05,
+      "loss": 2.2331,
+      "step": 132664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.468327478694299e-05,
+      "loss": 2.2264,
+      "step": 132672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.467839223114465e-05,
+      "loss": 2.2367,
+      "step": 132680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.467351030484434e-05,
+      "loss": 2.2461,
+      "step": 132688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.46686290081634e-05,
+      "loss": 2.2329,
+      "step": 132696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.466374834122317e-05,
+      "loss": 2.24,
+      "step": 132704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.465886830414502e-05,
+      "loss": 2.2315,
+      "step": 132712
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.465398889705024e-05,
+      "loss": 2.238,
+      "step": 132720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4649110120060124e-05,
+      "loss": 2.2279,
+      "step": 132728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4644231973296e-05,
+      "loss": 2.2282,
+      "step": 132736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.46393544568791e-05,
+      "loss": 2.2255,
+      "step": 132744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.463447757093069e-05,
+      "loss": 2.2336,
+      "step": 132752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.462960131557203e-05,
+      "loss": 2.2248,
+      "step": 132760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.462472569092432e-05,
+      "loss": 2.236,
+      "step": 132768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4619850697108766e-05,
+      "loss": 2.2256,
+      "step": 132776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.461497633424659e-05,
+      "loss": 2.2318,
+      "step": 132784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.461010260245893e-05,
+      "loss": 2.232,
+      "step": 132792
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.460522950186698e-05,
+      "loss": 2.2271,
+      "step": 132800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.460035703259188e-05,
+      "loss": 2.2302,
+      "step": 132808
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.459548519475473e-05,
+      "loss": 2.2273,
+      "step": 132816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.459061398847669e-05,
+      "loss": 2.2311,
+      "step": 132824
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.458574341387884e-05,
+      "loss": 2.2329,
+      "step": 132832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4580873471082245e-05,
+      "loss": 2.2182,
+      "step": 132840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.457600416020801e-05,
+      "loss": 2.2333,
+      "step": 132848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.457113548137714e-05,
+      "loss": 2.221,
+      "step": 132856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4566267434710696e-05,
+      "loss": 2.2261,
+      "step": 132864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.45614000203297e-05,
+      "loss": 2.2331,
+      "step": 132872
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.455653323835517e-05,
+      "loss": 2.2235,
+      "step": 132880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.455166708890807e-05,
+      "loss": 2.2327,
+      "step": 132888
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4546801572109365e-05,
+      "loss": 2.2402,
+      "step": 132896
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4541936688080046e-05,
+      "loss": 2.2383,
+      "step": 132904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.453707243694104e-05,
+      "loss": 2.2333,
+      "step": 132912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4532208818813255e-05,
+      "loss": 2.2282,
+      "step": 132920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4527345833817614e-05,
+      "loss": 2.2301,
+      "step": 132928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.452248348207504e-05,
+      "loss": 2.217,
+      "step": 132936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4517621763706366e-05,
+      "loss": 2.2284,
+      "step": 132944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.451276067883248e-05,
+      "loss": 2.2402,
+      "step": 132952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.450790022757423e-05,
+      "loss": 2.2524,
+      "step": 132960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.450304041005241e-05,
+      "loss": 2.2393,
+      "step": 132968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4498181226387894e-05,
+      "loss": 2.2419,
+      "step": 132976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.449332267670145e-05,
+      "loss": 2.2275,
+      "step": 132984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.448846476111386e-05,
+      "loss": 2.2145,
+      "step": 132992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4483607479745904e-05,
+      "loss": 2.2475,
+      "step": 133000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.447875083271831e-05,
+      "loss": 2.2269,
+      "step": 133008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.447389482015186e-05,
+      "loss": 2.2226,
+      "step": 133016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4469039442167236e-05,
+      "loss": 2.2178,
+      "step": 133024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4464184698885154e-05,
+      "loss": 2.232,
+      "step": 133032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.445933059042633e-05,
+      "loss": 2.2177,
+      "step": 133040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.445447711691139e-05,
+      "loss": 2.2261,
+      "step": 133048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4449624278461e-05,
+      "loss": 2.214,
+      "step": 133056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.444477207519584e-05,
+      "loss": 2.2307,
+      "step": 133064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.443992050723651e-05,
+      "loss": 2.2416,
+      "step": 133072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4435069574703615e-05,
+      "loss": 2.228,
+      "step": 133080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4430219277717776e-05,
+      "loss": 2.2288,
+      "step": 133088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.442536961639955e-05,
+      "loss": 2.2174,
+      "step": 133096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.44205205908695e-05,
+      "loss": 2.2319,
+      "step": 133104
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.441567220124817e-05,
+      "loss": 2.2227,
+      "step": 133112
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.441082444765612e-05,
+      "loss": 2.2349,
+      "step": 133120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.440597733021383e-05,
+      "loss": 2.2409,
+      "step": 133128
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4401130849041805e-05,
+      "loss": 2.24,
+      "step": 133136
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.439628500426054e-05,
+      "loss": 2.2295,
+      "step": 133144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4391439795990536e-05,
+      "loss": 2.2228,
+      "step": 133152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.438659522435218e-05,
+      "loss": 2.2192,
+      "step": 133160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.438175128946596e-05,
+      "loss": 2.2385,
+      "step": 133168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.437690799145227e-05,
+      "loss": 2.2279,
+      "step": 133176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.437206533043151e-05,
+      "loss": 2.2539,
+      "step": 133184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.43672233065241e-05,
+      "loss": 2.2224,
+      "step": 133192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4362381919850394e-05,
+      "loss": 2.227,
+      "step": 133200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.435754117053073e-05,
+      "loss": 2.2444,
+      "step": 133208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.435270105868549e-05,
+      "loss": 2.2338,
+      "step": 133216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4347861584434956e-05,
+      "loss": 2.2327,
+      "step": 133224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.434302274789948e-05,
+      "loss": 2.2226,
+      "step": 133232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4338184549199336e-05,
+      "loss": 2.2302,
+      "step": 133240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.433334698845479e-05,
+      "loss": 2.2254,
+      "step": 133248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.432851006578615e-05,
+      "loss": 2.246,
+      "step": 133256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.43236737813136e-05,
+      "loss": 2.247,
+      "step": 133264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.43188381351574e-05,
+      "loss": 2.2316,
+      "step": 133272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.431400312743778e-05,
+      "loss": 2.2443,
+      "step": 133280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4309168758274925e-05,
+      "loss": 2.2256,
+      "step": 133288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4304335027789e-05,
+      "loss": 2.2211,
+      "step": 133296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.42995019361002e-05,
+      "loss": 2.2222,
+      "step": 133304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.429466948332867e-05,
+      "loss": 2.2294,
+      "step": 133312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.428983766959454e-05,
+      "loss": 2.2153,
+      "step": 133320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.428500649501792e-05,
+      "loss": 2.2214,
+      "step": 133328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4280175959718927e-05,
+      "loss": 2.2141,
+      "step": 133336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.427534606381766e-05,
+      "loss": 2.2546,
+      "step": 133344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.427051680743415e-05,
+      "loss": 2.2353,
+      "step": 133352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.42656881906885e-05,
+      "loss": 2.2198,
+      "step": 133360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.426086021370073e-05,
+      "loss": 2.2451,
+      "step": 133368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.425603287659084e-05,
+      "loss": 2.2162,
+      "step": 133376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.425120617947889e-05,
+      "loss": 2.251,
+      "step": 133384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.424638012248483e-05,
+      "loss": 2.2348,
+      "step": 133392
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4241554705728636e-05,
+      "loss": 2.23,
+      "step": 133400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4236729929330306e-05,
+      "loss": 2.2459,
+      "step": 133408
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.423190579340974e-05,
+      "loss": 2.2327,
+      "step": 133416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.422708229808689e-05,
+      "loss": 2.2215,
+      "step": 133424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.422225944348168e-05,
+      "loss": 2.2285,
+      "step": 133432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4217437229713977e-05,
+      "loss": 2.2276,
+      "step": 133440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.421261565690369e-05,
+      "loss": 2.2151,
+      "step": 133448
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4207794725170675e-05,
+      "loss": 2.2187,
+      "step": 133456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4202974434634763e-05,
+      "loss": 2.2319,
+      "step": 133464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4198154785415815e-05,
+      "loss": 2.2174,
+      "step": 133472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.419333577763362e-05,
+      "loss": 2.2294,
+      "step": 133480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.418851741140798e-05,
+      "loss": 2.21,
+      "step": 133488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.418369968685872e-05,
+      "loss": 2.2286,
+      "step": 133496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4178882604105575e-05,
+      "loss": 2.2367,
+      "step": 133504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4174066163268313e-05,
+      "loss": 2.2306,
+      "step": 133512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4169250364466644e-05,
+      "loss": 2.2254,
+      "step": 133520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.416443520782032e-05,
+      "loss": 2.234,
+      "step": 133528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.415962069344905e-05,
+      "loss": 2.2462,
+      "step": 133536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.415480682147248e-05,
+      "loss": 2.2151,
+      "step": 133544
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.414999359201033e-05,
+      "loss": 2.217,
+      "step": 133552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.414518100518224e-05,
+      "loss": 2.2356,
+      "step": 133560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4140369061107835e-05,
+      "loss": 2.2434,
+      "step": 133568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.413555775990678e-05,
+      "loss": 2.2438,
+      "step": 133576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.413074710169865e-05,
+      "loss": 2.234,
+      "step": 133584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4125937086603047e-05,
+      "loss": 2.2242,
+      "step": 133592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4121127714739555e-05,
+      "loss": 2.2364,
+      "step": 133600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.411631898622772e-05,
+      "loss": 2.2288,
+      "step": 133608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4111510901187105e-05,
+      "loss": 2.2349,
+      "step": 133616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.410670345973724e-05,
+      "loss": 2.2363,
+      "step": 133624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4101896661997626e-05,
+      "loss": 2.2474,
+      "step": 133632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.409709050808777e-05,
+      "loss": 2.2273,
+      "step": 133640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.409228499812716e-05,
+      "loss": 2.2291,
+      "step": 133648
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.408748013223523e-05,
+      "loss": 2.231,
+      "step": 133656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4082675910531466e-05,
+      "loss": 2.2305,
+      "step": 133664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.40778723331353e-05,
+      "loss": 2.2304,
+      "step": 133672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.40730694001661e-05,
+      "loss": 2.2251,
+      "step": 133680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.406826711174335e-05,
+      "loss": 2.2058,
+      "step": 133688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4063465467986355e-05,
+      "loss": 2.2429,
+      "step": 133696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.405866446901451e-05,
+      "loss": 2.2228,
+      "step": 133704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.40538641149472e-05,
+      "loss": 2.2185,
+      "step": 133712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.404906440590374e-05,
+      "loss": 2.2271,
+      "step": 133720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.404426534200344e-05,
+      "loss": 2.2341,
+      "step": 133728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.403946692336562e-05,
+      "loss": 2.2395,
+      "step": 133736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.403466915010957e-05,
+      "loss": 2.2397,
+      "step": 133744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.402987202235455e-05,
+      "loss": 2.2292,
+      "step": 133752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.402507554021982e-05,
+      "loss": 2.2311,
+      "step": 133760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.402027970382462e-05,
+      "loss": 2.2227,
+      "step": 133768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.401548451328822e-05,
+      "loss": 2.2309,
+      "step": 133776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.401068996872975e-05,
+      "loss": 2.2222,
+      "step": 133784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.400589607026846e-05,
+      "loss": 2.2245,
+      "step": 133792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.400110281802352e-05,
+      "loss": 2.2283,
+      "step": 133800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.399631021211406e-05,
+      "loss": 2.2386,
+      "step": 133808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.399151825265926e-05,
+      "loss": 2.2373,
+      "step": 133816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.398672693977823e-05,
+      "loss": 2.2291,
+      "step": 133824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.398193627359008e-05,
+      "loss": 2.2355,
+      "step": 133832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.397714625421393e-05,
+      "loss": 2.2402,
+      "step": 133840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.397235688176882e-05,
+      "loss": 2.2124,
+      "step": 133848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.396756815637386e-05,
+      "loss": 2.2118,
+      "step": 133856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.396278007814807e-05,
+      "loss": 2.2479,
+      "step": 133864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3957992647210476e-05,
+      "loss": 2.2385,
+      "step": 133872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.395320586368014e-05,
+      "loss": 2.2263,
+      "step": 133880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3948419727676e-05,
+      "loss": 2.2383,
+      "step": 133888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.394363423931706e-05,
+      "loss": 2.2242,
+      "step": 133896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.393884939872234e-05,
+      "loss": 2.2191,
+      "step": 133904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.39340652060107e-05,
+      "loss": 2.2311,
+      "step": 133912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3929281661301135e-05,
+      "loss": 2.2281,
+      "step": 133920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3924498764712564e-05,
+      "loss": 2.2307,
+      "step": 133928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3919716516363885e-05,
+      "loss": 2.2407,
+      "step": 133936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3914934916373974e-05,
+      "loss": 2.2217,
+      "step": 133944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.39101539648617e-05,
+      "loss": 2.2288,
+      "step": 133952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.390537366194593e-05,
+      "loss": 2.2256,
+      "step": 133960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.39005940077455e-05,
+      "loss": 2.2476,
+      "step": 133968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.389581500237923e-05,
+      "loss": 2.2314,
+      "step": 133976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.389103664596591e-05,
+      "loss": 2.2307,
+      "step": 133984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.388625893862439e-05,
+      "loss": 2.2164,
+      "step": 133992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3881481880473366e-05,
+      "loss": 2.2413,
+      "step": 134000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.387670547163165e-05,
+      "loss": 2.2238,
+      "step": 134008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3871929712217966e-05,
+      "loss": 2.2419,
+      "step": 134016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.386715460235103e-05,
+      "loss": 2.2335,
+      "step": 134024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.386238014214957e-05,
+      "loss": 2.2187,
+      "step": 134032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3857606331732284e-05,
+      "loss": 2.2413,
+      "step": 134040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.385283317121781e-05,
+      "loss": 2.2251,
+      "step": 134048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.384806066072486e-05,
+      "loss": 2.2473,
+      "step": 134056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.384328880037204e-05,
+      "loss": 2.2322,
+      "step": 134064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3838517590278014e-05,
+      "loss": 2.2356,
+      "step": 134072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.383374703056137e-05,
+      "loss": 2.2374,
+      "step": 134080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3828977121340696e-05,
+      "loss": 2.2402,
+      "step": 134088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3824207862734613e-05,
+      "loss": 2.2445,
+      "step": 134096
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.381943925486164e-05,
+      "loss": 2.2373,
+      "step": 134104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.381467129784034e-05,
+      "loss": 2.2278,
+      "step": 134112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.380990399178926e-05,
+      "loss": 2.2305,
+      "step": 134120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.380513733682691e-05,
+      "loss": 2.2261,
+      "step": 134128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.380037133307177e-05,
+      "loss": 2.2268,
+      "step": 134136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.379560598064235e-05,
+      "loss": 2.2247,
+      "step": 134144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.37908412796571e-05,
+      "loss": 2.2197,
+      "step": 134152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.378607723023449e-05,
+      "loss": 2.233,
+      "step": 134160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3781313832492926e-05,
+      "loss": 2.2258,
+      "step": 134168
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.377655108655085e-05,
+      "loss": 2.2286,
+      "step": 134176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3771788992526646e-05,
+      "loss": 2.2258,
+      "step": 134184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.376702755053871e-05,
+      "loss": 2.2346,
+      "step": 134192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3762266760705423e-05,
+      "loss": 2.2363,
+      "step": 134200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.375750662314513e-05,
+      "loss": 2.2378,
+      "step": 134208
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3752747137976146e-05,
+      "loss": 2.2359,
+      "step": 134216
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.374798830531684e-05,
+      "loss": 2.2236,
+      "step": 134224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.374323012528547e-05,
+      "loss": 2.2302,
+      "step": 134232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.373847259800035e-05,
+      "loss": 2.2248,
+      "step": 134240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.373371572357976e-05,
+      "loss": 2.2468,
+      "step": 134248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.372895950214193e-05,
+      "loss": 2.2353,
+      "step": 134256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.372420393380511e-05,
+      "loss": 2.2262,
+      "step": 134264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.371944901868754e-05,
+      "loss": 2.2186,
+      "step": 134272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.37146947569074e-05,
+      "loss": 2.2152,
+      "step": 134280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.370994114858292e-05,
+      "loss": 2.2439,
+      "step": 134288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3705188193832246e-05,
+      "loss": 2.2252,
+      "step": 134296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.370043589277354e-05,
+      "loss": 2.2376,
+      "step": 134304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.369568424552496e-05,
+      "loss": 2.2406,
+      "step": 134312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3690933252204605e-05,
+      "loss": 2.2205,
+      "step": 134320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.36861829129306e-05,
+      "loss": 2.2373,
+      "step": 134328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.368143322782105e-05,
+      "loss": 2.2177,
+      "step": 134336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3676684196994024e-05,
+      "loss": 2.2221,
+      "step": 134344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3671935820567566e-05,
+      "loss": 2.2313,
+      "step": 134352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.366718809865975e-05,
+      "loss": 2.2247,
+      "step": 134360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3662441031388605e-05,
+      "loss": 2.2189,
+      "step": 134368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.365769461887211e-05,
+      "loss": 2.2231,
+      "step": 134376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.365294886122829e-05,
+      "loss": 2.2254,
+      "step": 134384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.364820375857512e-05,
+      "loss": 2.2289,
+      "step": 134392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3643459311030564e-05,
+      "loss": 2.2253,
+      "step": 134400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.363871551871255e-05,
+      "loss": 2.2325,
+      "step": 134408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.363397238173904e-05,
+      "loss": 2.2288,
+      "step": 134416
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3629229900227944e-05,
+      "loss": 2.2371,
+      "step": 134424
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.362448807429712e-05,
+      "loss": 2.2365,
+      "step": 134432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.361974690406451e-05,
+      "loss": 2.2269,
+      "step": 134440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.361500638964795e-05,
+      "loss": 2.2196,
+      "step": 134448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.361026653116527e-05,
+      "loss": 2.2226,
+      "step": 134456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.360552732873434e-05,
+      "loss": 2.2266,
+      "step": 134464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3600788782472956e-05,
+      "loss": 2.2347,
+      "step": 134472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3596050892498937e-05,
+      "loss": 2.2454,
+      "step": 134480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.359131365893005e-05,
+      "loss": 2.2528,
+      "step": 134488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.358657708188406e-05,
+      "loss": 2.2432,
+      "step": 134496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.358184116147874e-05,
+      "loss": 2.2129,
+      "step": 134504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3577105897831805e-05,
+      "loss": 2.2317,
+      "step": 134512
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3572371291060976e-05,
+      "loss": 2.2166,
+      "step": 134520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.356763734128399e-05,
+      "loss": 2.2245,
+      "step": 134528
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.356290404861847e-05,
+      "loss": 2.2312,
+      "step": 134536
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3558171413182123e-05,
+      "loss": 2.2223,
+      "step": 134544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3553439435092625e-05,
+      "loss": 2.2209,
+      "step": 134552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.354870811446759e-05,
+      "loss": 2.2332,
+      "step": 134560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.354397745142464e-05,
+      "loss": 2.2188,
+      "step": 134568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3539247446081344e-05,
+      "loss": 2.226,
+      "step": 134576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.353451809855536e-05,
+      "loss": 2.2203,
+      "step": 134584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.352978940896422e-05,
+      "loss": 2.2349,
+      "step": 134592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.352506137742546e-05,
+      "loss": 2.2271,
+      "step": 134600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3520334004056676e-05,
+      "loss": 2.2331,
+      "step": 134608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3515607288975347e-05,
+      "loss": 2.2341,
+      "step": 134616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.351088123229899e-05,
+      "loss": 2.2164,
+      "step": 134624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.35061558341451e-05,
+      "loss": 2.2248,
+      "step": 134632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.350143109463116e-05,
+      "loss": 2.2171,
+      "step": 134640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3496707013874596e-05,
+      "loss": 2.2288,
+      "step": 134648
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.349198359199288e-05,
+      "loss": 2.2264,
+      "step": 134656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3487260829103426e-05,
+      "loss": 2.2254,
+      "step": 134664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.348253872532363e-05,
+      "loss": 2.2227,
+      "step": 134672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.347781728077092e-05,
+      "loss": 2.2285,
+      "step": 134680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3473096495562615e-05,
+      "loss": 2.2425,
+      "step": 134688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.346837636981614e-05,
+      "loss": 2.2355,
+      "step": 134696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.34636569036488e-05,
+      "loss": 2.2353,
+      "step": 134704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3458938097177904e-05,
+      "loss": 2.2277,
+      "step": 134712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3454219950520825e-05,
+      "loss": 2.2162,
+      "step": 134720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.344950246379478e-05,
+      "loss": 2.229,
+      "step": 134728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.344478563711709e-05,
+      "loss": 2.2235,
+      "step": 134736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.344006947060503e-05,
+      "loss": 2.231,
+      "step": 134744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3435353964375796e-05,
+      "loss": 2.2327,
+      "step": 134752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.343063911854665e-05,
+      "loss": 2.2291,
+      "step": 134760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.34259249332348e-05,
+      "loss": 2.2241,
+      "step": 134768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.342121140855744e-05,
+      "loss": 2.2321,
+      "step": 134776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3416498544631745e-05,
+      "loss": 2.2177,
+      "step": 134784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.341178634157487e-05,
+      "loss": 2.2218,
+      "step": 134792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.340707479950398e-05,
+      "loss": 2.2192,
+      "step": 134800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.340236391853619e-05,
+      "loss": 2.2318,
+      "step": 134808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3397653698788614e-05,
+      "loss": 2.2276,
+      "step": 134816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.339294414037835e-05,
+      "loss": 2.2335,
+      "step": 134824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3388235243422506e-05,
+      "loss": 2.2309,
+      "step": 134832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.338352700803808e-05,
+      "loss": 2.2316,
+      "step": 134840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.337881943434219e-05,
+      "loss": 2.2354,
+      "step": 134848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.337411252245183e-05,
+      "loss": 2.2329,
+      "step": 134856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3369406272484e-05,
+      "loss": 2.2272,
+      "step": 134864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.336470068455574e-05,
+      "loss": 2.2372,
+      "step": 134872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.335999575878401e-05,
+      "loss": 2.2237,
+      "step": 134880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3355291495285756e-05,
+      "loss": 2.2308,
+      "step": 134888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.335058789417796e-05,
+      "loss": 2.2284,
+      "step": 134896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.334588495557752e-05,
+      "loss": 2.2314,
+      "step": 134904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.334118267960139e-05,
+      "loss": 2.2383,
+      "step": 134912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.333648106636645e-05,
+      "loss": 2.2332,
+      "step": 134920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.333178011598957e-05,
+      "loss": 2.232,
+      "step": 134928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.332707982858765e-05,
+      "loss": 2.2432,
+      "step": 134936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.332238020427749e-05,
+      "loss": 2.2253,
+      "step": 134944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.331768124317596e-05,
+      "loss": 2.2336,
+      "step": 134952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.331298294539989e-05,
+      "loss": 2.2171,
+      "step": 134960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.330828531106603e-05,
+      "loss": 2.2312,
+      "step": 134968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3303588340291195e-05,
+      "loss": 2.2289,
+      "step": 134976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3298892033192164e-05,
+      "loss": 2.2346,
+      "step": 134984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.329419638988567e-05,
+      "loss": 2.2292,
+      "step": 134992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3289501410488446e-05,
+      "loss": 2.2298,
+      "step": 135000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.328480709511721e-05,
+      "loss": 2.2269,
+      "step": 135008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.328011344388867e-05,
+      "loss": 2.238,
+      "step": 135016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3275420456919514e-05,
+      "loss": 2.2217,
+      "step": 135024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.327072813432639e-05,
+      "loss": 2.2321,
+      "step": 135032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3266036476225973e-05,
+      "loss": 2.2286,
+      "step": 135040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3261345482734894e-05,
+      "loss": 2.2414,
+      "step": 135048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.325665515396975e-05,
+      "loss": 2.2348,
+      "step": 135056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.325196549004717e-05,
+      "loss": 2.2197,
+      "step": 135064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.324727649108373e-05,
+      "loss": 2.2365,
+      "step": 135072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.324258815719599e-05,
+      "loss": 2.2372,
+      "step": 135080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.323790048850052e-05,
+      "loss": 2.2426,
+      "step": 135088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.323321348511384e-05,
+      "loss": 2.222,
+      "step": 135096
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3228527147152475e-05,
+      "loss": 2.2363,
+      "step": 135104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.322384147473294e-05,
+      "loss": 2.2267,
+      "step": 135112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.32191564679717e-05,
+      "loss": 2.2157,
+      "step": 135120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.321447212698524e-05,
+      "loss": 2.2205,
+      "step": 135128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3209788451890006e-05,
+      "loss": 2.2367,
+      "step": 135136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.320510544280243e-05,
+      "loss": 2.233,
+      "step": 135144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3200423099838965e-05,
+      "loss": 2.2249,
+      "step": 135152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.319574142311595e-05,
+      "loss": 2.2242,
+      "step": 135160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.319106041274983e-05,
+      "loss": 2.2145,
+      "step": 135168
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.318638006885696e-05,
+      "loss": 2.2522,
+      "step": 135176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.318170039155369e-05,
+      "loss": 2.2263,
+      "step": 135184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.317702138095634e-05,
+      "loss": 2.2246,
+      "step": 135192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.317234303718127e-05,
+      "loss": 2.2236,
+      "step": 135200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.316766536034476e-05,
+      "loss": 2.2301,
+      "step": 135208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.316298835056309e-05,
+      "loss": 2.232,
+      "step": 135216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.315831200795254e-05,
+      "loss": 2.2255,
+      "step": 135224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3153636332629377e-05,
+      "loss": 2.2197,
+      "step": 135232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.314896132470982e-05,
+      "loss": 2.2379,
+      "step": 135240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.314428698431009e-05,
+      "loss": 2.2343,
+      "step": 135248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.31396133115464e-05,
+      "loss": 2.2256,
+      "step": 135256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.313494030653495e-05,
+      "loss": 2.2333,
+      "step": 135264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3130267969391884e-05,
+      "loss": 2.2201,
+      "step": 135272
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.312559630023337e-05,
+      "loss": 2.2196,
+      "step": 135280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3120925299175566e-05,
+      "loss": 2.2158,
+      "step": 135288
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.311625496633455e-05,
+      "loss": 2.2322,
+      "step": 135296
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.311158530182648e-05,
+      "loss": 2.2442,
+      "step": 135304
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.31069163057674e-05,
+      "loss": 2.2326,
+      "step": 135312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.310224797827338e-05,
+      "loss": 2.232,
+      "step": 135320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.309758031946052e-05,
+      "loss": 2.2294,
+      "step": 135328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.309291332944482e-05,
+      "loss": 2.2371,
+      "step": 135336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.308824700834232e-05,
+      "loss": 2.2385,
+      "step": 135344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3083581356269014e-05,
+      "loss": 2.2183,
+      "step": 135352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.307891637334089e-05,
+      "loss": 2.2303,
+      "step": 135360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.307425205967395e-05,
+      "loss": 2.2297,
+      "step": 135368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.30695884153841e-05,
+      "loss": 2.2271,
+      "step": 135376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.30649254405873e-05,
+      "loss": 2.2342,
+      "step": 135384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3060263135399495e-05,
+      "loss": 2.2367,
+      "step": 135392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3055601499936564e-05,
+      "loss": 2.2358,
+      "step": 135400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.30509405343144e-05,
+      "loss": 2.2344,
+      "step": 135408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.304628023864888e-05,
+      "loss": 2.2391,
+      "step": 135416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.304162061305586e-05,
+      "loss": 2.2252,
+      "step": 135424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3036961657651175e-05,
+      "loss": 2.2588,
+      "step": 135432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.303230337255064e-05,
+      "loss": 2.2377,
+      "step": 135440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.302764575787008e-05,
+      "loss": 2.2395,
+      "step": 135448
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3022988813725266e-05,
+      "loss": 2.2357,
+      "step": 135456
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.301833254023197e-05,
+      "loss": 2.2506,
+      "step": 135464
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3013676937505966e-05,
+      "loss": 2.2197,
+      "step": 135472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.300902200566298e-05,
+      "loss": 2.2134,
+      "step": 135480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.300436774481873e-05,
+      "loss": 2.222,
+      "step": 135488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2999714155088934e-05,
+      "loss": 2.2365,
+      "step": 135496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.299506123658928e-05,
+      "loss": 2.2365,
+      "step": 135504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2990408989435405e-05,
+      "loss": 2.2366,
+      "step": 135512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2985757413743015e-05,
+      "loss": 2.2227,
+      "step": 135520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.298110650962772e-05,
+      "loss": 2.2224,
+      "step": 135528
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.297645627720515e-05,
+      "loss": 2.2117,
+      "step": 135536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2971806716590915e-05,
+      "loss": 2.2226,
+      "step": 135544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.296715782790059e-05,
+      "loss": 2.2328,
+      "step": 135552
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.296250961124977e-05,
+      "loss": 2.2245,
+      "step": 135560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2957862066753976e-05,
+      "loss": 2.2182,
+      "step": 135568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.295321519452876e-05,
+      "loss": 2.2234,
+      "step": 135576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.294856899468968e-05,
+      "loss": 2.2009,
+      "step": 135584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2943923467352175e-05,
+      "loss": 2.2199,
+      "step": 135592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2939278612631776e-05,
+      "loss": 2.2205,
+      "step": 135600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.293463443064396e-05,
+      "loss": 2.2274,
+      "step": 135608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.292999092150416e-05,
+      "loss": 2.2203,
+      "step": 135616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2925348085327835e-05,
+      "loss": 2.2415,
+      "step": 135624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2920705922230376e-05,
+      "loss": 2.2379,
+      "step": 135632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.291606443232722e-05,
+      "loss": 2.2443,
+      "step": 135640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2911423615733736e-05,
+      "loss": 2.2115,
+      "step": 135648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2906783472565294e-05,
+      "loss": 2.2439,
+      "step": 135656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2902144002937265e-05,
+      "loss": 2.2246,
+      "step": 135664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2897505206964974e-05,
+      "loss": 2.2245,
+      "step": 135672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2892867084763733e-05,
+      "loss": 2.2242,
+      "step": 135680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2888229636448875e-05,
+      "loss": 2.2186,
+      "step": 135688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2883592862135656e-05,
+      "loss": 2.2303,
+      "step": 135696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.287895676193936e-05,
+      "loss": 2.2083,
+      "step": 135704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.287432133597526e-05,
+      "loss": 2.2318,
+      "step": 135712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.286968658435857e-05,
+      "loss": 2.2265,
+      "step": 135720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2865052507204495e-05,
+      "loss": 2.2238,
+      "step": 135728
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.286041910462828e-05,
+      "loss": 2.2513,
+      "step": 135736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2855786376745076e-05,
+      "loss": 2.2288,
+      "step": 135744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.285115432367008e-05,
+      "loss": 2.2319,
+      "step": 135752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2846522945518435e-05,
+      "loss": 2.2292,
+      "step": 135760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2841892242405264e-05,
+      "loss": 2.2282,
+      "step": 135768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2837262214445724e-05,
+      "loss": 2.2332,
+      "step": 135776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.283263286175486e-05,
+      "loss": 2.2451,
+      "step": 135784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.28280041844478e-05,
+      "loss": 2.2432,
+      "step": 135792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.282337618263963e-05,
+      "loss": 2.2242,
+      "step": 135800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.281874885644535e-05,
+      "loss": 2.2364,
+      "step": 135808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2814122205980024e-05,
+      "loss": 2.2249,
+      "step": 135816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.280949623135868e-05,
+      "loss": 2.2172,
+      "step": 135824
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2804870932696316e-05,
+      "loss": 2.2431,
+      "step": 135832
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.280024631010791e-05,
+      "loss": 2.2298,
+      "step": 135840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2795622363708414e-05,
+      "loss": 2.2489,
+      "step": 135848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.279099909361281e-05,
+      "loss": 2.2193,
+      "step": 135856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.278637649993603e-05,
+      "loss": 2.2297,
+      "step": 135864
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.278175458279297e-05,
+      "loss": 2.2329,
+      "step": 135872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2777133342298535e-05,
+      "loss": 2.2245,
+      "step": 135880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.277251277856765e-05,
+      "loss": 2.2495,
+      "step": 135888
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2767892891715135e-05,
+      "loss": 2.225,
+      "step": 135896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.276327368185586e-05,
+      "loss": 2.2323,
+      "step": 135904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.275865514910466e-05,
+      "loss": 2.2237,
+      "step": 135912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.275403729357634e-05,
+      "loss": 2.2236,
+      "step": 135920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.274942011538572e-05,
+      "loss": 2.2164,
+      "step": 135928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.274480361464758e-05,
+      "loss": 2.2198,
+      "step": 135936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.274018779147666e-05,
+      "loss": 2.2172,
+      "step": 135944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.273557264598774e-05,
+      "loss": 2.2491,
+      "step": 135952
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.273095817829554e-05,
+      "loss": 2.2262,
+      "step": 135960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.272634438851479e-05,
+      "loss": 2.2258,
+      "step": 135968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.272173127676017e-05,
+      "loss": 2.2409,
+      "step": 135976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2717118843146355e-05,
+      "loss": 2.2239,
+      "step": 135984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.271250708778806e-05,
+      "loss": 2.2223,
+      "step": 135992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2707896010799875e-05,
+      "loss": 2.2271,
+      "step": 136000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.270328561229646e-05,
+      "loss": 2.2282,
+      "step": 136008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.269867589239245e-05,
+      "loss": 2.238,
+      "step": 136016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.26940668512024e-05,
+      "loss": 2.2184,
+      "step": 136024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2689458488840906e-05,
+      "loss": 2.2347,
+      "step": 136032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.268485080542255e-05,
+      "loss": 2.2129,
+      "step": 136040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.268024380106187e-05,
+      "loss": 2.2517,
+      "step": 136048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.267563747587339e-05,
+      "loss": 2.2432,
+      "step": 136056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.267103182997162e-05,
+      "loss": 2.2316,
+      "step": 136064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.266642686347108e-05,
+      "loss": 2.2356,
+      "step": 136072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.266182257648623e-05,
+      "loss": 2.2231,
+      "step": 136080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.265721896913153e-05,
+      "loss": 2.2135,
+      "step": 136088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2652616041521443e-05,
+      "loss": 2.2154,
+      "step": 136096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.264801379377039e-05,
+      "loss": 2.2189,
+      "step": 136104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.264341222599277e-05,
+      "loss": 2.2229,
+      "step": 136112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.263881133830301e-05,
+      "loss": 2.2044,
+      "step": 136120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.263421113081546e-05,
+      "loss": 2.2225,
+      "step": 136128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.262961160364448e-05,
+      "loss": 2.2263,
+      "step": 136136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.262501275690445e-05,
+      "loss": 2.2298,
+      "step": 136144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2620414590709666e-05,
+      "loss": 2.2153,
+      "step": 136152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2615817105174435e-05,
+      "loss": 2.2334,
+      "step": 136160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.261122030041308e-05,
+      "loss": 2.2389,
+      "step": 136168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.260662417653984e-05,
+      "loss": 2.2311,
+      "step": 136176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.260202873366901e-05,
+      "loss": 2.2534,
+      "step": 136184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.259743397191483e-05,
+      "loss": 2.2443,
+      "step": 136192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.259283989139149e-05,
+      "loss": 2.2208,
+      "step": 136200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2588246492213265e-05,
+      "loss": 2.2308,
+      "step": 136208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2583653774494275e-05,
+      "loss": 2.2614,
+      "step": 136216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.257906173834873e-05,
+      "loss": 2.2248,
+      "step": 136224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.25744703838908e-05,
+      "loss": 2.2401,
+      "step": 136232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.256987971123461e-05,
+      "loss": 2.2286,
+      "step": 136240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.256528972049428e-05,
+      "loss": 2.2349,
+      "step": 136248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.256070041178393e-05,
+      "loss": 2.2466,
+      "step": 136256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.255611178521765e-05,
+      "loss": 2.2399,
+      "step": 136264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2551523840909514e-05,
+      "loss": 2.2233,
+      "step": 136272
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2546936578973566e-05,
+      "loss": 2.2301,
+      "step": 136280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.254234999952386e-05,
+      "loss": 2.231,
+      "step": 136288
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.253776410267441e-05,
+      "loss": 2.2268,
+      "step": 136296
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.253317888853922e-05,
+      "loss": 2.2274,
+      "step": 136304
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.252859435723229e-05,
+      "loss": 2.2399,
+      "step": 136312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2524010508867596e-05,
+      "loss": 2.2143,
+      "step": 136320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.251942734355907e-05,
+      "loss": 2.2301,
+      "step": 136328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2514844861420665e-05,
+      "loss": 2.2208,
+      "step": 136336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.251026306256631e-05,
+      "loss": 2.2178,
+      "step": 136344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.250568194710989e-05,
+      "loss": 2.2355,
+      "step": 136352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2501101515165305e-05,
+      "loss": 2.2462,
+      "step": 136360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2496521766846426e-05,
+      "loss": 2.2158,
+      "step": 136368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.249194270226709e-05,
+      "loss": 2.222,
+      "step": 136376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.248736432154116e-05,
+      "loss": 2.2305,
+      "step": 136384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.248278662478242e-05,
+      "loss": 2.2202,
+      "step": 136392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.247820961210471e-05,
+      "loss": 2.2398,
+      "step": 136400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.24736332836218e-05,
+      "loss": 2.2273,
+      "step": 136408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.246905763944743e-05,
+      "loss": 2.2312,
+      "step": 136416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.246448267969541e-05,
+      "loss": 2.2385,
+      "step": 136424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2459908404479414e-05,
+      "loss": 2.2134,
+      "step": 136432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.245533481391319e-05,
+      "loss": 2.2274,
+      "step": 136440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.245076190811043e-05,
+      "loss": 2.2497,
+      "step": 136448
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.244618968718484e-05,
+      "loss": 2.2388,
+      "step": 136456
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.244161815125005e-05,
+      "loss": 2.2322,
+      "step": 136464
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.243704730041972e-05,
+      "loss": 2.2235,
+      "step": 136472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.24324771348075e-05,
+      "loss": 2.2306,
+      "step": 136480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2427907654526984e-05,
+      "loss": 2.2164,
+      "step": 136488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.242333885969177e-05,
+      "loss": 2.2344,
+      "step": 136496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.241877075041546e-05,
+      "loss": 2.2201,
+      "step": 136504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2414203326811596e-05,
+      "loss": 2.2365,
+      "step": 136512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.240963658899373e-05,
+      "loss": 2.2244,
+      "step": 136520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.24050705370754e-05,
+      "loss": 2.2337,
+      "step": 136528
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2400505171170115e-05,
+      "loss": 2.2209,
+      "step": 136536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.239594049139136e-05,
+      "loss": 2.2319,
+      "step": 136544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2391376497852624e-05,
+      "loss": 2.2064,
+      "step": 136552
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.238681319066737e-05,
+      "loss": 2.2261,
+      "step": 136560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.238225056994903e-05,
+      "loss": 2.2423,
+      "step": 136568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2377688635811045e-05,
+      "loss": 2.225,
+      "step": 136576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.237312738836681e-05,
+      "loss": 2.2025,
+      "step": 136584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.236856682772974e-05,
+      "loss": 2.2462,
+      "step": 136592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.23640069540132e-05,
+      "loss": 2.2293,
+      "step": 136600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.235944776733054e-05,
+      "loss": 2.2418,
+      "step": 136608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.235488926779513e-05,
+      "loss": 2.2201,
+      "step": 136616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2350331455520244e-05,
+      "loss": 2.222,
+      "step": 136624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2345774330619226e-05,
+      "loss": 2.225,
+      "step": 136632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.234121789320539e-05,
+      "loss": 2.2349,
+      "step": 136640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.233666214339195e-05,
+      "loss": 2.2218,
+      "step": 136648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.233210708129219e-05,
+      "loss": 2.2234,
+      "step": 136656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.232755270701937e-05,
+      "loss": 2.2371,
+      "step": 136664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.232299902068669e-05,
+      "loss": 2.2424,
+      "step": 136672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.231844602240737e-05,
+      "loss": 2.2264,
+      "step": 136680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2313893712294566e-05,
+      "loss": 2.2275,
+      "step": 136688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.230934209046148e-05,
+      "loss": 2.2289,
+      "step": 136696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.230479115702127e-05,
+      "loss": 2.2274,
+      "step": 136704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.230024091208703e-05,
+      "loss": 2.2245,
+      "step": 136712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2295691355771936e-05,
+      "loss": 2.2354,
+      "step": 136720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.229114248818905e-05,
+      "loss": 2.2305,
+      "step": 136728
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.228659430945147e-05,
+      "loss": 2.2283,
+      "step": 136736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.228204681967227e-05,
+      "loss": 2.2337,
+      "step": 136744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.22775000189645e-05,
+      "loss": 2.2239,
+      "step": 136752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.227295390744118e-05,
+      "loss": 2.2501,
+      "step": 136760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.226840848521534e-05,
+      "loss": 2.223,
+      "step": 136768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.226386375239998e-05,
+      "loss": 2.2171,
+      "step": 136776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.225931970910807e-05,
+      "loss": 2.2172,
+      "step": 136784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.22547763554526e-05,
+      "loss": 2.2138,
+      "step": 136792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.225023369154648e-05,
+      "loss": 2.2266,
+      "step": 136800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.224569171750268e-05,
+      "loss": 2.2445,
+      "step": 136808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.22411504334341e-05,
+      "loss": 2.2206,
+      "step": 136816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.223660983945362e-05,
+      "loss": 2.2313,
+      "step": 136824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.223206993567416e-05,
+      "loss": 2.2266,
+      "step": 136832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.222753072220854e-05,
+      "loss": 2.2228,
+      "step": 136840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.222299219916961e-05,
+      "loss": 2.2432,
+      "step": 136848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.221845436667024e-05,
+      "loss": 2.2415,
+      "step": 136856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2213917224823185e-05,
+      "loss": 2.2286,
+      "step": 136864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.220938077374127e-05,
+      "loss": 2.2291,
+      "step": 136872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.220484501353728e-05,
+      "loss": 2.227,
+      "step": 136880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.220030994432395e-05,
+      "loss": 2.2401,
+      "step": 136888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.219577556621405e-05,
+      "loss": 2.2203,
+      "step": 136896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.219124187932027e-05,
+      "loss": 2.2349,
+      "step": 136904
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2186708883755346e-05,
+      "loss": 2.2374,
+      "step": 136912
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2182176579631963e-05,
+      "loss": 2.227,
+      "step": 136920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2177644967062776e-05,
+      "loss": 2.2403,
+      "step": 136928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.217311404616047e-05,
+      "loss": 2.2277,
+      "step": 136936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.216858381703767e-05,
+      "loss": 2.2277,
+      "step": 136944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2164054279806994e-05,
+      "loss": 2.2077,
+      "step": 136952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.215952543458105e-05,
+      "loss": 2.2205,
+      "step": 136960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.215499728147244e-05,
+      "loss": 2.2099,
+      "step": 136968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.215046982059371e-05,
+      "loss": 2.2264,
+      "step": 136976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.214594305205742e-05,
+      "loss": 2.2264,
+      "step": 136984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.214141697597613e-05,
+      "loss": 2.2329,
+      "step": 136992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.213689159246232e-05,
+      "loss": 2.2378,
+      "step": 137000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.213236690162853e-05,
+      "loss": 2.2232,
+      "step": 137008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2127842903587205e-05,
+      "loss": 2.2307,
+      "step": 137016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2123319598450844e-05,
+      "loss": 2.2489,
+      "step": 137024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.211879698633189e-05,
+      "loss": 2.244,
+      "step": 137032
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.211427506734276e-05,
+      "loss": 2.2403,
+      "step": 137040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2109753841595904e-05,
+      "loss": 2.2144,
+      "step": 137048
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.210523330920367e-05,
+      "loss": 2.2196,
+      "step": 137056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2100713470278464e-05,
+      "loss": 2.2447,
+      "step": 137064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.209619432493267e-05,
+      "loss": 2.2285,
+      "step": 137072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.209167587327861e-05,
+      "loss": 2.2394,
+      "step": 137080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.208715811542862e-05,
+      "loss": 2.227,
+      "step": 137088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2082641051495e-05,
+      "loss": 2.2501,
+      "step": 137096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2078124681590065e-05,
+      "loss": 2.2443,
+      "step": 137104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.207360900582609e-05,
+      "loss": 2.2309,
+      "step": 137112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.206909402431531e-05,
+      "loss": 2.2372,
+      "step": 137120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.206457973717e-05,
+      "loss": 2.2419,
+      "step": 137128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.206006614450238e-05,
+      "loss": 2.2217,
+      "step": 137136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.205555324642462e-05,
+      "loss": 2.2065,
+      "step": 137144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2051041043048966e-05,
+      "loss": 2.232,
+      "step": 137152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2046529534487566e-05,
+      "loss": 2.2126,
+      "step": 137160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2042018720852575e-05,
+      "loss": 2.234,
+      "step": 137168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.203750860225614e-05,
+      "loss": 2.2478,
+      "step": 137176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.203299917881038e-05,
+      "loss": 2.2221,
+      "step": 137184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2028490450627375e-05,
+      "loss": 2.2274,
+      "step": 137192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.202398241781926e-05,
+      "loss": 2.2199,
+      "step": 137200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.201947508049807e-05,
+      "loss": 2.2324,
+      "step": 137208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.201496843877587e-05,
+      "loss": 2.2252,
+      "step": 137216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.20104624927647e-05,
+      "loss": 2.2434,
+      "step": 137224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2005957242576555e-05,
+      "loss": 2.2372,
+      "step": 137232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.200145268832348e-05,
+      "loss": 2.2297,
+      "step": 137240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1996948830117425e-05,
+      "loss": 2.2299,
+      "step": 137248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.199244566807035e-05,
+      "loss": 2.2257,
+      "step": 137256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.198794320229424e-05,
+      "loss": 2.224,
+      "step": 137264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1983441432900984e-05,
+      "loss": 2.2436,
+      "step": 137272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.197894036000252e-05,
+      "loss": 2.2256,
+      "step": 137280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.197443998371075e-05,
+      "loss": 2.224,
+      "step": 137288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.196994030413755e-05,
+      "loss": 2.2342,
+      "step": 137296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1965441321394776e-05,
+      "loss": 2.2381,
+      "step": 137304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.196094303559427e-05,
+      "loss": 2.2325,
+      "step": 137312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1956445446847875e-05,
+      "loss": 2.2302,
+      "step": 137320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.195194855526739e-05,
+      "loss": 2.225,
+      "step": 137328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.19474523609646e-05,
+      "loss": 2.2359,
+      "step": 137336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1942956864051305e-05,
+      "loss": 2.2382,
+      "step": 137344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1938462064639255e-05,
+      "loss": 2.231,
+      "step": 137352
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.193396796284016e-05,
+      "loss": 2.2291,
+      "step": 137360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.192947455876579e-05,
+      "loss": 2.2365,
+      "step": 137368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.192498185252784e-05,
+      "loss": 2.2143,
+      "step": 137376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1920489844237966e-05,
+      "loss": 2.225,
+      "step": 137384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.191599853400788e-05,
+      "loss": 2.2208,
+      "step": 137392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.191150792194924e-05,
+      "loss": 2.2417,
+      "step": 137400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.190701800817363e-05,
+      "loss": 2.2419,
+      "step": 137408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.190252879279272e-05,
+      "loss": 2.2329,
+      "step": 137416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.18980402759181e-05,
+      "loss": 2.2378,
+      "step": 137424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1893552457661345e-05,
+      "loss": 2.2333,
+      "step": 137432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.188906533813403e-05,
+      "loss": 2.2293,
+      "step": 137440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.188457891744771e-05,
+      "loss": 2.2464,
+      "step": 137448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.188009319571392e-05,
+      "loss": 2.2307,
+      "step": 137456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1875608173044154e-05,
+      "loss": 2.227,
+      "step": 137464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.187112384954992e-05,
+      "loss": 2.226,
+      "step": 137472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.186664022534273e-05,
+      "loss": 2.2201,
+      "step": 137480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.186215730053399e-05,
+      "loss": 2.2197,
+      "step": 137488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.185767507523518e-05,
+      "loss": 2.223,
+      "step": 137496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.185319354955774e-05,
+      "loss": 2.2189,
+      "step": 137504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1848712723613056e-05,
+      "loss": 2.2397,
+      "step": 137512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.184423259751255e-05,
+      "loss": 2.2377,
+      "step": 137520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1839753171367557e-05,
+      "loss": 2.2393,
+      "step": 137528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.183527444528947e-05,
+      "loss": 2.2477,
+      "step": 137536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1830796419389617e-05,
+      "loss": 2.2356,
+      "step": 137544
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.182631909377932e-05,
+      "loss": 2.2219,
+      "step": 137552
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.182184246856991e-05,
+      "loss": 2.226,
+      "step": 137560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.181736654387265e-05,
+      "loss": 2.2302,
+      "step": 137568
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.181289131979881e-05,
+      "loss": 2.2142,
+      "step": 137576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.180841679645966e-05,
+      "loss": 2.2342,
+      "step": 137584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1803942973966435e-05,
+      "loss": 2.2279,
+      "step": 137592
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1799469852430344e-05,
+      "loss": 2.2214,
+      "step": 137600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1794997431962607e-05,
+      "loss": 2.2347,
+      "step": 137608
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1790525712674394e-05,
+      "loss": 2.2224,
+      "step": 137616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.178605469467687e-05,
+      "loss": 2.2294,
+      "step": 137624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.17815843780812e-05,
+      "loss": 2.2347,
+      "step": 137632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.17771147629985e-05,
+      "loss": 2.2361,
+      "step": 137640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.17726458495399e-05,
+      "loss": 2.2188,
+      "step": 137648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1768177637816496e-05,
+      "loss": 2.227,
+      "step": 137656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.176371012793935e-05,
+      "loss": 2.2219,
+      "step": 137664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.175924332001956e-05,
+      "loss": 2.2379,
+      "step": 137672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.175477721416813e-05,
+      "loss": 2.2292,
+      "step": 137680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.175031181049611e-05,
+      "loss": 2.2507,
+      "step": 137688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1745847109114526e-05,
+      "loss": 2.2528,
+      "step": 137696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.174138311013433e-05,
+      "loss": 2.2266,
+      "step": 137704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.173691981366652e-05,
+      "loss": 2.2268,
+      "step": 137712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.173245721982207e-05,
+      "loss": 2.2338,
+      "step": 137720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.17279953287119e-05,
+      "loss": 2.2317,
+      "step": 137728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.172353414044694e-05,
+      "loss": 2.239,
+      "step": 137736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.171907365513808e-05,
+      "loss": 2.2237,
+      "step": 137744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.171461387289624e-05,
+      "loss": 2.2255,
+      "step": 137752
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.171015479383227e-05,
+      "loss": 2.2267,
+      "step": 137760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.170569641805701e-05,
+      "loss": 2.2438,
+      "step": 137768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1701238745681326e-05,
+      "loss": 2.2218,
+      "step": 137776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.169678177681602e-05,
+      "loss": 2.2253,
+      "step": 137784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.169232551157189e-05,
+      "loss": 2.2209,
+      "step": 137792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.168786995005973e-05,
+      "loss": 2.2439,
+      "step": 137800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.16834150923903e-05,
+      "loss": 2.2179,
+      "step": 137808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.167896093867433e-05,
+      "loss": 2.2375,
+      "step": 137816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1674507489022596e-05,
+      "loss": 2.2259,
+      "step": 137824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1670054743545776e-05,
+      "loss": 2.2346,
+      "step": 137832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.166560270235456e-05,
+      "loss": 2.2283,
+      "step": 137840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.166115136555965e-05,
+      "loss": 2.2209,
+      "step": 137848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.165670073327169e-05,
+      "loss": 2.2427,
+      "step": 137856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.165225080560134e-05,
+      "loss": 2.2228,
+      "step": 137864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.164780158265921e-05,
+      "loss": 2.2118,
+      "step": 137872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1643353064555905e-05,
+      "loss": 2.2284,
+      "step": 137880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.163890525140205e-05,
+      "loss": 2.2287,
+      "step": 137888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1634458143308165e-05,
+      "loss": 2.2247,
+      "step": 137896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.163001174038484e-05,
+      "loss": 2.2247,
+      "step": 137904
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.162556604274263e-05,
+      "loss": 2.2289,
+      "step": 137912
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1621121050492e-05,
+      "loss": 2.2344,
+      "step": 137920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1616676763743485e-05,
+      "loss": 2.2117,
+      "step": 137928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.161223318260759e-05,
+      "loss": 2.2135,
+      "step": 137936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.160779030719476e-05,
+      "loss": 2.245,
+      "step": 137944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1603348137615445e-05,
+      "loss": 2.2132,
+      "step": 137952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.159890667398007e-05,
+      "loss": 2.2247,
+      "step": 137960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.159446591639907e-05,
+      "loss": 2.2172,
+      "step": 137968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1590025864982845e-05,
+      "loss": 2.2319,
+      "step": 137976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1585586519841744e-05,
+      "loss": 2.2317,
+      "step": 137984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.158114788108617e-05,
+      "loss": 2.235,
+      "step": 137992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.157670994882644e-05,
+      "loss": 2.2323,
+      "step": 138000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1572272723172884e-05,
+      "loss": 2.2264,
+      "step": 138008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.156783620423582e-05,
+      "loss": 2.2173,
+      "step": 138016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.156340039212555e-05,
+      "loss": 2.2307,
+      "step": 138024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1558965286952324e-05,
+      "loss": 2.2371,
+      "step": 138032
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.155453088882642e-05,
+      "loss": 2.2261,
+      "step": 138040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.155009719785807e-05,
+      "loss": 2.232,
+      "step": 138048
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1545664214157475e-05,
+      "loss": 2.2153,
+      "step": 138056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1541231937834886e-05,
+      "loss": 2.2331,
+      "step": 138064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.153680036900044e-05,
+      "loss": 2.2282,
+      "step": 138072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1532369507764354e-05,
+      "loss": 2.2413,
+      "step": 138080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.152793935423675e-05,
+      "loss": 2.2162,
+      "step": 138088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.152350990852775e-05,
+      "loss": 2.239,
+      "step": 138096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1519081170747526e-05,
+      "loss": 2.2196,
+      "step": 138104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1514653141006106e-05,
+      "loss": 2.2288,
+      "step": 138112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1510225819413605e-05,
+      "loss": 2.2301,
+      "step": 138120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1505799206080104e-05,
+      "loss": 2.2211,
+      "step": 138128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1501373301115634e-05,
+      "loss": 2.2217,
+      "step": 138136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.149694810463022e-05,
+      "loss": 2.2254,
+      "step": 138144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.149252361673385e-05,
+      "loss": 2.2161,
+      "step": 138152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.148809983753657e-05,
+      "loss": 2.212,
+      "step": 138160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.148367676714832e-05,
+      "loss": 2.2304,
+      "step": 138168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.147925440567905e-05,
+      "loss": 2.2322,
+      "step": 138176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1474832753238726e-05,
+      "loss": 2.2278,
+      "step": 138184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.147041180993726e-05,
+      "loss": 2.2219,
+      "step": 138192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.146599157588455e-05,
+      "loss": 2.228,
+      "step": 138200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1461572051190504e-05,
+      "loss": 2.2236,
+      "step": 138208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1457153235964974e-05,
+      "loss": 2.2334,
+      "step": 138216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1452735130317815e-05,
+      "loss": 2.2224,
+      "step": 138224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1448317734358864e-05,
+      "loss": 2.2239,
+      "step": 138232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1443901048197945e-05,
+      "loss": 2.238,
+      "step": 138240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.143948507194483e-05,
+      "loss": 2.228,
+      "step": 138248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1435069805709335e-05,
+      "loss": 2.2306,
+      "step": 138256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1430655249601217e-05,
+      "loss": 2.2374,
+      "step": 138264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.14262414037302e-05,
+      "loss": 2.2188,
+      "step": 138272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.142182826820603e-05,
+      "loss": 2.215,
+      "step": 138280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1417415843138415e-05,
+      "loss": 2.2345,
+      "step": 138288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.141300412863706e-05,
+      "loss": 2.2296,
+      "step": 138296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1408593124811616e-05,
+      "loss": 2.2299,
+      "step": 138304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1404182831771755e-05,
+      "loss": 2.2302,
+      "step": 138312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1399773249627145e-05,
+      "loss": 2.2305,
+      "step": 138320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.139536437848735e-05,
+      "loss": 2.2212,
+      "step": 138328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1390956218462006e-05,
+      "loss": 2.2193,
+      "step": 138336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1386548769660716e-05,
+      "loss": 2.2295,
+      "step": 138344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.138214203219303e-05,
+      "loss": 2.2301,
+      "step": 138352
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.137773600616851e-05,
+      "loss": 2.2276,
+      "step": 138360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.137333069169666e-05,
+      "loss": 2.2147,
+      "step": 138368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.136892608888704e-05,
+      "loss": 2.2047,
+      "step": 138376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.136452219784913e-05,
+      "loss": 2.2269,
+      "step": 138384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.13601190186924e-05,
+      "loss": 2.2225,
+      "step": 138392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.135571655152634e-05,
+      "loss": 2.2321,
+      "step": 138400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.135131479646038e-05,
+      "loss": 2.2323,
+      "step": 138408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.134691375360393e-05,
+      "loss": 2.2215,
+      "step": 138416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.134251342306643e-05,
+      "loss": 2.2275,
+      "step": 138424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.133811380495727e-05,
+      "loss": 2.2078,
+      "step": 138432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.13337148993858e-05,
+      "loss": 2.2343,
+      "step": 138440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.13293167064614e-05,
+      "loss": 2.2287,
+      "step": 138448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.132491922629341e-05,
+      "loss": 2.1978,
+      "step": 138456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1320522458991135e-05,
+      "loss": 2.2234,
+      "step": 138464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1316126404663907e-05,
+      "loss": 2.2243,
+      "step": 138472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.131173106342097e-05,
+      "loss": 2.2392,
+      "step": 138480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.130733643537163e-05,
+      "loss": 2.2194,
+      "step": 138488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.130294252062513e-05,
+      "loss": 2.2361,
+      "step": 138496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.129854931929069e-05,
+      "loss": 2.226,
+      "step": 138504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1294156831477555e-05,
+      "loss": 2.2204,
+      "step": 138512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.128976505729488e-05,
+      "loss": 2.2214,
+      "step": 138520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.128537399685186e-05,
+      "loss": 2.2291,
+      "step": 138528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1280983650257694e-05,
+      "loss": 2.208,
+      "step": 138536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.127659401762147e-05,
+      "loss": 2.2167,
+      "step": 138544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.127220509905234e-05,
+      "loss": 2.2215,
+      "step": 138552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.126781689465944e-05,
+      "loss": 2.2222,
+      "step": 138560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.126342940455182e-05,
+      "loss": 2.2318,
+      "step": 138568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1259042628838576e-05,
+      "loss": 2.2245,
+      "step": 138576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.125465656762874e-05,
+      "loss": 2.2468,
+      "step": 138584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1250271221031386e-05,
+      "loss": 2.2233,
+      "step": 138592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.12458865891555e-05,
+      "loss": 2.2208,
+      "step": 138600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.124150267211011e-05,
+      "loss": 2.2436,
+      "step": 138608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.123711947000418e-05,
+      "loss": 2.2371,
+      "step": 138616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.123273698294671e-05,
+      "loss": 2.2178,
+      "step": 138624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.12283552110466e-05,
+      "loss": 2.2105,
+      "step": 138632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.122397415441282e-05,
+      "loss": 2.2274,
+      "step": 138640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.121959381315427e-05,
+      "loss": 2.2289,
+      "step": 138648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.121521418737984e-05,
+      "loss": 2.2298,
+      "step": 138656
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.121083527719842e-05,
+      "loss": 2.2262,
+      "step": 138664
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1206457082718875e-05,
+      "loss": 2.2173,
+      "step": 138672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.120207960405001e-05,
+      "loss": 2.2196,
+      "step": 138680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.119770284130071e-05,
+      "loss": 2.2174,
+      "step": 138688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.119332679457972e-05,
+      "loss": 2.2221,
+      "step": 138696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.118895146399588e-05,
+      "loss": 2.2296,
+      "step": 138704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.118457684965794e-05,
+      "loss": 2.2294,
+      "step": 138712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1180202951674634e-05,
+      "loss": 2.2418,
+      "step": 138720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.117582977015474e-05,
+      "loss": 2.2334,
+      "step": 138728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.117145730520693e-05,
+      "loss": 2.2139,
+      "step": 138736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1167085556939925e-05,
+      "loss": 2.2177,
+      "step": 138744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1162714525462436e-05,
+      "loss": 2.2226,
+      "step": 138752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1158344210883065e-05,
+      "loss": 2.2161,
+      "step": 138760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1153974613310505e-05,
+      "loss": 2.2218,
+      "step": 138768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.114960573285338e-05,
+      "loss": 2.2369,
+      "step": 138776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.114523756962029e-05,
+      "loss": 2.2177,
+      "step": 138784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.114087012371984e-05,
+      "loss": 2.2319,
+      "step": 138792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.113650339526058e-05,
+      "loss": 2.2339,
+      "step": 138800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.113213738435109e-05,
+      "loss": 2.2315,
+      "step": 138808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.112777209109991e-05,
+      "loss": 2.231,
+      "step": 138816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1123407515615545e-05,
+      "loss": 2.2089,
+      "step": 138824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1119043658006516e-05,
+      "loss": 2.217,
+      "step": 138832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.11146805183813e-05,
+      "loss": 2.2187,
+      "step": 138840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.111031809684836e-05,
+      "loss": 2.2219,
+      "step": 138848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.110595639351617e-05,
+      "loss": 2.2027,
+      "step": 138856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.110159540849313e-05,
+      "loss": 2.2357,
+      "step": 138864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1097235141887675e-05,
+      "loss": 2.2173,
+      "step": 138872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.10928755938082e-05,
+      "loss": 2.2135,
+      "step": 138880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.108851676436308e-05,
+      "loss": 2.2176,
+      "step": 138888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.108415865366068e-05,
+      "loss": 2.2216,
+      "step": 138896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1079801261809335e-05,
+      "loss": 2.2302,
+      "step": 138904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1075444588917374e-05,
+      "loss": 2.1962,
+      "step": 138912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1071088635093115e-05,
+      "loss": 2.2367,
+      "step": 138920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1066733400444837e-05,
+      "loss": 2.2049,
+      "step": 138928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.10623788850808e-05,
+      "loss": 2.2169,
+      "step": 138936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.105802508910929e-05,
+      "loss": 2.2304,
+      "step": 138944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.105367201263851e-05,
+      "loss": 2.2174,
+      "step": 138952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.104931965577669e-05,
+      "loss": 2.2107,
+      "step": 138960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.104496801863205e-05,
+      "loss": 2.2171,
+      "step": 138968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.104061710131273e-05,
+      "loss": 2.2387,
+      "step": 138976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.103626690392694e-05,
+      "loss": 2.218,
+      "step": 138984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.103191742658278e-05,
+      "loss": 2.212,
+      "step": 138992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.102756866938842e-05,
+      "loss": 2.2239,
+      "step": 139000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1023220632451956e-05,
+      "loss": 2.23,
+      "step": 139008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.101887331588146e-05,
+      "loss": 2.2243,
+      "step": 139016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1014526719785035e-05,
+      "loss": 2.2242,
+      "step": 139024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1010180844270734e-05,
+      "loss": 2.2263,
+      "step": 139032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.100583568944658e-05,
+      "loss": 2.2303,
+      "step": 139040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.100149125542061e-05,
+      "loss": 2.2034,
+      "step": 139048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.099714754230083e-05,
+      "loss": 2.2174,
+      "step": 139056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0992804550195194e-05,
+      "loss": 2.2031,
+      "step": 139064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0988462279211705e-05,
+      "loss": 2.2166,
+      "step": 139072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.09841207294583e-05,
+      "loss": 2.219,
+      "step": 139080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.09797799010429e-05,
+      "loss": 2.2025,
+      "step": 139088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0975439794073434e-05,
+      "loss": 2.2312,
+      "step": 139096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.09711004086578e-05,
+      "loss": 2.2304,
+      "step": 139104
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.096676174490385e-05,
+      "loss": 2.2231,
+      "step": 139112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0962423802919475e-05,
+      "loss": 2.214,
+      "step": 139120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.095808658281248e-05,
+      "loss": 2.231,
+      "step": 139128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.095375008469073e-05,
+      "loss": 2.2088,
+      "step": 139136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.094941430866201e-05,
+      "loss": 2.2199,
+      "step": 139144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.09450792548341e-05,
+      "loss": 2.2256,
+      "step": 139152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.094074492331481e-05,
+      "loss": 2.2266,
+      "step": 139160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.093641131421183e-05,
+      "loss": 2.225,
+      "step": 139168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0932078427632926e-05,
+      "loss": 2.2125,
+      "step": 139176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.092774626368583e-05,
+      "loss": 2.2193,
+      "step": 139184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0923414822478215e-05,
+      "loss": 2.2309,
+      "step": 139192
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0919084104117774e-05,
+      "loss": 2.218,
+      "step": 139200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.091475410871216e-05,
+      "loss": 2.2301,
+      "step": 139208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.091042483636902e-05,
+      "loss": 2.2271,
+      "step": 139216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0906096287195995e-05,
+      "loss": 2.2301,
+      "step": 139224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0901768461300664e-05,
+      "loss": 2.2322,
+      "step": 139232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0897441358790646e-05,
+      "loss": 2.2384,
+      "step": 139240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0893114979773495e-05,
+      "loss": 2.2192,
+      "step": 139248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.088878932435677e-05,
+      "loss": 2.2349,
+      "step": 139256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.088446439264802e-05,
+      "loss": 2.2119,
+      "step": 139264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0880140184754744e-05,
+      "loss": 2.228,
+      "step": 139272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.087581670078444e-05,
+      "loss": 2.2304,
+      "step": 139280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0871493940844616e-05,
+      "loss": 2.23,
+      "step": 139288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.086717190504272e-05,
+      "loss": 2.2373,
+      "step": 139296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0862850593486186e-05,
+      "loss": 2.2273,
+      "step": 139304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.085853000628247e-05,
+      "loss": 2.2246,
+      "step": 139312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0854210143538956e-05,
+      "loss": 2.2168,
+      "step": 139320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.084989100536305e-05,
+      "loss": 2.2429,
+      "step": 139328
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.084557259186213e-05,
+      "loss": 2.2195,
+      "step": 139336
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.084125490314353e-05,
+      "loss": 2.2152,
+      "step": 139344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.083693793931464e-05,
+      "loss": 2.2312,
+      "step": 139352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.083262170048272e-05,
+      "loss": 2.2232,
+      "step": 139360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.082830618675509e-05,
+      "loss": 2.2249,
+      "step": 139368
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.082399139823906e-05,
+      "loss": 2.2265,
+      "step": 139376
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0819677335041866e-05,
+      "loss": 2.2293,
+      "step": 139384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.081536399727076e-05,
+      "loss": 2.2333,
+      "step": 139392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0811051385032995e-05,
+      "loss": 2.2181,
+      "step": 139400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.080673949843577e-05,
+      "loss": 2.2347,
+      "step": 139408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.080242833758627e-05,
+      "loss": 2.2412,
+      "step": 139416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0798117902591674e-05,
+      "loss": 2.2258,
+      "step": 139424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.079380819355915e-05,
+      "loss": 2.2158,
+      "step": 139432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.078949921059583e-05,
+      "loss": 2.2304,
+      "step": 139440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.078519095380883e-05,
+      "loss": 2.2314,
+      "step": 139448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.078088342330526e-05,
+      "loss": 2.2456,
+      "step": 139456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.077657661919222e-05,
+      "loss": 2.2398,
+      "step": 139464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.077227054157674e-05,
+      "loss": 2.2402,
+      "step": 139472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.076796519056592e-05,
+      "loss": 2.2215,
+      "step": 139480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0763660566266755e-05,
+      "loss": 2.2304,
+      "step": 139488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.075935666878626e-05,
+      "loss": 2.2345,
+      "step": 139496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0755053498231445e-05,
+      "loss": 2.2181,
+      "step": 139504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0750751054709285e-05,
+      "loss": 2.2316,
+      "step": 139512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.074644933832673e-05,
+      "loss": 2.231,
+      "step": 139520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.074214834919074e-05,
+      "loss": 2.2537,
+      "step": 139528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0737848087408197e-05,
+      "loss": 2.2423,
+      "step": 139536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.073354855308606e-05,
+      "loss": 2.2456,
+      "step": 139544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.072924974633118e-05,
+      "loss": 2.2117,
+      "step": 139552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.072495166725043e-05,
+      "loss": 2.2248,
+      "step": 139560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.072065431595069e-05,
+      "loss": 2.2402,
+      "step": 139568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.071635769253874e-05,
+      "loss": 2.2295,
+      "step": 139576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.071206179712143e-05,
+      "loss": 2.2335,
+      "step": 139584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0707766629805566e-05,
+      "loss": 2.2225,
+      "step": 139592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.070347219069789e-05,
+      "loss": 2.2297,
+      "step": 139600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.069917847990518e-05,
+      "loss": 2.2457,
+      "step": 139608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.06948854975342e-05,
+      "loss": 2.226,
+      "step": 139616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0690593243691645e-05,
+      "loss": 2.2405,
+      "step": 139624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.068630171848423e-05,
+      "loss": 2.2295,
+      "step": 139632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.068201092201864e-05,
+      "loss": 2.2315,
+      "step": 139640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.067772085440156e-05,
+      "loss": 2.2237,
+      "step": 139648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.067343151573963e-05,
+      "loss": 2.2263,
+      "step": 139656
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.066914290613946e-05,
+      "loss": 2.2215,
+      "step": 139664
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.06648550257077e-05,
+      "loss": 2.2176,
+      "step": 139672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.066056787455095e-05,
+      "loss": 2.2226,
+      "step": 139680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.065628145277575e-05,
+      "loss": 2.2289,
+      "step": 139688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.065199576048871e-05,
+      "loss": 2.2134,
+      "step": 139696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0647710797796334e-05,
+      "loss": 2.2165,
+      "step": 139704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.064342656480516e-05,
+      "loss": 2.215,
+      "step": 139712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.06391430616217e-05,
+      "loss": 2.2343,
+      "step": 139720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0634860288352435e-05,
+      "loss": 2.2387,
+      "step": 139728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.063057824510383e-05,
+      "loss": 2.2374,
+      "step": 139736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0626296931982353e-05,
+      "loss": 2.2174,
+      "step": 139744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.062201634909443e-05,
+      "loss": 2.2335,
+      "step": 139752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.061773649654648e-05,
+      "loss": 2.2285,
+      "step": 139760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.06134573744449e-05,
+      "loss": 2.2326,
+      "step": 139768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0609178982896054e-05,
+      "loss": 2.2317,
+      "step": 139776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0604901322006336e-05,
+      "loss": 2.2388,
+      "step": 139784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.060062439188205e-05,
+      "loss": 2.2403,
+      "step": 139792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.059634819262954e-05,
+      "loss": 2.2415,
+      "step": 139800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.059207272435514e-05,
+      "loss": 2.2397,
+      "step": 139808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.058779798716508e-05,
+      "loss": 2.2236,
+      "step": 139816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0583523981165675e-05,
+      "loss": 2.2386,
+      "step": 139824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.057925070646315e-05,
+      "loss": 2.221,
+      "step": 139832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.057497816316378e-05,
+      "loss": 2.2285,
+      "step": 139840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.057070635137374e-05,
+      "loss": 2.2217,
+      "step": 139848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0566435271199236e-05,
+      "loss": 2.2261,
+      "step": 139856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.056216492274646e-05,
+      "loss": 2.2275,
+      "step": 139864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.055789530612157e-05,
+      "loss": 2.2216,
+      "step": 139872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0553626421430695e-05,
+      "loss": 2.2138,
+      "step": 139880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.054935826877998e-05,
+      "loss": 2.2354,
+      "step": 139888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0545090848275524e-05,
+      "loss": 2.2096,
+      "step": 139896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.054082416002341e-05,
+      "loss": 2.2214,
+      "step": 139904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0536558204129716e-05,
+      "loss": 2.2271,
+      "step": 139912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0532292980700496e-05,
+      "loss": 2.222,
+      "step": 139920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.052802848984176e-05,
+      "loss": 2.2279,
+      "step": 139928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.052376473165957e-05,
+      "loss": 2.2312,
+      "step": 139936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.051950170625988e-05,
+      "loss": 2.2296,
+      "step": 139944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0515239413748675e-05,
+      "loss": 2.2182,
+      "step": 139952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0510977854231938e-05,
+      "loss": 2.2334,
+      "step": 139960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0506717027815597e-05,
+      "loss": 2.2179,
+      "step": 139968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0502456934605584e-05,
+      "loss": 2.208,
+      "step": 139976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.04981975747078e-05,
+      "loss": 2.2405,
+      "step": 139984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0493938948228123e-05,
+      "loss": 2.229,
+      "step": 139992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0489681055272463e-05,
+      "loss": 2.2147,
+      "step": 140000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0485423895946617e-05,
+      "loss": 2.2272,
+      "step": 140008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0481167470356442e-05,
+      "loss": 2.2274,
+      "step": 140016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0476911778607786e-05,
+      "loss": 2.225,
+      "step": 140024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0472656820806385e-05,
+      "loss": 2.2303,
+      "step": 140032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.046840259705806e-05,
+      "loss": 2.217,
+      "step": 140040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0464149107468555e-05,
+      "loss": 2.2246,
+      "step": 140048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.045989635214362e-05,
+      "loss": 2.242,
+      "step": 140056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0455644331188982e-05,
+      "loss": 2.2471,
+      "step": 140064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0451393044710324e-05,
+      "loss": 2.2236,
+      "step": 140072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0447142492813367e-05,
+      "loss": 2.2304,
+      "step": 140080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0442892675603763e-05,
+      "loss": 2.2241,
+      "step": 140088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.043864359318715e-05,
+      "loss": 2.2426,
+      "step": 140096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0434395245669185e-05,
+      "loss": 2.2261,
+      "step": 140104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0430147633155472e-05,
+      "loss": 2.2396,
+      "step": 140112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0425900755751598e-05,
+      "loss": 2.2387,
+      "step": 140120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0421654613563157e-05,
+      "loss": 2.2272,
+      "step": 140128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0417409206695703e-05,
+      "loss": 2.239,
+      "step": 140136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0413164535254756e-05,
+      "loss": 2.221,
+      "step": 140144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.040892059934588e-05,
+      "loss": 2.2128,
+      "step": 140152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0404677399074556e-05,
+      "loss": 2.2335,
+      "step": 140160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0400434934546258e-05,
+      "loss": 2.2268,
+      "step": 140168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0396193205866485e-05,
+      "loss": 2.2162,
+      "step": 140176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.039195221314065e-05,
+      "loss": 2.2317,
+      "step": 140184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0387711956474236e-05,
+      "loss": 2.2241,
+      "step": 140192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.03834724359726e-05,
+      "loss": 2.2257,
+      "step": 140200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0379233651741155e-05,
+      "loss": 2.2368,
+      "step": 140208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.037499560388532e-05,
+      "loss": 2.2182,
+      "step": 140216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0370758292510386e-05,
+      "loss": 2.2298,
+      "step": 140224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.036652171772173e-05,
+      "loss": 2.2126,
+      "step": 140232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.036228587962468e-05,
+      "loss": 2.2431,
+      "step": 140240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0358050778324524e-05,
+      "loss": 2.2205,
+      "step": 140248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.035381641392656e-05,
+      "loss": 2.2266,
+      "step": 140256
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.034958278653603e-05,
+      "loss": 2.2336,
+      "step": 140264
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0345349896258206e-05,
+      "loss": 2.2491,
+      "step": 140272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0341117743198315e-05,
+      "loss": 2.2188,
+      "step": 140280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.033688632746155e-05,
+      "loss": 2.2375,
+      "step": 140288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0332655649153134e-05,
+      "loss": 2.2291,
+      "step": 140296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.032842570837823e-05,
+      "loss": 2.2257,
+      "step": 140304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0324196505241976e-05,
+      "loss": 2.2291,
+      "step": 140312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0319968039849533e-05,
+      "loss": 2.2271,
+      "step": 140320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0315740312306023e-05,
+      "loss": 2.2278,
+      "step": 140328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.031151332271653e-05,
+      "loss": 2.2225,
+      "step": 140336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.030728707118615e-05,
+      "loss": 2.23,
+      "step": 140344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0303061557819945e-05,
+      "loss": 2.2169,
+      "step": 140352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0298836782722957e-05,
+      "loss": 2.2268,
+      "step": 140360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.029461274600023e-05,
+      "loss": 2.2278,
+      "step": 140368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.029038944775675e-05,
+      "loss": 2.2284,
+      "step": 140376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0286166888097523e-05,
+      "loss": 2.2379,
+      "step": 140384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0281945067127532e-05,
+      "loss": 2.2126,
+      "step": 140392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0277723984951706e-05,
+      "loss": 2.2323,
+      "step": 140400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0273503641675022e-05,
+      "loss": 2.2489,
+      "step": 140408
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0269284037402347e-05,
+      "loss": 2.2501,
+      "step": 140416
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0265065172238604e-05,
+      "loss": 2.2213,
+      "step": 140424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.026084704628871e-05,
+      "loss": 2.2394,
+      "step": 140432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.025662965965746e-05,
+      "loss": 2.2238,
+      "step": 140440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0252413012449742e-05,
+      "loss": 2.2246,
+      "step": 140448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.024819710477038e-05,
+      "loss": 2.2137,
+      "step": 140456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.024398193672418e-05,
+      "loss": 2.2492,
+      "step": 140464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.023976750841593e-05,
+      "loss": 2.2342,
+      "step": 140472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.023555381995038e-05,
+      "loss": 2.2309,
+      "step": 140480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0231340871432317e-05,
+      "loss": 2.2286,
+      "step": 140488
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0227128662966464e-05,
+      "loss": 2.2186,
+      "step": 140496
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0222917194657514e-05,
+      "loss": 2.2333,
+      "step": 140504
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0218706466610202e-05,
+      "loss": 2.2196,
+      "step": 140512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0214496478929182e-05,
+      "loss": 2.2301,
+      "step": 140520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0210287231719116e-05,
+      "loss": 2.2245,
+      "step": 140528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0206078725084657e-05,
+      "loss": 2.2209,
+      "step": 140536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.020187095913043e-05,
+      "loss": 2.2381,
+      "step": 140544
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0197663933961017e-05,
+      "loss": 2.2278,
+      "step": 140552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0193457649681034e-05,
+      "loss": 2.2269,
+      "step": 140560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0189252106395044e-05,
+      "loss": 2.2286,
+      "step": 140568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0185047304207575e-05,
+      "loss": 2.2401,
+      "step": 140576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.018084324322319e-05,
+      "loss": 2.2268,
+      "step": 140584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0176639923546374e-05,
+      "loss": 2.2148,
+      "step": 140592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0172437345281645e-05,
+      "loss": 2.2275,
+      "step": 140600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.016823550853347e-05,
+      "loss": 2.2335,
+      "step": 140608
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.01640344134063e-05,
+      "loss": 2.2298,
+      "step": 140616
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0159834060004605e-05,
+      "loss": 2.2301,
+      "step": 140624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0155634448432757e-05,
+      "loss": 2.226,
+      "step": 140632
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0151435578795183e-05,
+      "loss": 2.2284,
+      "step": 140640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0147237451196297e-05,
+      "loss": 2.2394,
+      "step": 140648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0143040065740418e-05,
+      "loss": 2.2288,
+      "step": 140656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0138843422531904e-05,
+      "loss": 2.2315,
+      "step": 140664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0134647521675113e-05,
+      "loss": 2.2412,
+      "step": 140672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.013045236327433e-05,
+      "loss": 2.2335,
+      "step": 140680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0126257947433852e-05,
+      "loss": 2.2357,
+      "step": 140688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0122064274257944e-05,
+      "loss": 2.2186,
+      "step": 140696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0117871343850883e-05,
+      "loss": 2.2191,
+      "step": 140704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0113679156316886e-05,
+      "loss": 2.226,
+      "step": 140712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0109487711760174e-05,
+      "loss": 2.2213,
+      "step": 140720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0105297010284962e-05,
+      "loss": 2.2283,
+      "step": 140728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.010110705199542e-05,
+      "loss": 2.2326,
+      "step": 140736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0096917836995698e-05,
+      "loss": 2.2158,
+      "step": 140744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0092729365389966e-05,
+      "loss": 2.225,
+      "step": 140752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.008854163728234e-05,
+      "loss": 2.2275,
+      "step": 140760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.008435465277691e-05,
+      "loss": 2.2287,
+      "step": 140768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0080168411977794e-05,
+      "loss": 2.2438,
+      "step": 140776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0075982914989044e-05,
+      "loss": 2.225,
+      "step": 140784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0071798161914705e-05,
+      "loss": 2.2188,
+      "step": 140792
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0067614152858837e-05,
+      "loss": 2.2229,
+      "step": 140800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0063430887925424e-05,
+      "loss": 2.2417,
+      "step": 140808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0059248367218496e-05,
+      "loss": 2.2229,
+      "step": 140816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0055066590842008e-05,
+      "loss": 2.2201,
+      "step": 140824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.005088555889991e-05,
+      "loss": 2.2287,
+      "step": 140832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0046705271496183e-05,
+      "loss": 2.2254,
+      "step": 140840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0042525728734704e-05,
+      "loss": 2.2323,
+      "step": 140848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.003834693071939e-05,
+      "loss": 2.2199,
+      "step": 140856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0034168877554164e-05,
+      "loss": 2.2148,
+      "step": 140864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0029991569342836e-05,
+      "loss": 2.2259,
+      "step": 140872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.002581500618929e-05,
+      "loss": 2.2208,
+      "step": 140880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.002163918819734e-05,
+      "loss": 2.2316,
+      "step": 140888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.00174641154708e-05,
+      "loss": 2.2225,
+      "step": 140896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.001328978811348e-05,
+      "loss": 2.2193,
+      "step": 140904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0009116206229125e-05,
+      "loss": 2.2134,
+      "step": 140912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0004943369921517e-05,
+      "loss": 2.2219,
+      "step": 140920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.0000771279294383e-05,
+      "loss": 2.2152,
+      "step": 140928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9996599934451424e-05,
+      "loss": 2.2361,
+      "step": 140936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9992429335496374e-05,
+      "loss": 2.2226,
+      "step": 140944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9988259482532898e-05,
+      "loss": 2.2443,
+      "step": 140952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.998409037566464e-05,
+      "loss": 2.2104,
+      "step": 140960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9979922014995286e-05,
+      "loss": 2.2123,
+      "step": 140968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.997575440062843e-05,
+      "loss": 2.2197,
+      "step": 140976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9971587532667678e-05,
+      "loss": 2.207,
+      "step": 140984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9967421411216645e-05,
+      "loss": 2.2362,
+      "step": 140992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9963256036378884e-05,
+      "loss": 2.2154,
+      "step": 141000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9959091408257932e-05,
+      "loss": 2.2261,
+      "step": 141008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.995492752695735e-05,
+      "loss": 2.2209,
+      "step": 141016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.995076439258063e-05,
+      "loss": 2.2214,
+      "step": 141024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9946602005231307e-05,
+      "loss": 2.2259,
+      "step": 141032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.99424403650128e-05,
+      "loss": 2.2139,
+      "step": 141040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.993827947202861e-05,
+      "loss": 2.2159,
+      "step": 141048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.993411932638217e-05,
+      "loss": 2.2275,
+      "step": 141056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.992995992817689e-05,
+      "loss": 2.2359,
+      "step": 141064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.992580127751617e-05,
+      "loss": 2.2282,
+      "step": 141072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9921643374503433e-05,
+      "loss": 2.2315,
+      "step": 141080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.991748621924199e-05,
+      "loss": 2.2243,
+      "step": 141088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9913329811835228e-05,
+      "loss": 2.2328,
+      "step": 141096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9909174152386456e-05,
+      "loss": 2.2206,
+      "step": 141104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9905019240998994e-05,
+      "loss": 2.2215,
+      "step": 141112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9900865077776132e-05,
+      "loss": 2.2172,
+      "step": 141120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.989671166282113e-05,
+      "loss": 2.2204,
+      "step": 141128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9892558996237263e-05,
+      "loss": 2.2088,
+      "step": 141136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.988840707812776e-05,
+      "loss": 2.227,
+      "step": 141144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9884255908595825e-05,
+      "loss": 2.2053,
+      "step": 141152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.988010548774468e-05,
+      "loss": 2.2323,
+      "step": 141160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9875955815677485e-05,
+      "loss": 2.2134,
+      "step": 141168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.98718068924974e-05,
+      "loss": 2.2202,
+      "step": 141176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.986765871830759e-05,
+      "loss": 2.2348,
+      "step": 141184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9863511293211168e-05,
+      "loss": 2.2147,
+      "step": 141192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.985936461731122e-05,
+      "loss": 2.2371,
+      "step": 141200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9855218690710863e-05,
+      "loss": 2.2387,
+      "step": 141208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.985107351351315e-05,
+      "loss": 2.2271,
+      "step": 141216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9846929085821117e-05,
+      "loss": 2.2227,
+      "step": 141224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.984278540773783e-05,
+      "loss": 2.2245,
+      "step": 141232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9838642479366263e-05,
+      "loss": 2.225,
+      "step": 141240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9834500300809454e-05,
+      "loss": 2.2064,
+      "step": 141248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9830358872170327e-05,
+      "loss": 2.2212,
+      "step": 141256
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.982621819355187e-05,
+      "loss": 2.2276,
+      "step": 141264
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.982207826505703e-05,
+      "loss": 2.2302,
+      "step": 141272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9817939086788688e-05,
+      "loss": 2.2176,
+      "step": 141280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9813800658849768e-05,
+      "loss": 2.2317,
+      "step": 141288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9809662981343156e-05,
+      "loss": 2.2179,
+      "step": 141296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9805526054371718e-05,
+      "loss": 2.2341,
+      "step": 141304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.980138987803829e-05,
+      "loss": 2.22,
+      "step": 141312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.979725445244568e-05,
+      "loss": 2.2367,
+      "step": 141320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.979311977769673e-05,
+      "loss": 2.2309,
+      "step": 141328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9788985853894207e-05,
+      "loss": 2.2344,
+      "step": 141336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.978485268114088e-05,
+      "loss": 2.217,
+      "step": 141344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9780720259539513e-05,
+      "loss": 2.2247,
+      "step": 141352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9776588589192834e-05,
+      "loss": 2.2344,
+      "step": 141360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9772457670203536e-05,
+      "loss": 2.2485,
+      "step": 141368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9768327502674347e-05,
+      "loss": 2.2236,
+      "step": 141376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9764198086707927e-05,
+      "loss": 2.2434,
+      "step": 141384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9760069422406923e-05,
+      "loss": 2.2142,
+      "step": 141392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9755941509873996e-05,
+      "loss": 2.2362,
+      "step": 141400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9751814349211766e-05,
+      "loss": 2.2236,
+      "step": 141408
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9747687940522807e-05,
+      "loss": 2.2152,
+      "step": 141416
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.974356228390973e-05,
+      "loss": 2.2352,
+      "step": 141424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9739437379475077e-05,
+      "loss": 2.2326,
+      "step": 141432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9735313227321415e-05,
+      "loss": 2.2272,
+      "step": 141440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.973118982755127e-05,
+      "loss": 2.2317,
+      "step": 141448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9727067180267126e-05,
+      "loss": 2.2295,
+      "step": 141456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9722945285571513e-05,
+      "loss": 2.2257,
+      "step": 141464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9718824143566853e-05,
+      "loss": 2.2339,
+      "step": 141472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9714703754355627e-05,
+      "loss": 2.2162,
+      "step": 141480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.971058411804028e-05,
+      "loss": 2.2433,
+      "step": 141488
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9706465234723192e-05,
+      "loss": 2.2254,
+      "step": 141496
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9702347104506773e-05,
+      "loss": 2.2463,
+      "step": 141504
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9698229727493416e-05,
+      "loss": 2.2277,
+      "step": 141512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.969411310378547e-05,
+      "loss": 2.2284,
+      "step": 141520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9689997233485266e-05,
+      "loss": 2.2437,
+      "step": 141528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.968588211669513e-05,
+      "loss": 2.2187,
+      "step": 141536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9681767753517377e-05,
+      "loss": 2.2401,
+      "step": 141544
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9677654144054273e-05,
+      "loss": 2.2359,
+      "step": 141552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.967354128840808e-05,
+      "loss": 2.2183,
+      "step": 141560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9669429186681064e-05,
+      "loss": 2.2463,
+      "step": 141568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9665317838975436e-05,
+      "loss": 2.248,
+      "step": 141576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9661207245393406e-05,
+      "loss": 2.2318,
+      "step": 141584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9657097406037183e-05,
+      "loss": 2.222,
+      "step": 141592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.965298832100891e-05,
+      "loss": 2.2225,
+      "step": 141600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9648879990410756e-05,
+      "loss": 2.2398,
+      "step": 141608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9644772414344853e-05,
+      "loss": 2.2368,
+      "step": 141616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.964066559291332e-05,
+      "loss": 2.2155,
+      "step": 141624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.963655952621822e-05,
+      "loss": 2.2078,
+      "step": 141632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9632454214361683e-05,
+      "loss": 2.2293,
+      "step": 141640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9628349657445726e-05,
+      "loss": 2.2157,
+      "step": 141648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9624245855572418e-05,
+      "loss": 2.216,
+      "step": 141656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9620142808843758e-05,
+      "loss": 2.2182,
+      "step": 141664
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9616040517361753e-05,
+      "loss": 2.2298,
+      "step": 141672
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.961193898122841e-05,
+      "loss": 2.226,
+      "step": 141680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.960783820054565e-05,
+      "loss": 2.2199,
+      "step": 141688
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.960373817541544e-05,
+      "loss": 2.2419,
+      "step": 141696
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9599638905939736e-05,
+      "loss": 2.2349,
+      "step": 141704
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.95955403922204e-05,
+      "loss": 2.2359,
+      "step": 141712
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9591442634359347e-05,
+      "loss": 2.2104,
+      "step": 141720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.958734563245843e-05,
+      "loss": 2.2089,
+      "step": 141728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9583249386619528e-05,
+      "loss": 2.233,
+      "step": 141736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9579153896944456e-05,
+      "loss": 2.2301,
+      "step": 141744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9575059163535023e-05,
+      "loss": 2.2385,
+      "step": 141752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.957096518649304e-05,
+      "loss": 2.2106,
+      "step": 141760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.956687196592028e-05,
+      "loss": 2.2533,
+      "step": 141768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.956277950191848e-05,
+      "loss": 2.2298,
+      "step": 141776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9558687794589414e-05,
+      "loss": 2.2313,
+      "step": 141784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.955459684403479e-05,
+      "loss": 2.2206,
+      "step": 141792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.955050665035628e-05,
+      "loss": 2.2103,
+      "step": 141800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.954641721365561e-05,
+      "loss": 2.2365,
+      "step": 141808
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9542328534034426e-05,
+      "loss": 2.2195,
+      "step": 141816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.953824061159436e-05,
+      "loss": 2.2096,
+      "step": 141824
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9534153446437057e-05,
+      "loss": 2.2198,
+      "step": 141832
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9530067038664123e-05,
+      "loss": 2.2437,
+      "step": 141840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.952598138837712e-05,
+      "loss": 2.2112,
+      "step": 141848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.952189649567766e-05,
+      "loss": 2.2234,
+      "step": 141856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9517812360667256e-05,
+      "loss": 2.221,
+      "step": 141864
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9513728983447486e-05,
+      "loss": 2.2361,
+      "step": 141872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.95096463641198e-05,
+      "loss": 2.239,
+      "step": 141880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9505564502785736e-05,
+      "loss": 2.2287,
+      "step": 141888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9501483399546782e-05,
+      "loss": 2.2179,
+      "step": 141896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9497403054504346e-05,
+      "loss": 2.2263,
+      "step": 141904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.94933234677599e-05,
+      "loss": 2.235,
+      "step": 141912
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.948924463941487e-05,
+      "loss": 2.2391,
+      "step": 141920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9485166569570623e-05,
+      "loss": 2.2209,
+      "step": 141928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9481089258328564e-05,
+      "loss": 2.2287,
+      "step": 141936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9477012705790043e-05,
+      "loss": 2.2149,
+      "step": 141944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9472936912056426e-05,
+      "loss": 2.2342,
+      "step": 141952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9468861877229016e-05,
+      "loss": 2.2258,
+      "step": 141960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9464787601409118e-05,
+      "loss": 2.2175,
+      "step": 141968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.946071408469804e-05,
+      "loss": 2.2248,
+      "step": 141976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.945664132719703e-05,
+      "loss": 2.2315,
+      "step": 141984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9452569329007328e-05,
+      "loss": 2.2324,
+      "step": 141992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9448498090230204e-05,
+      "loss": 2.2219,
+      "step": 142000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.944442761096684e-05,
+      "loss": 2.217,
+      "step": 142008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.944035789131842e-05,
+      "loss": 2.2117,
+      "step": 142016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9436288931386148e-05,
+      "loss": 2.209,
+      "step": 142024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9432220731271154e-05,
+      "loss": 2.2288,
+      "step": 142032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9428153291074577e-05,
+      "loss": 2.2238,
+      "step": 142040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.942408661089755e-05,
+      "loss": 2.2056,
+      "step": 142048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9420020690841154e-05,
+      "loss": 2.2336,
+      "step": 142056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9415955531006474e-05,
+      "loss": 2.222,
+      "step": 142064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9411891131494578e-05,
+      "loss": 2.2292,
+      "step": 142072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9407827492406487e-05,
+      "loss": 2.229,
+      "step": 142080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.940376461384326e-05,
+      "loss": 2.2366,
+      "step": 142088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9399702495905855e-05,
+      "loss": 2.2392,
+      "step": 142096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9395641138695285e-05,
+      "loss": 2.2186,
+      "step": 142104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9391580542312526e-05,
+      "loss": 2.2334,
+      "step": 142112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9387520706858485e-05,
+      "loss": 2.2461,
+      "step": 142120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.938346163243412e-05,
+      "loss": 2.2413,
+      "step": 142128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.937940331914034e-05,
+      "loss": 2.2312,
+      "step": 142136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9375345767078034e-05,
+      "loss": 2.2225,
+      "step": 142144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9371288976348066e-05,
+      "loss": 2.2294,
+      "step": 142152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9367232947051275e-05,
+      "loss": 2.2299,
+      "step": 142160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.936317767928853e-05,
+      "loss": 2.2222,
+      "step": 142168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9359123173160613e-05,
+      "loss": 2.2264,
+      "step": 142176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9355069428768333e-05,
+      "loss": 2.2435,
+      "step": 142184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9351016446212472e-05,
+      "loss": 2.2328,
+      "step": 142192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9346964225593775e-05,
+      "loss": 2.2435,
+      "step": 142200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9342912767012982e-05,
+      "loss": 2.2292,
+      "step": 142208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.933886207057083e-05,
+      "loss": 2.233,
+      "step": 142216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9334812136368e-05,
+      "loss": 2.2254,
+      "step": 142224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.933076296450517e-05,
+      "loss": 2.2159,
+      "step": 142232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9326714555083035e-05,
+      "loss": 2.225,
+      "step": 142240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9322666908202207e-05,
+      "loss": 2.218,
+      "step": 142248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9318620023963316e-05,
+      "loss": 2.226,
+      "step": 142256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9314573902466986e-05,
+      "loss": 2.2165,
+      "step": 142264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9310528543813787e-05,
+      "loss": 2.2277,
+      "step": 142272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.930648394810428e-05,
+      "loss": 2.2388,
+      "step": 142280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.930244011543904e-05,
+      "loss": 2.2292,
+      "step": 142288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.929839704591857e-05,
+      "loss": 2.2381,
+      "step": 142296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.929435473964342e-05,
+      "loss": 2.2368,
+      "step": 142304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9290313196714032e-05,
+      "loss": 2.229,
+      "step": 142312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9286272417230897e-05,
+      "loss": 2.2365,
+      "step": 142320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9282232401294506e-05,
+      "loss": 2.2308,
+      "step": 142328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.927819314900524e-05,
+      "loss": 2.2285,
+      "step": 142336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.927415466046354e-05,
+      "loss": 2.229,
+      "step": 142344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9270116935769816e-05,
+      "loss": 2.2274,
+      "step": 142352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.926607997502443e-05,
+      "loss": 2.2471,
+      "step": 142360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.926204377832774e-05,
+      "loss": 2.2321,
+      "step": 142368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9258008345780084e-05,
+      "loss": 2.2278,
+      "step": 142376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.92539736774818e-05,
+      "loss": 2.2329,
+      "step": 142384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.924993977353318e-05,
+      "loss": 2.2153,
+      "step": 142392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.92459066340345e-05,
+      "loss": 2.2432,
+      "step": 142400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9241874259086034e-05,
+      "loss": 2.2143,
+      "step": 142408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9237842648788028e-05,
+      "loss": 2.2372,
+      "step": 142416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9233811803240695e-05,
+      "loss": 2.2237,
+      "step": 142424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9229781722544268e-05,
+      "loss": 2.2183,
+      "step": 142432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9225752406798912e-05,
+      "loss": 2.2409,
+      "step": 142440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.92217238561048e-05,
+      "loss": 2.231,
+      "step": 142448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.921769607056209e-05,
+      "loss": 2.2376,
+      "step": 142456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9213669050270914e-05,
+      "loss": 2.243,
+      "step": 142464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9209642795331365e-05,
+      "loss": 2.2359,
+      "step": 142472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.920561730584357e-05,
+      "loss": 2.2247,
+      "step": 142480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9201592581907565e-05,
+      "loss": 2.2232,
+      "step": 142488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.919756862362344e-05,
+      "loss": 2.2291,
+      "step": 142496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.919354543109121e-05,
+      "loss": 2.2293,
+      "step": 142504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.918952300441089e-05,
+      "loss": 2.2306,
+      "step": 142512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9185501343682503e-05,
+      "loss": 2.2214,
+      "step": 142520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9181480449005992e-05,
+      "loss": 2.2444,
+      "step": 142528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.917746032048133e-05,
+      "loss": 2.229,
+      "step": 142536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9173440958208485e-05,
+      "loss": 2.2358,
+      "step": 142544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9169422362287348e-05,
+      "loss": 2.2142,
+      "step": 142552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9165404532817823e-05,
+      "loss": 2.2207,
+      "step": 142560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9161387469899797e-05,
+      "loss": 2.2247,
+      "step": 142568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9157371173633152e-05,
+      "loss": 2.2258,
+      "step": 142576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.915335564411772e-05,
+      "loss": 2.2229,
+      "step": 142584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9149340881453312e-05,
+      "loss": 2.2322,
+      "step": 142592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9145326885739766e-05,
+      "loss": 2.2446,
+      "step": 142600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.914131365707685e-05,
+      "loss": 2.2409,
+      "step": 142608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.913730119556433e-05,
+      "loss": 2.2134,
+      "step": 142616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9133289501301987e-05,
+      "loss": 2.2258,
+      "step": 142624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9129278574389517e-05,
+      "loss": 2.2168,
+      "step": 142632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9125268414926638e-05,
+      "loss": 2.2133,
+      "step": 142640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9121259023013058e-05,
+      "loss": 2.2285,
+      "step": 142648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9117250398748447e-05,
+      "loss": 2.2103,
+      "step": 142656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9113242542232443e-05,
+      "loss": 2.229,
+      "step": 142664
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9109235453564708e-05,
+      "loss": 2.223,
+      "step": 142672
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9105229132844846e-05,
+      "loss": 2.2192,
+      "step": 142680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9101223580172444e-05,
+      "loss": 2.2155,
+      "step": 142688
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9097218795647103e-05,
+      "loss": 2.2113,
+      "step": 142696
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.909321477936835e-05,
+      "loss": 2.2179,
+      "step": 142704
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.908921153143577e-05,
+      "loss": 2.2139,
+      "step": 142712
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9085209051948848e-05,
+      "loss": 2.2323,
+      "step": 142720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9081207341007087e-05,
+      "loss": 2.218,
+      "step": 142728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9077206398710012e-05,
+      "loss": 2.2325,
+      "step": 142736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9073206225157025e-05,
+      "loss": 2.2345,
+      "step": 142744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9069206820447604e-05,
+      "loss": 2.2239,
+      "step": 142752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.906520818468119e-05,
+      "loss": 2.2275,
+      "step": 142760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.906121031795715e-05,
+      "loss": 2.21,
+      "step": 142768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9057213220374906e-05,
+      "loss": 2.2132,
+      "step": 142776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.90532168920338e-05,
+      "loss": 2.2358,
+      "step": 142784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9049221333033206e-05,
+      "loss": 2.2361,
+      "step": 142792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9045226543472448e-05,
+      "loss": 2.22,
+      "step": 142800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9041232523450813e-05,
+      "loss": 2.2225,
+      "step": 142808
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.903723927306763e-05,
+      "loss": 2.2326,
+      "step": 142816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9033246792422142e-05,
+      "loss": 2.2153,
+      "step": 142824
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.902925508161361e-05,
+      "loss": 2.2113,
+      "step": 142832
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9025264140741276e-05,
+      "loss": 2.208,
+      "step": 142840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9021273969904355e-05,
+      "loss": 2.2208,
+      "step": 142848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.901728456920203e-05,
+      "loss": 2.2333,
+      "step": 142856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.90132959387335e-05,
+      "loss": 2.2198,
+      "step": 142864
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9009308078597897e-05,
+      "loss": 2.2072,
+      "step": 142872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.900532098889437e-05,
+      "loss": 2.222,
+      "step": 142880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9001334669722052e-05,
+      "loss": 2.2304,
+      "step": 142888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8997349121180035e-05,
+      "loss": 2.2328,
+      "step": 142896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.899336434336738e-05,
+      "loss": 2.239,
+      "step": 142904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8989380336383183e-05,
+      "loss": 2.2204,
+      "step": 142912
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.898539710032645e-05,
+      "loss": 2.2116,
+      "step": 142920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8981414635296257e-05,
+      "loss": 2.2309,
+      "step": 142928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8977432941391546e-05,
+      "loss": 2.2246,
+      "step": 142936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.897345201871134e-05,
+      "loss": 2.2302,
+      "step": 142944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8969471867354614e-05,
+      "loss": 2.2362,
+      "step": 142952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8965492487420282e-05,
+      "loss": 2.2203,
+      "step": 142960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8961513879007285e-05,
+      "loss": 2.2118,
+      "step": 142968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.895753604221456e-05,
+      "loss": 2.2307,
+      "step": 142976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.895355897714095e-05,
+      "loss": 2.2164,
+      "step": 142984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.894958268388535e-05,
+      "loss": 2.208,
+      "step": 142992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8945607162546607e-05,
+      "loss": 2.2041,
+      "step": 143000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.894163241322356e-05,
+      "loss": 2.2479,
+      "step": 143008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8937658436015018e-05,
+      "loss": 2.2336,
+      "step": 143016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8933685231019762e-05,
+      "loss": 2.2132,
+      "step": 143024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8929712798336587e-05,
+      "loss": 2.2363,
+      "step": 143032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8925741138064238e-05,
+      "loss": 2.2336,
+      "step": 143040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8921770250301444e-05,
+      "loss": 2.2313,
+      "step": 143048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8917800135146932e-05,
+      "loss": 2.2378,
+      "step": 143056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8913830792699402e-05,
+      "loss": 2.2302,
+      "step": 143064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8909862223057516e-05,
+      "loss": 2.2015,
+      "step": 143072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8905894426319947e-05,
+      "loss": 2.2182,
+      "step": 143080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8901927402585333e-05,
+      "loss": 2.2177,
+      "step": 143088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.889796115195229e-05,
+      "loss": 2.2277,
+      "step": 143096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8893995674519425e-05,
+      "loss": 2.2342,
+      "step": 143104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8890030970385317e-05,
+      "loss": 2.2117,
+      "step": 143112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8886067039648522e-05,
+      "loss": 2.2307,
+      "step": 143120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8882103882407602e-05,
+      "loss": 2.2163,
+      "step": 143128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8878141498761063e-05,
+      "loss": 2.2238,
+      "step": 143136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8874179888807435e-05,
+      "loss": 2.228,
+      "step": 143144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.887021905264517e-05,
+      "loss": 2.2154,
+      "step": 143152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8866258990372745e-05,
+      "loss": 2.229,
+      "step": 143160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8862299702088646e-05,
+      "loss": 2.212,
+      "step": 143168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8858341187891237e-05,
+      "loss": 2.2236,
+      "step": 143176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.885438344787897e-05,
+      "loss": 2.2187,
+      "step": 143184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8850426482150234e-05,
+      "loss": 2.2305,
+      "step": 143192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8846470290803386e-05,
+      "loss": 2.2206,
+      "step": 143200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.884251487393679e-05,
+      "loss": 2.2244,
+      "step": 143208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8838560231648748e-05,
+      "loss": 2.2356,
+      "step": 143216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.883460636403761e-05,
+      "loss": 2.2346,
+      "step": 143224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8830653271201657e-05,
+      "loss": 2.2143,
+      "step": 143232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8826700953239145e-05,
+      "loss": 2.2213,
+      "step": 143240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.882274941024836e-05,
+      "loss": 2.2138,
+      "step": 143248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8818798642327527e-05,
+      "loss": 2.2121,
+      "step": 143256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8814848649574838e-05,
+      "loss": 2.2248,
+      "step": 143264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8810899432088524e-05,
+      "loss": 2.2138,
+      "step": 143272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8806950989966752e-05,
+      "loss": 2.2356,
+      "step": 143280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.880300332330767e-05,
+      "loss": 2.2175,
+      "step": 143288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.879905643220943e-05,
+      "loss": 2.2318,
+      "step": 143296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.879511031677015e-05,
+      "loss": 2.2371,
+      "step": 143304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.879116497708792e-05,
+      "loss": 2.2197,
+      "step": 143312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8787220413260842e-05,
+      "loss": 2.2288,
+      "step": 143320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8783276625386963e-05,
+      "loss": 2.222,
+      "step": 143328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.877933361356432e-05,
+      "loss": 2.2299,
+      "step": 143336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.877539137789096e-05,
+      "loss": 2.2053,
+      "step": 143344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8771449918464853e-05,
+      "loss": 2.2276,
+      "step": 143352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.876750923538404e-05,
+      "loss": 2.2261,
+      "step": 143360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.876356932874642e-05,
+      "loss": 2.2273,
+      "step": 143368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.875963019864998e-05,
+      "loss": 2.2128,
+      "step": 143376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8755691845192645e-05,
+      "loss": 2.2218,
+      "step": 143384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.875175426847231e-05,
+      "loss": 2.2393,
+      "step": 143392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8747817468586872e-05,
+      "loss": 2.2408,
+      "step": 143400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.874388144563419e-05,
+      "loss": 2.2209,
+      "step": 143408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8739946199712136e-05,
+      "loss": 2.2148,
+      "step": 143416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.873601173091852e-05,
+      "loss": 2.2272,
+      "step": 143424
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8732078039351157e-05,
+      "loss": 2.2393,
+      "step": 143432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8728145125107847e-05,
+      "loss": 2.2293,
+      "step": 143440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.872421298828636e-05,
+      "loss": 2.2333,
+      "step": 143448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8720281628984435e-05,
+      "loss": 2.2281,
+      "step": 143456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8716351047299832e-05,
+      "loss": 2.2315,
+      "step": 143464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8712421243330245e-05,
+      "loss": 2.2367,
+      "step": 143472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.870849221717338e-05,
+      "loss": 2.2313,
+      "step": 143480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8704563968926906e-05,
+      "loss": 2.2104,
+      "step": 143488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.870063649868849e-05,
+      "loss": 2.2325,
+      "step": 143496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8696709806555744e-05,
+      "loss": 2.2161,
+      "step": 143504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8692783892626323e-05,
+      "loss": 2.2196,
+      "step": 143512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8688858756997805e-05,
+      "loss": 2.2258,
+      "step": 143520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.868493439976776e-05,
+      "loss": 2.2221,
+      "step": 143528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8681010821033772e-05,
+      "loss": 2.2247,
+      "step": 143536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.867708802089335e-05,
+      "loss": 2.2241,
+      "step": 143544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.867316599944405e-05,
+      "loss": 2.2211,
+      "step": 143552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8669244756783366e-05,
+      "loss": 2.2008,
+      "step": 143560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8665324293008755e-05,
+      "loss": 2.2302,
+      "step": 143568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8661404608217717e-05,
+      "loss": 2.2273,
+      "step": 143576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.865748570250766e-05,
+      "loss": 2.2208,
+      "step": 143584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.865356757597602e-05,
+      "loss": 2.1996,
+      "step": 143592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8649650228720232e-05,
+      "loss": 2.2326,
+      "step": 143600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8645733660837632e-05,
+      "loss": 2.2282,
+      "step": 143608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8641817872425623e-05,
+      "loss": 2.2269,
+      "step": 143616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8637902863581533e-05,
+      "loss": 2.2095,
+      "step": 143624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8633988634402705e-05,
+      "loss": 2.2395,
+      "step": 143632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8630075184986436e-05,
+      "loss": 2.2184,
+      "step": 143640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.862616251543001e-05,
+      "loss": 2.2392,
+      "step": 143648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8622250625830715e-05,
+      "loss": 2.2219,
+      "step": 143656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8618339516285785e-05,
+      "loss": 2.2169,
+      "step": 143664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8614429186892454e-05,
+      "loss": 2.2282,
+      "step": 143672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.861051963774794e-05,
+      "loss": 2.2143,
+      "step": 143680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.860661086894943e-05,
+      "loss": 2.2233,
+      "step": 143688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8602702880594085e-05,
+      "loss": 2.2201,
+      "step": 143696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8598795672779077e-05,
+      "loss": 2.2194,
+      "step": 143704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8594889245601543e-05,
+      "loss": 2.214,
+      "step": 143712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8590983599158562e-05,
+      "loss": 2.2203,
+      "step": 143720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.858707873354727e-05,
+      "loss": 2.2286,
+      "step": 143728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8583174648864724e-05,
+      "loss": 2.2328,
+      "step": 143736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.857927134520797e-05,
+      "loss": 2.2247,
+      "step": 143744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8575368822674065e-05,
+      "loss": 2.2144,
+      "step": 143752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8571467081360004e-05,
+      "loss": 2.2322,
+      "step": 143760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.856756612136282e-05,
+      "loss": 2.2205,
+      "step": 143768
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8563665942779454e-05,
+      "loss": 2.2082,
+      "step": 143776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8559766545706873e-05,
+      "loss": 2.2083,
+      "step": 143784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8555867930242054e-05,
+      "loss": 2.2179,
+      "step": 143792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8551970096481852e-05,
+      "loss": 2.2466,
+      "step": 143800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.854807304452321e-05,
+      "loss": 2.2098,
+      "step": 143808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.854417677446302e-05,
+      "loss": 2.2249,
+      "step": 143816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8540281286398102e-05,
+      "loss": 2.2175,
+      "step": 143824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8536386580425334e-05,
+      "loss": 2.2185,
+      "step": 143832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8532492656641513e-05,
+      "loss": 2.2297,
+      "step": 143840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8528599515143467e-05,
+      "loss": 2.222,
+      "step": 143848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8524707156027976e-05,
+      "loss": 2.2134,
+      "step": 143856
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.852081557939177e-05,
+      "loss": 2.2222,
+      "step": 143864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.851692478533164e-05,
+      "loss": 2.2141,
+      "step": 143872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8513034773944293e-05,
+      "loss": 2.2289,
+      "step": 143880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8509145545326423e-05,
+      "loss": 2.2233,
+      "step": 143888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.850525709957474e-05,
+      "loss": 2.2184,
+      "step": 143896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8501369436785895e-05,
+      "loss": 2.2218,
+      "step": 143904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.849748255705653e-05,
+      "loss": 2.2303,
+      "step": 143912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8493596460483296e-05,
+      "loss": 2.2377,
+      "step": 143920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.848971114716279e-05,
+      "loss": 2.2244,
+      "step": 143928
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8485826617191586e-05,
+      "loss": 2.2122,
+      "step": 143936
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.848194287066628e-05,
+      "loss": 2.2261,
+      "step": 143944
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.847805990768341e-05,
+      "loss": 2.2184,
+      "step": 143952
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8474177728339497e-05,
+      "loss": 2.2128,
+      "step": 143960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8470296332731072e-05,
+      "loss": 2.2284,
+      "step": 143968
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.846641572095461e-05,
+      "loss": 2.2142,
+      "step": 143976
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8462535893106613e-05,
+      "loss": 2.236,
+      "step": 143984
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.845865684928349e-05,
+      "loss": 2.2169,
+      "step": 143992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8454778589581695e-05,
+      "loss": 2.2187,
+      "step": 144000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8450901114097667e-05,
+      "loss": 2.2323,
+      "step": 144008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8447024422927748e-05,
+      "loss": 2.2261,
+      "step": 144016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.844314851616835e-05,
+      "loss": 2.2076,
+      "step": 144024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8439273393915843e-05,
+      "loss": 2.2237,
+      "step": 144032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8435399056266514e-05,
+      "loss": 2.2179,
+      "step": 144040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.843152550331672e-05,
+      "loss": 2.2303,
+      "step": 144048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.842765273516273e-05,
+      "loss": 2.2405,
+      "step": 144056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8423780751900853e-05,
+      "loss": 2.2319,
+      "step": 144064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.841990955362732e-05,
+      "loss": 2.2238,
+      "step": 144072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8416039140438378e-05,
+      "loss": 2.2284,
+      "step": 144080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8412169512430248e-05,
+      "loss": 2.2072,
+      "step": 144088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.840830066969913e-05,
+      "loss": 2.2342,
+      "step": 144096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8404432612341198e-05,
+      "loss": 2.2207,
+      "step": 144104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8400565340452616e-05,
+      "loss": 2.2148,
+      "step": 144112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.839669885412953e-05,
+      "loss": 2.2242,
+      "step": 144120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8392833153468052e-05,
+      "loss": 2.239,
+      "step": 144128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8388968238564297e-05,
+      "loss": 2.2163,
+      "step": 144136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.838510410951433e-05,
+      "loss": 2.215,
+      "step": 144144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.838124076641422e-05,
+      "loss": 2.2352,
+      "step": 144152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8377378209360016e-05,
+      "loss": 2.2261,
+      "step": 144160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.837351643844774e-05,
+      "loss": 2.2042,
+      "step": 144168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.836965545377338e-05,
+      "loss": 2.2233,
+      "step": 144176
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8365795255432948e-05,
+      "loss": 2.2326,
+      "step": 144184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8361935843522385e-05,
+      "loss": 2.2243,
+      "step": 144192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8358077218137668e-05,
+      "loss": 2.214,
+      "step": 144200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.835421937937468e-05,
+      "loss": 2.228,
+      "step": 144208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.835036232732935e-05,
+      "loss": 2.2348,
+      "step": 144216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8346506062097588e-05,
+      "loss": 2.2103,
+      "step": 144224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.834265058377521e-05,
+      "loss": 2.2253,
+      "step": 144232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.833879589245809e-05,
+      "loss": 2.2418,
+      "step": 144240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8334941988242068e-05,
+      "loss": 2.2345,
+      "step": 144248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8331088871222945e-05,
+      "loss": 2.2394,
+      "step": 144256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.83272365414965e-05,
+      "loss": 2.206,
+      "step": 144264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8323384999158496e-05,
+      "loss": 2.2406,
+      "step": 144272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8319534244304714e-05,
+      "loss": 2.226,
+      "step": 144280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8315684277030864e-05,
+      "loss": 2.2459,
+      "step": 144288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8311835097432643e-05,
+      "loss": 2.2492,
+      "step": 144296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8307986705605766e-05,
+      "loss": 2.2203,
+      "step": 144304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8304139101645897e-05,
+      "loss": 2.2317,
+      "step": 144312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8300292285648677e-05,
+      "loss": 2.2163,
+      "step": 144320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.829644625770976e-05,
+      "loss": 2.2401,
+      "step": 144328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8292601017924742e-05,
+      "loss": 2.2264,
+      "step": 144336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8288756566389206e-05,
+      "loss": 2.2364,
+      "step": 144344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8284912903198757e-05,
+      "loss": 2.2225,
+      "step": 144352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.828107002844893e-05,
+      "loss": 2.227,
+      "step": 144360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.827722794223525e-05,
+      "loss": 2.2298,
+      "step": 144368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8273386644653253e-05,
+      "loss": 2.2429,
+      "step": 144376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8269546135798426e-05,
+      "loss": 2.2207,
+      "step": 144384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8265706415766223e-05,
+      "loss": 2.2186,
+      "step": 144392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8261867484652138e-05,
+      "loss": 2.2341,
+      "step": 144400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8258029342551578e-05,
+      "loss": 2.2376,
+      "step": 144408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8254191989559993e-05,
+      "loss": 2.2249,
+      "step": 144416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.825035542577273e-05,
+      "loss": 2.2199,
+      "step": 144424
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.82465196512852e-05,
+      "loss": 2.2229,
+      "step": 144432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8242684666192764e-05,
+      "loss": 2.2199,
+      "step": 144440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8238850470590738e-05,
+      "loss": 2.224,
+      "step": 144448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.823501706457446e-05,
+      "loss": 2.2381,
+      "step": 144456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8231184448239212e-05,
+      "loss": 2.2397,
+      "step": 144464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.822735262168029e-05,
+      "loss": 2.2292,
+      "step": 144472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.822352158499295e-05,
+      "loss": 2.2273,
+      "step": 144480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8219691338272425e-05,
+      "loss": 2.2346,
+      "step": 144488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8215861881613937e-05,
+      "loss": 2.2383,
+      "step": 144496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8212033215112698e-05,
+      "loss": 2.2358,
+      "step": 144504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8208205338863864e-05,
+      "loss": 2.225,
+      "step": 144512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8204378252962626e-05,
+      "loss": 2.2253,
+      "step": 144520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8200551957504107e-05,
+      "loss": 2.2288,
+      "step": 144528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8196726452583427e-05,
+      "loss": 2.2378,
+      "step": 144536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8192901738295708e-05,
+      "loss": 2.2396,
+      "step": 144544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.818907781473602e-05,
+      "loss": 2.2169,
+      "step": 144552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8185254681999416e-05,
+      "loss": 2.2238,
+      "step": 144560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.818143234018096e-05,
+      "loss": 2.2359,
+      "step": 144568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8177610789375662e-05,
+      "loss": 2.229,
+      "step": 144576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8173790029678527e-05,
+      "loss": 2.2296,
+      "step": 144584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.816997006118455e-05,
+      "loss": 2.2334,
+      "step": 144592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8166150883988678e-05,
+      "loss": 2.2254,
+      "step": 144600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8162332498185888e-05,
+      "loss": 2.2362,
+      "step": 144608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8158514903871057e-05,
+      "loss": 2.2421,
+      "step": 144616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8154698101139127e-05,
+      "loss": 2.2439,
+      "step": 144624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.815088209008499e-05,
+      "loss": 2.2219,
+      "step": 144632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8147066870803484e-05,
+      "loss": 2.2452,
+      "step": 144640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8143252443389467e-05,
+      "loss": 2.2233,
+      "step": 144648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8139438807937783e-05,
+      "loss": 2.2308,
+      "step": 144656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.813562596454321e-05,
+      "loss": 2.2297,
+      "step": 144664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.813181391330056e-05,
+      "loss": 2.2447,
+      "step": 144672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8128002654304574e-05,
+      "loss": 2.2251,
+      "step": 144680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8124192187650038e-05,
+      "loss": 2.226,
+      "step": 144688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.812038251343166e-05,
+      "loss": 2.2196,
+      "step": 144696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8116573631744135e-05,
+      "loss": 2.225,
+      "step": 144704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.811276554268218e-05,
+      "loss": 2.2297,
+      "step": 144712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.810895824634045e-05,
+      "loss": 2.2325,
+      "step": 144720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8105151742813586e-05,
+      "loss": 2.2336,
+      "step": 144728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.810134603219624e-05,
+      "loss": 2.2385,
+      "step": 144736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8097541114583004e-05,
+      "loss": 2.2199,
+      "step": 144744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8093736990068474e-05,
+      "loss": 2.2232,
+      "step": 144752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.808993365874723e-05,
+      "loss": 2.2293,
+      "step": 144760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.808613112071381e-05,
+      "loss": 2.2282,
+      "step": 144768
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8082329376062743e-05,
+      "loss": 2.2074,
+      "step": 144776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.807852842488856e-05,
+      "loss": 2.2361,
+      "step": 144784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8074728267285743e-05,
+      "loss": 2.2148,
+      "step": 144792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.807092890334876e-05,
+      "loss": 2.2451,
+      "step": 144800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.806713033317207e-05,
+      "loss": 2.2267,
+      "step": 144808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8063332556850092e-05,
+      "loss": 2.2271,
+      "step": 144816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8059535574477268e-05,
+      "loss": 2.223,
+      "step": 144824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8055739386147955e-05,
+      "loss": 2.2488,
+      "step": 144832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8051943991956543e-05,
+      "loss": 2.2144,
+      "step": 144840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8048149391997413e-05,
+      "loss": 2.2254,
+      "step": 144848
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8044355586364852e-05,
+      "loss": 2.2366,
+      "step": 144856
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8040562575153196e-05,
+      "loss": 2.238,
+      "step": 144864
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.803677035845676e-05,
+      "loss": 2.2164,
+      "step": 144872
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.803297893636978e-05,
+      "loss": 2.2172,
+      "step": 144880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8029188308986536e-05,
+      "loss": 2.2191,
+      "step": 144888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8025398476401256e-05,
+      "loss": 2.2278,
+      "step": 144896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8021609438708156e-05,
+      "loss": 2.2406,
+      "step": 144904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8017821196001444e-05,
+      "loss": 2.2303,
+      "step": 144912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.801403374837527e-05,
+      "loss": 2.201,
+      "step": 144920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8010247095923815e-05,
+      "loss": 2.224,
+      "step": 144928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.800646123874121e-05,
+      "loss": 2.2179,
+      "step": 144936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.800267617692156e-05,
+      "loss": 2.2353,
+      "step": 144944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7998891910558978e-05,
+      "loss": 2.2214,
+      "step": 144952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.799510843974753e-05,
+      "loss": 2.237,
+      "step": 144960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7991325764581265e-05,
+      "loss": 2.2173,
+      "step": 144968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7987543885154248e-05,
+      "loss": 2.195,
+      "step": 144976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7983762801560482e-05,
+      "loss": 2.2114,
+      "step": 144984
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7979982513893955e-05,
+      "loss": 2.2317,
+      "step": 144992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.797620302224866e-05,
+      "loss": 2.1975,
+      "step": 145000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.797242432671856e-05,
+      "loss": 2.2209,
+      "step": 145008
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7968646427397566e-05,
+      "loss": 2.2203,
+      "step": 145016
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7964869324379628e-05,
+      "loss": 2.216,
+      "step": 145024
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.796109301775862e-05,
+      "loss": 2.2169,
+      "step": 145032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7957317507628465e-05,
+      "loss": 2.2488,
+      "step": 145040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7953542794082964e-05,
+      "loss": 2.2352,
+      "step": 145048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7949768877215988e-05,
+      "loss": 2.2122,
+      "step": 145056
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7945995757121375e-05,
+      "loss": 2.2184,
+      "step": 145064
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.794222343389288e-05,
+      "loss": 2.2153,
+      "step": 145072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7938451907624303e-05,
+      "loss": 2.2365,
+      "step": 145080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.793468117840943e-05,
+      "loss": 2.2115,
+      "step": 145088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7930911246341966e-05,
+      "loss": 2.2155,
+      "step": 145096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.792714211151566e-05,
+      "loss": 2.2285,
+      "step": 145104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.792337377402418e-05,
+      "loss": 2.2374,
+      "step": 145112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7919606233961243e-05,
+      "loss": 2.2169,
+      "step": 145120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.791583949142049e-05,
+      "loss": 2.2246,
+      "step": 145128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7912073546495564e-05,
+      "loss": 2.2235,
+      "step": 145136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7908308399280093e-05,
+      "loss": 2.2111,
+      "step": 145144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7904544049867676e-05,
+      "loss": 2.2167,
+      "step": 145152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7900780498351886e-05,
+      "loss": 2.2403,
+      "step": 145160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.789701774482631e-05,
+      "loss": 2.2305,
+      "step": 145168
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7893255789384464e-05,
+      "loss": 2.217,
+      "step": 145176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7889494632119877e-05,
+      "loss": 2.2179,
+      "step": 145184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7885734273126068e-05,
+      "loss": 2.2098,
+      "step": 145192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7881974712496513e-05,
+      "loss": 2.2121,
+      "step": 145200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.787821595032465e-05,
+      "loss": 2.2178,
+      "step": 145208
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7874457986703962e-05,
+      "loss": 2.2133,
+      "step": 145216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7870700821727845e-05,
+      "loss": 2.221,
+      "step": 145224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7866944455489697e-05,
+      "loss": 2.2378,
+      "step": 145232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7863188888082937e-05,
+      "loss": 2.2112,
+      "step": 145240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7859434119600885e-05,
+      "loss": 2.2113,
+      "step": 145248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7855680150136932e-05,
+      "loss": 2.2272,
+      "step": 145256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.785192697978436e-05,
+      "loss": 2.2019,
+      "step": 145264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.784817460863649e-05,
+      "loss": 2.2268,
+      "step": 145272
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7844423036786628e-05,
+      "loss": 2.2266,
+      "step": 145280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7840672264327988e-05,
+      "loss": 2.221,
+      "step": 145288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7836922291353863e-05,
+      "loss": 2.2165,
+      "step": 145296
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7833173117957438e-05,
+      "loss": 2.2108,
+      "step": 145304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7829424744231957e-05,
+      "loss": 2.2249,
+      "step": 145312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7825677170270587e-05,
+      "loss": 2.2242,
+      "step": 145320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.782193039616648e-05,
+      "loss": 2.2312,
+      "step": 145328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7818184422012804e-05,
+      "loss": 2.2266,
+      "step": 145336
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7814439247902673e-05,
+      "loss": 2.2022,
+      "step": 145344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7810694873929182e-05,
+      "loss": 2.2277,
+      "step": 145352
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.780695130018544e-05,
+      "loss": 2.2282,
+      "step": 145360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.78032085267645e-05,
+      "loss": 2.2143,
+      "step": 145368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.77994665537594e-05,
+      "loss": 2.2318,
+      "step": 145376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.779572538126318e-05,
+      "loss": 2.2323,
+      "step": 145384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.779198500936885e-05,
+      "loss": 2.2203,
+      "step": 145392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7788245438169363e-05,
+      "loss": 2.2265,
+      "step": 145400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.778450666775773e-05,
+      "loss": 2.2368,
+      "step": 145408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7780768698226872e-05,
+      "loss": 2.2152,
+      "step": 145416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7777031529669705e-05,
+      "loss": 2.2139,
+      "step": 145424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7773295162179162e-05,
+      "loss": 2.2234,
+      "step": 145432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.776955959584811e-05,
+      "loss": 2.225,
+      "step": 145440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7765824830769434e-05,
+      "loss": 2.229,
+      "step": 145448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.776209086703596e-05,
+      "loss": 2.2064,
+      "step": 145456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7758357704740524e-05,
+      "loss": 2.2007,
+      "step": 145464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.775462534397595e-05,
+      "loss": 2.2326,
+      "step": 145472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.775089378483498e-05,
+      "loss": 2.2288,
+      "step": 145480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7747163027410416e-05,
+      "loss": 2.2126,
+      "step": 145488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7743433071795017e-05,
+      "loss": 2.2338,
+      "step": 145496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.773970391808147e-05,
+      "loss": 2.2341,
+      "step": 145504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7735975566362514e-05,
+      "loss": 2.2333,
+      "step": 145512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.773224801673082e-05,
+      "loss": 2.2267,
+      "step": 145520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7728521269279067e-05,
+      "loss": 2.2307,
+      "step": 145528
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7724795324099894e-05,
+      "loss": 2.2236,
+      "step": 145536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7721070181285917e-05,
+      "loss": 2.2259,
+      "step": 145544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7717345840929776e-05,
+      "loss": 2.2302,
+      "step": 145552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7713622303124037e-05,
+      "loss": 2.2217,
+      "step": 145560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7709899567961257e-05,
+      "loss": 2.2181,
+      "step": 145568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7706177635534004e-05,
+      "loss": 2.23,
+      "step": 145576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7702456505934803e-05,
+      "loss": 2.2137,
+      "step": 145584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7698736179256148e-05,
+      "loss": 2.2394,
+      "step": 145592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7695016655590544e-05,
+      "loss": 2.2096,
+      "step": 145600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7691297935030447e-05,
+      "loss": 2.2266,
+      "step": 145608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7687580017668302e-05,
+      "loss": 2.2238,
+      "step": 145616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.768386290359655e-05,
+      "loss": 2.2422,
+      "step": 145624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.76801465929076e-05,
+      "loss": 2.2321,
+      "step": 145632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7676431085693815e-05,
+      "loss": 2.2289,
+      "step": 145640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.767271638204759e-05,
+      "loss": 2.2241,
+      "step": 145648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7669002482061257e-05,
+      "loss": 2.221,
+      "step": 145656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7665289385827164e-05,
+      "loss": 2.2232,
+      "step": 145664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.766157709343759e-05,
+      "loss": 2.2269,
+      "step": 145672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7657865604984835e-05,
+      "loss": 2.2201,
+      "step": 145680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.765415492056119e-05,
+      "loss": 2.215,
+      "step": 145688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7650445040258865e-05,
+      "loss": 2.2272,
+      "step": 145696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7646735964170102e-05,
+      "loss": 2.2226,
+      "step": 145704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7643027692387137e-05,
+      "loss": 2.2232,
+      "step": 145712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7639320225002108e-05,
+      "loss": 2.2296,
+      "step": 145720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.763561356210722e-05,
+      "loss": 2.214,
+      "step": 145728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7631907703794602e-05,
+      "loss": 2.236,
+      "step": 145736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7628202650156405e-05,
+      "loss": 2.2137,
+      "step": 145744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7624498401284716e-05,
+      "loss": 2.2076,
+      "step": 145752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7620794957271624e-05,
+      "loss": 2.2263,
+      "step": 145760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.761709231820921e-05,
+      "loss": 2.2162,
+      "step": 145768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7613390484189516e-05,
+      "loss": 2.2128,
+      "step": 145776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7609689455304554e-05,
+      "loss": 2.2234,
+      "step": 145784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7605989231646356e-05,
+      "loss": 2.2379,
+      "step": 145792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7602289813306902e-05,
+      "loss": 2.2185,
+      "step": 145800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7598591200378148e-05,
+      "loss": 2.2083,
+      "step": 145808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7594893392952065e-05,
+      "loss": 2.2112,
+      "step": 145816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7591196391120567e-05,
+      "loss": 2.2342,
+      "step": 145824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7587500194975553e-05,
+      "loss": 2.2143,
+      "step": 145832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7583804804608925e-05,
+      "loss": 2.222,
+      "step": 145840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7580110220112554e-05,
+      "loss": 2.2223,
+      "step": 145848
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.757641644157827e-05,
+      "loss": 2.2256,
+      "step": 145856
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7572723469097916e-05,
+      "loss": 2.2329,
+      "step": 145864
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.756903130276329e-05,
+      "loss": 2.2404,
+      "step": 145872
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7565339942666198e-05,
+      "loss": 2.2283,
+      "step": 145880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.756164938889838e-05,
+      "loss": 2.2443,
+      "step": 145888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.755795964155159e-05,
+      "loss": 2.2198,
+      "step": 145896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7554270700717593e-05,
+      "loss": 2.206,
+      "step": 145904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7550582566488043e-05,
+      "loss": 2.2204,
+      "step": 145912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.754689523895465e-05,
+      "loss": 2.2317,
+      "step": 145920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.75432087182091e-05,
+      "loss": 2.2313,
+      "step": 145928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7539523004343008e-05,
+      "loss": 2.2184,
+      "step": 145936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7535838097448026e-05,
+      "loss": 2.2115,
+      "step": 145944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.753215399761574e-05,
+      "loss": 2.2208,
+      "step": 145952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.752847070493776e-05,
+      "loss": 2.2275,
+      "step": 145960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.752478821950564e-05,
+      "loss": 2.219,
+      "step": 145968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7521106541410913e-05,
+      "loss": 2.2237,
+      "step": 145976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.751742567074514e-05,
+      "loss": 2.2341,
+      "step": 145984
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7513745607599807e-05,
+      "loss": 2.2306,
+      "step": 145992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7510066352066388e-05,
+      "loss": 2.2276,
+      "step": 146000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.750638790423638e-05,
+      "loss": 2.2282,
+      "step": 146008
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7502710264201215e-05,
+      "loss": 2.2194,
+      "step": 146016
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7499033432052308e-05,
+      "loss": 2.221,
+      "step": 146024
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7495357407881082e-05,
+      "loss": 2.2328,
+      "step": 146032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7491682191778922e-05,
+      "loss": 2.2323,
+      "step": 146040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.748800778383718e-05,
+      "loss": 2.2244,
+      "step": 146048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.748433418414722e-05,
+      "loss": 2.2145,
+      "step": 146056
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7480661392800368e-05,
+      "loss": 2.2148,
+      "step": 146064
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7476989409887898e-05,
+      "loss": 2.2204,
+      "step": 146072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7473318235501146e-05,
+      "loss": 2.2242,
+      "step": 146080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.746964786973133e-05,
+      "loss": 2.246,
+      "step": 146088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7465978312669745e-05,
+      "loss": 2.2226,
+      "step": 146096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.746230956440756e-05,
+      "loss": 2.2384,
+      "step": 146104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.745864162503602e-05,
+      "loss": 2.2258,
+      "step": 146112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.745497449464632e-05,
+      "loss": 2.2215,
+      "step": 146120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7451308173329583e-05,
+      "loss": 2.209,
+      "step": 146128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7447642661176988e-05,
+      "loss": 2.216,
+      "step": 146136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7443977958279633e-05,
+      "loss": 2.2154,
+      "step": 146144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.744031406472865e-05,
+      "loss": 2.2397,
+      "step": 146152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.743665098061512e-05,
+      "loss": 2.2232,
+      "step": 146160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7432988706030083e-05,
+      "loss": 2.2233,
+      "step": 146168
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7429327241064614e-05,
+      "loss": 2.2448,
+      "step": 146176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.742566658580972e-05,
+      "loss": 2.2319,
+      "step": 146184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7422006740356404e-05,
+      "loss": 2.2209,
+      "step": 146192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7418347704795665e-05,
+      "loss": 2.2204,
+      "step": 146200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7414689479218458e-05,
+      "loss": 2.2342,
+      "step": 146208
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7411032063715712e-05,
+      "loss": 2.2258,
+      "step": 146216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.740737545837838e-05,
+      "loss": 2.2252,
+      "step": 146224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.740371966329735e-05,
+      "loss": 2.2271,
+      "step": 146232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.74000646785635e-05,
+      "loss": 2.2207,
+      "step": 146240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.73964105042677e-05,
+      "loss": 2.2254,
+      "step": 146248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7392757140500803e-05,
+      "loss": 2.2264,
+      "step": 146256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7389104587353604e-05,
+      "loss": 2.2343,
+      "step": 146264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7385452844916936e-05,
+      "loss": 2.2444,
+      "step": 146272
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7381801913281575e-05,
+      "loss": 2.2301,
+      "step": 146280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7378151792538264e-05,
+      "loss": 2.2268,
+      "step": 146288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.737450248277777e-05,
+      "loss": 2.2186,
+      "step": 146296
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7370853984090793e-05,
+      "loss": 2.2344,
+      "step": 146304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7367206296568065e-05,
+      "loss": 2.2101,
+      "step": 146312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.736355942030024e-05,
+      "loss": 2.2301,
+      "step": 146320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7359913355377976e-05,
+      "loss": 2.2187,
+      "step": 146328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7356268101891958e-05,
+      "loss": 2.2245,
+      "step": 146336
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7352623659932754e-05,
+      "loss": 2.2402,
+      "step": 146344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7348980029590998e-05,
+      "loss": 2.2344,
+      "step": 146352
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.734533721095725e-05,
+      "loss": 2.2154,
+      "step": 146360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.73416952041221e-05,
+      "loss": 2.2081,
+      "step": 146368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7338054009176066e-05,
+      "loss": 2.2405,
+      "step": 146376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7334413626209665e-05,
+      "loss": 2.2271,
+      "step": 146384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.733077405531341e-05,
+      "loss": 2.2213,
+      "step": 146392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.732713529657779e-05,
+      "loss": 2.2337,
+      "step": 146400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7323497350093237e-05,
+      "loss": 2.2213,
+      "step": 146408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.731986021595021e-05,
+      "loss": 2.2085,
+      "step": 146416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7316223894239134e-05,
+      "loss": 2.2161,
+      "step": 146424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7312588385050387e-05,
+      "loss": 2.2242,
+      "step": 146432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.730895368847437e-05,
+      "loss": 2.2289,
+      "step": 146440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7305319804601424e-05,
+      "loss": 2.2137,
+      "step": 146448
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.730168673352189e-05,
+      "loss": 2.2095,
+      "step": 146456
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7298054475326107e-05,
+      "loss": 2.214,
+      "step": 146464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.729442303010436e-05,
+      "loss": 2.2222,
+      "step": 146472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.729079239794691e-05,
+      "loss": 2.2378,
+      "step": 146480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7287162578944043e-05,
+      "loss": 2.223,
+      "step": 146488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7283533573185977e-05,
+      "loss": 2.209,
+      "step": 146496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7279905380762953e-05,
+      "loss": 2.2268,
+      "step": 146504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7276278001765134e-05,
+      "loss": 2.2205,
+      "step": 146512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7272651436282714e-05,
+      "loss": 2.2053,
+      "step": 146520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.726902568440588e-05,
+      "loss": 2.2086,
+      "step": 146528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7265400746224705e-05,
+      "loss": 2.2233,
+      "step": 146536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7261776621829348e-05,
+      "loss": 2.2247,
+      "step": 146544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7258153311309918e-05,
+      "loss": 2.2476,
+      "step": 146552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7254530814756452e-05,
+      "loss": 2.207,
+      "step": 146560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.725090913225903e-05,
+      "loss": 2.2034,
+      "step": 146568
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7247288263907677e-05,
+      "loss": 2.2278,
+      "step": 146576
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7243668209792416e-05,
+      "loss": 2.2295,
+      "step": 146584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.724004897000324e-05,
+      "loss": 2.2248,
+      "step": 146592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7236430544630114e-05,
+      "loss": 2.2206,
+      "step": 146600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7232812933763006e-05,
+      "loss": 2.2106,
+      "step": 146608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.722919613749185e-05,
+      "loss": 2.2152,
+      "step": 146616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.722558015590655e-05,
+      "loss": 2.2175,
+      "step": 146624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7221964989097004e-05,
+      "loss": 2.2177,
+      "step": 146632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.721835063715309e-05,
+      "loss": 2.2157,
+      "step": 146640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7214737100164645e-05,
+      "loss": 2.2175,
+      "step": 146648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7211124378221533e-05,
+      "loss": 2.209,
+      "step": 146656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7207512471413533e-05,
+      "loss": 2.2198,
+      "step": 146664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7203901379830453e-05,
+      "loss": 2.214,
+      "step": 146672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.720029110356207e-05,
+      "loss": 2.2309,
+      "step": 146680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.719668164269813e-05,
+      "loss": 2.2063,
+      "step": 146688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.719307299732836e-05,
+      "loss": 2.2335,
+      "step": 146696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7189465167542482e-05,
+      "loss": 2.2124,
+      "step": 146704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.718585815343017e-05,
+      "loss": 2.2242,
+      "step": 146712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.718225195508113e-05,
+      "loss": 2.2318,
+      "step": 146720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7178646572584965e-05,
+      "loss": 2.2176,
+      "step": 146728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.717504200603133e-05,
+      "loss": 2.2259,
+      "step": 146736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7171438255509856e-05,
+      "loss": 2.2071,
+      "step": 146744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7167835321110092e-05,
+      "loss": 2.2342,
+      "step": 146752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.716423320292162e-05,
+      "loss": 2.2144,
+      "step": 146760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.716063190103402e-05,
+      "loss": 2.2354,
+      "step": 146768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7157031415536765e-05,
+      "loss": 2.2267,
+      "step": 146776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7153431746519408e-05,
+      "loss": 2.2229,
+      "step": 146784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.714983289407142e-05,
+      "loss": 2.2415,
+      "step": 146792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7146234858282277e-05,
+      "loss": 2.2145,
+      "step": 146800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7142637639241425e-05,
+      "loss": 2.2136,
+      "step": 146808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7139041237038275e-05,
+      "loss": 2.221,
+      "step": 146816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7135445651762255e-05,
+      "loss": 2.2087,
+      "step": 146824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.713185088350274e-05,
+      "loss": 2.2119,
+      "step": 146832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.71282569323491e-05,
+      "loss": 2.2158,
+      "step": 146840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7124663798390688e-05,
+      "loss": 2.2249,
+      "step": 146848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7121071481716813e-05,
+      "loss": 2.2473,
+      "step": 146856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7117479982416793e-05,
+      "loss": 2.2138,
+      "step": 146864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7113889300579915e-05,
+      "loss": 2.2246,
+      "step": 146872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.711029943629544e-05,
+      "loss": 2.2153,
+      "step": 146880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7106710389652594e-05,
+      "loss": 2.2168,
+      "step": 146888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7103122160740634e-05,
+      "loss": 2.2233,
+      "step": 146896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7099534749648746e-05,
+      "loss": 2.239,
+      "step": 146904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7095948156466105e-05,
+      "loss": 2.2125,
+      "step": 146912
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.70923623812819e-05,
+      "loss": 2.2219,
+      "step": 146920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7088777424185248e-05,
+      "loss": 2.2099,
+      "step": 146928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7085193285265298e-05,
+      "loss": 2.2139,
+      "step": 146936
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7081609964611108e-05,
+      "loss": 2.2315,
+      "step": 146944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7078027462311807e-05,
+      "loss": 2.2306,
+      "step": 146952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.707444577845644e-05,
+      "loss": 2.2148,
+      "step": 146960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7070864913134037e-05,
+      "loss": 2.2317,
+      "step": 146968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7067284866433627e-05,
+      "loss": 2.2429,
+      "step": 146976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7063705638444213e-05,
+      "loss": 2.2289,
+      "step": 146984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7060127229254765e-05,
+      "loss": 2.2175,
+      "step": 146992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.705654963895426e-05,
+      "loss": 2.2305,
+      "step": 147000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7052972867631616e-05,
+      "loss": 2.21,
+      "step": 147008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7049396915375775e-05,
+      "loss": 2.2046,
+      "step": 147016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7045821782275617e-05,
+      "loss": 2.2417,
+      "step": 147024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7042247468420023e-05,
+      "loss": 2.2248,
+      "step": 147032
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.703867397389786e-05,
+      "loss": 2.2212,
+      "step": 147040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7035101298797964e-05,
+      "loss": 2.2166,
+      "step": 147048
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7031529443209142e-05,
+      "loss": 2.2227,
+      "step": 147056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.70279584072202e-05,
+      "loss": 2.1995,
+      "step": 147064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7024388190919914e-05,
+      "loss": 2.2211,
+      "step": 147072
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7020818794397022e-05,
+      "loss": 2.2317,
+      "step": 147080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.701725021774029e-05,
+      "loss": 2.2267,
+      "step": 147088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7013682461038427e-05,
+      "loss": 2.2147,
+      "step": 147096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.70101155243801e-05,
+      "loss": 2.2179,
+      "step": 147104
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7006549407854008e-05,
+      "loss": 2.2276,
+      "step": 147112
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7002984111548806e-05,
+      "loss": 2.2143,
+      "step": 147120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.699941963555311e-05,
+      "loss": 2.2236,
+      "step": 147128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6995855979955555e-05,
+      "loss": 2.2424,
+      "step": 147136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.699229314484471e-05,
+      "loss": 2.2099,
+      "step": 147144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6988731130309184e-05,
+      "loss": 2.2235,
+      "step": 147152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6985169936437487e-05,
+      "loss": 2.2189,
+      "step": 147160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6981609563318168e-05,
+      "loss": 2.243,
+      "step": 147168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6978050011039758e-05,
+      "loss": 2.2292,
+      "step": 147176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6974491279690712e-05,
+      "loss": 2.2149,
+      "step": 147184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6970933369359525e-05,
+      "loss": 2.224,
+      "step": 147192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.696737628013464e-05,
+      "loss": 2.2149,
+      "step": 147200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6963820012104486e-05,
+      "loss": 2.2377,
+      "step": 147208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6960264565357475e-05,
+      "loss": 2.2164,
+      "step": 147216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6956709939981983e-05,
+      "loss": 2.2225,
+      "step": 147224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6953156136066405e-05,
+      "loss": 2.2243,
+      "step": 147232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6949603153699067e-05,
+      "loss": 2.2139,
+      "step": 147240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6946050992968297e-05,
+      "loss": 2.2182,
+      "step": 147248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6942499653962418e-05,
+      "loss": 2.2411,
+      "step": 147256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6938949136769703e-05,
+      "loss": 2.2102,
+      "step": 147264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6935399441478414e-05,
+      "loss": 2.2246,
+      "step": 147272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6931850568176815e-05,
+      "loss": 2.2127,
+      "step": 147280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6928302516953118e-05,
+      "loss": 2.2331,
+      "step": 147288
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6924755287895525e-05,
+      "loss": 2.2351,
+      "step": 147296
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6921208881092234e-05,
+      "loss": 2.2346,
+      "step": 147304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6917663296631406e-05,
+      "loss": 2.2189,
+      "step": 147312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6914118534601168e-05,
+      "loss": 2.2111,
+      "step": 147320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.691057459508966e-05,
+      "loss": 2.2195,
+      "step": 147328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6907031478184986e-05,
+      "loss": 2.2132,
+      "step": 147336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.690348918397522e-05,
+      "loss": 2.2127,
+      "step": 147344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6899947712548423e-05,
+      "loss": 2.2208,
+      "step": 147352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6896407063992636e-05,
+      "loss": 2.2245,
+      "step": 147360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6892867238395896e-05,
+      "loss": 2.2231,
+      "step": 147368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6889328235846177e-05,
+      "loss": 2.2092,
+      "step": 147376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.688579005643147e-05,
+      "loss": 2.2249,
+      "step": 147384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6882252700239763e-05,
+      "loss": 2.2201,
+      "step": 147392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6878716167358946e-05,
+      "loss": 2.2104,
+      "step": 147400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.687518045787697e-05,
+      "loss": 2.2267,
+      "step": 147408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6871645571881716e-05,
+      "loss": 2.2267,
+      "step": 147416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.686811150946108e-05,
+      "loss": 2.2202,
+      "step": 147424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.686457827070291e-05,
+      "loss": 2.2255,
+      "step": 147432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6861045855695034e-05,
+      "loss": 2.2155,
+      "step": 147440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.685751426452528e-05,
+      "loss": 2.2238,
+      "step": 147448
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6853983497281443e-05,
+      "loss": 2.2179,
+      "step": 147456
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6850453554051286e-05,
+      "loss": 2.2303,
+      "step": 147464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6846924434922583e-05,
+      "loss": 2.2303,
+      "step": 147472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6843396139983057e-05,
+      "loss": 2.2341,
+      "step": 147480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6839868669320417e-05,
+      "loss": 2.2397,
+      "step": 147488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6836342023022366e-05,
+      "loss": 2.2157,
+      "step": 147496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6832816201176575e-05,
+      "loss": 2.2214,
+      "step": 147504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.682929120387069e-05,
+      "loss": 2.2341,
+      "step": 147512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.682576703119235e-05,
+      "loss": 2.2137,
+      "step": 147520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6822243683229165e-05,
+      "loss": 2.2261,
+      "step": 147528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6818721160068717e-05,
+      "loss": 2.2151,
+      "step": 147536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6815199461798596e-05,
+      "loss": 2.2137,
+      "step": 147544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6811678588506328e-05,
+      "loss": 2.2355,
+      "step": 147552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.680815854027947e-05,
+      "loss": 2.227,
+      "step": 147560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6804639317205495e-05,
+      "loss": 2.2329,
+      "step": 147568
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.680112091937192e-05,
+      "loss": 2.229,
+      "step": 147576
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6797603346866213e-05,
+      "loss": 2.2239,
+      "step": 147584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6794086599775797e-05,
+      "loss": 2.208,
+      "step": 147592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.679057067818811e-05,
+      "loss": 2.2414,
+      "step": 147600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.678705558219059e-05,
+      "loss": 2.2283,
+      "step": 147608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.678354131187057e-05,
+      "loss": 2.2205,
+      "step": 147616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6780027867315444e-05,
+      "loss": 2.228,
+      "step": 147624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6776515248612554e-05,
+      "loss": 2.2033,
+      "step": 147632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6773003455849223e-05,
+      "loss": 2.2161,
+      "step": 147640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6769492489112758e-05,
+      "loss": 2.2373,
+      "step": 147648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6765982348490427e-05,
+      "loss": 2.2213,
+      "step": 147656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6762473034069515e-05,
+      "loss": 2.2215,
+      "step": 147664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6758964545937248e-05,
+      "loss": 2.2208,
+      "step": 147672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6755456884180846e-05,
+      "loss": 2.2177,
+      "step": 147680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.675195004888753e-05,
+      "loss": 2.2258,
+      "step": 147688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.674844404014446e-05,
+      "loss": 2.2347,
+      "step": 147696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.674493885803879e-05,
+      "loss": 2.2162,
+      "step": 147704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6741434502657693e-05,
+      "loss": 2.2322,
+      "step": 147712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6737930974088254e-05,
+      "loss": 2.2213,
+      "step": 147720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6734428272417572e-05,
+      "loss": 2.2242,
+      "step": 147728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6730926397732753e-05,
+      "loss": 2.2407,
+      "step": 147736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6727425350120834e-05,
+      "loss": 2.2248,
+      "step": 147744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6723925129668844e-05,
+      "loss": 2.2462,
+      "step": 147752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6720425736463815e-05,
+      "loss": 2.231,
+      "step": 147760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6716927170592733e-05,
+      "loss": 2.2268,
+      "step": 147768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.671342943214259e-05,
+      "loss": 2.1959,
+      "step": 147776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.670993252120031e-05,
+      "loss": 2.2202,
+      "step": 147784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.670643643785284e-05,
+      "loss": 2.2096,
+      "step": 147792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6702941182187114e-05,
+      "loss": 2.2203,
+      "step": 147800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.669944675428999e-05,
+      "loss": 2.2047,
+      "step": 147808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6695953154248353e-05,
+      "loss": 2.2169,
+      "step": 147816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6692460382149075e-05,
+      "loss": 2.2421,
+      "step": 147824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6688968438078953e-05,
+      "loss": 2.1969,
+      "step": 147832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6685477322124825e-05,
+      "loss": 2.2235,
+      "step": 147840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.668198703437346e-05,
+      "loss": 2.2201,
+      "step": 147848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6678497574911636e-05,
+      "loss": 2.2339,
+      "step": 147856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.667500894382611e-05,
+      "loss": 2.2172,
+      "step": 147864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.667152114120359e-05,
+      "loss": 2.2296,
+      "step": 147872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6668034167130794e-05,
+      "loss": 2.2291,
+      "step": 147880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6664548021694418e-05,
+      "loss": 2.2215,
+      "step": 147888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6661062704981105e-05,
+      "loss": 2.2122,
+      "step": 147896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6657578217077523e-05,
+      "loss": 2.2318,
+      "step": 147904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6654094558070286e-05,
+      "loss": 2.2093,
+      "step": 147912
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6650611728045993e-05,
+      "loss": 2.2354,
+      "step": 147920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6647129727091246e-05,
+      "loss": 2.2268,
+      "step": 147928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6643648555292593e-05,
+      "loss": 2.2308,
+      "step": 147936
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6640168212736572e-05,
+      "loss": 2.224,
+      "step": 147944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.663668869950972e-05,
+      "loss": 2.2302,
+      "step": 147952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6633210015698528e-05,
+      "loss": 2.2111,
+      "step": 147960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.662973216138947e-05,
+      "loss": 2.2175,
+      "step": 147968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.662625513666903e-05,
+      "loss": 2.2446,
+      "step": 147976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6622778941623615e-05,
+      "loss": 2.2235,
+      "step": 147984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6619303576339682e-05,
+      "loss": 2.2134,
+      "step": 147992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.661582904090359e-05,
+      "loss": 2.2253,
+      "step": 148000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.661235533540173e-05,
+      "loss": 2.2265,
+      "step": 148008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6608882459920482e-05,
+      "loss": 2.2337,
+      "step": 148016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6605410414546146e-05,
+      "loss": 2.1978,
+      "step": 148024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6601939199365068e-05,
+      "loss": 2.2044,
+      "step": 148032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6598468814463516e-05,
+      "loss": 2.2166,
+      "step": 148040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6594999259927777e-05,
+      "loss": 2.2185,
+      "step": 148048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.659153053584411e-05,
+      "loss": 2.2293,
+      "step": 148056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6588062642298738e-05,
+      "loss": 2.2307,
+      "step": 148064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.658459557937788e-05,
+      "loss": 2.2214,
+      "step": 148072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6581129347167727e-05,
+      "loss": 2.221,
+      "step": 148080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6577663945754437e-05,
+      "loss": 2.2186,
+      "step": 148088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6574199375224185e-05,
+      "loss": 2.2062,
+      "step": 148096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6570735635663084e-05,
+      "loss": 2.2357,
+      "step": 148104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.656727272715724e-05,
+      "loss": 2.2093,
+      "step": 148112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6563810649792752e-05,
+      "loss": 2.2058,
+      "step": 148120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6560349403655683e-05,
+      "loss": 2.2334,
+      "step": 148128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6556888988832074e-05,
+      "loss": 2.2101,
+      "step": 148136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6553429405407965e-05,
+      "loss": 2.2258,
+      "step": 148144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6549970653469354e-05,
+      "loss": 2.2218,
+      "step": 148152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6546512733102215e-05,
+      "loss": 2.2139,
+      "step": 148160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.654305564439253e-05,
+      "loss": 2.2227,
+      "step": 148168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.653959938742624e-05,
+      "loss": 2.2119,
+      "step": 148176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6536143962289255e-05,
+      "loss": 2.2266,
+      "step": 148184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6532689369067486e-05,
+      "loss": 2.2209,
+      "step": 148192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6529235607846814e-05,
+      "loss": 2.2226,
+      "step": 148200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.652578267871311e-05,
+      "loss": 2.221,
+      "step": 148208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6522330581752197e-05,
+      "loss": 2.2333,
+      "step": 148216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6518879317049894e-05,
+      "loss": 2.23,
+      "step": 148224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.651542888469203e-05,
+      "loss": 2.2136,
+      "step": 148232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6511979284764337e-05,
+      "loss": 2.2135,
+      "step": 148240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.650853051735261e-05,
+      "loss": 2.234,
+      "step": 148248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.650508258254256e-05,
+      "loss": 2.2287,
+      "step": 148256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.650163548041993e-05,
+      "loss": 2.2358,
+      "step": 148264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.649818921107039e-05,
+      "loss": 2.2141,
+      "step": 148272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.649474377457963e-05,
+      "loss": 2.2188,
+      "step": 148280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6491299171033295e-05,
+      "loss": 2.2019,
+      "step": 148288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.648785540051703e-05,
+      "loss": 2.2262,
+      "step": 148296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6484412463116426e-05,
+      "loss": 2.2297,
+      "step": 148304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.64809703589171e-05,
+      "loss": 2.217,
+      "step": 148312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.647752908800461e-05,
+      "loss": 2.2226,
+      "step": 148320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.64740886504645e-05,
+      "loss": 2.2117,
+      "step": 148328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6470649046382322e-05,
+      "loss": 2.2153,
+      "step": 148336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6467210275843564e-05,
+      "loss": 2.2243,
+      "step": 148344
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.646377233893371e-05,
+      "loss": 2.2125,
+      "step": 148352
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6460335235738255e-05,
+      "loss": 2.2088,
+      "step": 148360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6456898966342625e-05,
+      "loss": 2.2192,
+      "step": 148368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6453463530832247e-05,
+      "loss": 2.2168,
+      "step": 148376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6450028929292524e-05,
+      "loss": 2.2125,
+      "step": 148384
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6446595161808858e-05,
+      "loss": 2.2223,
+      "step": 148392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.64431622284666e-05,
+      "loss": 2.2228,
+      "step": 148400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.643973012935108e-05,
+      "loss": 2.219,
+      "step": 148408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6436298864547638e-05,
+      "loss": 2.2159,
+      "step": 148416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.643286843414159e-05,
+      "loss": 2.2293,
+      "step": 148424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.642943883821818e-05,
+      "loss": 2.2409,
+      "step": 148432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6426010076862685e-05,
+      "loss": 2.217,
+      "step": 148440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.642258215016037e-05,
+      "loss": 2.2205,
+      "step": 148448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.641915505819641e-05,
+      "loss": 2.2084,
+      "step": 148456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6415728801056028e-05,
+      "loss": 2.2262,
+      "step": 148464
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.641230337882439e-05,
+      "loss": 2.2187,
+      "step": 148472
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6408878791586666e-05,
+      "loss": 2.2158,
+      "step": 148480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6405455039427984e-05,
+      "loss": 2.2156,
+      "step": 148488
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6402032122433452e-05,
+      "loss": 2.2226,
+      "step": 148496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6398610040688175e-05,
+      "loss": 2.2284,
+      "step": 148504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6395188794277226e-05,
+      "loss": 2.2236,
+      "step": 148512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6391768383285647e-05,
+      "loss": 2.2088,
+      "step": 148520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6388348807798487e-05,
+      "loss": 2.2229,
+      "step": 148528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6384930067900744e-05,
+      "loss": 2.2213,
+      "step": 148536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6381512163677402e-05,
+      "loss": 2.2078,
+      "step": 148544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.637809509521345e-05,
+      "loss": 2.2167,
+      "step": 148552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.637467886259382e-05,
+      "loss": 2.2122,
+      "step": 148560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6371263465903444e-05,
+      "loss": 2.2279,
+      "step": 148568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6367848905227238e-05,
+      "loss": 2.2288,
+      "step": 148576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.636443518065008e-05,
+      "loss": 2.2245,
+      "step": 148584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.636102229225683e-05,
+      "loss": 2.2319,
+      "step": 148592
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6357610240132354e-05,
+      "loss": 2.2282,
+      "step": 148600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6354199024361447e-05,
+      "loss": 2.2157,
+      "step": 148608
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6350788645028944e-05,
+      "loss": 2.2247,
+      "step": 148616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6347379102219598e-05,
+      "loss": 2.2324,
+      "step": 148624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.634397039601818e-05,
+      "loss": 2.2344,
+      "step": 148632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.634056252650945e-05,
+      "loss": 2.2242,
+      "step": 148640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.63371554937781e-05,
+      "loss": 2.2245,
+      "step": 148648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.633374929790884e-05,
+      "loss": 2.2311,
+      "step": 148656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.633034393898636e-05,
+      "loss": 2.2174,
+      "step": 148664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6326939417095297e-05,
+      "loss": 2.2215,
+      "step": 148672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6323535732320304e-05,
+      "loss": 2.2342,
+      "step": 148680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6320132884745985e-05,
+      "loss": 2.232,
+      "step": 148688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6316730874456947e-05,
+      "loss": 2.2127,
+      "step": 148696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.631332970153776e-05,
+      "loss": 2.2246,
+      "step": 148704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6309929366072972e-05,
+      "loss": 2.2218,
+      "step": 148712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6306529868147123e-05,
+      "loss": 2.22,
+      "step": 148720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.630313120784472e-05,
+      "loss": 2.2174,
+      "step": 148728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6299733385250248e-05,
+      "loss": 2.2266,
+      "step": 148736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6296336400448195e-05,
+      "loss": 2.2172,
+      "step": 148744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6292940253523002e-05,
+      "loss": 2.2245,
+      "step": 148752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6289544944559084e-05,
+      "loss": 2.2178,
+      "step": 148760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.628615047364088e-05,
+      "loss": 2.2302,
+      "step": 148768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6282756840852752e-05,
+      "loss": 2.2271,
+      "step": 148776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.627936404627906e-05,
+      "loss": 2.2195,
+      "step": 148784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.627597209000417e-05,
+      "loss": 2.2162,
+      "step": 148792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6272580972112404e-05,
+      "loss": 2.2207,
+      "step": 148800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.626919069268804e-05,
+      "loss": 2.2188,
+      "step": 148808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6265801251815392e-05,
+      "loss": 2.2082,
+      "step": 148816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6262412649578706e-05,
+      "loss": 2.2225,
+      "step": 148824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6259024886062243e-05,
+      "loss": 2.2304,
+      "step": 148832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.625563796135019e-05,
+      "loss": 2.2262,
+      "step": 148840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6252251875526758e-05,
+      "loss": 2.2108,
+      "step": 148848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6248866628676157e-05,
+      "loss": 2.222,
+      "step": 148856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6245482220882495e-05,
+      "loss": 2.2207,
+      "step": 148864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.624209865222994e-05,
+      "loss": 2.2258,
+      "step": 148872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6238715922802604e-05,
+      "loss": 2.2149,
+      "step": 148880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6235334032684568e-05,
+      "loss": 2.2272,
+      "step": 148888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6231952981959916e-05,
+      "loss": 2.2352,
+      "step": 148896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6228572770712705e-05,
+      "loss": 2.2171,
+      "step": 148904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6225193399026964e-05,
+      "loss": 2.2227,
+      "step": 148912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6221814866986714e-05,
+      "loss": 2.2262,
+      "step": 148920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6218437174675918e-05,
+      "loss": 2.2128,
+      "step": 148928
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.621506032217858e-05,
+      "loss": 2.2256,
+      "step": 148936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.621168430957863e-05,
+      "loss": 2.2257,
+      "step": 148944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6208309136959992e-05,
+      "loss": 2.2166,
+      "step": 148952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.620493480440659e-05,
+      "loss": 2.2331,
+      "step": 148960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6201561312002292e-05,
+      "loss": 2.2211,
+      "step": 148968
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6198188659830968e-05,
+      "loss": 2.2386,
+      "step": 148976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6194816847976483e-05,
+      "loss": 2.2155,
+      "step": 148984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6191445876522634e-05,
+      "loss": 2.2249,
+      "step": 148992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6188075745553232e-05,
+      "loss": 2.2183,
+      "step": 149000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6184706455152063e-05,
+      "loss": 2.2122,
+      "step": 149008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6181338005402888e-05,
+      "loss": 2.2063,
+      "step": 149016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.617797039638944e-05,
+      "loss": 2.2211,
+      "step": 149024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6174603628195443e-05,
+      "loss": 2.2101,
+      "step": 149032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6171237700904594e-05,
+      "loss": 2.2336,
+      "step": 149040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6167872614600585e-05,
+      "loss": 2.211,
+      "step": 149048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.616450836936704e-05,
+      "loss": 2.2246,
+      "step": 149056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6161144965287616e-05,
+      "loss": 2.2287,
+      "step": 149064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.615778240244594e-05,
+      "loss": 2.2224,
+      "step": 149072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6154420680925574e-05,
+      "loss": 2.2357,
+      "step": 149080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6151059800810118e-05,
+      "loss": 2.2206,
+      "step": 149088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6147699762183113e-05,
+      "loss": 2.2339,
+      "step": 149096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.614434056512808e-05,
+      "loss": 2.2181,
+      "step": 149104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.614098220972855e-05,
+      "loss": 2.2046,
+      "step": 149112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6137624696067987e-05,
+      "loss": 2.2247,
+      "step": 149120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6134268024229886e-05,
+      "loss": 2.2185,
+      "step": 149128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.613091219429768e-05,
+      "loss": 2.2105,
+      "step": 149136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6127557206354786e-05,
+      "loss": 2.2188,
+      "step": 149144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6124203060484635e-05,
+      "loss": 2.2281,
+      "step": 149152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6120849756770592e-05,
+      "loss": 2.227,
+      "step": 149160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6117497295296018e-05,
+      "loss": 2.2284,
+      "step": 149168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6114145676144275e-05,
+      "loss": 2.2272,
+      "step": 149176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6110794899398666e-05,
+      "loss": 2.2286,
+      "step": 149184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.610744496514249e-05,
+      "loss": 2.2289,
+      "step": 149192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.610409587345905e-05,
+      "loss": 2.225,
+      "step": 149200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6100747624431582e-05,
+      "loss": 2.2178,
+      "step": 149208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6097400218143332e-05,
+      "loss": 2.2429,
+      "step": 149216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6094053654677514e-05,
+      "loss": 2.2368,
+      "step": 149224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.609070793411733e-05,
+      "loss": 2.2042,
+      "step": 149232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6087363056545954e-05,
+      "loss": 2.2245,
+      "step": 149240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.608401902204653e-05,
+      "loss": 2.2104,
+      "step": 149248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6080675830702193e-05,
+      "loss": 2.2372,
+      "step": 149256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6077333482596082e-05,
+      "loss": 2.2223,
+      "step": 149264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6073991977811246e-05,
+      "loss": 2.2246,
+      "step": 149272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6070651316430784e-05,
+      "loss": 2.2451,
+      "step": 149280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6067311498537745e-05,
+      "loss": 2.2363,
+      "step": 149288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6063972524215135e-05,
+      "loss": 2.2142,
+      "step": 149296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6060634393545984e-05,
+      "loss": 2.2118,
+      "step": 149304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6057297106613266e-05,
+      "loss": 2.2217,
+      "step": 149312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6053960663499948e-05,
+      "loss": 2.2184,
+      "step": 149320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.605062506428899e-05,
+      "loss": 2.2133,
+      "step": 149328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.604729030906328e-05,
+      "loss": 2.2082,
+      "step": 149336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.604395639790576e-05,
+      "loss": 2.2219,
+      "step": 149344
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6040623330899295e-05,
+      "loss": 2.2233,
+      "step": 149352
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.603729110812673e-05,
+      "loss": 2.2207,
+      "step": 149360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.603395972967093e-05,
+      "loss": 2.2165,
+      "step": 149368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6030629195614697e-05,
+      "loss": 2.2113,
+      "step": 149376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.602729950604083e-05,
+      "loss": 2.2195,
+      "step": 149384
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.602397066103212e-05,
+      "loss": 2.2131,
+      "step": 149392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6020642660671307e-05,
+      "loss": 2.2259,
+      "step": 149400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6017315505041118e-05,
+      "loss": 2.2209,
+      "step": 149408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.601398919422429e-05,
+      "loss": 2.2252,
+      "step": 149416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.601066372830351e-05,
+      "loss": 2.209,
+      "step": 149424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6007339107361427e-05,
+      "loss": 2.2161,
+      "step": 149432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6004015331480718e-05,
+      "loss": 2.201,
+      "step": 149440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6000692400743998e-05,
+      "loss": 2.2213,
+      "step": 149448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5997370315233893e-05,
+      "loss": 2.2025,
+      "step": 149456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5994049075032956e-05,
+      "loss": 2.2012,
+      "step": 149464
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5990728680223786e-05,
+      "loss": 2.2157,
+      "step": 149472
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5987409130888924e-05,
+      "loss": 2.2133,
+      "step": 149480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5984090427110875e-05,
+      "loss": 2.2296,
+      "step": 149488
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.598077256897215e-05,
+      "loss": 2.2299,
+      "step": 149496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5977455556555253e-05,
+      "loss": 2.2261,
+      "step": 149504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5974139389942614e-05,
+      "loss": 2.2123,
+      "step": 149512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5970824069216694e-05,
+      "loss": 2.2136,
+      "step": 149520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.59675095944599e-05,
+      "loss": 2.2065,
+      "step": 149528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.596419596575464e-05,
+      "loss": 2.2184,
+      "step": 149536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5960883183183286e-05,
+      "loss": 2.2241,
+      "step": 149544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5957571246828187e-05,
+      "loss": 2.226,
+      "step": 149552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5954260156771704e-05,
+      "loss": 2.2218,
+      "step": 149560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5950949913096123e-05,
+      "loss": 2.1947,
+      "step": 149568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.594764051588374e-05,
+      "loss": 2.2182,
+      "step": 149576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5944331965216852e-05,
+      "loss": 2.2178,
+      "step": 149584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5941024261177683e-05,
+      "loss": 2.2311,
+      "step": 149592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.593771740384847e-05,
+      "loss": 2.2214,
+      "step": 149600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.593441139331143e-05,
+      "loss": 2.2198,
+      "step": 149608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5931106229648744e-05,
+      "loss": 2.2234,
+      "step": 149616
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5927801912942572e-05,
+      "loss": 2.2051,
+      "step": 149624
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.592449844327507e-05,
+      "loss": 2.2202,
+      "step": 149632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5921195820728368e-05,
+      "loss": 2.2086,
+      "step": 149640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.591789404538455e-05,
+      "loss": 2.2254,
+      "step": 149648
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5914593117325713e-05,
+      "loss": 2.2106,
+      "step": 149656
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5911293036633915e-05,
+      "loss": 2.2145,
+      "step": 149664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.590799380339122e-05,
+      "loss": 2.1994,
+      "step": 149672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5904695417679592e-05,
+      "loss": 2.2177,
+      "step": 149680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.590139787958107e-05,
+      "loss": 2.2168,
+      "step": 149688
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5898101189177637e-05,
+      "loss": 2.2073,
+      "step": 149696
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5894805346551216e-05,
+      "loss": 2.2096,
+      "step": 149704
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5891510351783766e-05,
+      "loss": 2.2434,
+      "step": 149712
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5888216204957204e-05,
+      "loss": 2.2214,
+      "step": 149720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.58849229061534e-05,
+      "loss": 2.2246,
+      "step": 149728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5881630455454252e-05,
+      "loss": 2.21,
+      "step": 149736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.587833885294159e-05,
+      "loss": 2.2221,
+      "step": 149744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5875048098697258e-05,
+      "loss": 2.2076,
+      "step": 149752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5871758192803063e-05,
+      "loss": 2.2357,
+      "step": 149760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5868469135340778e-05,
+      "loss": 2.2296,
+      "step": 149768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5865180926392185e-05,
+      "loss": 2.2211,
+      "step": 149776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.586189356603903e-05,
+      "loss": 2.2391,
+      "step": 149784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.585860705436302e-05,
+      "loss": 2.2065,
+      "step": 149792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.585532139144588e-05,
+      "loss": 2.2212,
+      "step": 149800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5852036577369276e-05,
+      "loss": 2.2191,
+      "step": 149808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5848752612214873e-05,
+      "loss": 2.2125,
+      "step": 149816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.584546949606431e-05,
+      "loss": 2.2225,
+      "step": 149824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5842187228999215e-05,
+      "loss": 2.2225,
+      "step": 149832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.583890581110117e-05,
+      "loss": 2.2133,
+      "step": 149840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5835625242451765e-05,
+      "loss": 2.211,
+      "step": 149848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5832345523132552e-05,
+      "loss": 2.2039,
+      "step": 149856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.582906665322505e-05,
+      "loss": 2.2104,
+      "step": 149864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.58257886328108e-05,
+      "loss": 2.2098,
+      "step": 149872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.582251146197126e-05,
+      "loss": 2.2069,
+      "step": 149880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5819235140787946e-05,
+      "loss": 2.2282,
+      "step": 149888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.581595966934226e-05,
+      "loss": 2.2082,
+      "step": 149896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5812685047715656e-05,
+      "loss": 2.2153,
+      "step": 149904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.580941127598955e-05,
+      "loss": 2.2053,
+      "step": 149912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.58061383542453e-05,
+      "loss": 2.2136,
+      "step": 149920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5802866282564296e-05,
+      "loss": 2.2136,
+      "step": 149928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.579959506102787e-05,
+      "loss": 2.2167,
+      "step": 149936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5796324689717344e-05,
+      "loss": 2.2044,
+      "step": 149944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.579305516871403e-05,
+      "loss": 2.2205,
+      "step": 149952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.57897864980992e-05,
+      "loss": 2.2315,
+      "step": 149960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.578651867795412e-05,
+      "loss": 2.2019,
+      "step": 149968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5783251708360024e-05,
+      "loss": 2.2274,
+      "step": 149976
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5779985589398115e-05,
+      "loss": 2.2302,
+      "step": 149984
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5776720321149616e-05,
+      "loss": 2.2125,
+      "step": 149992
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5773455903695694e-05,
+      "loss": 2.2098,
+      "step": 150000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5770192337117488e-05,
+      "loss": 2.2179,
+      "step": 150008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.576692962149615e-05,
+      "loss": 2.2381,
+      "step": 150016
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.576366775691278e-05,
+      "loss": 2.2242,
+      "step": 150024
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5760406743448463e-05,
+      "loss": 2.2194,
+      "step": 150032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5757146581184288e-05,
+      "loss": 2.2352,
+      "step": 150040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.575388727020129e-05,
+      "loss": 2.2139,
+      "step": 150048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5750628810580478e-05,
+      "loss": 2.2181,
+      "step": 150056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.574737120240289e-05,
+      "loss": 2.2209,
+      "step": 150064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5744114445749508e-05,
+      "loss": 2.2114,
+      "step": 150072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5740858540701274e-05,
+      "loss": 2.2212,
+      "step": 150080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5737603487339136e-05,
+      "loss": 2.2133,
+      "step": 150088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.573434928574402e-05,
+      "loss": 2.2351,
+      "step": 150096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.573109593599684e-05,
+      "loss": 2.2102,
+      "step": 150104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5727843438178443e-05,
+      "loss": 2.2254,
+      "step": 150112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.57245917923697e-05,
+      "loss": 2.2273,
+      "step": 150120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5721340998651472e-05,
+      "loss": 2.2219,
+      "step": 150128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5718091057104538e-05,
+      "loss": 2.2158,
+      "step": 150136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.571484196780971e-05,
+      "loss": 2.2284,
+      "step": 150144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.571159373084775e-05,
+      "loss": 2.2225,
+      "step": 150152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5708346346299427e-05,
+      "loss": 2.2308,
+      "step": 150160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5705099814245464e-05,
+      "loss": 2.2037,
+      "step": 150168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5701854134766557e-05,
+      "loss": 2.2183,
+      "step": 150176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.569860930794341e-05,
+      "loss": 2.2322,
+      "step": 150184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.569536533385668e-05,
+      "loss": 2.2199,
+      "step": 150192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5692122212587016e-05,
+      "loss": 2.2427,
+      "step": 150200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.568887994421505e-05,
+      "loss": 2.2266,
+      "step": 150208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5685638528821376e-05,
+      "loss": 2.2242,
+      "step": 150216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.568239796648657e-05,
+      "loss": 2.2183,
+      "step": 150224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5679158257291213e-05,
+      "loss": 2.2013,
+      "step": 150232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5675919401315833e-05,
+      "loss": 2.1997,
+      "step": 150240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5672681398640936e-05,
+      "loss": 2.2176,
+      "step": 150248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.566944424934704e-05,
+      "loss": 2.2182,
+      "step": 150256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5666207953514615e-05,
+      "loss": 2.2087,
+      "step": 150264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5662972511224097e-05,
+      "loss": 2.2037,
+      "step": 150272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.565973792255595e-05,
+      "loss": 2.2236,
+      "step": 150280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5656504187590562e-05,
+      "loss": 2.2138,
+      "step": 150288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5653271306408346e-05,
+      "loss": 2.2062,
+      "step": 150296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5650039279089645e-05,
+      "loss": 2.2238,
+      "step": 150304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5646808105714816e-05,
+      "loss": 2.2124,
+      "step": 150312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5643577786364216e-05,
+      "loss": 2.2196,
+      "step": 150320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5640348321118102e-05,
+      "loss": 2.231,
+      "step": 150328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.563711971005679e-05,
+      "loss": 2.2215,
+      "step": 150336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.563389195326056e-05,
+      "loss": 2.2069,
+      "step": 150344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.56306650508096e-05,
+      "loss": 2.2238,
+      "step": 150352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5627439002784182e-05,
+      "loss": 2.2147,
+      "step": 150360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5624213809264473e-05,
+      "loss": 2.2085,
+      "step": 150368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.562098947033068e-05,
+      "loss": 2.202,
+      "step": 150376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.561776598606294e-05,
+      "loss": 2.2325,
+      "step": 150384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5614543356541393e-05,
+      "loss": 2.2202,
+      "step": 150392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.561132158184616e-05,
+      "loss": 2.2163,
+      "step": 150400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5608100662057333e-05,
+      "loss": 2.226,
+      "step": 150408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5604880597254977e-05,
+      "loss": 2.2133,
+      "step": 150416
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5601661387519148e-05,
+      "loss": 2.22,
+      "step": 150424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5598443032929886e-05,
+      "loss": 2.2203,
+      "step": 150432
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5595225533567175e-05,
+      "loss": 2.2257,
+      "step": 150440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5592008889511026e-05,
+      "loss": 2.2159,
+      "step": 150448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5588793100841402e-05,
+      "loss": 2.219,
+      "step": 150456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5585578167638223e-05,
+      "loss": 2.2316,
+      "step": 150464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5582364089981454e-05,
+      "loss": 2.2026,
+      "step": 150472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.557915086795097e-05,
+      "loss": 2.2042,
+      "step": 150480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5575938501626644e-05,
+      "loss": 2.2129,
+      "step": 150488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5572726991088363e-05,
+      "loss": 2.2388,
+      "step": 150496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5569516336415937e-05,
+      "loss": 2.2166,
+      "step": 150504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5566306537689223e-05,
+      "loss": 2.2145,
+      "step": 150512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.556309759498797e-05,
+      "loss": 2.2288,
+      "step": 150520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5559889508391976e-05,
+      "loss": 2.2066,
+      "step": 150528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.555668227798101e-05,
+      "loss": 2.2149,
+      "step": 150536
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5553475903834767e-05,
+      "loss": 2.1916,
+      "step": 150544
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.555027038603298e-05,
+      "loss": 2.2299,
+      "step": 150552
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5547065724655357e-05,
+      "loss": 2.1988,
+      "step": 150560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5543861919781532e-05,
+      "loss": 2.2238,
+      "step": 150568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.554065897149116e-05,
+      "loss": 2.2228,
+      "step": 150576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5537456879863873e-05,
+      "loss": 2.2294,
+      "step": 150584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5534255644979283e-05,
+      "loss": 2.2179,
+      "step": 150592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5531055266916965e-05,
+      "loss": 2.2237,
+      "step": 150600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5527855745756467e-05,
+      "loss": 2.2135,
+      "step": 150608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.552465708157736e-05,
+      "loss": 2.2275,
+      "step": 150616
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5521459274459137e-05,
+      "loss": 2.2116,
+      "step": 150624
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5518262324481295e-05,
+      "loss": 2.2254,
+      "step": 150632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5515066231723335e-05,
+      "loss": 2.2295,
+      "step": 150640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5511870996264696e-05,
+      "loss": 2.2159,
+      "step": 150648
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5508676618184807e-05,
+      "loss": 2.2227,
+      "step": 150656
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5505483097563093e-05,
+      "loss": 2.2027,
+      "step": 150664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5502290434478944e-05,
+      "loss": 2.2307,
+      "step": 150672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5499098629011714e-05,
+      "loss": 2.2281,
+      "step": 150680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5495907681240773e-05,
+      "loss": 2.213,
+      "step": 150688
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.549271759124543e-05,
+      "loss": 2.2158,
+      "step": 150696
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5489528359105e-05,
+      "loss": 2.2204,
+      "step": 150704
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.548633998489877e-05,
+      "loss": 2.2235,
+      "step": 150712
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5483152468705995e-05,
+      "loss": 2.2159,
+      "step": 150720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5479965810605937e-05,
+      "loss": 2.2072,
+      "step": 150728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5476780010677787e-05,
+      "loss": 2.228,
+      "step": 150736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5473595069000757e-05,
+      "loss": 2.2123,
+      "step": 150744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5470410985654046e-05,
+      "loss": 2.2138,
+      "step": 150752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5467227760716776e-05,
+      "loss": 2.2171,
+      "step": 150760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5464045394268106e-05,
+      "loss": 2.2094,
+      "step": 150768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5460863886387142e-05,
+      "loss": 2.233,
+      "step": 150776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5457683237152968e-05,
+      "loss": 2.22,
+      "step": 150784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5454503446644678e-05,
+      "loss": 2.219,
+      "step": 150792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5451324514941294e-05,
+      "loss": 2.2131,
+      "step": 150800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5448146442121868e-05,
+      "loss": 2.2116,
+      "step": 150808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5444969228265404e-05,
+      "loss": 2.2253,
+      "step": 150816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5441792873450864e-05,
+      "loss": 2.2297,
+      "step": 150824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5438617377757247e-05,
+      "loss": 2.1875,
+      "step": 150832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5435442741263478e-05,
+      "loss": 2.227,
+      "step": 150840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5432268964048468e-05,
+      "loss": 2.2225,
+      "step": 150848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5429096046191135e-05,
+      "loss": 2.222,
+      "step": 150856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.542592398777036e-05,
+      "loss": 2.2237,
+      "step": 150864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.542275278886499e-05,
+      "loss": 2.2295,
+      "step": 150872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5419582449553864e-05,
+      "loss": 2.2302,
+      "step": 150880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5416412969915804e-05,
+      "loss": 2.2208,
+      "step": 150888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5413244350029585e-05,
+      "loss": 2.2288,
+      "step": 150896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5410076589974003e-05,
+      "loss": 2.2494,
+      "step": 150904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5406909689827797e-05,
+      "loss": 2.212,
+      "step": 150912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5403743649669696e-05,
+      "loss": 2.2229,
+      "step": 150920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5400578469578405e-05,
+      "loss": 2.2333,
+      "step": 150928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5397414149632616e-05,
+      "loss": 2.2342,
+      "step": 150936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5394250689911004e-05,
+      "loss": 2.2211,
+      "step": 150944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5391088090492188e-05,
+      "loss": 2.213,
+      "step": 150952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.538792635145481e-05,
+      "loss": 2.2249,
+      "step": 150960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5384765472877477e-05,
+      "loss": 2.2306,
+      "step": 150968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.538160545483874e-05,
+      "loss": 2.2206,
+      "step": 150976
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.537844629741719e-05,
+      "loss": 2.2154,
+      "step": 150984
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.537528800069135e-05,
+      "loss": 2.2267,
+      "step": 150992
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.537213056473973e-05,
+      "loss": 2.2174,
+      "step": 151000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5368973989640827e-05,
+      "loss": 2.2024,
+      "step": 151008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.536581827547311e-05,
+      "loss": 2.2211,
+      "step": 151016
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5362663422315048e-05,
+      "loss": 2.2136,
+      "step": 151024
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.535950943024506e-05,
+      "loss": 2.2247,
+      "step": 151032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5356356299341545e-05,
+      "loss": 2.2311,
+      "step": 151040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.535320402968291e-05,
+      "loss": 2.2091,
+      "step": 151048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5350052621347502e-05,
+      "loss": 2.2278,
+      "step": 151056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.534690207441367e-05,
+      "loss": 2.2298,
+      "step": 151064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.534375238895975e-05,
+      "loss": 2.217,
+      "step": 151072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5340603565064035e-05,
+      "loss": 2.2181,
+      "step": 151080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5337455602804793e-05,
+      "loss": 2.2156,
+      "step": 151088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5334308502260306e-05,
+      "loss": 2.2355,
+      "step": 151096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5331162263508794e-05,
+      "loss": 2.2385,
+      "step": 151104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.532801688662847e-05,
+      "loss": 2.221,
+      "step": 151112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.532487237169755e-05,
+      "loss": 2.208,
+      "step": 151120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.532172871879419e-05,
+      "loss": 2.2274,
+      "step": 151128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.531858592799654e-05,
+      "loss": 2.2283,
+      "step": 151136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5315443999382732e-05,
+      "loss": 2.2273,
+      "step": 151144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5312302933030878e-05,
+      "loss": 2.2268,
+      "step": 151152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.530916272901908e-05,
+      "loss": 2.2265,
+      "step": 151160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5306023387425376e-05,
+      "loss": 2.2279,
+      "step": 151168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5302884908327816e-05,
+      "loss": 2.2223,
+      "step": 151176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5299747291804456e-05,
+      "loss": 2.222,
+      "step": 151184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5296610537933245e-05,
+      "loss": 2.2322,
+      "step": 151192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5293474646792204e-05,
+      "loss": 2.232,
+      "step": 151200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5290339618459267e-05,
+      "loss": 2.2174,
+      "step": 151208
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.528720545301239e-05,
+      "loss": 2.2248,
+      "step": 151216
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5284072150529482e-05,
+      "loss": 2.2187,
+      "step": 151224
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5280939711088425e-05,
+      "loss": 2.2392,
+      "step": 151232
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5277808134767112e-05,
+      "loss": 2.2267,
+      "step": 151240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5274677421643383e-05,
+      "loss": 2.2265,
+      "step": 151248
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.527154757179506e-05,
+      "loss": 2.2383,
+      "step": 151256
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5268418585299976e-05,
+      "loss": 2.2426,
+      "step": 151264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.526529046223589e-05,
+      "loss": 2.2327,
+      "step": 151272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5262163202680584e-05,
+      "loss": 2.23,
+      "step": 151280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.52590368067118e-05,
+      "loss": 2.2128,
+      "step": 151288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5255911274407264e-05,
+      "loss": 2.2336,
+      "step": 151296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5252786605844657e-05,
+      "loss": 2.23,
+      "step": 151304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5249662801101687e-05,
+      "loss": 2.2232,
+      "step": 151312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5246539860256002e-05,
+      "loss": 2.2172,
+      "step": 151320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5243417783385214e-05,
+      "loss": 2.2231,
+      "step": 151328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.524029657056698e-05,
+      "loss": 2.2301,
+      "step": 151336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5237176221878865e-05,
+      "loss": 2.2184,
+      "step": 151344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.523405673739846e-05,
+      "loss": 2.2203,
+      "step": 151352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.523093811720329e-05,
+      "loss": 2.2135,
+      "step": 151360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.52278203613709e-05,
+      "loss": 2.2222,
+      "step": 151368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.522470346997881e-05,
+      "loss": 2.2114,
+      "step": 151376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5221587443104484e-05,
+      "loss": 2.2203,
+      "step": 151384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.521847228082539e-05,
+      "loss": 2.2274,
+      "step": 151392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5215357983219e-05,
+      "loss": 2.2205,
+      "step": 151400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.521224455036269e-05,
+      "loss": 2.2335,
+      "step": 151408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.52091319823339e-05,
+      "loss": 2.1936,
+      "step": 151416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5206020279209975e-05,
+      "loss": 2.2279,
+      "step": 151424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5202909441068308e-05,
+      "loss": 2.202,
+      "step": 151432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5199799467986208e-05,
+      "loss": 2.2152,
+      "step": 151440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.519669036004099e-05,
+      "loss": 2.2215,
+      "step": 151448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5193582117309973e-05,
+      "loss": 2.226,
+      "step": 151456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5190474739870395e-05,
+      "loss": 2.2147,
+      "step": 151464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5187368227799524e-05,
+      "loss": 2.2212,
+      "step": 151472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.518426258117459e-05,
+      "loss": 2.2174,
+      "step": 151480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.518115780007279e-05,
+      "loss": 2.2272,
+      "step": 151488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5178053884571306e-05,
+      "loss": 2.2187,
+      "step": 151496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5174950834747322e-05,
+      "loss": 2.2263,
+      "step": 151504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5171848650677964e-05,
+      "loss": 2.2376,
+      "step": 151512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5168747332440344e-05,
+      "loss": 2.2249,
+      "step": 151520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5165646880111594e-05,
+      "loss": 2.2142,
+      "step": 151528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5162547293768763e-05,
+      "loss": 2.2081,
+      "step": 151536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5159448573488908e-05,
+      "loss": 2.2017,
+      "step": 151544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5156350719349075e-05,
+      "loss": 2.2349,
+      "step": 151552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.515325373142627e-05,
+      "loss": 2.2291,
+      "step": 151560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.51501576097975e-05,
+      "loss": 2.2043,
+      "step": 151568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.514706235453971e-05,
+      "loss": 2.2255,
+      "step": 151576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5143967965729854e-05,
+      "loss": 2.2263,
+      "step": 151584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5140874443444883e-05,
+      "loss": 2.2243,
+      "step": 151592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.513778178776167e-05,
+      "loss": 2.2254,
+      "step": 151600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5134689998757123e-05,
+      "loss": 2.2242,
+      "step": 151608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5131599076508087e-05,
+      "loss": 2.2058,
+      "step": 151616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5128509021091408e-05,
+      "loss": 2.2154,
+      "step": 151624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5125419832583917e-05,
+      "loss": 2.2185,
+      "step": 151632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5122331511062387e-05,
+      "loss": 2.2243,
+      "step": 151640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.511924405660362e-05,
+      "loss": 2.2181,
+      "step": 151648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.511615746928436e-05,
+      "loss": 2.2272,
+      "step": 151656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.511307174918134e-05,
+      "loss": 2.2297,
+      "step": 151664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.510998689637126e-05,
+      "loss": 2.2278,
+      "step": 151672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5106902910930837e-05,
+      "loss": 2.2325,
+      "step": 151680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5103819792936707e-05,
+      "loss": 2.2124,
+      "step": 151688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5100737542465537e-05,
+      "loss": 2.2123,
+      "step": 151696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.509765615959396e-05,
+      "loss": 2.2309,
+      "step": 151704
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5094575644398543e-05,
+      "loss": 2.2154,
+      "step": 151712
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5091495996955903e-05,
+      "loss": 2.203,
+      "step": 151720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.50884172173426e-05,
+      "loss": 2.224,
+      "step": 151728
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5085339305635148e-05,
+      "loss": 2.2129,
+      "step": 151736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5082262261910083e-05,
+      "loss": 2.2201,
+      "step": 151744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.50791860862439e-05,
+      "loss": 2.2175,
+      "step": 151752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5076110778713065e-05,
+      "loss": 2.222,
+      "step": 151760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.507303633939403e-05,
+      "loss": 2.2124,
+      "step": 151768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5069962768363226e-05,
+      "loss": 2.2068,
+      "step": 151776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.506689006569709e-05,
+      "loss": 2.2189,
+      "step": 151784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5063818231471964e-05,
+      "loss": 2.212,
+      "step": 151792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5060747265764238e-05,
+      "loss": 2.2097,
+      "step": 151800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5057677168650273e-05,
+      "loss": 2.2216,
+      "step": 151808
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5054607940206353e-05,
+      "loss": 2.2286,
+      "step": 151816
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5051539580508815e-05,
+      "loss": 2.2203,
+      "step": 151824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5048472089633924e-05,
+      "loss": 2.2191,
+      "step": 151832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.504540546765792e-05,
+      "loss": 2.2199,
+      "step": 151840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5042339714657082e-05,
+      "loss": 2.2253,
+      "step": 151848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.503927483070758e-05,
+      "loss": 2.2246,
+      "step": 151856
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.503621081588564e-05,
+      "loss": 2.232,
+      "step": 151864
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5033147670267425e-05,
+      "loss": 2.2293,
+      "step": 151872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.503008539392907e-05,
+      "loss": 2.2203,
+      "step": 151880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5027023986946723e-05,
+      "loss": 2.2172,
+      "step": 151888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5023963449396486e-05,
+      "loss": 2.2218,
+      "step": 151896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5020903781354437e-05,
+      "loss": 2.2042,
+      "step": 151904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.501784498289665e-05,
+      "loss": 2.221,
+      "step": 151912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.501478705409916e-05,
+      "loss": 2.2121,
+      "step": 151920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.501172999503798e-05,
+      "loss": 2.2311,
+      "step": 151928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.500867380578913e-05,
+      "loss": 2.2233,
+      "step": 151936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.500561848642857e-05,
+      "loss": 2.2093,
+      "step": 151944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5002564037032255e-05,
+      "loss": 2.2321,
+      "step": 151952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4999510457676138e-05,
+      "loss": 2.2092,
+      "step": 151960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4996457748436113e-05,
+      "loss": 2.2241,
+      "step": 151968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4993405909388078e-05,
+      "loss": 2.2058,
+      "step": 151976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4990354940607887e-05,
+      "loss": 2.2299,
+      "step": 151984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4987304842171406e-05,
+      "loss": 2.2344,
+      "step": 151992
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4984255614154463e-05,
+      "loss": 2.2211,
+      "step": 152000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4981207256632843e-05,
+      "loss": 2.2212,
+      "step": 152008
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.497815976968234e-05,
+      "loss": 2.228,
+      "step": 152016
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4975113153378734e-05,
+      "loss": 2.2254,
+      "step": 152024
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.497206740779772e-05,
+      "loss": 2.2019,
+      "step": 152032
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4969022533015056e-05,
+      "loss": 2.2151,
+      "step": 152040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4965978529106423e-05,
+      "loss": 2.2173,
+      "step": 152048
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4962935396147484e-05,
+      "loss": 2.2131,
+      "step": 152056
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4959893134213908e-05,
+      "loss": 2.2294,
+      "step": 152064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4956851743381313e-05,
+      "loss": 2.2088,
+      "step": 152072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4953811223725323e-05,
+      "loss": 2.2159,
+      "step": 152080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4950771575321518e-05,
+      "loss": 2.2413,
+      "step": 152088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.494773279824546e-05,
+      "loss": 2.2225,
+      "step": 152096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.49446948925727e-05,
+      "loss": 2.2218,
+      "step": 152104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.494165785837876e-05,
+      "loss": 2.2312,
+      "step": 152112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4938621695739134e-05,
+      "loss": 2.2248,
+      "step": 152120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.493558640472931e-05,
+      "loss": 2.234,
+      "step": 152128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4932551985424743e-05,
+      "loss": 2.2296,
+      "step": 152136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4929518437900863e-05,
+      "loss": 2.2111,
+      "step": 152144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.49264857622331e-05,
+      "loss": 2.2245,
+      "step": 152152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.492345395849683e-05,
+      "loss": 2.2221,
+      "step": 152160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4920423026767423e-05,
+      "loss": 2.2299,
+      "step": 152168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4917392967120245e-05,
+      "loss": 2.2225,
+      "step": 152176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.491436377963061e-05,
+      "loss": 2.2305,
+      "step": 152184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4911335464373832e-05,
+      "loss": 2.2209,
+      "step": 152192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4908308021425182e-05,
+      "loss": 2.2264,
+      "step": 152200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.490528145085993e-05,
+      "loss": 2.2301,
+      "step": 152208
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.490225575275334e-05,
+      "loss": 2.214,
+      "step": 152216
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4899230927180593e-05,
+      "loss": 2.2301,
+      "step": 152224
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.48962069742169e-05,
+      "loss": 2.2186,
+      "step": 152232
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.489318389393746e-05,
+      "loss": 2.2287,
+      "step": 152240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4890161686417385e-05,
+      "loss": 2.2369,
+      "step": 152248
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4887140351731845e-05,
+      "loss": 2.2088,
+      "step": 152256
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.488411988995592e-05,
+      "loss": 2.214,
+      "step": 152264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4881100301164727e-05,
+      "loss": 2.2262,
+      "step": 152272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4878081585433323e-05,
+      "loss": 2.222,
+      "step": 152280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4875063742836738e-05,
+      "loss": 2.2306,
+      "step": 152288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4872046773450024e-05,
+      "loss": 2.2087,
+      "step": 152296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.486903067734816e-05,
+      "loss": 2.2218,
+      "step": 152304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.486601545460613e-05,
+      "loss": 2.208,
+      "step": 152312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.486300110529891e-05,
+      "loss": 2.2242,
+      "step": 152320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4859987629501426e-05,
+      "loss": 2.2171,
+      "step": 152328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4856975027288577e-05,
+      "loss": 2.2157,
+      "step": 152336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.485396329873528e-05,
+      "loss": 2.224,
+      "step": 152344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4850952443916403e-05,
+      "loss": 2.2223,
+      "step": 152352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.484794246290678e-05,
+      "loss": 2.2302,
+      "step": 152360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4844933355781265e-05,
+      "loss": 2.2203,
+      "step": 152368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4841925122614648e-05,
+      "loss": 2.2085,
+      "step": 152376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4838917763481706e-05,
+      "loss": 2.2323,
+      "step": 152384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4835911278457227e-05,
+      "loss": 2.2313,
+      "step": 152392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4832905667615937e-05,
+      "loss": 2.2044,
+      "step": 152400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.482990093103256e-05,
+      "loss": 2.2184,
+      "step": 152408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4826897068781775e-05,
+      "loss": 2.2151,
+      "step": 152416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4823894080938282e-05,
+      "loss": 2.2266,
+      "step": 152424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4820891967576744e-05,
+      "loss": 2.2322,
+      "step": 152432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4817890728771763e-05,
+      "loss": 2.2269,
+      "step": 152440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4814890364597974e-05,
+      "loss": 2.2088,
+      "step": 152448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.481189087512996e-05,
+      "loss": 2.2369,
+      "step": 152456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4808892260442272e-05,
+      "loss": 2.227,
+      "step": 152464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4805894520609485e-05,
+      "loss": 2.2191,
+      "step": 152472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4802897655706102e-05,
+      "loss": 2.222,
+      "step": 152480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4799901665806635e-05,
+      "loss": 2.2226,
+      "step": 152488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4796906550985567e-05,
+      "loss": 2.2297,
+      "step": 152496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.479391231131734e-05,
+      "loss": 2.2248,
+      "step": 152504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4790918946876417e-05,
+      "loss": 2.2297,
+      "step": 152512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.478792645773719e-05,
+      "loss": 2.2204,
+      "step": 152520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4784934843974057e-05,
+      "loss": 2.2249,
+      "step": 152528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4781944105661405e-05,
+      "loss": 2.2288,
+      "step": 152536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4778954242873574e-05,
+      "loss": 2.2194,
+      "step": 152544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4775965255684887e-05,
+      "loss": 2.2275,
+      "step": 152552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4772977144169658e-05,
+      "loss": 2.2079,
+      "step": 152560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4769989908402173e-05,
+      "loss": 2.207,
+      "step": 152568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4767003548456683e-05,
+      "loss": 2.2257,
+      "step": 152576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4764018064407445e-05,
+      "loss": 2.2307,
+      "step": 152584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4761033456328672e-05,
+      "loss": 2.1986,
+      "step": 152592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4758049724294556e-05,
+      "loss": 2.2179,
+      "step": 152600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4755066868379285e-05,
+      "loss": 2.2383,
+      "step": 152608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.475208488865699e-05,
+      "loss": 2.229,
+      "step": 152616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4749103785201844e-05,
+      "loss": 2.2006,
+      "step": 152624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4746123558087918e-05,
+      "loss": 2.2153,
+      "step": 152632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4743144207389306e-05,
+      "loss": 2.2213,
+      "step": 152640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.474016573318011e-05,
+      "loss": 2.2119,
+      "step": 152648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.473718813553433e-05,
+      "loss": 2.2232,
+      "step": 152656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.473421141452602e-05,
+      "loss": 2.2235,
+      "step": 152664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4731235570229167e-05,
+      "loss": 2.2186,
+      "step": 152672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4728260602717748e-05,
+      "loss": 2.2247,
+      "step": 152680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.472528651206574e-05,
+      "loss": 2.2344,
+      "step": 152688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4722313298347055e-05,
+      "loss": 2.2218,
+      "step": 152696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.471934096163563e-05,
+      "loss": 2.2319,
+      "step": 152704
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4716369502005343e-05,
+      "loss": 2.2293,
+      "step": 152712
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.471339891953006e-05,
+      "loss": 2.2333,
+      "step": 152720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.471042921428365e-05,
+      "loss": 2.2089,
+      "step": 152728
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.470746038633993e-05,
+      "loss": 2.2169,
+      "step": 152736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.47044924357727e-05,
+      "loss": 2.2189,
+      "step": 152744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.470152536265575e-05,
+      "loss": 2.2143,
+      "step": 152752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.469855916706284e-05,
+      "loss": 2.2237,
+      "step": 152760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4695593849067705e-05,
+      "loss": 2.2371,
+      "step": 152768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4692629408744074e-05,
+      "loss": 2.2092,
+      "step": 152776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.468966584616564e-05,
+      "loss": 2.2339,
+      "step": 152784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4686703161406062e-05,
+      "loss": 2.2379,
+      "step": 152792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.468374135453902e-05,
+      "loss": 2.2349,
+      "step": 152800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4680780425638126e-05,
+      "loss": 2.2193,
+      "step": 152808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4677820374776996e-05,
+      "loss": 2.2138,
+      "step": 152816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4674861202029203e-05,
+      "loss": 2.2158,
+      "step": 152824
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4671902907468323e-05,
+      "loss": 2.2103,
+      "step": 152832
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.466894549116792e-05,
+      "loss": 2.2272,
+      "step": 152840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.466598895320148e-05,
+      "loss": 2.2189,
+      "step": 152848
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4663033293642522e-05,
+      "loss": 2.2176,
+      "step": 152856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.466007851256453e-05,
+      "loss": 2.2325,
+      "step": 152864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.465712461004094e-05,
+      "loss": 2.2121,
+      "step": 152872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4654171586145203e-05,
+      "loss": 2.2014,
+      "step": 152880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4651219440950723e-05,
+      "loss": 2.2179,
+      "step": 152888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4648268174530893e-05,
+      "loss": 2.2119,
+      "step": 152896
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4645317786959086e-05,
+      "loss": 2.2267,
+      "step": 152904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4642368278308642e-05,
+      "loss": 2.2312,
+      "step": 152912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4639419648652896e-05,
+      "loss": 2.2066,
+      "step": 152920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4636471898065142e-05,
+      "loss": 2.2221,
+      "step": 152928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4633525026618656e-05,
+      "loss": 2.2261,
+      "step": 152936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4630579034386716e-05,
+      "loss": 2.2214,
+      "step": 152944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.462763392144255e-05,
+      "loss": 2.2236,
+      "step": 152952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4624689687859364e-05,
+      "loss": 2.2429,
+      "step": 152960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4621746333710372e-05,
+      "loss": 2.2128,
+      "step": 152968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.461880385906873e-05,
+      "loss": 2.2211,
+      "step": 152976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4615862264007595e-05,
+      "loss": 2.2201,
+      "step": 152984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4612921548600092e-05,
+      "loss": 2.2142,
+      "step": 152992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.460998171291934e-05,
+      "loss": 2.2202,
+      "step": 153000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4607042757038392e-05,
+      "loss": 2.2128,
+      "step": 153008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.460410468103035e-05,
+      "loss": 2.2307,
+      "step": 153016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4601167484968233e-05,
+      "loss": 2.231,
+      "step": 153024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4598231168925067e-05,
+      "loss": 2.2058,
+      "step": 153032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.459529573297384e-05,
+      "loss": 2.2111,
+      "step": 153040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4592361177187524e-05,
+      "loss": 2.2191,
+      "step": 153048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4589427501639108e-05,
+      "loss": 2.2328,
+      "step": 153056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.458649470640148e-05,
+      "loss": 2.2118,
+      "step": 153064
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4583562791547562e-05,
+      "loss": 2.2287,
+      "step": 153072
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4580631757150264e-05,
+      "loss": 2.2221,
+      "step": 153080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.457770160328242e-05,
+      "loss": 2.2266,
+      "step": 153088
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.457477233001689e-05,
+      "loss": 2.2341,
+      "step": 153096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4571843937426506e-05,
+      "loss": 2.2283,
+      "step": 153104
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4568916425584042e-05,
+      "loss": 2.2228,
+      "step": 153112
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4565989794562297e-05,
+      "loss": 2.2158,
+      "step": 153120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4563064044434017e-05,
+      "loss": 2.2302,
+      "step": 153128
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.456013917527195e-05,
+      "loss": 2.2262,
+      "step": 153136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4557215187148796e-05,
+      "loss": 2.2264,
+      "step": 153144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4554292080137243e-05,
+      "loss": 2.2253,
+      "step": 153152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.455136985430997e-05,
+      "loss": 2.2219,
+      "step": 153160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4548448509739628e-05,
+      "loss": 2.2202,
+      "step": 153168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4545528046498816e-05,
+      "loss": 2.2074,
+      "step": 153176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4542608464660167e-05,
+      "loss": 2.2147,
+      "step": 153184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.453968976429625e-05,
+      "loss": 2.2185,
+      "step": 153192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.453677194547962e-05,
+      "loss": 2.2098,
+      "step": 153200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4533855008282823e-05,
+      "loss": 2.2098,
+      "step": 153208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4530938952778366e-05,
+      "loss": 2.217,
+      "step": 153216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.452802377903874e-05,
+      "loss": 2.2256,
+      "step": 153224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4525109487136432e-05,
+      "loss": 2.2211,
+      "step": 153232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4522196077143884e-05,
+      "loss": 2.2395,
+      "step": 153240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.451928354913352e-05,
+      "loss": 2.2218,
+      "step": 153248
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.451637190317774e-05,
+      "loss": 2.2084,
+      "step": 153256
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4513461139348932e-05,
+      "loss": 2.2294,
+      "step": 153264
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4510551257719483e-05,
+      "loss": 2.2362,
+      "step": 153272
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4507642258361695e-05,
+      "loss": 2.2187,
+      "step": 153280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4504734141347914e-05,
+      "loss": 2.224,
+      "step": 153288
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4501826906750418e-05,
+      "loss": 2.2278,
+      "step": 153296
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4498920554641482e-05,
+      "loss": 2.2167,
+      "step": 153304
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4496015085093372e-05,
+      "loss": 2.2218,
+      "step": 153312
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4493110498178303e-05,
+      "loss": 2.2152,
+      "step": 153320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4490206793968496e-05,
+      "loss": 2.2276,
+      "step": 153328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4487303972536135e-05,
+      "loss": 2.2219,
+      "step": 153336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4484402033953374e-05,
+      "loss": 2.2271,
+      "step": 153344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4481500978292374e-05,
+      "loss": 2.2177,
+      "step": 153352
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.447860080562524e-05,
+      "loss": 2.2443,
+      "step": 153360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4475701516024065e-05,
+      "loss": 2.2308,
+      "step": 153368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.447280310956095e-05,
+      "loss": 2.2181,
+      "step": 153376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4469905586307934e-05,
+      "loss": 2.2009,
+      "step": 153384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4467008946337043e-05,
+      "loss": 2.2071,
+      "step": 153392
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4464113189720307e-05,
+      "loss": 2.2356,
+      "step": 153400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.44612183165297e-05,
+      "loss": 2.2155,
+      "step": 153408
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.445832432683719e-05,
+      "loss": 2.2199,
+      "step": 153416
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.445543122071473e-05,
+      "loss": 2.2096,
+      "step": 153424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4452538998234238e-05,
+      "loss": 2.2147,
+      "step": 153432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.44496476594676e-05,
+      "loss": 2.2202,
+      "step": 153440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4446757204486727e-05,
+      "loss": 2.2273,
+      "step": 153448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.444386763336345e-05,
+      "loss": 2.2249,
+      "step": 153456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4440978946169628e-05,
+      "loss": 2.2044,
+      "step": 153464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4438091142977038e-05,
+      "loss": 2.2217,
+      "step": 153472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.44352042238575e-05,
+      "loss": 2.2199,
+      "step": 153480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4432318188882786e-05,
+      "loss": 2.2096,
+      "step": 153488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4429433038124615e-05,
+      "loss": 2.217,
+      "step": 153496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4426548771654745e-05,
+      "loss": 2.2312,
+      "step": 153504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4423665389544857e-05,
+      "loss": 2.2162,
+      "step": 153512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4420782891866637e-05,
+      "loss": 2.2312,
+      "step": 153520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4417901278691747e-05,
+      "loss": 2.2057,
+      "step": 153528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4415020550091823e-05,
+      "loss": 2.2331,
+      "step": 153536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4412140706138485e-05,
+      "loss": 2.2085,
+      "step": 153544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.440926174690332e-05,
+      "loss": 2.2107,
+      "step": 153552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4406383672457896e-05,
+      "loss": 2.224,
+      "step": 153560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4403506482873772e-05,
+      "loss": 2.2057,
+      "step": 153568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.440063017822247e-05,
+      "loss": 2.2312,
+      "step": 153576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4397754758575497e-05,
+      "loss": 2.2279,
+      "step": 153584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4394880224004335e-05,
+      "loss": 2.2363,
+      "step": 153592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.439200657458045e-05,
+      "loss": 2.2363,
+      "step": 153600
+    }
+  ],
+  "logging_steps": 8,
+  "max_steps": 160000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 1600,
+  "total_flos": 2.828530564580839e+21,
+  "train_batch_size": 26,
+  "trial_name": null,
+  "trial_params": null
+}